AOBPreview originally published online on September 5, 2007
Annals of Botany 2007 100(6):1143-1154; doi:10.1093/aob/mcm200
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A Preliminary Study of Genetic Variation in Populations of Monstera adansonii var. klotzschiana (Araceae) from North-East Brazil, Estimated with AFLP Molecular Markers
1 Programa de Pós-Graduação em Botânica, Universidade Estadual de Feira de Santana (UEFS), Departamento de Ciências Biológicas, Avenida Universitária s/n, 44031–460, Feira de Santana, Bahia, Brazil
2 Herbarium, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AE, UK
3 Genetics Section, Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3DS, UK
* For correspondence. E-mail s.mayo{at}kew.org
Received: 23 March 2007 Returned for revision: 5 June 2007 Accepted: 4 July 2007 Published electronically: 6 September 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY INFORMATIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Background and Aims: This study sought genetic evidence of long-term isolation in populations of Monstera adansonii var. klotzschiana (Araceae), a herbaceous, probably outbreeding, humid forest hemi-epiphyte, in the brejo forests of Ceará (north-east Brazil), and clarification of their relationships with populations in Amazonia and the Atlantic forest of Brazil.
Methods: Within-population genetic diversity and between-population dissimilarity were estimated using AFLP molecular markers in 75 individuals from eight populations located in Ceará, the Brazilian Atlantic Forest and Amazonia.
Key Results: The populations showed a clinal pattern of weak genetic differentiation over a large geographical region (FST = 0·1896). A strong correlation between genetic and geographical distance (Mantel test: r = 0·6903, P = 0·002) suggests a historical pattern of isolation by distance. Genetic structure analysis revealed at least two distinct gene pools in the data. The two isolated Ceará populations are significantly different from each other (pairwise 
PT = 0·137, P = 0·003) and as diverse (Nei's gene diversity, average He = 0·1832, 0·1706) as those in the Atlantic and Amazon forest regions. The population in southern Brazil is less diverse (Nei's gene diversity, average He = 0·127) than the rest. The Ceará populations are related to those of the Atlantic forest rather than those from Amazonia (AMOVA, among-groups variation = 11·95 %, P = 0·037).
Conclusions: The gene pools detected within an overall pattern of clinal variation suggest distinct episodes of gene flow, possibly correlated with past humid forest expansions. The Ceará populations show no evidence of erosion of genetic diversity, although this was expected because of their isolation. Their genetic differentiation and relatively high diversity reinforce the importance of conserving the endangered brejo forests.
Key words: AFLP, Amazonia, Araceae, Atlantic forest, brejo forest, French Guiana, Monstera adansonii var. klotzschiana, north-east Brazil, population genetic variation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY INFORMATIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
The brejo forests have long interested palaeobiogeographers (e.g. Ab'Sáber, 1982; Andrade-Lima, 1982) because they are considered to represent relicts of former continuous humid forest in the dry north-east of Brazil. Studies of angiosperm species composition have so far shown only low levels of endemism (Sales et al., 1998), especially when compared with the similarly fragmented, montane ‘campo rupestre’ vegetation of central Bahia (Stannard, 1995). In this study a population-level approach was taken and with a focus on the root-climbing aroid Monstera adansonii Schott var. klotzschiana (Schott) Madison, a widespread species ecologically restricted to humid forests (Mayo, 1984). Using AFLP markers, we looked for evidence of genetic distinction and divergence between the brejo populations of this taxon in Ceará (Fig. 1A: KCEpa, KCEub), where they are most isolated ecologically, and between them and conspecific populations in the Atlantic (Fig. 1A: KBA, KES, KPER, KRJSP) and Amazon (Fig. 1A: KFG, KPA) forest regions.
|
The brejo forests: biogeography and history
The predominant ecosystem in north-east Brazil is the caatinga thorn forest, characterized by its dry climate and high temperatures (Sampaio et al., 2002). Within it lies a mosaic of humid forested montane areas above 500 m elevation with a mean annual rainfall over 1200 mm, known as ‘brejos’ (Sales et al., 1998; Pôrto et al., 2004). The brejo forests are classified as humid evergreen montane forest within the Brazilian Atlantic forest system (Rodal et al., 1998; Sales et al., 1998). In Ceará State there are seven brejo forest fragments designated by Figueiredo (1997) as tropical semi-evergreen moist forest, of which we studied two. Biological studies of the brejos of north-east Brazil have mostly focused on species composition and have shown many disjunctions with the Amazon or Atlantic forests and some cases of endemism (Andrade-Lima, 1966; Plowman, 1986; Sales et al., 1998; Borges-Nojosa and Caramaschi, 2003; Pôrto et al., 2004).
The brejo forests are thought to be natural fragments representing relicts of previous expansions of the Atlantic and Amazon forests during more humid periods of the Pleistocene and Holocene (e.g. Ab'Sáber 1982; Andrade-Lima, 1982; Bigarella and Andrade-Lima, 1982; Borges-Nojosa and Caramaschi, 2003). In a Quaternary pollen study, Oliveira et al. (1999) discussed two hypothesized corridors of humid forest that could have linked the Amazon and Atlantic forests during cooler and more humid periods of the Holocene. Route A (Fig. 1A) followed the north-eastern coast from Maranhão to the region around Salvador in Bahia and Route B traversed the interior of southern Maranhão, Piauí, Bahia and central Brazil and corresponds to the gallery forest links proposed by Oliveira-Filho and Ratter (1995). Borges-Nojosa and Caramaschi (2003), based on lizard and worm lizard (amphisbaenid) biogeography, suggested that the Ceará brejos are relicts from at least two expansions of humid forest from Amazonia and the Atlantic forest during the Pleistocene (40 000–23 000 years BP) and the Holocene (6000–5000 years BP). According to Oliveira et al. (1999), from 11 000 to 5000 years BP the palaeoclimates hypothesized for the south-east, south and central regions of Brazil (Ledru, 1993; Behling, 1998) were similar to those for the north-east, but from about 4000 years BP north-east Brazil became much drier, while rainfall in the other regions gradually increased.
