Biodiversity and Conservation (2006) 15:4357–4373 DOI 10.1007/s10531-005-3741-5
Springer 2006 Genetic variation in three endangered species of Encholirium (Bromeliaceae) from Cadeia do Espinhaço, Brazil, detected using RAPD markers MARCELO MATTOS CAVALLARI1,*, RAFAELA CAMPOSTRINI FORZZA2, ELIZABETH ANN VEASEY1, MARIA IMACULADA ZUCCHI1 and GIANCARLO CONDE XAVIER OLIVEIRA1 1 Departamento de Genética, Escola Superior de Agricultura ‘Luiz de Queiroz’ (ESALQ), Universidade de São Paulo, CP 83, 13400-970 Piracicaba, São Paulo, Brazil; 2Instituto de Pesquisas Jardim Botânico do Rio de Janeiro (JBRJ), Rua Pacheco Leão 915, 22460-030 Rio de Janeiro, Brazil; *Author for correspondence (e-mail: mmcavall@ibb.unesp.br; phone: +55-14-3881-3036; fax: +55-14-3815-3131/+55-14-3811-6147) Received 23 February 2005; accepted in revised form 22 September 2005 Key words: Bromeliaceae, Conservation biology, Encholirium, Population genetic structure, RAPD, Rare plants Abstract. Encholirium is a Brazilian genus of Bromeliaceae that occurs exclusively in rocky landscapes. This work aimed to generate basic information for the conservation of three Encholirium species that are endemic to the rocky mountains of Cadeia do Espinhaço, employing population genetic analyses. E. pedicellatum and E. biflorum have only one very small population each, both occurring in unprotected, private land sites, being critically endangered. E. subsecundum is more widespread, and some of its populations dwell in protected areas. Five Random Amplified Polymorphic DNA (RAPD) primers generated approximately 60 polymorphic bands for each species. This technique demonstrated the presence of a single RAPD profile for every individual sampled (except for one clone found in E. biflorum). High levels of genetic variability were not expected, due to the clonal habit of the plants and small size of the populations. Populations of E. biflorum and E. pedicellatum presented, respectively, 16.06% (Ust = 0.16, p < 0.001) and 8.44% (Ust = 0.08, p < 0.001) of the total genetic diversity attributable to genetic differences among groups within the populations. In E. subsecundum, 14.52% (Ust = 0.15, p < 0.001) of the total genetic diversity was found among populations. Estimates of the Shannon’s Diversity Index provided similar results. These results are valuable for the development of conservation strategies. Introduction Population genetics studies of rare and threatened flora have become imperative for plant conservation (Holsinger and Gottlieb 1991). Small, isolated populations are particularly subject to inbreeding and genetic drift, and their genetic variation is expected to be low in comparison to that of larger populations (Tansley and Brown 2000). Similarly, endemic and rare species typically exhibit low levels of genetic variation (Hamrick and Godt 1989). Loss of 4358 genetic variation is potentially thought to lead to a decrease in the species ability to survive environmental changes and demographic fluctuations (Ellstrand and Elan 1993). Random amplified polymorphic DNA (RAPD) (Welsh and McClelland 1990) has been used to identify clones in vascular plants (Fischer et al. 2000; Kreher et al. 2000; Persson and Gustavsson 2001) and to characterize the genetic diversity of natural plant populations (Lacerda et al. 2001; Allnutt et al. 2003). It is an ideal technique for studying conservation biology and genetic parameters in rare and endangered species, where genetic information and plant material are often scarce (James and Ashburner 1997). Encholirium is an endemic Brazilian genus of Bromeliaceae which comprises a total of 23 species (Forzza 2005). The rocky mountains of Cadeia do Espinhaço (Espinhaço Mountains) in the State of Minas Gerais are the habitat of nine endemic species of Encholirium and are considered to be the diversity center of the genus. These species dwell in the ‘campos rupestres’ (literally, ‘rocky fields’), shrubby savanna habitats that occur at an elevation gradient between 700 and 2000 m. The ‘campos rupestres’ are well-known by their plant species diversity (more than 3000 species) and by their high degree of endemism, estimated to occur in more than 30% of their plant species (Giulietti et al. 1997). Despite its taxonomic diversity, only a few areas of Cadeia do Espinhaço are protected by law. Of the nine Encholirium species endemic of this site, only three can be found in protected areas (E. subsecundum, E. heloisae and E. vogelii) (Forzza et al. 2003). E. pedicellatum and E. biflorum are only found in one very small population of each species, both occurring in privately owned sites. These small populations are composed of less than 100 individuals, being critically endangered. E. subsecundum is a more widespread species in Cadeia do Espinhaço, and some of its populations dwell in protected areas. According to the IUCN categories, E. pedicellatum, E. biflorum and E. subsecundum are considered ‘critically endangered’, ‘endangered’ and ‘vulnerable’, respectively (Forzza et al. 2003). These three species show clonal reproduction and seedling recruitment may be a rare event in natural