Molecular Ecology (2003) 12, 2875 –2883 doi: 10.1046/j.1365-294X.2003.01950.x Population genetic structure of mahogany (Swietenia macrophylla King, Meliaceae) across the Brazilian Amazon, based on variation at microsatellite loci: implications for conservation Blackwell Publishing Ltd. M A R I S T E R R A R . L E M E S ,*† R O G É R I O G R I B E L ,* J O H N P R O C T O R † and D A R I O G R A T T A P A G L I A ‡§ *Laboratório de Genética e Biologia Reprodutiva de Plantas, Instituto Nacional de Pesquisas da Amazônia, C.P. 478, 69011-970 Manaus-AM, Brazil, †Department of Biological Sciences, University of Stirling, Stirling FK9 4LA, UK, ‡Laboratório de Genética de Plantas, EMBRAPA-Recursos Genéticos e Biotecnologia, C.P. 02372, 70770-900 Brasília-DF, Brazil, §Laboratório de Biotecnologia Genômica, Pós-Graduação em Ciências Genômicas, Universidade Católica de Brasília, SGAN 916 Mod. B, Asa Norte, 70790-160, Brasília-DF, Brazil Abstract Mahogany (Swietenia macrophylla, Meliaceae) is the most valuable and intensively exploited Neotropical tree. No information is available regarding the genetic structure of mahogany in South America, yet the region harbours most of the unlogged populations of this prized hardwood. Here we report on the genetic diversity within and the differentiation among seven natural populations separated by up to 2100 km along the southern arc of the Brazilian Amazon basin. We analysed the variation at eight microsatellite loci for 194 adult individuals. All loci were highly variable, with the number of alleles per locus ranging from 13 to 27 (mean = 18.4). High levels of genetic diversity were found for all populations at the eight loci (mean HE = 0.781, range 0.754 –0.812). We found moderate but statistically significant genetic differentiation among populations considering both estimators of FST and RST, θ = 0.097 and ρ = 0.147, respectively. Estimates of θ and ρ were significantly greater than zero for all pairwise population comparisons. Pairwise ρ-values were positively and significantly correlated with geographical distance under the isolation-by-distance model. Furthermore, four of the populations exhibited a significant inbreeding coefficient. The finding of local differentiation among Amazonian mahogany populations underscores the need for in situ conservation of multiple populations of S. macrophylla across its distribution in the Brazilian Amazon. In addition, the occurrence of microgeographical genetic differentiation at a local scale indicates the importance of maintaining populations in their diverse habitats, especially in areas with mosaics of topography and soil. Keywords: Amazon, conservation genetics, genetic structure, mahogany, microsatellites, Swietenia macrophylla, tropical tree Received 24 March 2003; revision received 6 June 2003; accepted 10 July 2003 Introduction The destruction of tropical forests world-wide has increased dramatically in recent decades (Whitmore 1997; Bawa & Seidler 1998), posing a significant threat to the maintenance of biodiversity and biological processes in Correspondence: Maristerra R. Lemes. Fax: + 55 (92) 6433285; E-mail: mlemes@inpa.gov.br © 2003 Blackwell Publishing Ltd tropical forest ecosystems (Bawa 1994; Young et al. 1996). The genetic threat to tropical trees results from the loss of genetic diversity associated with the extinction of local populations, reduced population sizes, and the disruption of mutualisms with pollinators and seed-dispersing animals (Bawa 1994; Hall et al. 1996; Nason et al. 1997; Aldrich et al. 1998; Dick 2001). Moreover, selective logging may promote dysgenic selection as a result of the continuous exploitation of large, superior individuals and may increase levels 2876 M . R . L E M E S E T A L . of inbreeding as a result of reduction in stand density (Bawa 1994; Murawski et al. 1994). To evaluate and mitigate the genetic effects of deforestation and logging, it has become a priority to obtain information on the natural levels and distribution of genetic variation in populations of tropical trees. Population genetic studies of tropical trees have shown that most of the species investigated are outcrossed, exhibit high levels of genetic diversity and gene flow, and carry much of the variation within rather than among populations (Hamrick & Loveless 1989; Alvarez-Buylla et al. 1996). However, the great majority of these studies were developed over relatively small spatial scales, and employed isozymes as the primary genetic markers (Hamrick & Loveless 1986; Loveless 1992, 1998; Alvarez-Buylla & Garay 1994; Hall et al. 1994). In recent years, the development of microsatellites for an increasing number of tropical trees (White & Powell 1997; Aldrich et al. 1998; Brondani et al. 1998; Collevatti et al. 1999; Dayanandan et al. 1999; Dick & Hamilton 1999; Gaiotto et al. 2001; Lemes et al. 2002) have allowed larger scale and more refined studies of population genetic structure (e.g. White et al. 1999; Collevatti et al. 2001). The central aim of this work was to characterize and understand the genetic structure of natural populations of mahogany (Swietenia macrophylla, Meliaceae) across a 2100km transect of the Brazilian Amazon using microsatellite loci recently developed for this species (Lemes et al. 2002). Despite the perceived importance of the Amazon basin for tree species diversity, and as a repository for half the world’s remaining rain forest, our investigation of mahogany is the first population genetic analysis of a tree species distributed across this vast region. Furthermore, mahogany is of considerable interest to resource managers as it is by far the most valuable Neotropical hardwood species. One cubic metre of export-quality sawn mahogany is valued at about US$ 700 on the international market (Verissimo et al. 1995), and