Abstract

Madagascar has a highly distinctive flora and is one of the world biodiversity hot spots. There are more than 170 species of palms, the majority of which are vulnerable, endangered or critically endangered endemics. Palms are utilized for many human uses, many of which lead to plant death or seed harvesting. Combined with reduced populations resulting from extensive forest clearing, those species which are harvested from the wild are under additional threat of extinction. Species recovery programmes have the potential to save some of the most iconic species before it is too late. This study documented the current known populations of the critically endangered palm Beccariophoenix madagascariensis, a species utilized for both local and international purposes. The study specifically investigated the genetic diversity and inbreeding within populations and the potential differentiation between populations and with the newly described species B. alfredii. We found that despite critically small population sizes there was considerable genetic diversity within populations. We also found that ecologically and or geographically distinct populations were genetically distinct. Populations within 3 km of each other exhibited considerable gene flow, probably owing to seed dispersal. The populations were inbred but reproductive viability had been maintained. Conservation and recovery options are discussed.

INTRODUCTION

Madagascar has a highly diverse and distinctive flora and fauna and is recognized as one of the most ecologically rich countries in the world with over 80% of its flora being endemic (Gautier & Goodman, 2003). It is the fourth largest island in the world and one is of the world biodiversity hotspots (Meyers et al., 2000). There are more than 170 species of palms in 16 genera, nearly all of which are Madagascan endemics (97%; Dransfield & Beentje, 1995). This number is nearly three times the number of species of palm on the whole of the African continent. The majority (116) of Madagascan palms are classified as vulnerable, endangered or critically endangered (Dransfield & Beentje, 1995).

Four major human-induced processes are responsible for the present biodiversity crisis; over-exploitation of species; environmental deterioration leading to decreased habitat quality; introduction of exotic species and fragmentation of habitats (Turner, 1996; Honnay et al., 2005). Palms in Madagascar are used for a large number of applications which make them an important resource but can also endanger them (Byg & Balslev, 2001). They are especially important for subsistence farmers in the palm-rich eastern escarpment (Byg & Balslev, 2001). Many species have been highly prized for horticultural use internationally and some of these are reduced to a few individuals in the wild (Dransfield & Beentje, 1995).

The landscape in Madagascar has been greatly modified owing to human activities and much of the forested landscape is now highly fragmented (Simsik, 2004; McConnell & Sweeney, 2005). This has resulted in many species becoming much less abundant or having a reduced geographical extent, and many species are now threatened with extinction (Meyers et al., 2000). Habitat fragmentation can be seen as having three major components; pure loss of habitat; reduction in patch size; and increasing isolation of patches (Honnay et al., 2005). Whereas, in practice broadscale conservation of target areas has been the most cost-effective approach in Madagascar (McConnell & Sweeney, 2005), restoration of forest fragments is now becoming a target for conservation efforts (Pareliussen et al., 2006).

Tropical and subtropical rainforests have high species diversity and contain a large proportion of rare plant species (Tracey, 1981; Hubbell, 2001). Communities are typically characterized by a small number of common species and a much larger number of rare species (Hubbell, 2001; Murray & Lepschi, 2004). Rare species are the first to be lost from fragmented patches, and Hill and Curran (2003) found that tropical forest patch area, shape and isolation affected diversity and composition of fragments; in particular, they found that larger patches contained both more species, and more rare tree species. Thus, to maintain the diversity in humid tropical forests (rainforest) we need to maintain the diversity of rare species.

Because of the great diversity of endemic species and the relatively small number of resident scientists and naturalists, much is still unknown about distributions and ecology of individual species despite considerable interest from the international community (Dransfield & Beentje, 1995; Gautier & Goodman, 2003). The inaccessibility of many areas has meant that some areas have been poorly sampled in botanical surveys. Palms have been poorly sampled in the past and new species are still being discovered and known ranges of species extended with increased survey effort (Rakotoarinivo, Ranarivelo & Dransfield, 2007). This limits the effectiveness and sophistication of conservation programmes but also promises that new populations of some endangered species are likely to be found with increased search effort.

Conservation of individual species, however, requires a species-level approach with an understanding of population biology and the levels and partitioning of genetic variation across its geographical range (Coates & Hopper, 2000). Rare species are predicted to have lower reproductive success, lower genetic diversity and higher inbreeding compared with more abundant and more widespread species (Frankham, 1996). The remaining populations are more spatially isolated, reducing the potential for population demographic or genetic replenishment from outside the population (Hanski & Gilpin, 1997). These populations are now more sensitive to chance effects leading to demographic and or genetic decline. Therefore, it is expected that rare species in small, isolated populations may lose genetic diversity through drift and become less fit owing to increased inbreeding (Ellstrand & Elam, 1993; Byer & Waller, 1999). While increased spatial isolation is predicted to lead to erosion of variation within populations, it is also expected to lead to increased genetic differentiation among populations (Young & Clarke, 2000). However, studies of a rainforest palm Carpentaria acuminata showed that diversity was maintained in small populations within an archipelago of fragments by dispersal of seeds by mobile frugivores (Shapcott, 2000).

Thus, in addition to reduction of immediate population loss, species recovery programs may be aimed at increasing or maintaining the demographic population size of existing populations, increasing the genetic diversity of populations, creating or reintroducing new populations to increase the number and decrease the isolation of existing populations or enhancing connectivity among populations. An understanding of existing genetic diversity and differentiation among populations is thus a significant component of recovery programmes to ensure success. Unfortunately Madagascar has too many highly iconic species that are close to extinction in the wild (Britt, Clubbe & Ranarivelo, 2004). This project was part of a larger set of projects aimed at investigating the viability of some of the most threatened and iconic species of horticultural interest (Britt et al., 2004). This study specifically investigated the critically endangered (IUCN, 2004) palm Beccariophoenix madagascariensis to enable the establishment of a species recovery programme. This species was known to display morphological variation across its range, and thus an understanding of genetic differentiation within the species is important for developing conservation and recovery guidelines for this species.

Study species

Beccariophoenix madagascariensis is a large majestic palm growing to approximately 12 m tall with a trunk up to 30 cm diameter; the genus is endemic to Madagascar and belongs to subfamily Arecoideae, tribe Cocoseae, subtribe Attaleinae (Dransfield et al., 2005). It is classified as critically endangered (IUCN, 2004) with only a few known populations each consisting of a few mature individuals (Table 1). The species also has considerable cultural value with many forms of utilization by local people and is locally known as maroala (Dransfield & Beentje, 1995). Uses include harvest of ‘heart of palm’ for human consumption, large stems for house construction, fronds for weaving, e.g. in the making of ‘manarano’ hats (Dransfield & Beentje, 1995), and recently for making fish traps (Rakotoarinivo, 2005). However, as this species is single-stemmed, harvesting for some uses results in plant death. In addition, since its rediscovery in 1986 (Dransfield, 1988), the species has been highly prized for cultivation outside Madagascar; especially sought after are specimens that have a distinctive leaf form in the juvenile plants in which the leaf apex is composed of many folds, marginally split into short lobes, and basally split to produce ‘windows’. Such incomplete splitting of the apical folds is present in all populations of the genus, but most display only two or three ‘windows’ and the apex of the leaf is not nearly so distinctive and attractive. The form with broad apical leaf-lobes is only known from populations in one area (Table 1). The trade in palm seeds of B. madagascariensis has led to the felling of adult trees at some sites; its subsequent placement on CITES Schedule 2 (on which trade in seed is not prohibited) has had little if any effect on the trade of this species.

Table 1.

Summary of populations of Beccariophoenix, their sizes, number of samples collected, distance to nearest known population; environmental parameters including elevation, climate zone, vegetation type and rock type are given

Population code Location Population size
 
No. of samples Distance to nearest population Elevation (m a.s.l.) Climate zone Vegetation type Rock type 
Adults Total 
B. madagascariensis 
BM7756 Mantadia 1  16 3.2 km 1081 Humid Escarpment forest Basement rocks (Ign and Met) 
BM7758 Mantadia 2  12 3.2 km 1064 Humid Escarpment forest Basement rocks (Ign and Met) 
BM1009 Ranomafana Est 1w  20 km 54 Perhumid Escarpment forest Basement rocks (Ign and Met) Lateritic soils 
BM0168 Ranomafana Est 2w 20 km  Perhumid Escarpment forest Basement rocks (Ign and Met) Lateritic soils 
BM7765 Tolagnaro (St Luce 1) 40 11 0.5 km 10 Humid Littoral forest White sands 
 St Luce 2 90 0.5 km 21 Humid Littoral forest White sands 
 St Luce 3 39 2.5 km Humid Littoral forest White sands 
 St Luce 4 30 2.5 km 33 Humid Littoral forest White sands 
BM7768 Manentenina  63 km 28 Humid Littoral forest Basement rocks (Ign and Met) 
 Vondrozo 100   695 Humid Lowland humid forest White sands 
Total  125  44      
B. alfredii 
BA0136 Antsirabe1 1500 3000 1 km 1259 Subhumid with mist Gallery forest Quartzites and alluvial 
BA1136 Antsirabe2   14 1 km  Subhumid with mist Gallery forest Quartzites and alluvial 
Population code Location Population size
 
No. of samples Distance to nearest population Elevation (m a.s.l.) Climate zone Vegetation type Rock type 
Adults Total 
B. madagascariensis 
BM7756 Mantadia 1  16 3.2 km 1081 Humid Escarpment forest Basement rocks (Ign and Met) 
BM7758 Mantadia 2  12 3.2 km 1064 Humid Escarpment forest Basement rocks (Ign and Met) 
BM1009 Ranomafana Est 1w  20 km 54 Perhumid Escarpment forest Basement rocks (Ign and Met) Lateritic soils 
BM0168 Ranomafana Est 2w 20 km  Perhumid Escarpment forest Basement rocks (Ign and Met) Lateritic soils 
BM7765 Tolagnaro (St Luce 1) 40 11 0.5 km 10 Humid Littoral forest White sands 
 St Luce 2 90 0.5 km 21 Humid Littoral forest White sands 
 St Luce 3 39 2.5 km Humid Littoral forest White sands 
 St Luce 4 30 2.5 km 33 Humid Littoral forest White sands 
BM7768 Manentenina  63 km 28 Humid Littoral forest Basement rocks (Ign and Met) 
 Vondrozo 100   695 Humid Lowland humid forest White sands 
Total  125  44      
B. alfredii 
BA0136 Antsirabe1 1500 3000 1 km 1259 Subhumid with mist Gallery forest Quartzites and alluvial 
BA1136 Antsirabe2   14 1 km  Subhumid with mist Gallery forest Quartzites and alluvial 
w

Populations containing the ‘window’ form are indicated.