populations (Forzza personal communication). It is generally assumed that clonally reproducing plant species have a lower level of genetic diversity than nonclonal species, even so as much genetic diversity can be found in clonal as in nonclonal plants (Persson and Gustavsson 2001). Under a strict natural history standpoint, the small population size of these Encholirium species, the clonal habit of the plants and the apparently rare occurrence of sexual reproduction suggest that populations are composed of a few locally adapted clones. The purpose of the present study was to test the hypothesis that the populations are composed of few clones and to partition the genetic diversity among its hierarchical levels. The information generated by population genetic analyses is intended to be useful for the conservation of these endangered Encholirium species. 4359 Materials and methods Sampling strategy In studies describing clonally reproducing plants, a specific definition of individual is desirable. An individual, in this study, consists of visibly and physically connected rosettes (i.e., rosettes connected by stems). A stem segment previously connecting two parts of an individual may rotten and disappear, interrupting the connection between the rosettes, but even in this case we have marked the existence of two distinct individuals, due to the fact that the individual history is not detectable by field inspection. Four populations of E. subsecundum were sampled along the way between the city of Diamantina (4335¢41¢¢ W, 1814¢51¢¢ S), in the Diamantina Plateau, and Serra do Cipó National Park, near the city of Santana do Riacho, (4342¢51¢¢ W, 1910¢07¢¢ S), Minas Gerais State (Figure 1). Since we selected only populations alongside the road, many populations of this species could not be included. We have sampled the only known population of both E. biflorum and E. pedicellatum, near the city of Diamantina. The sampling strategy was influenced by the spatial distribution of the individuals in the populations. In all the populations, but in population 2 of E. subsecundum, they showed an aggregated type of distribution. This means that the populations are formed by a variable number of groups of individuals, each group including a variable number of individuals closely located in space (Figure 2) whose genetic similarity is a priori unknown. The groups are separated from each other by different distance lengths (Figure 2; Table 1). In each population, the groups were identified prior to sampling. One juvenile leaf per individual was randomly collected from nearly half of the individuals of each group. Thus samples were structured hierarchically in species, populations within species (only in E. subsecundum), groups within populations and individuals within groups, so that the distribution of genetic variability could be studied at several levels. Approximately half of the individuals of the population 2 of E. subsecundum were randomly sampled, since their regular distribution did not allow us to identify groups. A total of 31 individuals of E. biflorum distributed in three groups, 31 individuals of E. pedicellatum distributed in five groups and 77 individuals of E. subsecundum distributed in four populations and 10 groups (Figure 2) were sampled. Leaf tissue was dried in sealed plastic bags containing silica gel and stored at 20 C prior to DNA isolation. DNA isolation DNA was isolated using the CTAB extraction method adapted from Doyle and Doyle (1990). Dried plant tissue (0.1 g) was ground to a fine powder in a 1.5 ml microtube using liquid nitrogen. After allowing liquid nitrogen 4360 Figure 1. Map showing the localization of the populations sampled. Circles = E. subsecundum populations; Triangle = E. biflorum and E. pedicellatum populations (source: adapted from http:// www.nmnh.si.edu/botany/projects/cpd/sa/map56.htm). to evaporate, 0.9 ml of CTAB isolation buffer was added (2% CTAB, 8.18% NaCl, 0.5 M EDTA pH 8.0, 1 M Tris–HCl pH 8.0, 5 % 2-b-mecaptoethanol) and the samples were incubated at 65 C for 1 h. The mixture was centrifuged and the upper phase was transferred to a new microtube, shaken with 0.5 ml of chloroform/isoamyl alcohol (24:1) and centrifuged to separate the phases and remove the aqueous top layer. Isopropyl alcohol (0.4 ml) was added to the final aqueous extract and mixed gently to precipitate the DNA. The samples were stored at 20 C for 15 h. The precipitated was centrifuged to pellet the DNA, washed twice in 70% ethanol and once in 95% ethanol. DNA was dissolved in 75 ll TE buffer (Tris–HCl 1 M pH 8.0, EDTA 0.5 M pH 8.0). Samples were treated with 20 mg of RNAse (Ribonuclease A Bovine Pancreas, USBiological) and stored at 20 C. DNA concentration was 4361 Figure 2. Sampling scheme, showing the distances between the groups (gr) or populations (pop) of E. biflorum (bif), E. pedicellatum (ped), and E. subsecundum (sub), and the number of individuals sampled in each group/population (n). Table 1. Average distances between the groups/populations sampled. Average distance among E. E. E. E. E. E. biflorum groups pedicellatum groups subsecundum