Monstera adansonii: ecology and reproductive biology
Monstera adansonii var. klotzschiana is a root-climbing hemi-epiphyte in the family Araceae (Mayo et al., 1997) that occurs in humid forest environments. It is a common taxon in the eastern half of Amazonia, including the Guianas, and in the Brazilian Atlantic forest as defined by Fundação SOS Mata Atlântica (2007). It has not been observed in caatinga or other kinds of seasonal forest and is rare in undisturbed primary forest. Typically it flowers on large tree boles in secondary moist tropical forest with a well-developed canopy. Within this habitat, it flowers most conspicuously where light levels are higher, at the edge of clearings and along forest edges. It exploits natural clearings caused by tree-fall and is most abundant in habitats affected by human disturbance, but is intolerant of low humidity and unshaded positions. Our field observations in Ceará confirmed that this plant occurs only in habitats near waterfalls and streams that are permanent throughout the year, and where there is no exposure to dry winds or full sun.
The growth cycle of this taxon has been studied by Andrade and Mayo (1998) in the Atlantic forest of Pernambuco (NE Brazil). When conditions are favourable, the plant becomes robust and produces the mature perforated leaf blades typical of many Monstera species. Flagelliform branches of very different morphology descend to the ground and seek out new tree boles to colonize, usually forming small-leaved, terrestrial populations and hence extensive clones.
The reproductive biology of Monstera adansonii var. klotzschiana has not yet been studied in detail. The chromosome number is reported as 2n = 60 by Ramalho (1994). The inflorescence is a dense spike (spadix) of small, bisexual flowers without a perianth (perigone) and is protogynous as is normal in Araceae (Mayo et al., 1997). The stigmas are wet at anthesis and by the time the anthers ripen and shed their pollen, the stigmas have shrivelled and are probably no longer receptive. Our assumption, based on these observations, is that outbreeding is predominant in this taxon. This is supported by Chouteau et al. (2006), who studied pollen/ovule ratios in Araceae from French Guiana and reported results for Monstera adansonii that are consistent with xenogamy.
According to Gibernau (2003), pollination in Monstera has been reported by only three authors (Madison, 1977; Ramirez and Gomez, 1978; Ramirez, 1980; he also mentions Grayum, unpublished), the pollinators being either trigonid bees or scarabaeid beetles. Recently, Chouteau et al. (M. Gibernau, Toulouse, pers. comm.) have demonstrated that the morphological and phenological syndrome of the inflorescence in Monstera corresponds to that of Araceae genera that are known to be bee-pollinated. Grayum (1990 : p. 663, and pers. comm.), however, is doubtful that trigonid bees play any part in pollination of M. oreophila Madison in Panama, which appears to be pollinated before the spathe opens. However, field observations by the first author (IMA) show that the stigmas of flowering spadices of M. adansonii var. klotzschiana from Maranguape in Ceará are still wet and presumably receptive when the spathe is open, with at least the anterior part of the spadix exposed.
Madison reports the ripe infructescences and fruits of M. adansonii var. klotzschiana as medium-yellow to orange, but our observations and those of Ramalho (1994) in north-east Brazil are that they are white. The infructescences are held erect as dense spikes of fruits, which ripen by shedding the stylar portion of the pistils to reveal black seeds embedded in pulp. Andrade and Mayo (1998) reported the abundant production of fertile seeds of M. adansonii var. klotzschiana, which germinated freely on the forest floor. Data on dispersal vectors is sketchy. Scott Mori and Tatyana Lobova (New York Botanical Garden, pers. comm.) have commented that the white fruits of M. adansonii could well be dispersed by bats, but there is no hard evidence as yet. Madison (1977) reported that the seeds of M. adansonii and M. deliciosa lose their viability rapidly if allowed to dry, but when moist they can remain viable for several months.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY INFORMATIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Sampling
Seventy-five individuals were sampled from eleven humid forest localities in French Guiana, Pará, Ceará and the Brazilian Atlantic forest (Table 1, Fig. 1A). A further 24 individuals were sampled but gave only partial results and hence were later omitted. Smaller samples from different but neighbouring localities were amalgamated to form single population units (KBA, KCEub, KES, KFG, KRJSP). Because of the likelihood of extensive clonal growth in this species, care was taken to sample from plants that were widely distant in the local populations visited, and to avoid sampling from plants on the same host tree; this reduced the scope for larger samples in many populations.
|
Specimen identification
Monstera adansonii var. klotzschiana is characterized by its floral and vegetative morphology and preference for shaded habitats. The stigma is elliptic and raised on a shallow but distinct stylar proeminence at the apex of the gynoecium; the mature leaf blade is typically leathery, more than twice as long as wide and ovate-elliptic to oblong-elliptic with a somewhat asymmetric, acute-to-rounded leaf base. In all populations, fertile vouchers were prepared, but when fertile individuals were not seen at the time of sampling, identification of plants sampled for DNA was made in the field from living, non-flowering material using vegetative and habitat features.
Voucher specimens of all populations sampled (see Supplementary Information available online) are deposited at the following herbaria: Herbário, Universidade Federal do Ceará, Fortaleza (EAC); Herbário, Universidade Estadual de Feira de Santana (HUEFS); Herbarium, Royal Botanic Gardens, Kew (K). Species names in Araceae follow CATE Araceae (2007).