The inflorescence may be 50 cm long and has many rachillae; flowers occur as triads of a central female and two lateral males in the basal part of the rachillae and paired or solitary males towards the tips, with more than a thousand flowers on the inflorescence (Rakotoarinivo, 2005). Male flowers open first, are orange and produce a strong scent that attracts many insects; later the female flowers are receptive as the male flowers start to degenerate (Rakotoarinivo, 2005). Pollinators are not yet known but bees, beetles and ants were observed visiting flowers; the timing of beetle visitors indicating they were the most likely pollinator (Rakotoarinivo, 2005). Fleshy fruit are approx 2.5 cm in diameter and 180 fruits per infructescence have been recorded (Rakotoarinivo, 2005). Observations suggest that most seed is dispersed within 5 m of the parent plant (Rakotoarinivo, 2005). However, rare long distance dispersal seems possible as the fruits fall within the food size and colour preferences of potential local frugivores including lemurs, flying foxes and pigeons (Dew & Wright, 1998; Birkinshaw, 2001).

Beccariophoenix madagascariensis grows in evergreen humid forest types (rainforest) in low nutrient soils in two disjunct areas within Madagascar (Fig. 1, Table 1). The southern populations occur in littoral forest in the St Luce area; this has been identified as one of the most threatened ecosystem types in Madagascar (Bollen & Donati, 2005; Du Puy & Moat, 1996; Ramanamanjato & Ganzhorn, 2001). The northern populations occur at higher elevation, in evergreen, mountain ridge top forest, whereas the ‘window form’ is found at lower elevation in the perhumid climate zone on lateritic soils (Dransfield & Beentje, 1995; Table 1).

Figure 1.

Map of Madagascar indicating the locations of all known Beccariophoenix populations and some major or significant place names and or locations. ●, locations sampled in this study; ○, previously documented population location; ▪, newly discovered population not sampled. The population codes for the sampled locations are indicated on the map. Allelic frequencies at the CAC2 microsatellite locus (8 alleles) in Beccariophoenix populations are shown in pie charts indicating their relative spatial locations. Species and population codes are given for each pie chart displayed. Alleles are represented by differing shading as indicated in the legend.

Figure 1.

Map of Madagascar indicating the locations of all known Beccariophoenix populations and some major or significant place names and or locations. ●, locations sampled in this study; ○, previously documented population location; ▪, newly discovered population not sampled. The population codes for the sampled locations are indicated on the map. Allelic frequencies at the CAC2 microsatellite locus (8 alleles) in Beccariophoenix populations are shown in pie charts indicating their relative spatial locations. Species and population codes are given for each pie chart displayed. Alleles are represented by differing shading as indicated in the legend.

Beccariophoenix was thought to be a monotypic endemic genus until recently when a second species, B. alfredii, was discovered (Rakotoarinivo et al., 2007) in a highly contrasting environment which is considerably drier (Fig. 1; Table 1). Beccariophoenix alfredii is found in deep, protected, alluvial gullies in an otherwise fire-maintained grassland, where it often forms the dominant species in gallery forest (M. Rakotoarinivo, pers. observ.).

Aims

The study aimed to test the hypotheses that the morphological variation observed in the species was also reflected in genetic distinctiveness. Specific aims were to find genetic evidence to support the new species, B alfredii, as a distinct genetic and taxonomic entity and to determine whether the Ranamofana ‘open window’ form was a distinct genetic entity or, alternatively, if the genetic evidence indicated a variable species complex. The study aimed to identify whether or not the major disjunction between populations in the north and south was reflected in genetic distinctiveness and, if so, what the conservation and recovery implications would be. Given the small population sizes, the study aimed to investigate how much genetic diversity was present in the remaining populations of the species, and if they have the potential for longer term genetic viability, and how the genetic variation is partitioned within the genus and within the species B. madagascariensis. We were also interested to investigate whether or not the populations were inbred and what the conservation and recovery implications were based on those findings.

METHODS

Field methods

Beccariophoenix was sampled from five locations representing the entire known distribution of the genus at the time of sampling (Fig. 1). At each site the entire adult population was sampled as well as a proportion of the seedling population. Because several populations were extremely small, at some sites every individual was sampled (Table 1). Some additional samples were utilized that had been made independently as part of DNA collection material for gene banking from older collections; however, these usually consisted of a single specimen. The new B. alfredii population was sampled opportunistically; a single adult originally sampled as a DNA voucher (BA0136) and used for this study and subsequent seedlings were obtained from a nursery collection of the same species presumed to originate from approximately the same location (Table 1; BA1136). These last samples were used to test for potential taxonomic differentiation within the genus and cross-checked with representative samples previously analysed. Locations of sample sites were recorded and voucher specimens or photographic vouchers were lodged at RBG Kew for new populations. In the southern region one site was sampled near St Luce; however, three additional nearby subpopulations were identified after genetic sampling was completed and were subject to detailed demographic and phenological studies (Rakotoarinivo, 2005; Table 1). Additional site data were obtained from GIS databases (RBG Kew), field observations and other reference material. Samples for genetic analysis consisted of mature leaf material that was cut from the frond into smaller pieces, placed in individually labelled zip-locked plastic bags containing silica gel (Chase & Hills, 1991), and stored at ambient temperatures out of direct sunlight until the DNA was extracted.

Laboratory methods

Total genomic DNA was extracted from 0.1 to 0.4 g samples of silica-dried material using a modified 2 × CTAB procedure (Doyle & Doyle, 1987). Extracted DNA was then purified using QIAquick silica columns using the manufacturer's protocol. The purified total DNA was quantified using an Eppendorf biophotometer, using AFLPs and microsatellites as genetic markers.

AFLPs were conducted according to the AFLP plant mapping protocol of Applied Biosystems Inc. The quantity of each sample required to obtain 500 ng of DNA was determined. This volume was then dried in a vacuum oven at 60 °C and redissolved in sterile distilled water. This was then digested with restriction enzymes in restriction–ligation reactions (R–L reactions) using EcoRI (a relatively rare six-base cutter) and MseI (a frequent four-base cutter). Adaptors with a 5′-tail were ligated to the fragments using a ligase enzyme. Preselective PCR was undertaken using an EcoR1 primer based on the EcoR1 adaptor and an Mse1 adaptor based on the Mse1 primer but with an additional 3′ base. The product from this PCR was used in a selective PCR reaction in which the EcoR1 primer had a 5′ fluorescent label and two extra bases and the Mse1 primer had a further two 3′ bases. Twelve combinations of selective bases of EcoRI and MseI primers with three selective bases were tested. Three combinations of selective bases (ACT–CTC, AGG–CTA, AAC–CAC) yielded good results in the primer trial and were therefore used in the final analysis. EcoRI primers were labelled with fluorescent dyes. Electrophoresis of amplification reaction products to separate bands was undertaken on an ABI 3100 Genetic Analyser. Data were extracted from the Genescan files using 3100 DATA COLLECTION v.1.1 and ABI GENOTYPER v.3.7 (2.0) was used to score the fragments. Amplified fragments from 50 to 400 base pairs were scored as either present (1) or absent (0). All band assignment and scoring was cross checked between the computer program interpretation and the AFLP traces.

Four microsatellite loci previously developed for Cocos nucifera by Perera et al. (1999) were tested for cross amplification in Beccariophoenix using the methods of Perera et al. (1999), but with the concentrations of dNTPs (200 µm) and Taq polymerase (1 U) modified as per their use by Meerow et al. (2003). Two loci (CAC2, CAC6) were able to be successfully cross-amplified in Beccariopheonix. Florescent labelled CAC2 and CAC6 were used for analysis. The PCR product fragments were then run under electrophoresis conditions using 3100 GENETIC ANALYSER software. Data were extracted and 3100 DATA COLLECTION v.1.1 and ABI GENOTYPER v.3.7 were used to score the fragments. The size each of the microsatellite fragments was determined and recorded as presence or absence of each band (allele) at each loci enabling genotypes to be determined.

Statistical methods

Some samples did not successfully produce consistent, readable AFLP bands and so were not included in analyses of AFLP data. As a result 122 AFLP bands were scored for presence or absence within the B. madagascariensis samples analysed. When the additional B. alfredii (1136) samples were analysed for taxonomic distinctiveness of the new taxon the AFLP primer combination AGG–CTA did not work satisfactorily for the reference B. madagascariensis samples, so a subset of bands, excluding those generated from this primer combination, were used for the analysis of taxonomic distinction. Reference B. madagascariensis samples were compared with the bands recorded for them from the previous analysis and any additional bands were removed from analysis of B. alfredii samples. Any bands previously scored but which were not resolved in this analysis for the reference samples compared with their original analysis were also removed from the analysis of B. alfredii. Any new bands outside these limitations that were strongly expressed in B. alfredii samples were then added to the analysis and scored for all samples for the taxonomic comparison. Only two of the 14 B. alfredii samples were able to be resolved and subsequently added to the rest of the samples with the amended bands scored and used in the taxonomic analysis; 100 bands were thus used in the AFLP taxonomic analysis.