populations subsecundum pop. 1 subsecundum pop. 3 subsecundum pop. 4 11.3 m 88.0 m 116.6 km 3.06 km 30.0 m 4.16 m determined by comparison to standards on agarose gels and the DNA was diluted in TE buffer in order to be used as 0.2 ng DNA/ll for RAPD reactions. RAPD reactions The concentrations and conditions of the PCR reaction were optimized by means of preliminary assays for random samples to give repeatable markers. As an internal control, 10 individuals and one primer pair were randomly chosen, and four replicated independent reactions were performed in a different thermocycler. As band patterns were identical, the RAPD protocol 4362 was considered adequate. RAPD amplification reactions were performed in a final volume of 25 ll, containing 1 ng of template DNA, 0.5 lM of a 10-base pair random primer (Operon Technologies), 1 U of Taq polymerase (Fermentas Life Sciences), 0.5 mM of each dNTP, 6 mM MgCl2, 750 mM Tris–HCl (pH 8.8) and 200 mM (NH4)2SO4. The reactions were performed in a Primus 96-PLUS (MWG Biotech) thermocycler (internal control reactions were performed in a PTC-100, MJ Research) with a 5-min denaturation step at 94 C followed by 55 cycles of 1-min denaturation at 94 C, 1 min 45 s annealing at 35 C, and 2 min extension at 72 C, and a final 7-min elongation step at 72 C. Amplification products were separated by electrophoresis in 1.4% w/v agarose gels with 1· TBE buffer, stained with ethidium bromide. After a run of approximately 4 h (110 V), the gels were visualized by illumination with ultraviolet light and photographed for analysis. The molecular weight of the fragments was estimated using a molecular marker ladder of 100-bp (Fermentas Life Technologies). Ninety-two 10-mer random primers chosen by chance from the kits OPA, OPB, OPC, OPE, OPG, OPP, OPX and OPAX (Operon Technologies) were evaluated. Sixteen primers generated clear and distinct band patterns. The best five primers were used for the analysis (Table 2). Bootstrap analyses, carried out using the Dboot Software (Coelho 2000), showed that additional polymorphic bands would not affect the variation coefficient (data not shown), suggesting that the utilization of additional primers would have not changed the results. Data analysis Amplification products were scored manually and each band in the RAPD profile was considered an independent locus with two alleles: presence (1) or absence (0) of a band. Thus, a binary matrix based on the polymorphic bands was generated. Each species was analyzed separately, although the same primers and reaction conditions were used for all species. Table 2. Primers used for the genetic analysis, their sequences and the number of bands produced for each species. Primer Sequence (5¢ fi 3¢) E. biflorum E. pedicellatum E. subsecundum OPC-06 OPE-04 OPG-08 OPP-14 OPP-16 Total GAACGGACTC GTGACATGCC TCACGTCCAC CCAGCCGAAC CCAAGCTGCC 17 10 10 13 08 58 15 09 15 10 10 59 16 14 08 15 07 60 (1) (1) (0) (1) (2) (5) (0) (1) (0) (1) (4) (6) The number of monomorphic bands is shown between parentheses. (0) (0) (0) (0) (0) (0) 4363 The software NTSYS (Rohlf 1989) was utilized to generate UPGMA (unweighted pair-group method with arithmetic averages) dendrograms. The stability of the clusters was tested by 10,000 bootstrap re-samplings using the software Bood-P (Coelho 2003). The software NTSYS was also employed to identify the clones. A Jaccard coefficient between all pairs of individuals was generated. Jaccard coefficient is given by the equation Sij = a/a + b + c, where a is the number of bands present in both individuals, b is the number of bands present only in individual i and c is the number of bands present only in individual j. The individual plants were considered clones when the Jaccard coefficient between them was Sij = 1. The Analysis of Molecular Variance (AMOVA; Excoffier et al. 1992) was performed to estimate variance components, partitioning of the variation of each species among populations (applicable only for E.subsecundum), among groups within populations and among individuals. Variance components were inferred from metric distances among RAPD fragment patterns, estimated by E = n[1 2nxy/2n], where nxy is the number of bands shared by the individuals x and y, and n is the total number of polymorphic markers (Huff et al. 1993). Significance levels for variance component estimates were calculated by nonparametric permutation procedures, determining the probability of obtaining a more extreme variance component than the value observed by chance alone (1000 permutations). AMOVA analyses were performed using the software Arlequin 2.0 (Schneider et al. 2000). An additional measure of partitioning of the genetic variation was obtained P by Shannon’s Diversity Index (Lewontin, 1972): H = Pi log2 Pi, where Pi is the frequency of a given RAPD band. The index was calculated for each locus (H0). The average Shannon’s index for each group (H¢group) was then calculated by averaging H0 for all markers within the group. The average H¢group within a species was designated Hgroup. The average of all markers within a population, regardless of the groups, was