Molecular techniques
Genomic DNA was extracted from leaf fragments dried in silica gel following the 2x CTAB protocol of Doyle and Doyle (1987) with modifications (10 mL extraction with 40 µL ß-mercaptoethanol, followed by precipitation in isopropanol). All samples were then cleaned using QIAquick columns (QIAgen Inc., Warrington, Cheshire, UK). AFLP reactions (Vos et al., 1995) were conducted according to the AFLPTM Plant Mapping Protocol of PE Applied Biosystems Inc. (Applied Biosystems, 1996), using EcoRI and MseI. Eight primer combinations (with three and four selective bases on the EcoRI- and MseI-based primers, respectively) were tested on a small number of samples. The two combinations that gave the best results in terms of clarity of traces, numbers of bands and numbers of polymorphic bands, were EcoRI-ACT/MseI-CTAG and EcoRI-AGG/MseI-CTAG, and these were then used with all samples. Fragments were separated using an ABI 3100 sequencer and analysed using Genescan and Genotyper 2·0 software (PE Applied Biosystems Inc.). Amplified fragments between 50 and 500 base pairs were scored by visual inspection for presence (1) or absence (0) of peaks in the output traces. Only distinct peaks were scored as present and the manual scoring procedure was repeated three times on separate occasions to reduce to a minimum inconsistencies in scoring.
Genetic data analysis
The resulting binary matrices of AFLP bands were used to carry out various genetic similarity analyses. Jaccard's similarity coefficient (Sneath and Sokal, 1973) was computed for all pairs of individuals using the SIMQUAL module in NTSYSpc ver. 2·20d (Rohlf, 2005), and then transformed into a dissimilarity measure as (1 – Jaccard's similarity).
Agglomerative hierarchical cluster analysis was carried out on this dissimilarity matrix using the SAHN module in NTSYSpc 2·20d with the Flexible (FLEXI) and UPGMA (unweighted pair-group method, arithmetic average) clustering methods. The Flexible clustering method was originally introduced by Lance and Williams (1967) and is defined by the parameter values of a general recurrence formula, which determines the shortest distance between a group k and a group previously formed by the fusion of two other groups (i and j) (Everitt et al., 2001). We used a Beta parameter value of –0·25, the default setting in NTSYSpc.
A 1000 replicate bootstrap analysis of the dissimilarity matrix was carried out in NTSYSpc ver. 2·2d using the RESAMPLE and SAHN modules successively to produce 1000 bootstrapped FLEXI trees. The CONSENS module was then used to compute a majority-rule consensus tree. The original non-bootstrapped FLEXI tree with the >50 % bootstrap values added can be seen in Fig. 2.
|
We analysed the data using Ward's method in the PAST package (PAST ver. 1·42, Hammer et al., 2001), based on a Euclidean distance matrix computed from the original binary matrix. Ward's method is an agglomerative algorithm that minimizes the increase in total error sum of squares when a new group is added to the existing one (Everitt et al., 2001).
Optimization clustering was carried out on the binary matrix using the K-means algorithm implemented in PAST (Hammer et al., 2001), in which, starting from a random assignment of individuals to clusters, the individuals are iteratively moved to the cluster with the closest (in Euclidean distance) centroid. K-means clustering requires an a priori choice of cluster number (k), and we repeated the analysis separately for k values of 2, 3, 4 and 5.
Two model-based (Bayesian) approaches were used to investigate structure at the level of the individuals. The software package STRUCTURE version 2·1 (Pritchard et al., 2000) was applied with a ‘no admixture’ model, 50 000 Burn-in period length and 50 000 MCMC replicates after Burn-in. This approach requires that the number of clusters (k) is predefined, and the analysis then assigns the individuals to the clusters probabilistically. We performed five runs for each value of k (2 to 12). The output (see Supplementary Information available online) uses colour coding to show the assignments of individuals in each population to the clusters. STRUCTURE is widely used for inferring gene-pool structure in genetic data.
The other model-based cluster analysis carried out used the software package MCLUST Version 3 for R (Fraley and Raftery, 2006; R Development Core Team, 2006). The initial data frame consisted of a matrix of scores on the first 33 principal coordinate axes (encompassing 99·42 % of the variance), derived from a principal coordinate analysis (PCoA) carried out in NTSYSpc ver. 2·2d (modules DCENTER, EIGEN, MXPLOT) on the dissimilarity (1 – Jaccard's similarity) matrix. The function mclustBIC was used, specifying a prior distribution (prior = priorControl()) to provide a complete set of values for BIC (Bayesian Information Criterion), which is the parameter used to assess which of the ten models tested fits the data best.
ARLEQUIN ver. 2·000 (Schneider et al., 2000) was used to compute the number of AFLP phenotypes observed in each population, using the setting ‘Infer haplotype definitions from distance matrix’. Diversity measures for each population were calculated using (1) GenAlEx 6 (Peakall and Smouse, 2006), which computes average expected heterozygosity (He, equivalent to Nei's gene diversity) for each population from binary data assuming Hardy–Weinberg equilibrium, and (2) POPGENE (Population Genetic Analysis) version 1·31 (Yeh et al., 1999), which computed percentage of polymorphic loci (p) and Nei's (1973) gene diversity (H).
Population genetic structure was studied using AFLP-SURV version 1·0 (Vekemans, 2002; Vekemans et al., 2002). The programme estimates allele frequencies at each marker locus in each population, assuming that markers are dominant and that there are two alleles per locus (presence of the band being dominant and absence recessive). We used the default option, i.e. Bayesian method with non-uniform prior distribution of allele frequencies (Zhivotovsky, 1999), and assumed Hardy–Weinberg equilibrium for the data. Wright's fixation index FST (Hartl and Clark, 1997) was computed using the method of Lynch and Milligan (1994) and tested by a permutation procedure of 1000 replicates, which randomly permuted individuals among the populations and then recalculated FST for each permutation; the observed value of FST was then compared to the distribution of randomized FST values.