Brazil alone exports about 500 000 m3/year. A previous population genetic study of mahogany was limited to Central America where the species is commercially extinct in most regions (Gillies et al. 1999). The recent inclusion of mahogany in CITES (Convention for International Trade in Endangered Species, Appendix II 2002) highlights international concern regarding the future of South American populations. Most natural populations of mahogany have been logged and there is evidence that the species does not regenerate in areas of intense exploitation (Gullison et al. 1996). Thus there is an urgent need for effective conservation and management of the remnant populations. To this end estimates of population genetic parameters are essential. The variability observed at microsatellite loci provides estimates of inbreeding, heterozygosity, gene flow and outcrossing rate, all of which are important measures for assessing the conservation and management status of tropical trees under intense human pressure. The specific goals of our mahogany research were: (i) to quantify the genetic diversity within and among a sample of natural populations at the regional and topographic scale of the Brazilian Amazon; (ii) to test for the association between genetic and geographical distances among populations; and (iii) to provide recommendations for the establishment of in situ reserves and/or ex situ germplasm collections in Brazil. Materials and methods Population sites, collection of samples and DNA extraction Adult trees of Swietenia macrophylla were sampled from seven natural populations near the southern boundary of the Brazilian Amazon. Sample sites were located at distances between 8 and 2103 km apart (Fig. 1). Populations were selected on the basis of accessibility and to maximize regional representation for the species. Leaf samples were collected from 24 to 34 individuals per population. Sample sizes varied among populations because of the relative ease with which individualized tree crowns could be reached. The leaves were dried in silica gel, and stored at −20 °C until DNA extraction. Total genomic DNA was extracted from the leaves following a standard CTAB procedure (Doyle & Doyle 1987). DNA quantification was performed by comparison with known concentrations of a DNA standard (Lambda DNA) in ethidium bromide-stained 1% agarose gels. Fig. 1 Geographical distribution of Swietenia macrophylla in the Brazilian Amazon (dashed lines) and localization of the seven populations sampled: (1) A. Azul; (2) Maraj; (3) P. Bueno; (4) Cach. A; (5) Cach. E; (6) C. Mendes and (7) P. Lacerda. © 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 2875–2883 G E N E T I C S T R U C T U R E O F B R A Z I L I A N A M A Z O N M A H O G A N Y 2877 Microsatellite analysis Microsatellite marker analysis of 194 individuals representing the seven populations was carried forward using eight marker loci developed and optimized for S. macrophylla (Lemes et al. 2002). Microsatellite locus amplifications, electrophoresis conditions and allele-size determinations by fluorescence detection in multiplexed assays were carried out as described by Lemes et al. (2002). To avoid incomplete + A addition, minimize stutter peaks, and allow the multiplexing of several loci in the same polymerase chain reaction (PCR), a hot start PCR procedure and a final elongation step at 72 °C for 45 min were used. Significance levels were determined after 1000 bootstraps with 95% nominal confidence intervals. Permutation tests (Lynch & Crease 1990) were carried out to determine if observed values of ρ were significantly different from zero. Finally, the hypothesis that populations are differentiated because of the isolation-by-distance (Wright 1943) was tested by correlating pairwise θ and ρ against the pairwise geographical distance. The Spearman Rank correlation coefficient was calculated and significance was determined with 1000 permutations using the Mantel procedure (Mantel 1967). The Mantel test was carried out using the software genepop version 3.1.c (Raymond & Rousset 1998). Results Data analysis To estimate overall levels of genetic diversity, the following measures were calculated for all populations using the gda software (Lewis & Zaykin 1999): mean number of alleles per locus (A), and mean observed (HO) and mean expected (HE) heterozygosity. Tests for departure from Hardy–Weinberg equilibrium were performed using the U-test (Raymond & Rousset 1998) considering the hypotheses of heterozygote deficiency and excess, using genepop 3.1.c (Raymond & Rousset 1998). Exact P-values were determined by a Markov chain method (Guo & Thompson 1992) implemented in genepop 3.1.c (Raymond & Rousset 1998). The extent and significance of the genetic differentiation among populations was investigated by estimating the fixation indices based on two models: (i) the infinite allele model (Kimura & Crow 1964) and (ii) the stepwisemutation model (Ohta & Kimura 1973). Unbiased estimates of Wright F-statistics (Weir & Cockerham 1984) were obtained under the infinite allele model using fstat version 2.9.1 (Goudet 2000). We estimated θ, an estimator of Wright’s fixation index FST, over all populations and for each pairwise population comparison; f, the withinpopulation inbreeding, which measures the correlation of allele frequencies among individuals within populations; and F the overall inbreeding that measures the correlation of allele frequencies within individuals in different populations (Cockerham 1969). The statistical significance of θ, f and F were tested by bootstrapping over loci with a 95% nominal confidence interval (Goudet 2000). Significance tests of multilocus pairwise θ were carried out using the software fstat version 2.9.1 (Goudet 2000) with Bonferroni corrections. Genetic differentiation