The percentage of polymorphic AFLP bands (%P) was calculated for each population as well as the number of bands and number of private bands (pB) per population. The significance of genetic partitioning of AFLP variation among populations within the species was tested by AMOVA using GENALEX v.6 (Peakall & Smouse, 2005). A small subset of the samples did not produce bands for the microsatellite loci analysed and hence were not included in those analyses. GENALEX v.6 (Peakall & Smouse, 2005) was used to analyse the following: allelic frequencies at each site, the number of alleles per locus (A), the effective number of alleles per locus (Ne), the information index (I), the mean observed (Ho) and mean expected (He) heterozygosity based on Hardy–Weinberg assumptions, the allelic fixation index F (Wright, 1965) and the number and frequency of private alleles (Ap) for each population. Species level estimates of mean number of alleles per locus (As). The partitioning of genetic diversity (FIS, FIT and FST) was investigated using F statistics (Wright, 1965). AMOVA was undertaken to determine the apportionment of microsatellite genetic diversity within and among populations using the test statistic ΦPT to enable comparisons with the AFLP analysis (Peakall & Smouse, 2005). The average gene flow among sites (Nm) was estimated from FST values (Peakall & Smouse, 2005).

Relationships amongst populations for both AFLP and microsatellite data were investigated by calculating Nei's genetic distances to generate genetic distance matrices (using GENALEX v.6). These were then used to investigate relationships among populations in upgma analyses to generate dendrograms and in nonmetric multidimensional scaling analysis (MDS with 999 permutations; using the Primer 5 program). Genetic relationships among all individuals were also investigated to assess the distinctiveness of populations using Principal Coordinates Analysis (PCoA; Orloci, 1978) where genetic distance measures were standardized and 999 bootstrap permutations used undertaken on GENALEX v.6 program (Peakall & Smouse, 2005) and using nonmetric multidimensional scaling (MDS).

For each species population allelic frequencies at individual loci were plotted as pie graphs and overlaid onto maps indicating the geographical relationships among sites to investigate relationships among populations. Maps of Madagascar indicating the locations of known populations were generated using ArcGIS and relevant map layers. Site environmental data were derived from GIS data layers available for these locations (J. Moat, pers. comm.)

RESULTS

The known distribution of Beccariophoenix madagascariensis populations is noticeably disjunct with distinct northern and southern clusters (Table 1; Fig. 1). The southern population cluster is clearly in the best condition as far as adult population size and evidence of recruitment of seedlings to the adult population (Table 1; Rakotoarinivo, 2005). Most of the southern populations cluster in the Tolagnaro area (BM7765); the population BM7768 has been reduced to a single seedling following burning at the last count. Herbarium records indicate another previously recorded population nearby, but this has not been refound. Searches for further populations of this palm are continuing; indeed, while this paper was in preparation a new population was found close to the southern cluster and is now the largest known population (M. Rakotoarinivo, pers. comm.).

The northern populations are split into two distinct groups in regards to both habitat (Table 1) and vegetative form, but population sizes of all these populations are critically small and mostly composed of immature plants and seedlings (Table 1). All northern populations, however, have significant seedling populations indicating the potential for regeneration if left intact. Data from field surveys along with horticultural evidence from seed collectors indicate that the remaining individuals are not limited by reproductive failure or lack of seedling establishment but rather factors which directly kill adult or subadult plants. The Ranomafana Est populations (BM1008, BM168) form a geographical cluster within the northern populations and also inhabit a distinctively different habitat, being the only populations found in the drier perhumid zone, with differing soil and forest assemblages (Table 1). These populations have the distinctive ‘window’ form sought by the horticultural trade; the fact that this leaf form remains when grown in cultivation under a variety of environmental conditions indicates it has a genetic basis, thus making these populations genetically distinctive from all other populations. Unfortunately none of the ‘window’ form samples from Ranomafana Est could be resolved with the genetic markers used.

The new species B. alfredii occupies a highly disjunct location from the B. madagascariensis populations and highly distinctive habitat, occurring in a completely different climate zone in moist, protected gullies and on completely differing soil types (Table 1; Fig. 1).

The major geographical disjunction of greater than 600 km between northern and southern populations is also a genetic disjunction. The AFLP data clearly show that the two southern populations (BM7765, BM7768) are genetically distinct from the northern populations sampled (BM7756, BM7758). This was consistent in the dendrograms and using the two cluster methods PCA and MDS (Figs 2, 3). The Tolagnaro/St Luce (BM7765) population contained nine private AFLP bands indicating its differentiation from the northern populations (Table 2). The Manentenina sample (BM7768) clearly clusters with the Tolagnaro population (BM7765), which is the closest population geographically, located approximately 42 km away (Figs 1–3). However, unlike the AFLP data when all individuals were analysed as one group, the microsatellite loci did not segregate the northern and southern populations as clearly (Fig. 2). The southern population BM7765 contained three private microsatellite alleles clearly differentiating it from the northern populations (Figs 1, 4). The pie chart map overlays clearly show that for the CAC2 locus the two southern populations (BM7765, BM7768) contain an allele (219 bp) not found in northern populations (Fig. 1). The BM7768 population is not as clearly separated genetically from the northern populations (BM7756, BM7758) as it shares one CAC6 loci allele (127 bp) with these populations (Fig. 4). Estimates of gene flow when averaged across all populations were quite low (Nm = 0.302) consistent with the observed north–south population differentiation. However, when calculated on a pairwise basis, Nm values between each northern population and the southern BM7765 population (Nm = 1.317 and 1.519, respectively) were sufficient to support the hypothesis that the northern and southern populations differentiated owing to drift.

Figure 2.

Principal coordinates analysis (PCoA) of Beccariophoenix samples using genetic distance matrices with data standardization for both PCoA of AFLP. A, data and microsatellite data. B, individuals from B. madagascariensis populations BM7756, BM7758, B7765 and BM7768 are indicated by symbols illustrated and the B. alfredii sample is indicated BA0136. The PCoA of AFLP data (A) axis 1 accounts for 43% and axis 2 22.3% of the variation in the data, while the first three axes combined account for 80% of the variation. Microsatellite PCoA (B) axis 1 accounts for 26%, axis 2 accounts for 22.3% and combined the first three axes account for 64% of variation in the data.

Figure 2.

Principal coordinates analysis (PCoA) of Beccariophoenix samples using genetic distance matrices with data standardization for both PCoA of AFLP. A, data and microsatellite data. B, individuals from B. madagascariensis populations BM7756, BM7758, B7765 and BM7768 are indicated by symbols illustrated and the B. alfredii sample is indicated BA0136. The PCoA of AFLP data (A) axis 1 accounts for 43% and axis 2 22.3% of the variation in the data, while the first three axes combined account for 80% of the variation. Microsatellite PCoA (B) axis 1 accounts for 26%, axis 2 accounts for 22.3% and combined the first three axes account for 64% of variation in the data.

Figure 3.

Principal coordinates analysis (PCoA) of Beccariophoenix madagascariensis (BM) and B. alfredii (BA) samples using genetic distance matrices with data standardization (top). Individuals from B. madagascariensis populations BM7756, BM7758, BM7765 and BM7768 are indicated by symbols illustrated and the B. alfredii samples are indicated by BA1136. The data for the two species are encircled. The PCoA of AFLP data (A) co-ordinate axis 1 accounts for 54% and axis 2, 24% of the variation in the data, while the first three axes combined account for 87% of the variation. The bottom figure is the dendrogram from the same AFLP data (100 bands) using Nei's genetic distance between B. madagascariensis and B. alfredii populations (codes given).

Figure 3.

Principal coordinates analysis (PCoA) of Beccariophoenix madagascariensis (BM) and B. alfredii (BA) samples using genetic distance matrices with data standardization (top). Individuals from B. madagascariensis populations BM7756, BM7758, BM7765 and BM7768 are indicated by symbols illustrated and the B. alfredii samples are indicated by BA1136. The data for the two species are encircled. The PCoA of AFLP data (A) co-ordinate axis 1 accounts for 54% and axis 2, 24% of the variation in the data, while the first three axes combined account for 87% of the variation. The bottom figure is the dendrogram from the same AFLP data (100 bands) using Nei's genetic distance between B. madagascariensis and B. alfredii populations (codes given).

Table 2.

Summary of genetic variation within Beccriophoenix populations. Microsatellite data for two loci are summarized including; mean number of alleles per locus (A), Mean number of effective alleles (Ae), Mean Information index (I), Mean observed (Ho) and expected heterozygosity (He), and mean Fixation Index (F) are given; total F is mean FIS value. AFLP diversity is also summarized; total number of bands per population (bands), the percentage of polymorphic bands per population; the number of private bands per population (pB). Sample sizes (N), for both microsatellite and AFLP analyses are given. are indicated by

Population N Microsatellite
 
AFLP
 
%P pB 
Ae Ne I Ho He F N bands 
B. madagascariensis 
BM7756 13 3.05 1.30 0.231 0.672 0.65* 15 102 47.5 
BM7758 4.5 3.28 1.33 0.188 0.691 0.72* 11 103 56.6 
BM7765 11 4.5 2.47 1.13 0.364 0.595 0.39* 11 104 54.1 
BM7768 1.0 1.00 0.00 0.000 0.000 NA 84 0.0 
B. alfredii 
BA0136 1.0 1.00 0.00 0.000 0.000 NA NA NA NA NA 
Total 32 NA NA NA NA 0.599 38 122 NA 
Population N Microsatellite
 
AFLP
 
%P pB 
Ae Ne I Ho He F N bands 
B. madagascariensis 
BM7756 13 3.05 1.30 0.231 0.672 0.65* 15 102 47.5 
BM7758 4.5 3.28 1.33 0.188 0.691 0.72* 11 103 56.6 
BM7765 11 4.5 2.47 1.13 0.364 0.595 0.39* 11 104 54.1 
BM7768 1.0 1.00 0.00 0.000 0.000 NA 84 0.0 
B. alfredii 
BA0136 1.0 1.00 0.00 0.000 0.000 NA NA NA NA NA 
Total 32 NA NA NA NA 0.599 38 122 NA 
*

Populations where loci deviated significantly from Hardy–Weinberg expectations; NA, not applicable.

Figure 4.

Allelic frequencies at the CAC6 microsatellite locus (8 alleles) in Beccariophoenix populations are shown in pie charts indicating their relative spatial locations on a map of Madagascar. Species (BM or BA) and population codes are given for each pie chart displayed and population locations are indicated. Alleles are represented by differing shading as indicated in the legend.

Figure 4.