designated H¢pop and the average H¢pop within a species was designated Hpop. Finally, the average of all markers within a species, regardless of the groups and populations, was designated Hsp. Following the convention outlined by Lewontin (1972), genetic variation can be measured at several levels. The component within-groups was calculated by Hgroup/Hsp; the among-groups-within-populations component was calculated by (Hpop Hgroup)/Hsp; the among-populations (E. subsecundum) component were given by (Hsp Hpop)/Hsp. Results Encholirium biflorum A total of 58 RAPD bands were obtained. Of those, 53 (89.80%) were polymorphic (Table 2). Eight bands were restricted to specific groups (data not 4364 Table 3. AMOVA results for E. biflorum. Source of variation df SSD % of total variance p U-statistics Among groups Within groups Total 2 28 30 28.179 141.434 169.613 16.06% 83.94% (<0.001) (<0.001) 1 UST = 0.1606 UST = 0.8394 shown). Two adjacent individuals in the field presented the same RAPD profile, and the Jaccard coefficient between this individuals was Sij = 1. This observation suggests that these individuals are clones and that a connecting stem segment had rotten and disappeared, being no longer visible. Analysis of molecular variation showed that the vast majority (83.94%) of the genetic variation of E. biflorum is due to differences among individuals within groups (p < 0.001). A significant proportion (16.06%) was due to the differences among groups (p < 0.001) (Table 3). Estimates of Shannon’s Diversity Index also indicate that, on average, most of the diversity (64%) occurs within groups (Table 4), although the among-groups component [(Hpop Hgroup)/Hsp] contributes to a more significant proportion (35%) of the genetic variation according to this index. UPGMA dendrogram did not clustered E. biflorum groups according to their spatial distribution (Figure 3). Encholirium pedicellatum Fifty-nine RAPD bands were obtained for E. pedicellatum. Of those, 53 (91.37%) were polymorphic. The bands were not restricted to specific groups. Each individual presented a unique RAPD profile. Analysis of molecular variance showed that a significant proportion (8.44%) of the variation was due to the differences among groups (p < 0.001), however, most of the variation (91.56%) was due to differences among individuals within the groups (p < 0.001) (Table 5). According to the estimates of Shannon’s Diversity Index, most of the variation (93%) was due to differences among individuals within groups (Table 4). Results of the partition of E. pedicellatum genetic diversity were almost identical when comparing Shannon’s Diversity Index and Table 4. Diversity partitioning into its components using Shannon’s Diversity Index. Hgroup Hpop Hsp Hgroup/Hsp (Hpop Hgroup)/Hsp (Hsp Hpop)/Hsp E. subsecundum E. biflorum E. pedicellatum 0.24 0.33 0.47 0.59 0.20 0.20 0.21 0.32 0.32 0.64 0.35 – 0.29 0.27 0.27 0.93 0.06 – 4365 Figure 3. UPGMA dendrograms. Numbers below the dendrogram are values of Jaccard Coefficient (S). Numbers in percentage represents the stability of the node. 4366 Table 5. AMOVA results for E. pedicellatum. Source of variation df SSD % of total variance p U-statistics Among groups Within groups Total 4 26 30 26.560 110.214 136.774 8.44% 91.56% (<0.001) (<0.001) 1 UST = 0.084 UST = 0.9156 AMOVA. Groups of E. pedicellatum were not clustered in the UPGMA dendrogram according to their spatial distribution (Figure 3). Encholirium subsecundum A total of 60 RAPD bands were obtained, and all of them were polymorphic (Table 2). No fragments were restricted to specific groups or populations, but some of them were absent from some of the populations (data not shown). Each individual presented a unique RAPD profile. Analysis of the molecular variation indicated that the greatest variation percentage (84.38%) was due to the differences among individuals within the groups (p < 0.001). Among-group differences were not statistically significant (p > 0.050) and accounted for 1.10% of the genetic diversity, whereas among-population variation contributed with 14.52% (p < 0.001) (Table 6). Shannon’s Diversity Index indicated that among-group differences accounted for 20% of the total genetic diversity of E. subsecundum, whereas among-population variation contributed with 20% (Table 4). The UPGMA dendrogram groupedE. subsecundum populations and groups according to the geographical distribution, as expected (Figure 3). Discussion The small size of the populations, the clonal habit of the plants and the apparently infrequent occurrence of sexual reproduction provided indications that the populations of Encholirium studied were composed of a few locally Table 6. AMOVA results for E. subsecundum. Source of variation df SSD % of total variance p Among populations 3 71.675 14.52% Among groups within populations 7 41.648 1.10% Within groups 66 362.433 84.38% Total 76 475.156 U-statistics (<0.001) UCT = 0.1452 (>0.05) USC = 0.0128 (<0.001) 1 UST = 0.8438 4367 adapted clones, but this hypothesis was rejected by our analyses. Each individual presented a unique RAPD profile (except for one clone found