under the stepwisemutation model was assessed by ρ, an estimator of Slatkin’s RST (Slatkin 1995) and analogous to FST. The estimate of ρ was obtained by calculating the between- and withinpopulation components of variance of allele sizes for all loci, populations and pairwise population comparisons using the software RST calc, version 2.2 (Goodman 1997). © 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 2875–2883 Genetic diversity and Hardy–Weinberg equilibrium All microsatellite loci surveyed were highly polymorphic. The mean number of alleles per locus was 18.4 (range 13– 27) (Table 1), whereas the mean number of alleles observed per locus per population was 9.5 (range 7.6 –10.7) (Table 2). The mean expected heterozygosity (HE) was generally higher than the mean observed heterozygosity (HO), with only one locus showing the same value for both estimates (Table 2). Of the 56 tests of conformity to Hardy–Weinberg proportions, seven showed a significant departure from expected proportions at the 5% level. Based on the estimates of the inbreeding coefficient ( f ) all significant deviations were due to a deficit of heterozygotes (Table 2). Population differentiation, structuring and isolation by distance Over all loci and populations, the mean coefficient of inbreeding ( f ) was low but significantly different from Table 1 Characterization of eight microsatellite loci, pooling individuals from seven populations of Swietenia macrophylla in the Brazilian Amazon Locus N A HE (range) HO (range) sm01 sm22 sm31 sm32 sm40 sm46 sm47 sm51 190 189 192 193 188 193 194 190 18 17 27 17 13 17 17 21 0.847 (0.222–0.904) 0.830 (0.525–0.837) 0.926 (0.807–0.922) 0.913 (0.774–0.900) 0.763 (0.629–0.769) 0.885 (0.790–0.901) 0.799 (0.379–0.836) 0.845 (0.715–0.903) 0.679 (0.235–0.920) 0.698 (0.500–0.823) 0.833 (0.783–0.967) 0.767 (0.559–0.880) 0.734 (0.592–0.840) 0.824 (0.760–0.875) 0.696 (0.360–0.853) 0.763 (0.500–0.909) Mean over all loci 18.4 0.851 0.749 The SSR locus name; N, number of individuals; A, total number of alleles; HE, expected heterozygosity; HO, observed heterozygosity. HE and HO range estimated across populations. 2878 M . R . L E M E S E T A L . Table 2 Microsatellite diversity in seven populations of Swietenia macrophylla in the Brazilian Amazon averaged over loci Population N A HE HO f A. Azul Cach. A Maraj P. Lacerda C. Mendes Cach. E P. Bueno Over all populations 29.4 32.0 25.0 23.3 34.0 24.0 23.5 27.3 8.4 10.7 9.2 9.7 10.6 10.2 7.6 9.5 0.761 0.785 0.793 0.812 0.754 0.810 0.754 0.781 0.753 0.781 0.740 0.812 0.709 0.780 0.680 0.750 0.012 0.005 0.068** − 0.004 0.060*** 0.042* 0.100** 0.038*** N, mean sample size per locus; A, mean number of alleles per locus; HO, mean observed heterozygosity; HE, mean expected heterozygosity, and within population coefficient of inbreeding ( f ). Significant departures from Hardy–Weinberg expectations at *P < 0.05, **P < 0.01, ***P < 0.0001. Table 3 Single locus and overall locus estimates of Wright’s Fstatistics and genetic differentiation for the seven populations of Swietenia macrophylla in the Brazilian Amazon Locus f F θ ρ sm01 sm22 sm31 sm32 sm40 sm46 sm47 sm51 Over all loci Upper bound** Lower bound 0.015 0.073 0.072 0.095 − 0.039 0.019 − 0.008 0.055 0.038* 0.065 0.007 0.226 0.174 0.106 0.166 0.051 0.071 0.150 0.105 0.132* 0.169 0.095 0.214 0.109 0.037 0.079 0.087 0.053 0.157 0.053 0.097* 0.140 0.062 0.191 0.089 0.139 0.195 0.146 0.084 0.145 0.190 0.147* 0.192 0.136 *P < 0.0001. f, inbreeding coefficient; F, over all inbreeding coefficient, θ; fixation index; ρ; population differentiation based on the step-wise mutation model. **95% confidence interval bounds on the estimates over all loci. zero ( f = 0.038, P < 0.0001). Four populations exhibited a significantly positive inbreeding coefficient, suggesting nonrandom mating of individuals within these populations (Table 2). The value of F, the measure of inbreeding that considers both the effects of nonrandom mating within and among populations, was significantly different from zero (F = 0.132, P < 0.0001), giving evidence of population structure (Table 3). The two overall measures of population differentiation θ and ρ were both significantly greater than zero (θ = 0.097 and ρ = 0.147, P < 0.0001), indicating a moderate but signi- Table 4 Pairwise multilocus estimates of genetic differentiation, θ and ρ, among seven populations of Swietenia macrophylla in the Brazilian Amazon Population comparison Distance (km) θ* ρ** Cach. A–Cach. E Cach. A–P. Bueno Cach. E–P. Bueno A. Azul–Maraj P. Lacerda–Cach. E Cach. A–P. Lacerda P. Lacerda–P. Bueno Cach. A–C. Mendes C. Mendes– Cach. E C. Mendes–P. Bueno P. Lacerda–C. Mendes Maraj–P. Lacerda A. Azul–P. Lacerda Maraj–P. Bueno Cach. A–Maraj Maraj–Cach. E A. Azul–P. Bueno A. Azul–Cach. A A. Azul–Cach. E Maraj–C. Mendes A. Azul–C. Mendes 8 17 24 107 375 381 389 882 884 884 1216 1258 1307 1323 1334 1337 1342 1355 1358 2101 2103 0.130 0.046 0.122 0.034 0.082 0.095 0.103 0.156 0.087 0.144 0.085 0.052 0.086 0.071 0.076 0.081 0.066 0.089 0.114 0.115 0.126 0.135 0.129 0.087 0.036 0.074 0.053 0.106 0.207 0.079 0.078 0.117 0.169 0.148 0.160 0.160 0.096 0.139 0.133 0.097 0.180 0.223 *For all pairwise comparisons of θ, nonadjusted P-values, P < 0.0001. For corrected P-values using standard Bonferroni procedure, P < 0.005. P-values obtained after 21000 permutations. **For all pairwise comparisons of ρ, P < 0.0001, except for pair A. Azul × Maraj, where P < 0.01. ficant degree of genetic differentiation among populations of Swietenia macrophylla in the Brazilian Amazon (Table 3). The estimates of ρ were numerically greater than the estimates of θ for 15 out of the 21 pairwise population comparisons (Table 4). The correlation between ρ and the geographical distance for the 21 pairwise comparisons among the seven