Allelic frequencies at the CAC6 microsatellite locus (8 alleles) in Beccariophoenix populations are shown in pie charts indicating their relative spatial locations on a map of Madagascar. Species (BM or BA) and population codes are given for each pie chart displayed and population locations are indicated. Alleles are represented by differing shading as indicated in the legend.

The estimate of FST based on microsatellite loci among all populations was 0.348. While a significant 35% of the variation is among populations (BM7756, BM7758, BM7765), 65% of AFLP variation is found within populations (Table 3). Similarly the microsatellite data analysis found that 17% of variation was among populations (BM7756, BM7758, BM7765) while 83% of the variation was found within populations (Table 3). Thus, despite the very small population sizes and small sample sizes there was considerable genetic diversity within each of the populations sampled, seen by the scatter in the clusters (Fig. 2) clearly indicating that populations are not genetically depauperate. Of the122 AFLP bands scored for presence or absence, all three populations were similarly diverse with 102–104 of the 122 bands found within each population, and approximately 50% were variable in the three populations surveyed (Table 2). For each of the two microsatellite loci surveyed, eight alleles were recorded in the species (As) and both loci were polymorphic for all populations that consisted of more than one individual (Table 2). Genetic diversity as measured by mean expected heterozygosity (He) across the two loci was also high for all three populations measured, ranging between 0.691 and 0.585 (Table 2).

Table 3.

Summary table of amova for Beccariophoenix madagascariensis populations BM7756, BM7758 and BM7765 indicating the partitioning of variation within and among populations and its significance for AFLP data (122 bands) and microsatellite data (2 loci). Test statistic is ÖPT enabling comparison of results for different markers

Source d.f. SS MS Est. var. ΦPT comparison
 
Value Prob 
AFLP 
Among pops. 158.977 79.488 5.658 35%   
Within pops. 34 357.564 10.517 10.517 65% 0.349826 0.001 
Microsatellite 
Among pops. 14.239 7.119 0.462 17%   
Within pops. 29 66.230 2.284 2.284 83% 0.168231 0.002 
Source d.f. SS MS Est. var. ΦPT comparison
 
Value Prob 
AFLP 
Among pops. 158.977 79.488 5.658 35%   
Within pops. 34 357.564 10.517 10.517 65% 0.349826 0.001 
Microsatellite 
Among pops. 14.239 7.119 0.462 17%   
Within pops. 29 66.230 2.284 2.284 83% 0.168231 0.002 

The northern Mantadia populations BM7756 and BM7758 are located only 3.3 km apart and there was some variation between them. For example, BM7756 had two private AFLP bands and BM7758 had one private band (Table 2). Their allelic composition for the two microsatellite loci was also slightly different, each containing a private allele (Figs 1, 4). However, owing to the variation within each population all cluster analyses indicated a complete genetic overlap between the two populations clearly indicating they are not genetically distinct and essentially genetically they behave as one subdivided population (Fig. 2). There is a very high pairwise gene flow (Nm = 15.4) between populations BM7756 and BM7758, indicating considerable gene flow due either to pollen or seed dispersal.

Whilst the number of samples was limited (2), the AFLP results for B. alfredii (BA1136) show it to be highly distinct from all of the B. madagascariensis populations, with 25 private bands not found in any other population (Table 2). The figures produced from genetic distances by PCoA and the upgma dendrogram clearly indicate a far greater genetic distinctiveness of these samples (BA1136) compared with the internal diversity of the northern and southern B. madagascariensis populations (Fig. 3). Thus, these data clearly support separation of B. alfredii as a distinct taxon. However, the single sample from B. alfredii (BA0136) did not segregate from the B. madagascariensis populations at the two microsatellite loci analysed (Figs 1, 2, 4). The loci surveyed link the single sample of B. alfredii (BA0136) more closely to the northern populations (BM7756, BM7758) at the CAC6 loci as it contains an allele (127 bp) found in all populations except the southern BM7765 (Fig. 4). The B. alfredii population is located closer to (c. 250 km) the northern B. madagascariensis populations (BM7756, BM7758) compared with the BM7765 population (c. 500 km; Fig. 1). The AFLP PCoA results also place B. alfredii closer to the northern than the southern populations of B. madagascariensis (Fig. 3). Unfortunately, the samples obtained for the Ranomafana Est ‘window’ form were not successfully resolved with the microsatellite loci or the AFLP loci tested owing to technical difficulties, and thus we are not able to confirm whether or not these populations differentiated from others using these genetic markers.

The microsatellite loci enabled measures of population-level inbreeding and deviation from Hardy–Weinberg conditions to be determined. All three populations (BM7756, BM7758, BM7765) deviated significantly from Hardy–Weinberg equilibrium at each of the two loci studied (Table 2). Allelic fixation indices within populations were high (significantly greater than 0) with an excess of homozygotes compared with expected indicating that all populations are effectively inbred (Table 2). The two northern populations BM7756 and BM7758 were, however, considerably more inbred than the southern BM7765 population with fixation indices approximately twice as large (Table 2). This would be consistent with differences in population structure. Population BM7756 has only three mature plants plus immature plants present and population BM7758 consists of only immature plants whereas population BM7765 has six mature plants and a total population size of 40. Population BM7765 is thus larger with more mating possibilities (Table 1). It is also known that there are an additional three populations within 3 km of BM7765 with which it could potentially outcross if pollinators are able to move such distances (Table 1). The gene flow data from BM7756 and BM7758 suggest that pollinators and seed dispersers can potentially move distances of 3 km between populations.

DISCUSSION

Where a species has highly restricted distribution there is an increased chance of significant portions of its global distribution being lost as a direct result of deforestestation, making such a species highly vulnerable to extinction owing to fragmentation (Terborgh, 1992; Turner, 1996). Other direct effects of fragmentation include: reduction of population size, increased isolation of populations and increased likelihood of local extinction resulting from catastrophes such as fires (Terborgh, 1992; Turner, 1996). Diversity within populations is expected to be lost owing to the direct sampling effect of reduction in demographic population size and subsequent loss in later generations following genetic drift (Ellstrand & Elam, 1993). Correlations between decreasing population size and diversity have been found in some species (e.g. Armstrong & DeLange, 2005). However, the impact of decreased population size on within-population diversity is expected to take generations to lead to significant loss due to drift (Ellstrand & Elam, 1993).

Despite being critically endangered with very few individuals remaining in known populations, B. madagascariensis displays considerable diversity in the remaining populations (Tables 1, 2). Few studies of endangered palms have been published; however, Dowe, Benzie and Ballment (1997) studied a similarly endangered palm Carpoxylon macrospermum from the islands of Vanuatu and found very low genetic diversity in similarly small populations. Extremely low genetic diversity was also found in natural populations of the endangered palm Ptychosperma bleeeseri from northern Australia (Shapcott, 1998b) and low diversity was found in the threatened Phoenix canariensis from the Canary Islands (González-Pérez, Caujape-Castells & Sosa, 2004) contrasting with the results for B. madagascariensis. Alternatively, studies by Henderson, Billote and Pintaud (2006) of the restricted endemic species Phoenix atlantica from islands off the western coast of Africa found greater variation in microsatellite loci than the diversity found in this study. The genetic diversity found in several rare understorey Pinanga palms from Brunei was also relatively high (Shapcott, 1999).

Gitzendanner and Soltis (2000) stressed the importance of comparative studies when assessing diversity in rare species as taxonomy can have a strong effect on levels of diversity. Beccariophoenix is an isolated genus and few studies are available to enable a basis to determine the significance of diversity levels within the palm family. Studies using allozyme markers found genetic diversity levels to be quite variable among the palm species studied to date. Several widespread arboreal palm species studied had relatively low levels of within-population diversity for example, Australian Carpentaria acuminata and Archontophoenix cunninghamiana (Shapcott, 1998a, 1999), Mexican Astrocaryum mexicanum (Eguiarte, Perez-Nasser & Piñero, 1992) and the Californian Washingtonia filifera (McClenaghan & Beauchamp, 1986). Alternatively high genetic diversity was reported for Acrocomia aculeata populations (Lopes et al., 1992).

Two commercial species are amongst the groups of palms thought to be most closely related to B. madagascariensis (Dransfield & Beentjie, 1995). Teulat et al. (2000) found diversity in the coconut palm (Cocos nucifera) that was comparable to B. madagascarienis. High levels of genetic diversity have been found in both wild and cultivated populations of the commercial peach palm Bactris gasipaes (e.g. Adin et al., 2004; Couvreur et al., 2005). Moderate levels of genetic variation within populations have been recorded in other commercial palms including the oil palm (Elaeis guineensis, e.g. Hayati et al., 2004; Maizura et al., 2006) and the date palm (Phoenix dactylifera, e.g. Al-Khalifah & Askari, 2003; El-Assar et al., 2005). Contrasting results have been found in rattan palms; moderate genetic diversity was found in remnant populations of Calamus subinermis in Sabah by Bon, Joly and Alloysiu (1999), whereas low diversity was found in C. manan by Wickneswari et al. (2002). The Brazilian heart of palm (Euterpe edulis) is similar to B. madagascariensis in that land degradation and wild harvesting have greatly reduced its populations in the wild. Cardoso et al. (2000) found moderate genetic variation within populations (57.4%) of this species, but several populations studied were relatively genetically depauperate.

Drummond et al. (2000) studied a critically endangered species of Myrtaceae with similarly small adult population sizes in New Zealand and also found considerable diversity within populations. They suggested that this was a legacy of previously more extensive species populations. Thus it would seem that the high diversity recorded within populations is an indication that B. madagascariensis may also have been recently more abundant and has experienced relatively few generations since population fragmentation occurred. Indeed, the recent finding of the new population at Vondrozo is an indication that this prediction may be correct. The two samples of the new species B. alfredii that have been analysed were variable and the known population size is quite large, although currently only from one location. These results suggest that the new species may also contain significant within-population genetic diversity.