in the Encholirium biflorum population). The number of RAPD profiles found was compatible to the number of collected samples (except for E. biflorum that presented 30 RAPD profiles from 31 collected samples). Previous studies reported the occurrence of a great number of different RAPD band patterns found in clonal species. Fischer et al. (2000), for example, found 124 RAPD phenotypes in 127 samples of Ranunculus reptans, a clonal plant from gravel lakeshores in the Alps. The high number of different RAPD profiles was unexpected due to the apparently reduced sexual reproduction observed for these species. There are only three collections of E. biflorum deposited in worldwide herbaria (Forzza 2005), since flowers or fruits are rarely observed in this species. Periodic field inspections from 1998 to 2004 yielded only one E. biflorum individual with few fruits and unfertilized old flowers (Forzza personal communication). In E. pedicellatum, we found several individuals with mature fruits, but most of them were severely attacked by borers (Lepidoptera), and the number of viable seeds was extremely reduced. We observed also that few juvenile individuals could be found in the populations of E. subsecundum, although most of the mature individuals presented thousands of apparently viable seeds. Years of observation indicated that seedling recruitment is likely to be a rare event in natural populations of Encholirium species, especially in view of the high seed production observed in some species (Forzza personal communication). The reduced sexual reproduction in clonal species has been reported in other studies. Persson and Gustavsson (2001) reported that no seedling recruitment was observed in established populations of Vaccinium vitis-ideae, a clonal Ericaceae. Esselman et al. (1999) reported the lack of reports of viable seed production in natural populations of the Poaceae Calamagrostis porteri spp. insperata. The source and maintenance of genetic diversity within clonal plant populations have been questioned and reviewed in previous studies. Recruitment of individuals from seed or the occurrence of somatic mutations would add genetic diversity. Given the possibly long life of clonal plants, even a modest degree of sexual reproduction would produce an amount of genetic diversity accumulated for many years (Esselman et al. 1999), and a reasonable amount of the genetic variability that was available within a population from its establishment may be maintained even in the absence of novel genotype recruitment (Kreher et al. 2000). Moreover, given the clonal habit of the plants, the effects of genetic drift are less pronounced than in exclusively sexual species, contributing to the maintenance of several genotypes in the population (Ayres and Ryan 1999; Fischer et al. 2000). The analysis of molecular variance indicated that each species has a different pattern of distribution of the genetic diversity. Of the total variation detected in E. biflorum, 16.06% was due to differences among the groups. The same hierarchical level in E. pedicellatum accounted for half of the fore mentioned value (8.44%). Note that the average distance among groups of E. biflorum is 4368 11.33 m, whereas the groups of E. pedicellatum are located 88 m apart on the average (Table 1). The results suggest that gene flow among E. pedicellatum groups is more intense than among E. biflorum groups, despite the larger distances between E. pedicellatum groups. Shannon’s Diversity Index indicated that 35% of the total variance detected in E. biflorum was due to the differences among the groups, whereas the same hierarchical level in E. pedicellatum accounted only for 6%. Although these values are different from those of the AMOVA, the same observations about the gene flow among groups can be made: it is less intense among E. biflorum groups than among E. pedicellatum groups. Limited seed and pollen dispersal are determinants of the spatial genetic structure within populations (Chung et al. 1999). Features of E. biflorum reproduction are unknown, since a few fertile individuals have been found in years of observation, as mentioned above. It is possible that flowering is irregular, with few individuals at anthesis at the same time, which would contribute to the limited gene flow. In E. pedicellatum, however, at sampling moment, several individuals presented mature seeds, suggesting the occurrence of flowering synchrony among individuals from different groups. As the putative pollinators (hummingbirds) can easily cover the distances between the groups, the pollen flow would diminish the genetic divergences among the groups. The ability of the pollinators to easily cover the distances between the groups might explain the UPGMA dendrogram obtained for E. pedicellatum, where the groups were not clustered according to their spatial distribution. The same was observed in E. biflorum. However, since the distances between E. biflorum groups are very small (11.3 m on average) any type of clustering could be observed regardless of the spatial distribution of the groups. Although the dendrogram of