populations (Fig. 2) was positive and significant (r = 0.617; P = 0.022), suggesting a pattern of isolation by distance among the S. macrophylla populations in the Brazilian Amazon. No significant correlation with geographical distance was demonstrated based on pairwise θ-values (P = 0.396). Discussion Genetic diversity of Swietenia macrophylla in the Brazilian Amazon Mahogany in South America is typically found in aggregations of several tens to hundreds of mature trees along seasonal streambeds and densities within aggregations may vary between 0.1 and 3 mature trees/ha (Grogan 2001; © 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 2875–2883 G E N E T I C S T R U C T U R E O F B R A Z I L I A N A M A Z O N M A H O G A N Y 2879 Fig. 2 Relationship between pairwise RST and geographical distance among populations of Swietenia macrophylla in the Brazilian Amazon (Mantel test of correlation, r = 0.617, P = 0.022). Verissimo et al. 1995; Gullison et al. 1996). The aggregations are scattered in the forest matrix, so that there may be tens of kilometres separating the aggregations. Our sampling of around 30 adult trees per population (aggregation) therefore is likely to be a representative subsample of the genetic variability occurring in these restricted populations. The high levels of genetic diversity observed for S. macrophylla follow the pattern found in other microsatellite studies of tropical tree species (Aldrich et al. 1998; Dayanandan et al. 1999; White et al. 1999; Collevatti et al. 2001). It is noteworthy that the levels of genetic diversity (A, the mean number of alleles per locus; and HE, the average gene diversity) for S. macrophylla in Brazil (A = 18.4 and HE = 0.78) were higher than those found for the conspecific Central America populations (A = 13.0 and HE = 0.66, Novick et al. 2003) analysed with seven out of the eight loci used in this study. It suggests that the more continuous forests in the Brazilian Amazon, where topographic barriers to gene flow seem to be minimal and there were more stable climatic conditions in South America during the Pleistocene (Whitmore & Prance 1987), may enhance the maintenance of the diversity. As expected for microsatellite loci with high mutation rates, the genetic diversity measures for S. macrophylla are high when compared to those derived from other kinds of markers such as isozymes (Loveless 1992). At least two explanations may account for the deficit of heterozygotes observed in four populations. One possible explanation is the occurrence of null alleles, which fail to amplify because of mutations in the flanking primer sequences (Callen et al. 1993). Some studies have documented null alleles at microsatellite loci at frequencies of up to 15% (Callen et al. 1993; Paetkau & Strobeck 1995; Pemberton et al. 1995; Jarne & Lagoda 1996). However, this seems to be an unlikely explanation because amplification failures that would reflect null/null homozygotes were rare at all loci in the present study (maximum failure of 3% © 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 2875–2883 at locus sm40; Table 1). Furthermore stronger evidence for the absence of null alleles was also seen for these microsatellites in a mating system study we carried out in the Marajoara population (Lemes 2000). In that study, the allelic transmission from 25 mother trees to 400 progeny individuals was analysed for these same eight loci, and all offspring displayed at least one maternal allele. The second explanation, which we favour in the case of Amazon mahogany, is that assortative mating, caused by spatial clustering or coincidence in flowering time among related groups of trees, has led to inbreeding and homozygote excess. The Marajoara mating system study (Lemes 2000) showed that S. macrophylla is predominantly outcrossed, but that some trees exhibit a considerable degree of selfing. Nonetheless, selfing is generally averted in mahogany because anthesis of male and female flowers is not usually synchronous within a tree (Styles 1972). Genetic differentiation of S. macrophylla in the Brazilian Amazon Both multilocus estimates of genetic differentiation, θ = 0.097 and ρ = 0.147 indicate a moderate but significant degree of differentiation among populations of S. macrophylla in the Brazilian Amazon. Theory suggests that population differentiation is more accurately estimated by RST, because this measure better accounts for the high mutation rate of microsatellite markers (Hedrick 1999). In contrast, FST often underestimates population differentiation at microsatellite loci. Despite the high mutation rate of microsatellite loci, the value of θ (0.097) obtained for S. macrophylla is similar to the mean GST (0.11) value found for 37 different tropical taxa based on isozymes (Loveless 1992). However, the significance of the similarity in θ and GST is hard to assess given the differences in geographical scale, species life history traits and genetic markers among studies. Relatively few studies have assessed genetic variation in natural populations of tropical tree species using microsatellites (Chase et al. 1996; Aldrich et al. 1998; Dayanandan et al. 1999; White et al. 1999; Collevatti et al. 2001). Two additional species in the same family as mahogany (Meliaceae) have been studied with microsatellites, Carapa guianensis in Costa Rica (Dayanandan et al. 1999), and Swietenia humilis in Honduras (White et al. 1999). Both species exhibited much lower levels of genetic differentiation (ρ = 0.041 and 0.032, respectively) among populations than found here for S. macrophylla. The spatial scale of the C. guianensis and S. humilis studies, with a maximum distance between populations of 44 km, stands in contrast to the 2100 km geographical scale of S. macrophylla examined here. Nonetheless, high levels of genetic differentiation were also observed between Amazonian populations Cach. A– Cach. E or Cach. A–P. Bueno, less than 20 km apart (Table 4). 