The present study found that a major geographical and ecological disjunction between northern and southern B. madagascariensis populations was matched by significant genetic differentiation. Significant geographical structuring of species diversity has been recorded in other palm species with significant distances among populations. For example, in the restricted endemic species Phoenix atlanticaHenderson et al. (2006) found evidence of geographical structuring of diversity according to different islands and geographical distance. In general, genetic distance was correlated with geographical distance in Euterpe edulis (Cardoso et al., 2000). Teulat et al. (2000) found significant geographical distinctness between Cocos nucifera populations from different geographical regions and Australian Carpenatria acuminata and Archontophoenix cunninghamiana also have considerable population differentiation (Shapcott, 1998a, 1999). Loo et al. (1999) found significant diversity among populations of the fan palm Licuala glabra var. glabra that corresponded to morphological distinctiveness and locations on differing mountains, indicating that mountains can potentially act as dispersal barriers for some palm species.

Relationships between geographical disjunctions and genetic distinctiveness have been observed in several other species (e.g. Tremetsberger et al., 2004; Jang, Mullner & Greimler, 2005). Schönswetter et al. (2003) found that the two disjunct populations of alpine herbs were genetically distinct from those in the main population. They hypothesized that these populations were glacial survivors as the two areas are congruent with hotspots of rare relictual plant taxa. In this study the two disjunct regions are floristically different (Du Puy & Moat, 1996) and the Tolagnaro/St Luce area is known to have a highly localized endemic flora (Bollen & Donati, 2005). This suggests not only geographical disjunction, but also ecological distinctiveness which would provide differing selection pressures and indicates relatively old ecological differentiation.

In the present study two populations were located within 3 km of each other and exhibited limited genetic differentiation; they were thus effectively akin to subpopulations of one population. Low genetic differentiation has been observed in other palms where populations are in close proximity, e.g. among populations of Pinanga palms (Shapcott, 1999) and among Astrocaryum mexicanum populations in close proximity (Eguiarte et al., 1992). High levels of gene flow were reported among Euterpe edulis populations (Gaiotto, Grattapaglia & Vencovsky, 2003), which would reduce the amount of population differentiation.

The fleshy fruits of B. madagascariensis are green, ripening to yellowish or purple, whereas B. alfredii seeds ripen to a reddish to black colour. They are of a moderate size (2.5 cm diameter; Rakotoarinivo, 2005; Rakotoarinivo et al., 2007). Palms represent an important food source for many species of frugivores (Peres, 1994; Gaiotto et al., 2003; Price, 2004). The asynchrony and aseasonality of fruit production means that palms can provide a food source at times of year when fruits are relatively scarce (Peres, 1994; Gaiotto et al., 2003; Price, 2004). Beccariophoenix madagascariensis fruits are green in January, ripen by March and by June most fruit have fallen from trees (Rakotoarinivo, 2005). A study of the phenology of a St Luce littoral forest (Bollen & Donati, 2005) identifies generally low ripe fruit abundance in March, June and July; thus B. madagascariensis is producing ripe fruit at a time of relatively scarcity in this forest.

Lemurs play an important role as seed dispersers in Madagascar; black lemurs, for example, have been seen eating the fruit of 70 species (Dew & Wright, 1998; Birkinshaw, 2001). Lemurs are typically associated with seed dispersal of species with fruits larger than 2 cm length (Birkinshaw, 2001). Lemurs prefer fruit coloured green, brown, tan, purplish, red or some combination and 10% of fruit eaten are yellowish (Dew & Wright, 1998). Germination trials found that seeds which passed through the gut of lemurs sprouted faster and in greater numbers than seeds not eaten by lemurs (Dew & Wright, 1998).Thus, it seems highly likely that lemurs are the main seed dispersers of B. madagascariensis and B. alfredii. The number of lemur species has been shown to increase with increasing number of tree species and a minimum number of different species seems to be required to assure year round food availability (Ganzhorn et al., 1997). Thus maintaining forest tree diversity is important for maintaining lemur populations.

The dispersal potential of lemurs is limited by terrain, for example Rivers can form barriers to dispersal for lemurs (Goodman & Ganzhorn, 2004). Several aerial frugivores are potential seed dispersers of B. madagascariensis with less restriction on movement. These include the Madagascar blue pigeon (Alectroenus madagascariensis), Madagascar green pigeon (Treron australis) and Madagascar bulbul (Hypsipetes madagascariensis; Birkinshaw, 2001). Some of these species have been recorded swallowing moderate-sized fruit and voiding them in a viable state (Birkinshaw, 2001). Malagasy rainforest bird distribution shows little latitudinal variation, with most birds being found over most of the length of the island, but great elevational variation (Hawkins, 1999). Mammalian frugivores potentially capable of long-distance dispersal of Beccariophoenix fruits include the Madagascan fruit bat (Pteropus rufus) and straw coloured fruit bat (Eiodlon dupreanum) (Birkinshaw, 2001).

The low diversity and high levels of gene flow observed between the two northern populations of B. madagascariensis could be a result of relatively frequent seed dispersal by frugivores capable of moving across the altered landscape. Alternatively, the relatively high diversity found within all populations in the species may indicate that genetic diversity is maintained by gene flow between a network of adjacent populations that has delayed loss of diversity within small populations as a result of drift. Studies of the rainforest palm Carpentaria acuminata have shown that diversity was maintained in small populations within an archipelago of fragments by dispersal by mobile frugivores (Shapcott, 2000). Gaiotto et al. (2003) reported dispersal of 22 km between juveniles and seed parents of Euterpe edulis, indicating that long-range dispersal in palms may be significant at the landscape scale.

Increased isolation of populations can also a result of the changed nature of the matrix between populations, such that it becomes inhospitable or not traversable to some species, or their dispersal vectors (Soons & Ozinga, 2005; Aguilar et al., 2006). The species B. alfredii occupies a completely different habitat and is geographically several hundred kilometres from B. madagascariensis populations. Thus, a combination of potentially strong selective pressures and geographical isolation are likely to have led to the differentiation of B. alfredii and B. madagascariensis. Whilst frugivores could potentially act to slow divergence, given the differing environment it seems unlikely that the same species would have moved between the different environments (Hawkins, 1999). Fruit bats seem the most likely frugivore capable of such movements across the landscape (Birkinshaw, 2001; Price, 2004).

Large-seeded palms usually rely on medium–large birds and mammals for dispersal (Zona & Henderson, 1989), but this fauna is seriously threatened by increased hunting pressure and habitat loss. Galetti et al. (2006) found that large-seeded palms were significantly affected by loss of seed dispersers resulting from forest fragmentation and hunting. They found that the probability of seeds being predated by insects in small fragmented sites where the fauna had been reduced was higher than in more intact forest. Galetti et al. (2006) predicted that palms that rely on large-bodied animals are more prone to extinction following fragmentation than those dependent on small animals (Galetti et al., 2006). Such findings may have implications for future survival of Beccariophoenix populations in small fragments.

Populations that are small and isolated are expected to become more inbred leading to decline in reproductive output and hence population growth (Ellstrand & Elam, 1993). These features leave them more susceptible to demographic stochastic events. As might be expected, given the very limited number of adults, all populations of B. madagascariensis studied were significantly inbred (Tables 1, 2). High levels of inbreeding have been recorded in several other palms, e.g. in Carpentaria acuminata (Shapcott, 1998a), rare understorey Pinanga species (Shapcott, 1999) and Calamus subinermis (Bon et al., 1999). However, higher levels of outcrossing seem to be more common in arborescent palm species (e.g. Gaiotto et al., 2003), and heterozygote excesses have been recorded in some species (Eguiarte et al., 1992; González-Pérez et al., 2004). Despite high outcrossing rates, some palm species have been demonstrated to be self compatible (Burquez, Sarukhan & Pedroza, 1987; Eguiarte et al., 1992). Because the isolated single tree at Ranomafana Est produces viable seed (used in the seed trade) and the nearest known populations are approximately 20 km away, either the pollinators can travel 20 km to pollinate this individual or it is capable of self pollination.

Dichogamy is widespread in palms and protandry, where the female flower matures only after the male stamens have been shed, is the most common reproductive state (Loo et al., 2006). This was observed in Beccariophoenix (Rakotoarinivo, 2005). Within-flower self pollination is not possible because flowers are unisexual, but variation in timing between flowers on the same plant can enable selfing (Loo et al., 2006). In Euterpe edulis, for example, opening of male flowers before females promotes outcrossing; however, with one to three flowerings per reproductive cycle there is the possibility of inbreeding between flowers on the same plant (Cardoso et al., 2000). Self fertilization has been reported in Bactris species (Listabarth, 1996), but, similarly to Astrocaryum (Eguiarte et al., 1992), there is variation in self fertility. Rakotoarinivo (2005) found that seed set compared with flower production seemed likely to be pollen-limited owing to the limited reproductive synchrony between the male and female flowers. Cunningham (1996) also found that pollen limitation could reduce fruit set in palms.

Listabarth (1996) indicated that many palm species are beetle pollinated, and floral scent is important in beetle-pollinated species (Ervik, Tollsten & Knudsen et al., 1999). Beccariophoenix flowers have a strong odour which attracts a massive number of insects (Rakotoarinivo, 2005).The male phase lasts approximately 10 days while female flowers are only receptive for 2 days. Beetles were the only insect in contact with both pollen and receptive female flowers (Rakotoarinivo, 2005). Thus, selfing is possible but given the timeframes it is also likely the beetles move between plants. Beetle pollinators were resident within the Bactris inflorescence Listabarth (1996), however, Eguiarte et al. (1992) hypothesized that the high outcrossing rate in Astrocaryum mexicanum was due to the long distance movement of its beetle pollinators.

Authors have found that in some arborescent palms that wind, in addition to insects, plays a role in pollination (Scariot & Lleras, 1991; Gaiotto et al., 2003; Adin et al., 2004). Scariot and Lleras (1991) found Acrocomia aculeata exhibited both selfing and outcrossing and the combination of two pollination strategies, wind and insects, enabled greater reproductive success. It is unknown whether or not wind plays a role in the pollination of B. madagascariensis, but clearly flexible mating and pollination can enhance survivorship in a fragmented environment. Cunningham (1996) found that bats were also potential palm pollinators, but phenological studies of B. madagascariensis were undertaken during the day (Rakotoarinivo, 2005), so we do not know if bats also contribute to pollination in this species.