E. subsecundum groups revealed clusters according to their spatial distances, any probable clustering could have been obtained due to the small distances among the groups within the populations. Nevertheless, in Population 1, where the groups are 3 km apart on average, group clustering according to the spatial distribution was expected. Seed dispersal is very low in Encholirium species. Seeds reach the soil very close to the mother-plant, unless when they are dispersed by the runoff water. We observed that many individuals of E. pedicellatum are growing in flat depressions where water accumulates before dry out. This observation suggests that some groups may have been founded by seeds carried by water from another group situated ‘upstream’. This behavior would result in closely related individuals within groups and among groups. In the E. biflorum population, located in flat terrains, the seeds can only germinate next to the mother-plants, resulting in more genetically differentiated groups. The limited seed dispersal is likely to be the major feature determining the observed genetic structure of the populations of E. biflorum and E. pedicellatum. The analysis of molecular variance of E. subsecundum revealed that the genetic differences among groups accounted only for 1.1% of the total variation (p > 0.05). Results of Shannon’s Diversity Index, however, revealed that a 4369 significant proportion of the total genetic diversity is attributable to differences among groups (20%). The percentage of the total variation due to differences among populations of E. subsecundum (14.52% according to AMOVA and 20% according to the Shannon’s Diversity Index) that are 116.6 km apart, on average, is similar to the values observed for the differences among the groups of E. biflorum (only 11.6 m apart, on average). However, a great part of the gene flow in E. subsecundum may occur among populations that were not sampled. The high number of individuals with mature seeds, and the fact that this species blooms all year round, concentrating flowering between March and December, may have facilitate the gene flow among populations. The synchrony of flowering among populations of other species of Encholirium (E. vogelii and E. heloisae), was reported by Christianini et al. in preparation. E. subsecundum is apparently an outcrossing species, as it shows protogyny (Sazima et al. 1989). Outcrossing plant species usually have between 10 and 20% of the total variance among populations (Hamrick and Godt 1989). The results of AMOVA (14.52%) and of Shannon’s Diversity Index (20%) in E. subsecundum were, therefore, coherent. Many previous studies in outcrossing species revealed similar results (e.g. Huff et al. 1993; Bekessy et al. 2002). In E. pedicellatum and E. biflorum, similar results were found at the amonggroups level. The Hsp value obtained from estimates of Shannon Diversity Index was greater in the widespread E. subsecundum than in the narrowly restricted E. pedicellatum and E. biflorum (Table 4). One possible reason for this result is that geographically restricted species tends to have smaller effective population size than their widespread relatives (Maki and Horie 1999). Implications for conservation Before this study, these species, particularly E. biflorum and E. pedicellatum, were thought to have a small amount of genetic variability. However, the genetic analyses showed that every individual has a unique RAPD profile. AMOVA indicated that the genetic differences among the individuals contributed for the vast majority of the total variance (higher than 80% in all three species). If we assume RAPD variation to be representative of total genetic variation, this result implies that no individual should be taken from the natural populations, even for ex situ conservation, since this would represent the loss of a great portion of the genetic diversity. Germplasm banks, if considered, must be based on seeds. The impact of seed harvest may be very low, since a few seeds germinate in the field. On the other hand, studies have indicated that germination rates of Encholirium seeds in the laboratory are higher than 70%. Seed dehydration and freezing have been shown to promote an increase in the germination rates (Erika Borges, Rio de Janeiro State University, UERJ, personal communication). In the case of 4370 E. biflorum, where seeds are difficult to obtain, it is possible to collect rosettes from the largest individuals. Nevertheless, most of the rosettes from the same individual must be left in the native populations, in order to avoid serious damage to both the individual and the gene pool. Individuals without clonal shooting should not be removed unless the population they belong to is under imminent threat of destruction. Despite the possibility of ex situ conservation, in situ conservation must be considered as a priority. The conservation of the unique habitat of E. biflorum and E. pedicellatum is essential for the survival of these species. More extensive studies must be carried out to delineate areas protected by law that effectively help to preserve these species. These