2880 M . R . L E M E S E T A L . Considering all Brazilian mahogany populations, the genetic distance measured by RST is significantly correlated with geographical distance, whereas pairwise estimates based on θ were not. Despite the debate on the accuracy of measures based on variance in allele frequencies under the infinite allele model vs. allele size under the stepwise mutation model, theoretical studies suggest that the latter seems to be more appropriate for quantifying levels of genetic differentiation with microsatellites (Valdes et al. 1993; Slatkin 1995; Goldstein & Pollock 1997). Genetic differentiation among Amazonian mahogany populations probably reflects the interplay of ecological, evolutionary and biogeographic factors, such as pollen and seed dispersal mechanisms, demographic history and geographical barriers to gene flow (Alvarez-Buylla et al. 1996). Considering the large geographical scale of this study, our data indicate lower than expected levels of differentiation among Amazonian populations of S. macrophylla, given that it is a patchily distributed forest tree species that is pollinated by nonspecialist insects and whose seeds are dispersed by wind. Factors that should limit gene flow in mahogany are the following. First, short-distance pollination is thought to generally restrict opportunities for gene exchange between populations. The minute flowers of mahogany are pollinated by a diverse array of generalist insects, such as small bees and moths (Styles 1972), which have limited foraging ranges compared with other more specialized vectors, such as bats, large or medium-sized euglossine bees (Frankie et al. 1976; Bawa 1990), or the small wind-dispersed wasps that pollinate fig trees (Nason & Hamrick 1997). Second, wind dispersal of seeds correlates with higher levels of genetic differentiation (Loveless 1992). While most tropical trees have animal-dispersed seeds, in mahogany, median wind seed-dispersal distances are only 32–36 m (Gullison et al. 1996). Notwithstanding predictions of restricted gene flow based on mahogany life history traits, recent data on Swietenia humilis in a fragmented forest mosaic in Honduras have shown pollen movement at distances > 4.5 km (White et al. 2002). Such long-distance pollination promoted by nonspecialist insects may be one factor behind the relatively weak population genetic structure of Amazonian mahogany. In addition, the distance travelled by winddispersed seeds may be underestimated by current methods, particularly in light of the high frequency of storms and blow-downs in the Amazon (Nelson et al. 1994). Our results suggest that landscape topography may play a role at least as important as life history in establishing the population structure of Brazilian mahogany. Pairwise population comparisons indicated low levels of genetic differentiation (ρ = 0.04) between the easternmost populations sampled (Marajoara and A. Azul), which are 107 km apart but located in a flat region with no notable geographical barrier between them. In contrast, populations from the Serra dos Parecis mountains (Cach. A, Cach. E and P. Bueno), exhibited considerably higher pairwise differentiation (ρ = 0.09–0.13) despite the short distance (8 –24 km) between them. The pairwise ρ estimates between the Serra dos Parecis populations are only slightly lower than those found between eastern (Marajoara and A. Azul) and western populations (P. Bueno, Cach. A, Cach. E and C. Mendes) of mahogany, which lie 1323 –2103 km apart and are separated by the Tapajos and Xingu rivers. The modest pairwise divergence between western vs. eastern populations of mahogany suggests that major Amazonian rivers have not effectively isolated populations on either side, perhaps owing to headwater gene exchange or interfluvial gene flow during the drier and colder periods of the Pleistocene in the Amazon, when the large tributaries were probably much narrower (Maslin & Burns 2000). It is possible also that most populations share fairly recent ancestry as a result of recolonization of the current range from restricted glacial refugia. Our data suggest that mountains may represent a much more effective physical barrier to gene flow among S. macrophylla populations than the major Amazonian rivers. This is in accordance with findings in Mesoamerica (Novick et al. 2003), where the Talamanca mountains have apparently acted as a genetic barrier separating the Pacific and Atlantic populations of S. macrophylla. Implications for conservation Mahogany is threatened throughout its range in South America as a result of over-exploitation and habitat destruction, which have clearly reduced local population sizes and led many populations to local extinction (Verissimo et al. 1995; Grogan 2001). The distribution of S. macrophylla along the southern boundary of the Brazilian Amazon coincides with areas of higher deforestation rates collectively referred to as the ‘Arc of Deforestation’ including the Brazilian states of Pará, Tocantins, northern Mato Grosso, Rondônia and Acre. Habitat degradation caused by selective logging and, most importantly, through conversion of forest into soybean plantations and cattle ranches with recurrent use of fire is likely to reduce the colonization of new sites, despite the ability of mahogany to regenerate in disturbed habitats (Snook 1996). Deforestation and forest degradation have been so intense in this vast region that the genetic diversity of many remnant populations may already be compromised by genetic drift and inbreeding. Even establishing reserves provides no safeguard for mahogany as illegal extractions of this hardwood have been reported from