A distance matrix of FST between every pair of populations was calculated in AFLP-SURV as a measure of interpopulation genetic differentiation, from which 1000 bootstrapped replicate matrices were then computed. The software package PHYLIP version 3·66 (Felsenstein, 2006) was used to carry out the bootstrap analysis: the module NEIGHBOR was used to compute UPGMA dendrograms for all bootstrapped matrices and an extended majority-rule consensus tree was produced using the module CONSENSE (see Fig. 3). Interpopulation genetic distances were also computed and tested using the software package GenAlEx 6 (Peakall and Smouse, 2006), which computes PT, a metric analogous to FST and especially suitable for binary data. The significance of these interpopulation PT distances was evaluated by a permutation test of 1000 permutations.
|
Analysis of Molecular Variance (AMOVA) was carried out using ARLEQUIN version 2·000 (Schneider et al., 2000). Genetic structure and F statistics are computed from a matrix of Euclidean squared distances between every pair of individuals. Four models of the eight populations were tested to investigate regional relationships. The first treated all populations as a single group, i.e. a two-level analysis, to obtain a value for FST as an overall measure of population divergence. The other three models divided the populations into two higher groups, which were subjected to three-level hierarchical analyses; the groupings differed as follows: (1) Ceará and Atlantic forest populations together; (2) Ceará and Amazonian populations together; and (3) Ceará, Pernambuco and Amazonian populations together (Table 2).
|
Isolation-by-distance was investigated by computing the correlation between geographical and genetic (FST) population pair-wise distance matrices and applying the Mantel test, using NTSYSpc version 2·2d (Rohlf, 2005). The significance of the Mantel z-statistic value was tested non-parametrically by creating a null distribution of z using 10 000 random permutations and comparing the observed z value.
Population structure was also investigated by the Bayesian software HICKORY version 1.0.4 (Holsinger et al., 2002; Holsinger and Lewis, 2003), in order to assess the importance of inbreeding in the data and the assumption of Hardy–Weinberg equilibrium. HICKORY makes it possible to evaluate departures from Hardy–Weinberg equilibrium in dominant as well as co-dominant markers. The AFLP data were fitted to four models, ‘full model’, which allows for inbreeding, ‘f = 0’ model, which implies lack of inbreeding, ‘theta = 0’ model, which implies a zero-valued FST analogue (i.e. no differentiation between populations), and ‘f free’ model, which decouples the estimates of f and theta. Computations were carried out using the default values, as recommended in the manual (Holsinger and Lewis, 2003): burn-in = 5000, number of samples = 25 000 and thinning factor = 5. To estimate the best fit of the four models, the criteria DIC (deviance information criterion), Dbar, Dhat and pD were used, with interpretations following the recommendations given in the manual.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY INFORMATIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
The matrix of 75 individuals from which satisfactory results were obtained in both primer combinations was made up of 223 fragments; 88·79 % of the scored marker bands were polymorphic. All AFLP phenotypes recorded were unique to an individual and we tentatively concluded that no resampling of clones had occurred during the field collections (see Discussion).
Hierarchical and model-based cluster analysis
Agglomerative hierarchical cluster analysis using flexible method (FLEXI in NTSYSpc) and Ward's method (in PAST) resulted in dendrograms in which either two or three major clusters were differentiated with the same composition of individuals (Clusters 1–4, Fig. 2); cluster 1 had 90·5 % bootstrap support (1000 replicates). Optimization methods (K-means analysis in PAST) and the Bayesian analysis package STRUCTURE also found these clusters for settings of k = 2 and k = 3 (k = number of clusters; Figs 1B and 2). Principal coordinate analysis (PCoA, see Supplementary Information available online) showed a clear differentiation of cluster 1 and a lesser distinction of clusters 2 and 3; the first two principal coordinate axes accounted for a total of 34·2 % of the total variance in the data.
In the genetic structure analysis (STRUCTURE), the highest estimate of the likelihood of the data, given the number of clusters chosen, was obtained when k = 8, i.e. with eight clusters. The diagram showing assignment of individuals to the clusters revealed a clinal structure to the data (Fig. 1C, see Supplementary Information available online). However, for models with more than three clusters (k > 3) cluster composition became variable between runs. Reflecting this, the variance values of the five runs fluctuated considerably once the k value exceeded k = 4 and increased progressively; the largest k value for which cluster composition was constant was found to be k = 3 (Table 3). Figure 1A shows the occurrence of the clusters (gene pools) in the sampled populations for the three-cluster model, and the assignment of individuals to both three- and eight-cluster models (Fig. 1B, C). Gene pool 3 is the most widespread pattern, represented in all populations and the only one found in the southernmost population (KRJSP). Gene pool 2 is the most restricted, occurring only in north-east Brazil and Espírito Santo. Gene pool 1 is absent from the southern Brazilian and Bahian populations.
|
Model-based clustering using MCLUST found the best fit with a two-cluster solution (Fig. 2, Clusters 1 and 4), when scores from all 33 principal coordinates were used (comprising 99·42 % of total variance). The highest values for BIC all favoured the VEI model (Volume = variable, Shape = equal, Orientation = coordinate axes) and were, in descending order, 6734·082 for a two-cluster solution, 6663·384 for a three-cluster solution and 6583·755 for a four-cluster solution. The two clusters of the optimal model correspond in composition to Cluster 1 and Cluster 4 in the hierarchical cluster analyses (Fig. 2). Analyses of scores from the first ten principal coordinates (comprising 67·81 % of total variance), found an optimal six-cluster solution (see Supplementary Information available online), in which the groups corresponding to clusters 2 and 3 in the other analyses were further subdivided. In a similar way to the eight-group STRUCTURE solution, this seems to reflect the presence of further substructure in the data, which is not sufficiently well-defined to emerge consistently in all analyses.