The continuous growth pattern in palms lends itself to asynchrony of inflorescence production, with larger palms producing more inflorescences per year (DeSteven et al., 1987). Rakotoarinivo (2005) found that in B. madagascariensis the number of inflorescences was positively correlated with plant age and that at any one time approximately 12% of plants had synchronously open flowers, with mature inflorescences present over a six-month period. This finding has implications for these very small palm populations: as all remaining contain fewer than ten adult plants, they are unlikely to contain synchronously reproductively mature plants, thus eliminating outcrossing within populations. The ability to self fertilize becomes very important for species survival in such small populations. The lower levels of inbreeding in the St Luce populations suggest that historically there has been some outcrossing between plants or possibly between populations, indicating either some synchrony in flowering between plants within populations or between populations. The lower inbreeding may also indicate that the adult populations have recently declined and the offspring represent the offspring of mating from a previously larger population. The relatively high levels of genetic diversity indicate the potential for long-term survival of B. madagascariensis, provided that demographic viability is achieved. The relative abundance of juvenile B. madagascariensis plants (Table 1) suggests that the species viability is not determined by reproductive failure. Dowe et al. (1997) also found natural regeneration not the limiting factor for an endangered palm in Vanuatu.

Conservation

Byg and Balslev (2001) found that use of palm heart for food in Madagascar was the single most common use of palms followed by their use for construction. They found a significant correlation between number of uses, use value and species importance to the community. Thus, Becariophoenix clearly has high local importance value as it is known (the species has a local name) and has many uses including food and construction (Dransfield & Beentje, 1995). Whilst exploitation has clearly been part of the reason populations are in decline, it should also mean that there is more likelihood of a local interest in its conservation (Byg & Balslev, 2001). The new population of B. madagascariensis just discovered at Vondrozo is much larger than any others and in this area the palm is not utilized by the local people (M. Rakotoarinivo, pers. observ.). Clearly the population size has been reduced to critical levels and at present the remaining populations of B. madagascariensis cannot sustain continued utilization of adult plants. If the species is to survive then the existing populations need some sort of protection until a new cohort of adults matures.

Remote sensing data show that, despite considerable habitat loss, the Eastern humid forest cover is still relatively continuous in Madagascar (McConnell & Sweeney, 2005), extending between the northern and southern populations of B. madagascariensis. Thus, it was not surprising that a further B. madagacariensis population was found. Considerable fragmentation of rainforest in Madagascar is a result of using fire as a management tool to both maintain agricultural land and expand it (Kull, 2002). The history in Madagascar has resulted in poor fire management consensus between subsistence farmers and government agencies (Kull, 2002). Whilst repeated burning can reduce the area of forest, adult palms are resistant to single burns owing to their stem physiology (Tomlinson, 2006). Thus, where remaining populations consist only of juvenile plants they are likely to become locally extinct if subject to fire, as appears likely to happen to the Mantenenina BM7768 population. Furthermore, while the adult plants may survive a low-intensity fire given the demographic need for survivorship of juveniles to adult stages for the species survival, fires in any remaining populations would result in significant demographic set backs for the species.

Satellite images show reduction in conversion of forest to agricultural land via burning in the last decade; however, a considerable amount of forest is being converted from natural forest to plantations (McConnell & Sweeny, 2005). The majority of plantations are reported to be composed of Eucalyptus species native to Australia and are often grown contiguous with existing forest (McConnell & Sweeny, 2005). Eucalyptus species have been planted in the St Luce area (Ramanamanjato & Ganzhorn, 2001). The littoral forest fragments in this area range in size from 3 to 377 hectares are among the most intact Malagasy littoral forests; they have very high species diversity and contain several endemic plant species (Bollen & Donati, 2005) including B. madagascariensis populations. The presence of Eucalyptus in the Madagascan landscape is likely to significantly increase the flammability of the landscape, causing hotter fires and carrying them further into remaining humid forest fragments (Bowman, 2000; Whelan et al., 2002). Thus these southern populations are at risk from reduction by wildfires.

In northern Australia rainforest fragments have traditionally been maintained by indigenous people by cool back burning to reduce the fuel load adjacent to the forest boundary (Bowman, 2000; A. Shapcott, pers. observ.). These concepts have formed an integral part of modern fire management in Australia (e.g. Kirkpatrick, 1994; Bowman, 2000; Bradstock, Williams & Gill, 2002). Local Malagasy subsistence farmers potentially have the fire management skills to protect forest fragments from encroachment by hotter wildfires (Kull, 2002; Simsik, 2004).

A large portion of the remaining humid forest in Madagascar is represented in a variety of reserves (McConnell & Sweeny, 2005). However, commercial interests trading in timber and agricultural products are known to gain access and extract timber, leading to degradation within reserves (Simsik, 2004). In addition, local villagers source forest products from adjacent forests (Simsik, 2004). The northern Mantadia populations of B. madagascariensis are located within a National Park, but despite this security they have been utilized in recent years leading to their decline. Byg and Balslev (2001) argue that people will be more motivated to conserve species that are more important to them than species which are less so. Given the local significance of B. madagascariensis and its critically low population sizes yet substantial seed crop, it is a potential candidate for an active recovery programme. In Madagascar it has been documented that local people resent conservation activities that fail to include them as part of the programme and are frustrated by conservation programmes that victimize them and criminalize their traditional livelihood activities (Simsik, 2004). Thus, in order to gain the support for active conservation management by locals it is expected that a B. madagascariensis recovery programme would not only aim to conserve and expand the existing populations, but also establish plantings for future sustainable use by local villagers to divert utilization pressure on wild populations.

The genetic and morphological evidence suggests that B. madagascariensis populations exist as genetically and geographically distinctive groupings in addition to the new species B. alfredii. The results also suggest that within each grouping small populations can be maintained when clustered and that effective gene flow can occur over distances of 3–5 km. The data indicate that populations do not require genetic enhancement but the Ranomafana Est populations may require demographic enhancement. It is suggested that enhancement can be undertaken using seed or seedlings originating from any populations within the same genetically/geographically distinct region. The substantial juvenile populations of the Mantadia populations suggest that these populations could become both genetically and demographically viable if protected from fire or destructive harvesting. Mantadia populations are naturally small and located on isolated ridges/hills within a National Park and so are potential candidates for both population protection and reintroduction of new populations on nearby suitable ridge/hills to reduce the potential for local extinction due to stochastic events. The new populations will have the greatest benefit for the existing populations if established within 5 km using seeds derived from the adjacent populations given the evidence for local gene flow.

Pareliussen et al. (2006) found that tree seedlings planted closer to existing forest edges had a higher survivorship than those planted in grassy openings and recommended that expansion of existing forest edges would be more successful than creating new fragments in Madagascar. Thus, any new populations of B. madagascariensis created should ideally be within or on the edges of existing forest. Given the proximity of the Mantadia populations to local villages, seed from these populations could be utilized to create plantings of B. madagascariensis, and the species could even be locally utilized in amenity and street plantings. This strategy may also be utilized in the Ranomafana Est ‘window’ form population area. Here it is known that a single adult tree is protected as a commercial investment by a local villager. Ideally clusters of small populations would be either conserved or created, with some new populations earmarked for sustainable utilization by local villagers. They would be created from seeds collected from within the same geographical area to maintain the genetically distinct forms of the species.

The southern B. madagascariensis populations near Tolgnaro/St Luce seem to be very significant for the species because of the existence of several populations in close proximity to each other that are actively regenerating. Some of the southern populations are earmarked for conservation protection by a local ilmenite mining venture and thus their future appears relatively secure, but others are likely to be lost to the ilmenite mining activities (Rakotoarinivo, 2005). Given the significance of these populations for the survival of B. madagascariensis in the wild, the possibility of extending the planned conservation areas to include the populations within the affected area should be considered. There are existing conservation groups active in the St Luce area and a strong argument could be made that any populations of B. madagascariensis earmarked for destruction by mining activities should be compensated for by the creation of new populations within the local area, given the significance of these populations. These forests have also been shown to be highly significant for many species (Hawkins, 1999; Ramanamanjato & Ganzhorn, 2001; Bollen & Donati, 2005); thus the maintenance of local species in situ and their incorporation in post mining regeneration activities should be strongly encouraged. In Australia which also has a highly endemic flora and fauna, sand mining companies are required to revegetate post mining using local species and the techniques have been relatively well developed. The same standards of post mining operations should be required in this highly significant area of Madagascar.

The populations of the new species B. alfredii are significantly larger than those of B. madagascariensis (Table 1). Whilst B. alfredii does not grow within closed mesic humid forest types, it is found within deep, topographically protected gallery forest where it forms the dominant canopy species. In northern Australia several Carpentaria acuminata populations can also be found in similar topographically fire-protected, deep, gallery forests where populations have persisted for thousands of years in an otherwise fire-prone environment (Shapcott, 1998a). Thus, the isolation and topographically protected nature of the population of this species should ensure survival of this species. However, if human access to the populations is increased they would be at risk of unsustainable harvesting practises.

CONCLUSION

This study found that in B. madagascariensis ecological and geographical differentiation among populations reflected genetic differentiation. Populations growing in different forest types and geologies should be treated as genetically distinct and any recovery programmes should recognize this. The new species B. alfredii represents an extreme end of this genetic differentiation process. Despite critically small population sizes B. madagascariensis retains considerable genetic diversity and was probably more widespread in the recent past. We predict that with further searching more populations within the existing species range will be found, such as the one found in the later stages of this study.

This study found that the populations were inbred and that population demographic size needs to be increased to increase the potential for outcrossing but that the ability to self pollinate had enabled critically small populations to maintain reproductive viability. The results indicate that high levels of gene flow can be maintained between close populations and is likely to be facilitated by mobile frugivores such as lemurs, fruit bats and fruit eating pigeons. Thus, further habitat modification which endangers these species will directly affect the long-term viability of B. madagascariensis. Furthermore, the surrounding land use and modification is likely to impact on the viability of both existing and future populations within existing fragments. This study illustrates how even species on the brink of extinction may have the potential for recovery. However, it is clear that for species such as this, utilization can lead to both endangerment and protection, and any recovery program needs to factor in ongoing human utilization to be successful.