National Parks and Indian reserves in Brazil and other countries (Rodan et al. 1992; Grogan 2001). Clearly, the long-term survival of mahogany requires immediate protection of representative populations across © 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 2875–2883 G E N E T I C S T R U C T U R E O F B R A Z I L I A N A M A Z O N M A H O G A N Y 2881 the species’ geographical range. Our documentation of population genetic structure of Brazilian mahogany coupled with demographic data (Verissimo et al. 1995; Grogan 2001) provides a blueprint for designing conservation and management policies to maintain the genetic diversity of this valuable timber species. The very high level of genetic variation found within all populations, and the low densities of adult trees, indicate that in situ conservation strategies should be designed to preserve large areas to minimize the loss of diversity due to genetic drift. The predominantly allogamous mating system of S. macrophylla (Lemes 2000) suggests that breeding populations are large and gene flow is extensive in mahogany species, thus reinforcing the need to protect large forest areas. The isolation by distance observed among Amazonian mahogany populations suggests that in situ reserves should be distributed evenly across the species’ range in order to conserve maximally the regional genotypic diversity. Since mahogany is predominantly allogamous and resilient to some level of habitat disturbance and fragmentation (Lemes 2000; White et al. 2002), we recommend that in situ reserves be linked by smaller, managed areas in private lands, extractive reserves, or Indian territories, where limited selective logging may be allowed. These areas can act as corridors and stepping-stones for gene flow, and would provide a metapopulation structure to the managed and preserved populations. Regions that are topographically complex, like Serra dos Parecis, may warrant a more detailed conservation strategy, since populations in these areas tend to exhibit higher genetic differentiation. The occurrence of such microgeographical differentiation emphasizes the importance of maintaining populations in their diverse habitats, especially in areas with mosaics of topography and soils. Furthermore, in situ genetic conservation initiatives for mahogany should be associated with community-based conservation strategies aiming to protect samples of the rich, seasonal forests of the south Amazon boundary. If the rate of deforestation and mahogany logging in remnant populations continues at current rates, especially along the southern boundary of the Brazilian Amazon, ex situ conservation policies should be urgently implemented. As mahogany seed viability decays rapidly, conservation should involve germplasm preservation realized through planted trees. Seed collection strategies for the establishment of ex situ seed banks should follow the same principle suggested for in situ conservation: sampling of several open-pollinated progeny arrays from broadly spaced trees representing populations covering a wide geographical range and later maintaining the identity of those provenances in the ex situ collection. More intensive sampling will be required in regions of high topographic relief and heterogeneous soil types. © 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 2875–2883 A number of studies in recent years have used high throughput DNA marker technologies to describe the genetic diversity of Neotropical tree species at increasingly larger geographical scales and finer population structure (e.g. Chase et al. 1996; Aldrich et al. 1998; Dayanandan et al. 1999; White et al. 1999; Collevatti et al. 2001; White et al. 2002). Although such studies have contributed significantly to the understanding of the general patterns of genetic variation and dynamics of tropical trees, efforts to translate such knowledge into effective conservation practices have been timid. By showing that populations of Brazilian Amazon mahogany are isolated by distance and by regional topography, this study attempts to provide the basis for sound in situ and ex situ conservation planning. It should be kept in mind, however, that conservation of tropical trees involves not only theoretical considerations based on the genetics of the species, but also complex social and economical issues. Acknowledgements The study was funded by the Brazilian Ministry of Science and Technology (CNPq/PADCT grant no.62.00059/97-4 to D.G., CNPq/RHAE fellowship no. 260021/94.6 to M.R.L.), World Wildlife Fund — Brazil (Program Natureza e Sociedade, grant no. CSR 95–033 to M.R.L.), and Fundação Botânica Margareth Mee, whom we gratefully acknowledge. We also acknowledge partial financial support for this study from the European Commission, DGXII, International Cooperation with Developing Countries Programme contract number ERBIC18CT970149, and the Pilot Program to Protect the Brazilian Rain Forests (PPG-7, European Union and Brazilian Ministry of Science and Technology). Thanks to José Ribeiro, Jimmy Grogan, Jurandir Reis Galvão, Orlandino Candiotto and Jerônimo M. de Souza for their kind assistance during field work. Our thanks to Jaime Tadeu França (IBAMA) for his help in arranging plant collections in many areas and Rosana P. V. Brondani for technical and laboratory support. We are grateful to Eldredge Bermingham and Chris Dick for valuable comments on the manuscript. We also acknowledge the support of IBAMA (Brazilian Ministry of Environment) and EMBRAPA (Brazilian Ministry for Agriculture). References Aldrich PR, Hamrick JL, Chavarriaga P, Kochert G (1998) Microsatellite analysis of demographic genetic structure in fragmented populations of the tropical tree Symphonia globulifera. Molecular Ecology, 7, 933–944. Alvarez-Buylla ER, Garay AA (1994) Population genetic structure of Cecropia obtusifolia, a tropical pioneer tree species. Evolution, 48, 436–453. Alvarez-Buylla ER, García-Barros R, Lara-Moreno C, MartínezRamos M (1996) Demographic and genetic models in conservation biology: applications and perspectives for tropical rain forest tree species. Annual Review of Ecology and Systematics, 27, 387– 421. Bawa KS (1990) Plant–pollinator interactions in tropical rain forests. Annual Review of Ecology and Systematics, 20, 399–422. 