The two-cluster structure (Fig. 2, Clusters 1 and 4) is clear in all the different types of cluster analysis and consistent in assignment of individuals to the clusters. All analytical approaches except MCLUST also agreed in the composition of the three-cluster structure. It thus seems reasonable to infer that there are at least two distinct historical gene pools represented within the overall range sampled and probably three (Fig. 1).
Population genetic diversity
The POPGENE and GenAlEx 6 packages agreed precisely in values for average expected heterozygosity or Nei's gene diversity (Table 4). The results showed relatively higher diversity for the two Ceará populations and lowest diversity for the southernmost population (KRJSP). Comparison with the analyses of individuals (Fig. 1A, B) suggests that this may be due to the presence in the KRJSP population of only one gene pool.
|
Hierarchical cluster analysis (UPGMA), using interpopulation distance matrices of population pair-wise FST (see Supplementary Information available online), grouped the Ceará populations with those in the Atlantic forest, the two Amazonian populations forming a separate subgroup (Fig. 3). The generally low bootstrap support, except for the two pairs KCEpa/KPER (89·5 %) and KBA/KES (72·5 %), shows that over the large region sampled (3232 km maximum extent), geographical structure is relatively weak.
The permutation test of inter-population distance using GenAlEx 6 (PT, 1000 permutations), found all but two pairs (KCEpa/KPER, KPA/KPER) significantly different (P < 0·02). The genetic distance separating the two populations from Ceará (pairwise FST = 0·137, P = 0·003) was statistically significant, although they are geographically closer than any other pair (257 km). KPA is inconsistent in pairing with KFG in the UPGMA population cluster analysis, but preferentially with KPER in the PT pairwise distance analyses. This also suggests generally weak genetic structure at the population level.
The two-level AMOVA in ARLEQUIN found an FST value of 0·1896 (P < 0·001), with 18·96 % of variation among populations and 81·04 % within populations (Table 2). Three-level hierarchical AMOVA analyses of the two-group models (Table 2) supported the results of cluster analysis in favouring the association of the Ceará populations with those of the Atlantic Forest rather than with the Amazonian ones. Percentage variation among groups was 11·95 % when the Ceará populations are grouped with those of the Atlantic forest, in comparison to values of 2·88 % and 4·44 % for the other two models, which grouped the Ceará populations with those in Amazonia.
A Mantel test showed strong correlation between the geographical distance matrix and the matrix of pair-wise FST derived from AFLP-SURV [Fig. 4, P (random z 
observed z) = 0·002, r = 0·6903]. This result allows us to infer that over the sampled range, M. adansonii var. klotzschiana shows a historical pattern of isolation-by-distance (Wright, 1943, 1946), i.e. that during the period of time represented by these patterns, gene flow and genetic drift were in equilibrium.
|
The Bayesian software HICKORY results showed that inbreeding was unlikely to be a major driving force in determining the gene pool structure in this taxon (Table 5). Although the DIC parameter was lower in the Full model than in the f = 0 model (difference in DIC value = 17·74), the difference was more than accounted for by the number of parameters in the two models (difference in pD = 33·85), and the other two parameters (Dbar and Dhat) were lower in the f = 0 model (Dbar = 3591·52, Dhat = 2866·17) than in Full model (Dbar = 3607·62, Dhat = 2916·11). Because the model with the lowest Dbar value is that which best fits the data, this pattern of results allows the f = 0 model (inbreeding equals zero) to be considered at least as likely as the Full model, according to the HICKORY manual (Holsinger and Lewis, 2003).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY INFORMATIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
This study is the first known to us that compares genetic patterns in plants at population level in the brejo forests of north-east Brazil and as such is preliminary, given the absence of strictly comparable work in the region and the small sample sizes we had available. Instead of showing strong local differentiation in the brejo forest populations and marked differences between the Amazonian and Atlantic forest, the results present a picture of a widespread taxon with a clinal pattern of genetic variability and rather low differentiation over a subcontinental geographical region. The overall FST value of 0·1896 in M. adansonii var. klotzschiana is comparable with other plants having outcrossing or mixed mating systems, reviewed by Morjan and Rieseberg (2004 : Table 3 and Supplementary Information), although rather lower than in cases sampled across comparably large geographical regions. The low FST value represents high within-population variation and low differentiation between populations. Assogbadjo et al. (2006) reported a comparable result for Adansonia digitata (Malvaceae,
PT = 0·176, 17·63 % variation among populations, 82·37 % within populations), considered an outbreeder on other evidence. Cardoso et al. (2005), in a study of isolated populations of Caesalpinia echinata (Leguminosae) from eastern Brazil, found a strong correlation between genetic and geographical distance, but over half (61 %) of total genetic variation partitioned between populations. They proposed that this could have resulted from ancient fragmentation of seasonally dry vegetation within a predominant matrix of moist Atlantic forest, a comparable but inverse scenario to that of Monstera adansonii. The strong pattern of isolation-by-distance and the lack of evidence for inbreeding suggest that this is a historical pattern brought about by extensive gene flow which, as the distance between localities increases, has had a decreasing effect in cancelling out genetic differentiation caused by genetic drift. In a study of frog molecular phylogeography, Carnaval (2002) concluded that geologically recent, but not present-day, gene flow had taken place between brejo forest and coastal populations in the Atlantic forest of the north-eastern Brazilian states of Pernambuco and Alagoas.