ACKNOWLEDGEMENTS

This project was funded by the Friends of Kew Threatened Plants of Madagascar Appeal, the University of the Sunshine Coast funded A.S. to visit RBG Kew to undertake the laboratory work. Many assisted in the laboratory, especially Robyn Cowan, Dion Devey and Jeff Joseph, and in the field; Justin Moat assisted with GIS and mapping for the figures. Many thanks to Bill Baker for hosting me at Kew.

REFERENCES

Adin
A
,
Weber
JC
,
Soleto Montes
C
,
Vidaurre
H
,
Vosman
B
,
Smulders
MJM
2004
.
Genetic differentiation and trade among populations of peach palm (Bactris gasipaes Kunth) in the Peruvian Amazon-implications for genetic resource management
.
Theoretical and Applied Genetics
 
108
:
1564
1573
.
Aguilar
R
,
Ashworth
L
,
Galetto
L
,
Aizen
MA
2006
.
Plant reproductive susceptibility to habitat fragmentation; a review and synthesis through a meta-analysis
.
Ecology Letters
 
9
:
968
980
.
Al-Khalifah
NS
,
Askari
E
2003
.
Molecular phylogeny of date palm (Phoenix dactylifera L.) cultivars from Saudi Arabia by DNA fingerprinting
.
Theoretical and Applied Genetics
 
107
:
1266
1270
.
Armstrong
TTJ
,
DeLange
PJ
2005
.
Conservation genetics of Hebe speciosa (Plantaginaceae) and endangered New Zealand shrub
.
Botanical Journal of the Linnean Society
 
149
:
229
239
.
Birkinshaw
C
2001
.
Fruit characteristics of species dispersed by black lemur (Eulemur macaco) in the Lokobe forest Madagascar
.
Biotropica
 
33
:
478
486
.
Bollen
A
,
Donati
G
2005
.
Phenology of the littoral forest of St Luce Southastern Madagascar
.
Biotropica
 
37
:
32
43
.
Bon
M-C
,
Joly
HI
,
Alloysius
D
1999
.
Genetic variation in remnant populations of a rattan species (Calamus subinermis) in Sabah
.
Electrophoresis
 
17
:
1248
1252
.
Bowman
DMJS
2000
.
Australian rainforests; islands of green in a land of fire
 .
Cambridge
:
Cambridge University Press
.
Bradstock
RA
,
Williams
JE
,
Gill
AM
2002
.
Flammable Australia: the fire regimes and biodiversity of a continent
 .
Cambridge
:
Cambridge University Press
.
Britt
A
,
Clubbe
C
,
Ranarivelo
T
2004
.
Conserving Madagascars's plant diversity; Kew's Madagascar Threatened Plants Project
.
Kew Bulletin
 
21
:
258
264
.
Burquez
A
,
Sarukhan
J
,
Pedroza
AL
1987
.
Floral biology in a primary forest palm Astrocaryum mexicanum Liebm
.
Botanical Journal of the Linnean Society
 
94
:
407
419
.
Byer
DL
,
Waller
DM
1999
.
Do plant populations purge their genetic load? Effects of population size and mating history in inbreeding depression
.
Annual Review of Ecology and Systematics
 
30
:
479
513
.
Byg
A
,
Balslev
H
2001
.
Diversity and use of palms in Zahamena eastern Madagascar
.
Biodiversity and Conservation
 
10
:
951
970
.
Cardoso
SRS
,
Eloy
NB
,
Provan
J
,
Cardoso
MA
,
Ferreira
PCG
2000
.
Genetic differentiation of Euterpe edulis Mart. populations estimated by AFLP analysis
.
Molecular Ecology
 
9
:
1753
1760
.
Chase
MW
,
Hills
HG
1991
.
Silica gel: an ideal material for field preservation of leaf samples for DNA studies
.
Taxon
 
40
:
215
220
.
Coates
DJ
,
Hopper
SD
2000
.
Genetics and conservation of Australian flora
.
Australian Journal of Botany
 
48
:
1
3
.
Couvreur
TLP
,
Billotte
N
,
Risterucci
A-M
,
Lara
C
,
Vigouroux
Y
,
Ludena
B
,
Pham
J-L
,
Pintaud
J-C
2005
.
Close genetic proximity between cultivated and wild Bactris gasipaes Kunth revealed by microsatellite markers in Western Equador
.
Genetic Resources and Crop Evolution
 
00
:
1
13
.
Cunningham
SA
1996
.
Pollen supply limits fruit initiation by a rain forest understorey palm
.
Journal of Ecology
 
84
:
185
194
.
DeSteven
D
,
Windsor
DM
,
Putz
FE
,
De Leon
B
1987
.
Vegetative and reproductive phenologies of a palm assemblage in Panama
.
Biotropica
 
19
:
342
356
.
Dew
JL
,
Wright
P
1998
.
Frugivory and seed dispersal by four species of primates in Madagascar's eastern rain forest
.
Biotropica
 
30
:
425
437
.
Dowe
JL
,
Benzie
J
,
Ballment
E
1997
.
Ecology and genetics of Carpoxylon macrospermum H.Wendl. & Drude (Arecaceae), and endangered palm from Vanuatu
.
Biological Conservation
 
79
:
205
216
.
Doyle
JJ
,
Doyle
JL
1987
.
A rapid DNA isolation procedure from small quantities of fresh leaf tissue
.
Phytochemical Bulletin, Botanical Society of America
 
19
:
11
15
.
Dransfield
J
1988
.
Beccariophoenix madagascariensis
.
Principes
 
32
(
2
):
59
68
.
Dransfield
J
,
Beentje
H
1995
.
The Palms of Madagascar
 .
London
:
Royal Botanic Gardens Kew/ International Palm Society, Kew
.
Dransfield
J
,
Uhl
NW
,
Asmussen
CB
,
Baker
WJ
,
Harley
MM
,
Lewis
CE
2005
.
A new phylogenetic classification of the palm family, Arecaceae
.
Kew Bulletin
 
60
:
43
50
.
Drummond
RSM
,
Keeling
DJ
,
Richardson
TE
,
Gardner
RC
,
Wright
SD
2000
.
Genetic analysis and conservation of 31 surviving individuals of a rare New Zealand tree Metrosideros bartlettii (Myrtaceae)
.
Molecular Ecology
 
9
:
1149
1157
.
Du Puy
DJ
,
Moat
J
1996
.
A refined classification of the primary vegetation of Madagascar based on the underlying geology: using GIS to map its distribution and to assess its conservation status
. In:
Lourenco
WR
ed.
Biogeography of Madagascar.
 
Paris
:
ORSTOM
,
205
218
.
Eguiarte
LE
,
Perez-Nasser
N
,
Piñero
D
1992
.
Genetic structure, outcrossing rate and heterosis in Astrocaryum mexicanum (tropical palm): implications for evolution and conservation
.
Heredity
 
69
:
217
228
.
El-Assar
AM
,
Krueger
RR
,
Devanand
PS
,
Chao
C-CT
2005
.
Genetic analysis of Egyptian date (Pheonix dactylifera L.) accessions using AFLP markers
.
Genetic Resources and Crop Evolution
 
52
:
601
607
.
Ellstrand
NC
,
Elam
DR
1993
.
Population genetic consequences of small population size: Implications for plant conservation
.
Annual Review of Ecology and Systematics
 
24
:
217
242
.
Ervik
F
,
Tollsten
L
,
Knudsen
J
1999
.
Floral scent chemistry and pollination ecology of phytelephantoid palms (Arecaceae)
.
Plant Systematic Evolution
 
217
:
27
297
.
Frankham
R
1996
.
Relationship of genetic variation to population size in wildlife
.
Conservation Biology
 
10
:
1500
1508
.
Gaiotto
FA
,
Grattapaglia
D
,
Vencovsky
R
2003
.
Genetic structure, mating system and long-distance gene flow in heart of palm (Euterpe edulis Mart.)
.
Journal of Heredity
 
94
:
399
406
.
Galetti
M
,
Donatti
CI
,
Pires
AS
,
Guimaraes
PR
Jr
,
Jordano
P
2006
.
Seed survival and dispersal of an endemic Atlantic forest palm: the combined effect of defaunation and forest fragmentation
.
Botanical Journal of the Linnean Society
 
151
:
141
149
.
Ganzhorn
JU
,
Malcomber
S
,
Andrianantoanina
O
,
Goodman
SM
1997
.
Habitat characteristics and lemur species richness in Madagascar
.
Biotropica
 
29
:
331
343
.
Gautier
L
,
Goodman
SM
2003
.
Introduction to the flora of Madagascar, pp
.
229
250
. In:
Goodman
SM
,
Benstead
JP
eds.
The Natural history of Madagascar
 .
Chicago
:
University of Chicago Press
.
Gitzendanner
MA
,
Soltis
P
2000
.
Patterns of genetic variation in rare and widespread plant congeners
.
American Journal of Botany
 
87
:
783
792
.
González-Pérez
MA
,
Caujape-Castells
J
,
Sosa
PA
2004
.
Allozyme variation and structure of the Canarian endemic palm tree Phoenix canariensis (Arecaceae): implications for conservation
.
Heredity
 
93
:
307
315
.
Goodman
SM
,
Ganzhorn
JU
2004
.
Biogeography of lemurs in the humid forests of Madagascar; the role of elevational distribution and rivers
.
Journal of Biogeography
 
31
:
47
55
.
Hanski
I
,
Gilpin
ME
1997
.
Metapopulation ecology, genetics and evolution
 .
San Diego CA
:
Academic Press
.
Hawkins
AFA
1999
.
Altitudinal and latitudinal distribution of east Malagasy forest bird communities
.
Journal of Biogeography
 
26
:
447
458
.
Hayati
A
,
Wickneswari
R
,
Maizura
I
,
Rajanaidu
N
2004
.
Genetic diversity of oil palm (Elais guineensis Jacq) germplasm collections from Africa; implications for improvement and conservation of genetic resources
.
Theoretical and Applied Genetics
 
108
:
1274
1284
.
Henderson
SA
,
Billote
N
,
Pintaud
J-C
2006
.
Genetic isolation of Cape Verde Island Phoenix atlantica (Arecaceae) revealed by microsatellite markers
.
Conservation Genetics
 