studies are beyond the immediate goals of this paper, but the following considerations must be made. The endemism level of an ecosystem is an important criteria to the definition of areas potentially effective in biodiversity conservation (Gentry 1986). The Cadeia do Espinhaço is well-known as a center of endemism for plants (Giulietti et al. 1997). For instance, among the 538 endangered species of plants in the state of Minas Gerais, 358 (66.5%) are from the ecosystem ‘campos rupestres’ (Costa et al. 1998). Therefore, ‘campos rupestres’ from Cadeia do Espinhaço are strong candidates to legal protection. However, only a few sites are protected by law, which leaves most of these ecosystems vulnerable (Giulietti et al. 1997). Cadeia do Espinhaço comprise numerous low height mountains or ‘serras’ (900–2000 m high), interrupted by extensive and deep river valleys. It encompasses 50–100 km wide mountains in the states of Bahia and Minas Gerais, between latitudes 1000¢–2035¢ S and longitudes 4010¢–4430¢ W, in an area of approximately 6000–7000 km2. The ‘campos rupestres’, which occur at higher slopes (1000–2000 m high, or less in some areas), are composed of a continuous herbaceous stratum. However, instead of being a homogeneous vegetation type, these ‘campos’ are an assembly of communities forming a rich mosaic, under control of the local topography, the nature of the substrate, and the microclimate (Giulietti et al. 1997). There are two national parks at Espinhaço Mountains: Chapada Diamantina National Park (Bahia), which comprises 1520 km2 and Serra do Cipó National Park, in the state of Minas Gerais (338 km2). Additionally, there are several other small conservation areas. However, the infrastructure of the parks is not sufficient for the effective protection of the ecosystem (Giulietti et al. 1997). Considering the high endemism of the flora and fauna, these areas consist only of a small portion of the areas that need to be protected in the region (Giulietti et al. 1997). Areas in Minas Gerais, such as on Diamantina Plateau in the central part of the mountains (where Encholirium species occur), are one of the centers of extremely high diversity that still remains unprotected. Costa et al. (1998) suggested that Cadeia do Espinhaço must be protected as a whole unit. Extensive reservation areas are usually assumed to be more efficient to preserve biomes. However, smaller and non-continuous areas must be considered in this case. As each ‘serra’ harbors several different plant and 4371 animal species, every little piece of land that come to be protected will contain a large amount of biodiversity. Such an example is found in Encholirium species. Encholirium biflorum and Encholirium pedicellatum populations are separated by 3 km. Another endemic species, E. magalhaesii, is found in the same region. Thus, the protection of a not very extensive area would efficiently help in the conservation of three non-protected endangered species. As a concluding remark, the genetic structure of rare species must be analyzed prior to the formulation of ex situ conservation strategies. Sample (leaf) collection did not cause serious damage to the individuals nor the species, and the results are very useful. Our unexpected results showed that every individual is genetically different from the others. As they are very few, every individual represents a significant part of the total variation of the species. Collecting individuals for germplasm banks, particularly from the single-population species, would be a disastrous strategy. Acknowledgements The authors would like to thank World Wildlife Foundation (WWF) and the Ford Foundation for part of the financial support; CNPq (National Research Council of Brazil) for a fellowship granted to the first author; Rio de Janeiro Botanical Garden, Rio de Janeiro, Brazil; Alexandre Sebbenn, from Instituto Florestal do Estado de São Paulo, Piracicaba, SP, Brazil; Roland Vencovsky, from the Department of Genetics, ESALQ/University of São Paulo. References Allnutt T.R., Newton A.C., Premoli A. and Lara A. 2003. Genetic variation in the threatened South American conifer Pilgerodendron uviferum (Cupressaceae), detected using RAPD markers. Biol. Conserv. 114: 245–253. Ayres D.R. and Ryan F.J. 1999. Genetic diversity and structure of the narrow endemic Wyethia reticulata and its congener W. bolanderi (Asteraceae) using RAPD and allozyme techniques. Am. J. Bot. 86: 344–353. Bekessy S.A., Allnutt T.R., Premoli A.C., Lara A., Ennos R.A., Burgman M.A., Cortes M. and Newton A.C. 2002. Genetic variation in the vulnerable and endemic Monkey Puzzle tree, detected using RAPDs. Heredity 88: 243–249. Chung M.G., Chung J.M. and Epperson B.K. 1999. Spatial genetic structure of allozyme polymorphisms within populations of Rhus trichocarpa (Anacardiaceae). Silvae Genetica 48: 223– 227. Coelho A.S.G. 2000. Dboot: avaliação dos erros associados a estimativas de distâncias/ similaridades genéticas através do procedimento de bootstrap com número variável de marcadores (software).