Genetic structure analysis of the individuals shows evidence that two distinct gene pools are represented in the data, suggesting that different historical phases of widespread gene flow have taken place in the region sampled. Many of the analyses also indicated the presence of a third gene pool, and the eight-cluster result from the Bayesian STRUCTURE analysis even hinted at gradual and successive replacement of locally occurring, but weakly differentiated, genetic patterns from French Guiana in the north-west to São Paulo in the south-east (Fig. 1C). Monstera adansonii var. klotzschiana is restricted to humid forest habitats and although information on its ecology and population biology is sparse, it nevertheless seems unlikely that gene flow by pollen or seed dispersal can occur between populations separated by semi-arid environments. We therefore tentatively conclude that the hypothesized gene flow episodes could have been correlated with cycles of humid forest expansion across eastern South America.
The two brejo forest populations sampled in Ceará were found to be well differentiated from each other and as diverse or more so than those elsewhere. Our initial expectation had been that the Ceará populations would show lower-than-average diversity, due to effects of genetic drift and restricted gene flow in these small, ecologically and geographically isolated, natural forest ‘refugia’. Instead, it was the southernmost population from the Rio de Janeiro–São Paulo region that showed lower-than-average genetic diversity. Further studies are needed to explore these patterns, but there is at least an indication in our results that successive population expansions involving either Amazonia or the Atlantic forest, or both, have led to an accumulation of diversity in the forest fragments in Ceará that has counteracted genetic erosion.
The among-population relationships show that the Ceará populations are related to those of the Atlantic forest in north-east Brazil rather than to Amazonian ones. Although bootstrap values are mostly rather low in the UPGMA tree (Fig. 3), this interpretation is supported by the hierarchical AMOVA analyses (Table 2). Our results are in agreement with the current classification of the Ceará brejo forests as part of the Atlantic forest of Brazil (Fundação SOS Mata Atlântica, 2007). A similar conclusion was reached for the herpetofauna of the Baturité massif in Ceará by Borges-Nojosa (1991), who reported a biogeographic pattern of disjunctions dominated by Atlantic forest taxa.
From the conservation standpoint the results show that in north-east Brazil isolated populations in small and threatened areas of forest with low plant species endemism have unique intraspecific genetic diversity patterns and population diversity at least as high as that in conspecific populations from known biodiversity hotspots, such as the Atlantic forest and the Belém region of Amazonia. Future genetic investigations of other long-lived and widely distributed brejo forest species such as the trees Schefflera morototoni (Araliaceae), Tapirira guianensis and Thyrsodium schomburgkianum (both Anacardiaceae) would be interesting to compare with our results.
Our interpretations and conclusions are necessarily tentative at this stage, but the study is a useful basis from which to develop more comprehensive and detailed work in the future. Larger samples are needed to confirm the gene pool cluster patterns and population diversity levels reported here. Analysis of duplicates from the same individuals is needed to improve error estimates in AFLP band scoring and to inform the evaluation of clonality in the populations. More sampling localities are required, especially in the southern Atlantic forest, where genetic diversity was found to be least, and in Amazonia, where Monstera species diversity is much greater. Future work should focus on co-dominant markers such as nuclear microsatellites. Given the preliminary nature of this project, we chose to use AFLP as an efficient technique for obtaining a broad picture of molecular patterns in a previously unstudied taxon. However, AFLP markers are dominant and provide no information on heterozygote frequencies and hence no direct estimates of inbreeding. Although we found no evidence of inbreeding from HICKORY software analysis, the use of co-dominant markers would provide confirmation of this and better understanding of the processes that have conditioned the observed genetic structure. The current patchy understanding of the mating system also prevented us from inferring the contribution of inbreeding to the genetic structure observed, pointing to the need for new research on reproductive ecology in M. adansonii.
In conclusion, the AFLP markers in populations of M. adansonii var. klotzschiana show a pattern of relatively high within-population diversity and relatively low, although still significant, among-population divergence. For the populations in Ceará, these results were unexpected given their geographical and ecological isolation, since their within-population diversity was as high as or higher than that of populations in the Atlantic forest and Amazonia. The widespread pattern of isolation-by-distance and the distinct gene pool patterns found over the range sampled suggest that episodes of extensive gene flow have occurred involving populations in Amazonia, Ceará and the Atlantic forest of Brazil. We hypothesize that this is a legacy from an epoch when gene flow within continuous humid forest habitat was possible, rather than evidence of present-day long-distance gene dispersal, and suggest that our results are consistent with theories proposed by various previous authors of Holocene migration of forest biota between Amazonia and the Atlantic forest. The genetic differentiation and relatively high diversity of the separate brejo populations in Ceará emphasize their biological uniqueness and scientific importance, despite their threatened status and highly disturbed condition. Measures to reinforce conservation of these forests are urgently needed.
|
SUPPLEMENTARY INFORMATION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY INFORMATIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Supplementary information is available online at http://aob.oxfordjournals.org/ and gives the following: (1) Voucher herbarium specimens documenting the populations sampled; (2) Gene pool cluster diagrams from STRUCTURE ver. 2·1, showing clinal structure in three separate runs of the eight-cluster model; (3) Principal Coordinates Analysis (PCoA) ordination showing six-cluster model computed with MCLUST ver. 3; (4) PCoA ordination showing the three-cluster model; and (5) Tables of pairwise inter-population FST,
PT and geographical distance. |
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY INFORMATIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
We are most grateful to the following organizations and persons for their essential help: FUNCAP (Fundaçao Cearense de Amparo à Pesquisa), FAPESB (Fundação de Amparo à Pesquisa do Estado da Bahia),WWF Brasil, Kew Latin America Research Fellowships Programme (Royal Botanic Gardens, Kew), Margaret Mee Fellowships Programme (Fundação Botânica Margaret Mee–Rio de Janeiro, Royal Botanic Gardens, Kew), Ministério do Meio Ambiente (IBAMA, permit number 001/2004, 042/04), Conselho Nacional de Desenvolvimento Científica e Tecnológica (CNPq, MCT permit number, 168/01), Dr Ana Maria Giulietti and staff at Universidade Estadual de Feira de Santana (UEFS), Robyn Cowan, Anna Haigh, Dr Denis Barabé, Dr Jean-Jacques de Granville, Vanessa Hequet, Françoise Crozier, M.F. Prévost and other staff of the CAY herbarium, Bertrand Goguillon and staff of IRD (French Guiana), Muriel Nugent, Joep Moonen, Sr. Thales Alves Ribeiro, Sr. José Teodoro Soares, Rector of the Universidade Estadual do Vale do Acaraú (UVA), Sobral. We are grateful for the suggestions for improvement received from two anonymous reviewers.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY INFORMATIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
-
Ab'Sáber AN. The paleoclimate and paleoecology of Brazilian Amazônia. In: Biological diversification in the tropics—Prance GT, ed. (1982) New York: Columbia University Press. 41–59.