7
:
213
223
.
Hill
JL
,
Curran
PJ
2003
.
Area, shape and isolation of tropical forest fragments; effects of tree species diversity and implications for conservation
.
Journal of Biogeography
 
30
:
1391
1403
.
Honnay
O
,
Jacquemyn
H
,
Bossurt
B
,
Hermy
M
2005
.
Forest fragmentation effects on patch occupancy and population viability of herbaceous plant species
.
New Phytologist
 
166
:
723
736
.
Hubbell
SP
2001
.
Unified neutral theory of biodiversity and biogeography
 .
Princeton, NJ
:
Princeton University Press
.
IUCN
.
2004
.
Red list of endangered species
 .
Gland
:
IUCN
.
Jang
C-G
,
Mullner
AN
,
Greimler
J
2005
.
Conflicting patterns of genetic and morphological variation in European Gentianella section Gentianella
.
Botanical Journal of the Linnean Society
 
148
:
175
187
.
Kirkpatrick
J
1994
.
A continent transformed: human impact on the natural vegetation of Australia
 .
Oxford
:
Oxford University Press
.
Kull
CA
2002
.
Empowering pyromaniacs in Madagascar; ideology and legitimacy in community-based natural resource management
.
Development and Change
 
33
:
57
78
.
Listabarth
C
1996
.
Pollination of Bactris by Phyllotrox and Epurea. Implications of the palm breeding beetles on pollination at the community level
.
Biotropica
 
28
:
69
81
.
Loo
AHB
,
Dransfield
J
,
Chase
MW
,
Baker
WJ
2006
.
Low-copy nuclear DNA, phylogeny and the evolution of dichogamy in the betel nut palms and their relatives (Arecinae; Arecaceae)
.
Molecular Phylogenetics and Evolution
 
39
:
598
618
.
Loo
AHB
,
Tan
HTW
,
Kumar
PP
,
Saw
LG
1999
.
Population analysis of Licuala glabra Griff. var glabra (Palmae) using RAPD profiling
.
Annals of Botany
 
84
:
421
427
.
Lopes
CR
,
Dos Reis
SF
,
Ferriera
MA
,
Mortzsohn
MC
1992
.
Genetics of the genus Acrocomia (Palmae) III Microgeographical genetic variability in isozyme frequencies
.
Journal of Genetics and Breeding
 
46
:
9
14
.
Maizura
I
,
Rajanaidu
N
,
Zakri
AH
,
Cheah
SC
2006
.
Assessment of genetic diversity in oil palm (Elais guineensis Jacq.) using Restriction Fragment Length Polymorphism (RFLP)
.
Genetic Resources and Crop Evolution
 
53
:
187
195
.
McClenaghan
LR
Jr
,
Beauchamp
AC
1986
.
Low genetic differentiation among populations of the California fan palm (Washingtonia filifera)
.
Evolution
 
40
:
315
322
.
McConnell
WJ
,
Sweeney
SP
2005
.
Challenges of forest governance in Madagascar
.
Geographical Journal
 
171
:
223
238
.
Meerow
AW
,
Wisser
RJ
,
Brown
MS
,
Kuhn
DN
,
Schnell
RJ
,
Broschat
TK
2003
.
Analysis of genetic diversity and population structure within Florida coconut (Cocos nucifera L.) germplasm using microsatellite DNA, with special emphasis on the Fiji Dwarf cultivar
.
Theoretical and Applied Genetics
 
106
:
715
726
.
Meyers
N
,
Mittermeier
RA
,
Mittermeier
CG
,
Da Fonseca
GAB
,
Kents
J
2000
.
Biodiversity hot spots for conservation priorities
.
Nature
 
403
:
853
858
.
Murray
BR
,
Lepschi
B
2004
.
Are locally rare species abundant elsewhere in their geographical range?
Austral Ecology
 
29
:
287
293
.
Orloci
L
1978
.
Multivariate analysis in vegetation research
 .
The Hague
:
Dr W. Junk
.
Pareliussen
I
,
Gunilla
E
,
Olsson
A
,
Armbruster
WS
2006
.
Factors limiting the survival of native tree seedlings used in conservation efforts at the edges of forest fragments in upland Madagascar
.
Restoration Ecology
 
14
:
196
203
.
Peakall
R
,
Smouse
PE
2005
.
GenAlEx V6: Genetic Analysis in Excel. Population genetic software for teaching and research.
 
Canberra
:
Australian National University
.
Perera
L
,
Russell
JR
,
Provan
J
,
Powell
W
1999
.
Identification and characterisation of microsatellite loci in coconut and analysis of coconut populations in Sri Lanka
.
Molecular Ecology
 
8
:
335
346
.
Peres
CA
1994
.
Composition, density and fruiting phenology of arborescent palms in an Amazonian Terre Firme forest
.
Biotropica
 
26
:
285
294
.
Price
OF
2004
.
Indirect evidence that frugivorous birds track fluctuating fruit resources among rainforest patches in the Northern Territory, Australia
.
Austral Ecology
 
29
:
137
144
.
Rakotoarinivo
M
2005
.
Etude Demographique de Beccariophoenix madagascariensis Jum. Et H.Perrir. (Palmae) A Sainte Luce (Fort-Dauphin) en vue de la conservation de l'espece
 .
MSc thesis
 , University of Antananarivo, Madagascar.
Rakotoarinivo
M
,
Ranarivelo
T
,
Dransfield
J
2007
.
A new species of Beccariophoenix from the High Plateau of Madagascar
.
Palms
 
51
:
62
75
.
Ramanamanjato
J-B
,
Ganzhorn
JU
2001
.
Effects of forest fragmentation introduced Rattus rattus and the role of exotic tree plantations and secondary vegetation for the conservation of an endemic rodent and a small lemur in littoral forests of southeastern Madagascar
.
Animal Conservation
 
4
:
175
183
.
Scariot
AO
,
Lleras
E
1991
.
Reproductive biology of the palm Acrocomia aculeata in Central Brazil
.
Biotropica
 
23
:
12
22
.
Schönswetter
P
,
Tribsch
A
,
Schneeweis
GM
,
Niklfeld
H
2003
.
Disjunctions in relict alpine plants: phylography of Androsace brevis and A.wulfeniana (Primulaceae)
.
Botanical Journal of the Linnean Society
 
141
:
437
446
.
Shapcott
A
1998a
.
The patterns of genetic diversity in Carpentaria acuminata (Arecaceae), and rainforest history in northern Australia
.
Molecular Ecology
 
7
:
833
847
.
Shapcott
A
1998b
.
The genetics of Ptychosperma bleeseri a rare palm from the Northern Territory, Australia
.
Biological Conservation
 
85
:
203
209
.
Shapcott
A
1999
.
Comparison of the population genetics and densities of five Pinanga palm species at Kuala Belalong, Brunei
.
Molecular Ecology
 
8
:
1641
1654
.
Shapcott
A
2000
.
Conservation and genetics in the fragmented monsoon rainforest in the Northern Territory, Australia: a case study of three frugivore dispersed species
.
Australian Journal of Botany
 
48
:
397
407
.
Simsik
MJ
2004
.
The political ecology of biodiversity conservation on the Malagasy Highlands
.
Geojournal
 
58
:
233
242
.
Soons
MB
,
Ozinga
WA
2005
.
How important is long-distance seed dispersal for the regional survival of plant species?
Diversity and Distributions
 
11
:
165
172
.
Terborgh
JR
1992
.
Maintenance of diversity in tropical forests
.
Biotropica
 
24
:
283
292
.
Teulat
B
,
Aldam
C
,
Trehin
R
,
Lebrun
P
,
Barker
JHA
,
Arnold
GM
,
Karp
A
,
Baudouin
L
,
Rognon
F
2000
.
An analysis of genetic diversity in coconut (Cocos nucifera) populations from across the geographic range using sequence-tagged microsatellites (SSR's) and AFLPs
.
Theoretical and Applied Genetics
 
100
:
764
771
.
Tomlinson
PB
2006
.
The uniqueness of palms
.
Botanical Journal of the Linnean Society
 
151
:
5
14
.
Tracey
JG
1981
.
Australia's rainforest: where are the rare plants and how do we keep them?
In:
Synge
H
ed.
The biological aspects of rare plant conservation
 .
London
:
John Wiley
,
165
178
.
Tremetsberger
K
,
Talavera
S
,
Stuessy
TF
,
Ortiz
MA
,
Weiss-Schneeweis
H
,
Kadlec
G
2004
.
Relationship of Hypochaeris salzmanniana (Asteraceae, Lactuceae) an endangered species of the Iberian Peninsula to H. radicata and H. glabra and biogeographical implications
.
Botanical Journal of the Linnean Society
 
146
:
79
95
.
Turner
IM
1996
.
Species loss in fragments of tropical forest: a review of the evidence
.
Journal of Applied Ecology
 
33
:
200
209
.
Whelan
RJ
,
Rodgerson
L
,
Dickman
CR
,
Sutherland
EF
2002
.
Critical life cycles of plants and animals: developing a process-based understanding of population changes in fire-prone landscapes
. In:
Bradstock
RA
,
Williams
JG
,
Gill
AM
eds.
Flammable Australia: the fire regimes and biodiversity of a continent
 .
Cambridge
:
Cambridge University Press
,
94
124
.
Wickneswari
R
,
Siti Salwana
H
,
Norwati
M
,
Nur Supardi
MN
,
Minuddin
M
2002
.
Genetic diversity in potential seed sources of Calamus manna Miq. In peninsular Malaysia
.
Malaysian Applied Biology
 
31
:
49
58
.
Wright
S
1965
.
The interpretation of population structure by F-statistics with special regard to systems of mating
.
Evolution
 
19
:
395
420
.
Young
AG
,
Clarke
GM
2000
.
Genetics, demography, and viability of fragmented populations.
 
Cambridge
:
Cambridge University Press
.
Zona
S
,
Henderson
A
1989
.
A review of animal mediated seed dispersal of palms
.
Selbyana
 
11
:
6
21
.