2882 M . R . L E M E S E T A L . Bawa KS (1994) Effects of deforestation and forest fragmentation on genetic diversity in tropical tree populations. In: Proceedings of International Symposium on Genetic Conservation and Production of Tropical Forest Tree Seed (eds Drysdale RM, John SET, Yapa AC), pp. 10–16. ASEAN-Canada Forest Tree Seed Centre Project, Muak-lek, Saraburi, Thailand. Bawa KS, Seidler R (1998) Natural forest management and conservation of biodiversity in tropical forests. Conservation Biology, 12, 46 –55. Brondani RPV, Brondani C, Tarchini R, Grattapaglia D (1998) Development, characterization and mapping of microsatellite markers in Eucalyptus grandis and E. urophylla. Theoretical Applied Genetics, 97, 816– 827. Callen DF, Thompson AD, Shen Y (1993) Incidence and origin of ‘null’ alleles in the (AC)n microsatellite markers. American Journal of Human Genetics, 52, 922– 927. Chase M, Kesseli R, Bawa K (1996) Microsatellite markers for population and conservation genetics of tropical trees. American Journal of Botany, 83, 51–57. Cockerham CC (1969) Variance of gene frequencies. Evolution, 23, 72–84. Collevatti RG, Brondani RV, Grattapaglia D (1999) Development and characterization of microsatellite markers for genetic analysis of a Brazilian endangered tree species Caryocar brasiliense. Heredity, 83, 748–756. Collevatti RG, Grattapaglia D, Hay JD (2001) Population genetic structure of the endangered tropical tree species Caryocar brasiliense, based on variability at microsatellite loci. Molecular Ecology, 10, 349–356. Dayanandan S, Dole J, Bawa K, Kesseli R (1999) Population structure delineated with microsatellite markers in fragmented populations of a tropical tree, Carapa guianensis (Meliaceae). Molecular Ecolology, 10, 1585 –1592. Dick C (2001) Genetic rescue of remnant tropical trees by an alien pollinator. Proceedings of the Royal Society, London B., 268, 2391– 2397. Dick CW, Hamilton MB (1999) Microsatellites from the Amazonian tree Dinizia excelsa (Fabaceae). Molecular Ecology, 8, 1753 – 1768. Doyle JJ, Doyle JL (1987) Isolation of plant DNA from fresh tissue. Focus, 12, 13 –15. Frankie GW, Opler PA, Bawa KS (1976) Foraging behavior of solitary bees: implications for outcrossing of a neotropical forest tree species. Journal of Ecology, 64, 1049 –1057. Gaiotto FA, Brondani RPV, Grattapaglia D (2001) Microsatellite markers for Heart of Palm — Euterpe edulis and E. oleracea Mart. (Arecaceae). Molecular Ecology Notes, 1/2, 86 –88. Gillies ACM, Navarro C, Lowe AJ, Newton AC, Hernández M, Wilson J, Cornelius JP (1999) Genetic diversity in Mesoamerican populations of mahogany (Swietenia macrophylla), assessed using RAPDs. Heredity, 83, 722–732. Goldstein DB, Pollock DD (1997) Launching microsatellites: a review of mutation processes and methods of phylogenetic inference. Journal of Heredity, 88, 335– 342. Goodman SJ (1997) RST CALC: a collection of computer program for calculating estimates of genetic differentiation from microsatellite data and determining their significance. Molecular Ecology, 6, 881–885. Goudet J (2000) FSTAT, a program to estimate and test gene diversities and fixation indices (Version 2.9.1). Available from http://www.unil.ch/izea/softwares/fstat.html. Updated from Goudet (1995). Grogan JE (2001) Bigleaf mahogany (Swietenia macrophylla King) in southeast Pará, Brazil. A life history study with management guidelines for sustained production from natural forests. PhD Thesis, Yale University School of Forestry & Environmental Studies, New Haven, CT, USA. Gullison RE, Panfil SN, Strouse JJ, Hubbell S (1996) Ecology and management of mahogany (Swietenia macrophylla King) in the Chimanes Forest, Beni, Bolivia. Botanical Journal of the Linnean Society, 122, 9–34. Guo SW, Thompson EA (1992) Performing the exact test for Hardy–Weinberg proportions for multiple alleles. Biometrics, 48, 2868–2872. Hall P, Chase MR, Bawa KS (1994) Low genetic variation but high population differentiation in a common tropical forest tree species. Conservation Biology, 8, 471–482. Hall P, Walker S, Bawa KS (1996) Effects of forest fragmentation on diversity and mating systems in a tropical tree Pithecellobium elegans. Conservation Biology, 10, 757–768. Hamrick JL, Loveless MD (1986) Isoenzyme variation in tropical trees: procedures and preliminary results. Biotropica, 18, 201– 207. Hamrick JL, Loveless MD (1989) The genetic structure of tropical tree populations: associations with reproductive biology. In: The Evolutionary Biology of Plants (eds Bock JH, Linhart YB), pp. 129– 146. Westview Press, Boulder, CO. Hedrick PW (1999) Highly variable loci and their interpretation in evolution and conservation. Evolution, 53, 313–318. Jarne P, Lagoda PJL (1996) Microsatellites, from molecules to populations and back. Trends in Ecology and Evolution, 11, 424–429. Kimura M, Crow JF (1964) The number of alleles that can be maintained in a finite population. Genetics, 49, 725–738. Lemes MR (2000) Population genetic structure and mating system of Swietenia macrophylla King (Meliaceae) in the Brazilian Amazon: Implications for conservation. PhD Thesis. University of Stirling, Stirling, UK. Lemes MR, Brondani RPV, Grattapaglia D (2002) Multiplexed systems of microsatellite markers for genetic analysis in mahogany, Swietenia macrophylla King (Meliaceae), a threatened