Andrade IM, Mayo SJ. Dynamic shoot morphology in Monstera adansonii Schott var. klotzschiana (Schott) Madison (Araceae). Kew Bulletin (1998) 53:399–417.
Andrade-Lima D. Esboço fitoecológico de alguns brejos de Pernambuco. Boletim Técnico do Instituto de Pesquisas Agronômicas de Pernambuco, nova série (1966) 8:3–10.
Andrade-Lima D. Present-day forest refuges in northeastern Brasil. In: Biological Diversification in the tropics—Prance GT, ed. (1982) New York: Columbia University Press. 245–251.
Applied Biosystems. AFLP plant mapping protocol (1996) Foster City, CA: Applied Biosystems. Part Number 4303146C.
Assogbadjo AE, Kyndt T, Sinsin B, Gheysen G, Van Damme P. Patterns of genetic and morphometric diversity in Baobab (Adansonia digitata) populations across different climatic zones of Benin (West Africa). Annals of Botany (2006) 97:819–830.
Behling H. Late Quaternary vegetational and climate changes in Brazil. Review of Paleobotany and Palynology (1998) 99:143–156.[CrossRef]
Bigarella JJ, Andrade-Lima D. Paleoenvironmental changes in Brazil. In: Biological diversification in the tropics—Prance GT, ed. (1982) New York: Columbia University Press. 27–40.
Borges-Nojosa DM. Herpetofauna do Maciço de Baturité. Estado do Ceará: Composição, Ecologia e Considerações zoogeográficas (1991) João Pessoa: Dissertação de Mestrado, Universidade Federal da Paraíba.
Borges-Nojosa DM, Caramaschi U. Composição e Análise. Comparativa da Diversidade e das Afinidades Biogeográficas dos Lagartos e Anfisbenídeos (Squamata) dos brejos Nordestinos. In: Ecologia e Conservação da Caatinga—Leal I, Silva JMC, Tabarelli M, eds. (2003) 1, 1st edn. 489–540.
Cardoso SRS, Provan J, Lira CF, Pereira LOR, Ferreira PCG, Cardoso MA. High levels of genetic structuring as a result of population fragmentation in the tropical tree species Caesalpinia echinata Lam. Biodiversity and Conservation (2005) 14:1047–1057.[CrossRef][Web of Science]
Carnaval ACOQ. Phylogeography of four frog species in forest fragments of north-eastern Brazil – a preliminary study. Integrative and Comparative Biology (2002) 42:913–921.
CATE Araceae. Creating a taxonomic e-science: Araceae (2007) http://www.cate-araceae.org.
Chouteau M, Barabé D, Gibernau M. Pollen–ovule ratios in some Neotropical Araceae and their putative significance. Plant Systematics and Evolution (2006) 257:147–157.[CrossRef][Web of Science]
Doyle JJ, Doyle JL. Isolation of plant DNA from fresh tissue. Focus (1987) 12:13–15.
Everitt BS, Landau S, Leese M. Cluster analysis (2001) 4th edn. London: Arnold.
Felsenstein J. PHYLIP: Phylogeny inference package, version 3·66, July 2006. (2006) Seattle, WA: University of Washington. Available for download at. http://evolution.gs.washington.edu/phylip.html.
Figueiredo MA. Vegetação. In: Atlas do Ceará—Anon, ed. (1997) Fortaleza, Brazil: Fundação Instituto de Planejamento do Ceará (IPLANCE). 28–29.
Fraley C, Raftery AE. MCLUST version 3 for R: normal mixture modeling and model-based clustering (2006) 9. University of Washington: Department of Statistics. Technical Report no. 504 2006 Available for download at http://www.stat.washington.edu/mclust/.
Fundação SOS Mata Atlântica. (2007) accessed 21 March 2007). http://www.sosmatatlantica.org.br/.
Gibernau M. Pollinators and visitors of aroid inflorescences. Aroideana (2003) 26:66–83.
Grayum MH. Evolution and phylogeny of the Araceae. Annals of the Missouri Botanical Garden (1990) 77:628–697.[CrossRef][Web of Science]
Hammer Ø, Harper DAT, Ryan PD. PAST: palaeontological statistics software package for education and data analysis. Palaeontologia Electronica (2001) 4(1):9. Download from http://folk.uio.no/ohammer/past.
Hartl DL, Clark AG. Principles of population genetics (1997) 3rd edition. Sunderland, MA: Sinauer Associates.
Holsinger KE, Lewis PO. Hickory: a package for analysis of population genetic data v1·0 (2003) University of Connecticut, Storrs: Distributed by the authors. Version 1.0.4 available for download at http://darwin.eeb.uconn.edu/hickory/hickory.html.
Holsinger KE, Lewis PO, Dey DK. A Bayesian approach to inferring population structure from dominant markers. Molecular Ecology (2002) 11:1157–1164.[CrossRef][Medline]
Lance GN, Williams WT. A general theory of classificatory sorting strategies. 1. Hierarchical systems. Computer Journal (1967) 9:373–380.