[Dboot: Assessment of Errors Associated to Estimates of Genetic Distances/Similarities by Bootstrapping with a Variable Number of Markers (software)]. Laboratório de Genética Vegetal, Instituto de Ciências Biológicas, Universidade Federal de Goiás, Goiânia, Brazil. 4372 Coelho A.S.G. 2003. Bood-P: avaliação de dendrogramas baseados em estimativas de distâncias/ similaridades genéticas através do procedimento de bootstrap (software). [Bood-P: Assessment of Dendrograms based on Estimates of Genetic Distances/Similarities by Bootstrapping (software)]. Laboratório de Genética Vegetal, Instituto de Ciências Biológicas, Universidade Federal de Goiás, Goiânia. Costa C.M.R., Hermann G., Martins C.S., Lins L.V. and Lamas I.R. 1998. Biodiversidade em Minas Gerais: um Atlas para sua conservação. Fundação Biodiversitas, Belo Horizonte. Doyle J.J. and Doyle J.L. 1990. Isolation of plant DNA from fresh tissue. Focus 12: 13–15. Ellstrand N.C. and Elam D.R. 1993. Population genetic consequences of small population size: implications for plant conservation. Annu. Rev. Ecol. Syst. 24: 217–242. Esselman E.J., Jianqiang L., Crawford D.J., Winduss J.L. and Wolfe A.D. 1999. Clonal diversity in the rare Calamagrostis porteri spp insperata (Poaceae): comparative results for allozymes and random amplified polymorphic DNA (RAPD) and intersimple sequence repeat (ISSR) markers. Mol. Ecol. 8: 443–451. Excoffier L., Smouse P.E. and Quattro J.M. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes – application to human mitochondrial – DNA restriction data. Genetics 131: 479–491. Fischer M., Husi R., Prati D., Peintinger M., van Kleunen M. and Schmid B. 2000. RAPD variation among and within small and large populations of the rare clonal plant Ranunculus reptans (Ranunculaceae). Am. J. Bot. 87: 1128–1137. Forzza R.C. 2005. Revisão taxonômica de Encholirium Mart. ex. Schult. & Schult. f. (Pitcairnioideae – Bromeliaceae). Boletim de Botânica da Universidade de São Paulo 23: 1–60. Forzza R.C., Christianini A.V. and Wanderley M.G.L. 2003. Encholirium (Pitcairnioideae – Bromeliaceae): conhecimento atual e sugestões para conservação. Vidalia 1: 7–20. Gentry A.H. 1986. Endemism in tropical versus temperate plant communities. In: Soulé M.E. (ed.), Conservation Biology – The Science of Scarcity and Diversity. Sinauer Associates, Sunderland Massachusetts, pp. 153–181. Giulietti A.M., Pirani J.R. and Harley R.M. 1997. Espinhaço range region. In: Davis S.D., Heywood V.H., Herrera-Macbryde O., Villa-Lobos J. and Hamilton A. (eds), Centres of Plant Diversity: A Guide and Strategy for Their Conservation. Volume 3: The Americas. IUCN Publications Unit, Cambridge, pp. 397–404. Hamrick J.L. and Godt J.W. 1989. Allozyme diversity in plant species. In: Brown A.H.D., Clegg M.T. and Kahler A.L. (eds), Plant Population Genetics, Breeding, and Genetic Resources. Sinauer Associates, Sunderland, Massachusetts, pp. 43–63. Holsinger K.E. and Gottlieb L.D. 1991. Conservation of rare and endangered plants: principles and prospects. In: Falk D.A. and Holsinger K.E. (eds), Genetics and Conservation of Rare Plants. Oxford, Oxford University Press, pp. 195–223. Huff D.R., Peakall R. and Smouse P.E. 1993. RAPD variation within and among natural populations of outcrossing buffalograss [Buchloe dactyloides (Nutt.) Engelm.]. Theor. Appl. Genet. 86: 927–934. James E.A. and Ashburner G.R. 1997. Intraspecific variation in Astelia australiana (Liliaceae) and implications for the conservation of this Australian species. Biol. Conserv. 82: 253–261. Kreher S.A., Foré S.A. and Collins B.S. 2000. Genetic variation within and among patches of the clonal species Vaccinium stamineum L. Mol. Ecol. 9: 1247–1252. Lacerda D.R., Acedo M.D.P., Lemos-Filho J.P. and Lovato M.B. 2001. Genetic diversity and structure of natural populations of Plathymenia reticulata (Mimosoideae), a tropical tree from Brazilian Cerrado. Mol. Ecol. 10: 1143–1152. Lewontin R.C. 1972. The Apportionment of Human Diversity. Evol. Biol. 6: 381–494. Maki M. and Horie S. 1999. Random amplified polymorphic DNA (RAPD) markers reveal less genetic variation in the endangered plant Cerastium fischerianum var. molle than in the widespread conspecific C. fischerianum var. fischerianum (Caryophyllaceae). Mol. Ecol. 8: 145–150. 4373 Persson H.A. and Gustavsson B.A. 2001. The extent of clonality and genetic diversity in lingonberry (Vaccinium vitis- idaea L.) revealed by RAPDs and leaf-shape analysis. Mol. Ecol. 10: 1385–1397. Rohlf F.J. 1989. NTSYS-Pc: Numerical Taxonomy and Multivariate Analysis System. Exerter Publisher, New York, 210pp. Sazima I., Vogel S. and Sazima M. 1989. Bat pollination of Encholirium glaziovii, a terrestrial bromeliad. Plant Syst. Evol. 168: 167–179. Schneider S., Roessli D. and Excoffier L. 2000. Arlequin ver. 2000: A Software for Population Data Analysis (software). Genetic and Biometry Laboratory, University of Geneva, Geneva. Tansley S.A. and Brown C.R. 2000. RAPD variation in the rare and endangered Leucadendron elimense (Proteaceae): implications for their conservation. Biol. Conserv. 95: 39–48. Welsh J. and McClelland M. 1990. Fingerprinting genomes using PCR with arbitrary primers. Nucleic Acids Res. 18: 303–306.