Netropical timber species. Journal of Heredity, 93, 287–290. Lewis PO, Zaykin D (1999) Genetic Data Analysis: Computer program for the analysis of allelic data, Version 1.0 (d12). Free program over the internet GDA Home Page at http:// chee.unm.edu/gda/. Loveless MD (1992) Isozyme variation in tropical trees: patterns of genetic organization. New Forests, 6, 67–94. Loveless MD (1998) Population structure and mating system in Tachigali versicolor, a monocarpic Neotropical tree. Heredity, 81, 134–143. Lynch M, Crease TJ (1990) The analysis of population survey data on DNA sequence variation. Molecular Biology and Evolution, 7, 377– 394. Mantel N (1967) The detection of disease clustering and a generalised regression approach. Cancer Research, 27, 209–220. Maslin MA, Burns SJ (2000) Reconstruction of the Amazon basin effective moisture availability over the past 14,000 years. Science, 290, 2285–2287. Murawski DA, Gunatilleke IAU, Bawa K (1994) The effects of selective logging on inbreeding in Shorea megistophylla (Dipterocarpaceae) from Sri Lanka. Conservation Biology, 8, 997–1002. Nason JD, Hamrick JL (1997) Reproductive and genetic consequences of forest fragmentation: two case studies of Neotropical canopy trees. Journal of Heredity, 88, 264–276. © 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 2875–2883 G E N E T I C S T R U C T U R E O F B R A Z I L I A N A M A Z O N M A H O G A N Y 2883 Nason JD, Aldrich PR, Hamrick JL (1997) Dispersal of genetic structure in fragmented tropical tree populations. In: Tropical Forest Remnants; Ecology, Management, and Conservation of Fragmented Communities (eds Laurence WF, Bierregard R), pp. 304 –320. The University of Chicago Press, Chicago. Nelson BW, Kapos V, Adams JB, Oliveira WJ, Braun OPG, DoAmaral IL (1994) Forest disturbance by large blowdowns in the Brazilian Amazon. Ecology, 75, 853– 858. Novick RR, Dick, C, Lemes, MR, Navarro C, Caccone A, Bermingham E (2003) Genetic struture of Mesoamerican populations of big-leaf mahogany (Swietenia macrophylla) inferred from microsatellite analysis. Molecular Ecology, 12, in press. Ohta T, Kimura M (1973) The model of mutation appropriate to estimate the number of electrophoretically detectable alleles in a genetic population. Genetical Research, 22, 201–204. Paetkau D, Strobeck C (1995) The molecular basis and evolutionary history of a microsatellite null allele in bears. Molecular Ecology, 4, 519–520. Pemberton JM, Slate J, Bancroft DR, Barrett JA (1995) Nonamplifying alleles at microsatellite loci: a caution for parentage studies. Molecular Ecology, 4, 249–252. Raymond M, Rousset F (1998) genepop (Version 3.1c) an updated, Version of genepop V.1.2 (1995): population genetics software for exact tests and ecumenicism. Journal of Heredity, 86, 248– 249. Rodan BD, Newton AC, Verissimo A (1992) Mahogany conservation: status and policy initiatives. Environmental Conservation, 19, 331–338. Slatkin M (1995) A measure of population subdivision based on microsatellite allele frequencies. Genetics, 139, 457–462. Snook LK (1996) Catastrophic disturbance, logging and the ecology of mahogany (Swietenia macrophylla King): grounds for listing a major tropical timber species in CITES. Botanical Journal of the Linnean Society, 122, 35 – 46. Styles BT (1972) The flower biology of the Meliaceae and its bearing on tree breeding. Silvae Genetica, 21, 175–182. Valdes AM, Slatkin M, Freimer NB (1993) Allele frequencies at microsatellite loci: The stepwise mutation model revisited. Genetics, 133, 737–749. Verissimo AC, Barreto P, Tarifa R, Uhl C (1995) Extraction of a © 2003 Blackwell Publishing Ltd, Molecular Ecology, 12, 2875–2883 high-value natural resource in Amazonia: the case of mahogany. Forest Ecology and Management, 72, 39 –60. Weir BS, Cockerham CC (1984) Estimating F-statistics for the analysis of population structure. Evolution, 38, 1358–1370. White G, Powell W (1997) Isolation and characterization of microsatellite loci in Swietenia humilis (Meliaceae): an endangered tropical hardwood species. Molecular Ecology, 6, 851–860. White GM, Boshier DH, Powell W (1999) Genetic variation within a fragmented population of Swietenia humilis Zucc. Molecular Ecology, 11, 1899–1910. White GM, Boshier DH, Powell W (2002) Increased pollen flow counteracts fragmentation in a tropical dry forest: an example from Swietenia humilis Zuccarini. Proceedings of the National Academy of Sciences USA, 99, 2038–2042. Whitmore TC, Prance GT (1987) Biogeography and Quaternary History in Tropical America. Oxford Science Publications, Clarendon Press, Oxford. Whitmore TC (1997) Tropical forest disturbance, disappearance, and species loss. In: Tropical Forest Remnants; Ecology, Management, and Conservation of Fragmented Communities (eds Laurence WF, Bierregaard R), pp. 3–12. The University of Chicago Press, Chicago. Wright S (1943) Isolation by distance. Genetics, 28, 114–138. Young A, Boyle T, Brown T (1996) The population genetic consequences of habitat fragmentation for plants. Trends in Ecology and Evolution, 11, 413–418. This paper is part of the PhD dissertation of Maristerra R. Lemes at the University of Stirling, UK on the population genetics and conservation of mahogany in the Brazilian Amazon. She is a researcher at the Brazilian Institute for Amazon Research (INPA) whose main interests are population genetics using molecular tools for the conservation of tropical trees. Rogério Gribel is an ecologist at INPA working on genetics, reproductive biology and conservation of tropical trees. Professor John Proctor is a plant ecologist at the University of Stirling, UK working for many years in the tropics. Dario Grattapaglia is a geneticist involved in the development and use of molecular markers for population and quantitative genetics and molecular breeding of forest trees.