Ledru MP. Late Quaternary environmental and climatic changes in Central Brazil. Quaternary Research (1993) 39:90–98.
Lynch M, Milligan BG. Analysis of population genetic structure with RAPD markers. Molecular Ecology (1994) 3:91–99.[Medline]
Madison MT. A revision of Monstera (Araceae). Contributions to the Gray Herbarium (1977) 207:1–101.
Mayo SJ. Aspectos da fitogeografia das Araceae bahianas. Anais do XXXIV Congesso Nacional de Botânica, Porto Alegre, Brasil (1983) (1984) 2:215–227.
Mayo SJ, Bogner J, Boyce PC. The genera of Araceae (1997) Kew, London: Royal Botanic Gardens.
Morjan CL, Rieseberg LH. How species evolve collectively: implications of gene flow and selection for the spread of advantageous alleles. Molecular Ecology (2004) 13:1341–1356.[CrossRef][Medline]
Nei M. Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Sciences (1973) 70:3321–3323.
Oliveira PE, Barreto AMF, Suguio K. Late Pleistocene/Holocene climatic and vegetational history of the Brazilian Caatinga: the fossil dunes of the middle São Francisco River. Palaeogeography, Palaeoclimatology, Palaeoecology (1999) 152:319–337.[CrossRef]
Oliveira-Filho AT, Ratter JA. A study of the origin of Central Brazilian forests by the analysis of plant species distribution patterns. Edinburgh Journal of Botany (1995) 52:141–194.
Peakall R, Smouse PE. GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. In: Molecular Ecology Notes (2006) 6:288–295. Available for downloading at http://www.anu.edu.au/BoZo/GenAlEx/.[CrossRef][Web of Science]
Plowman T. Four new species of Erythroxylum (Erythroxylaceae) from north-eastern Brazil. Brittonia (1986) 38:189–200.[CrossRef][Web of Science]
Pôrto KC, Cabral JJP, Tabarelli M. Brejos de altitude em Pernambuco e Paraíba: história natural, ecologia e conservação. (2004) Brasília: Ministério do Meio Ambiente. 324. (pdf available at http://www.mma.gov.br/index.php?ido=conteudo.monta&idEstrutura=14&idConteudo=1989.
Pritchard JK, Stephens M, Donnelly P. Inference of population structure from multilocus genotype data. In: Genetics (2000) 155:945–959. Software available for download at http://pritch.bsd.uchicago.edu/software/structure2_1.html.
R Development Core Team. R: A language and environment for statistical computing (2006) Vienna, Austria: R Foundation for Statistical Computing. http://www.R-project.org.
Ramalho FC. Taxonomia e número cromossômico de representantes da família Araceae em Pernambuco (1994) Recife: Universidade Federal Rural de Pernambuco. M.Sc. dissertation.
Ramirez WB. Informe cientifico. Brenesia (1980) 18:367–368.
Ramirez WB, Gomez LPD. Production of nectar and gums by flowers of Monstera deliciosa (Araceae) and of some species of Clusia (Guttiferae) collected by New World Trigona bees. Brenesia (1978) 14–15:407–412.
Rodal MJN, Sales MF, Mayo SJ. Florestas Serranas de Pernambuco: Localização dos remanescentes dos brejos de altitude. (1998) Recife: Imprensa Universitária, Universidade Federal Rural de Pernambuco. 25.
Rohlf FJ. NTSYSpc: numerical taxonomy system (2005) Setauket, NY: Exeter Publishing, Ltd. ver. 2·20d.
Sales MF, Mayo SJ, Rodal MJN. Plantas vasculares das florestas serranas de Pernambuco: Um checklist da flora ameaçada dos brejos de altitude, Pernambuco Brasil. (1998) Recife: Imprensa Universitária, Universidade Federal Rural de Pernambuco.
Sampaio VSB, Giulietti AM, Virginio J, Gamarra-Rojas FL. Vegetação e flora da Caatinga. (2002) Recife: Associação Plantas do Nordeste, APNE Centro Nordestino de Informação Sobre Plantas, CNIP.
Schneider S, Roessli D, Excoffier L. Arlequin: a software for population genetic data: genetics and biometry laboratory (2000) Switzerland: University of Geneva. Available for downloading at http://lgb.unige.ch/arlequin/.
Sneath PHA, Sokal RR. Numerical taxonomy: the principles and practice of numerical classification (1973) San Francisco: WH Freeman.
Stannard BL. Flora of the Pico das Almas, Chapada Diamantina: Bahia, Brazil. (1995) Kew, London: Royal Botanic Gardens.
Vekemans X. AFLP-SURV version 1·0 (2002) Belgium: Distributed by the author. Laboratoire de Génétique et Ecologie Végétale, Université Libre de Bruxelles. Available for downloading at http://www.ulb.be/sciences/lagev/aflp-surv.html.
Vekemans X, Beauwens T, Lemaire M, Roldan-Ruiz I. Data from amplified fragment length polymorphism (AFLP) markers show indication of size homoplasy and of a relationships between degree of homoplasy and fragment size. Molecular Ecology (2002) 11:139–151.[CrossRef][Medline]
Vos P, Hogers R, Bleeker M, Reijans M, Van de Lee T, Hornes M, et al. AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research (1995) 23:4407–4414.
Wright S. Isolation by distance. Genetics (1943) 28:114–138.
Wright S. Isolation by distance under diverse systems of mating. Genetics (1946) 31:39–59.
Yeh FC, Yang R, Boyle T. POPGENE: Microsoft Windows-based freeware for population genetic analysis, version 1·31: Quick User Guide (1999) Edmonton: University of Alberta. 29.
Zhivotovsky LA. Estimating population structure in diploids with multilocus dominant DNA markers. Molecular Ecology (1999) 8:907–913.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||