Conservation Genetics

, Volume 10, Issue 2, pp 317–327

Low genetic diversity and recovery implications of the vulnerable Bankoualé Palm Livistona carinensis (Arecaceae), from North-eastern Africa and the Southern Arabian Peninsula


    • Faculty of Science Health and EducationUniversity of the Sunshine Coast
  • J. L. Dowe
    • Australian Centre for Tropical Freshwater ResearchJames Cook University
  • H. Ford
Research Article

DOI: 10.1007/s10592-008-9582-5

Cite this article as:
Shapcott, A., Dowe, J.L. & Ford, H. Conserv Genet (2009) 10: 317. doi:10.1007/s10592-008-9582-5


The Bankoualé Palm, Livistona carinensis is the only known species of Livistona occurring in Africa and is currently classified as vulnerable (IUCN 2004). This extreme outlier species of the genus is restricted to Yemen, Somalia and Djibouti, where all populations are in rapid decline. In Djibouti the palm is confined to three valley systems within the upland plateau of the Goda Massif. This study used microsatellite markers to investigate the genetic diversity and relationships within the species. At the species level L. carinensis contained very low genetic diversity. Most variation was due to the variation between the samples from Yemen and Somalia compared with those in Djibouti. The Djibouti populations were almost monomorphic across the nine loci tested. Interestingly, and despite the small sample sizes, the individuals from botanic gardens collections of the Yemen and Somalia populations were more genetically diverse than the Djibouti populations. This study indicates that the populations in Yemen and Somalia are highly significant for the conservation of the species genetic diversity. Given the lack of genetic diversity both within and among L. carinensis populations in Djibouti, plants could be cultivated for in-situ population enhancement from any seed that is available from within Djibouti with no significant genetic impacts of provenance mismatch. Clearly the populations from Djibouti, Somalia and Yemen are different genetic provenances raising some issues for the conservation and recovery of L. carinensis.


Species recoveryConservation geneticsRare speciesMicrosatellitesArecaceae


The Bankoualé Palm, Livistona carinensis (Chiov.) J. Dransf and N. W. Uhl, is the only species of Livistona (36 spp.) occurring in Africa (Uhl and Dransfield 1987; Dowe 2001) and is an extreme outlier in the distribution of the genus (Fig. 1). The Bankoualé palm has a ‘vulnerable’ conservation status in the IUCN Red List of Threatened Species (Johnson 1998; IUCN 2004), and is restricted to the Yemen, Somalia and Djibouti (Welch and Welch 1998). The rarity and vulnerability of L. carinensis was first recognised by Moore (1971). Population surveys and conservation assessments were undertaken by Langlois (1976), Moore (1977), in Somalia by Barbier (1985) and Thulin (1995), in Yemen by Bazara’a et al. (1990) and in Djibouti by Welch and Welch (1998), Ford and Bealy (2004) and Ford et al. (2008). The palm has undergone a significant and well-documented decline in its natural populations in Djibouti, Somalia and Yemen (Welch and Welch 1998; Ford and Bealy 2004; Ford et al. 2008) and a re-evaluation of the species conservation status is necessary.
Fig. 1

Global distribution of the genus Livistona with the location of the distribution of L. carinensis in Djibouti, Yemen and Somalia circled

Coates and Dixon (2007) advocate that plant conservation biology be an integration of population biology using genetics and demography to make conservation decisions to inform ex situ conservation. Studies have shown that genetic variation and fitness in plants both increase with increasing population size (Leimu et al. 2006). Rare species often have lower genetic diversity and reproductive success compared with more abundant and widespread species (Ellstrand and Elam 1993; Byer and Waller 1999; Gitzendanner and Soltis 2000; Frankham et al. 2002). Rare species are predicted to be even more susceptible to the extinction vortex resulting from decreased population sizes than common species (Leimu et al. 2006). Increased spatial isolation is predicted to lead to increased genetic differentiation among populations (Young and Clarke 2000; Ouborg et al. 2006). The effective long-term management and conservation of rare and threatened plants requires an understanding of population biology, and the way in which genetic variation is partitioned within and among populations of the managed species (Godt et al. 1994; Coates and Hopper 2000; Ouborg et al. 2006). However, few studies on rare or threatened palms have been published (e.g. Dowe et al. 1997; Shapcott 1998a, 1999; González-Pérez et al. 2004; Shapcott et al. 2007).

Increasingly managers of threatened species are considering population reintroductions and enhancements in order to ensure species viability (e.g. McKay et al. 2005). Population enhancement can have either beneficial or harmful impacts on the viability and fitness of populations and needs to be undertaken with consideration of the genetic impacts (Fischer and Matthies 1997; Montalvo and Ellstrand 2001; McKay et al. 2005; Jones and Monaco 2007). Many plant species are locally adapted over a range of geographic scales (McKay et al. 2005). O’Brian et al. (2007) found variation among provenances and suggested the importance of an environmental match of sources for restoration projects. Gustafson et al. (2004) also stressed the use of local sources for translocations or restoration activities. However, Keller et al. (2000) found evidence of reduced fitness from introducing plants from widely distant geographic regions.

In palms, genetic studies have been used to advise conservation priorities and management issues for several rare species (Dowe et al. 1997; Shapcott 1998a; Shapcott et al. 2007). Eguiarte et al. (1992) were also able to use the results of genetic studies to inform conservation management requirements for Astrocaryum mexicanum, while Cardoso et al. (2000) used AFLP studies of Euterpe edulis to identify conservation priorities for wild populations and provenance for commercial cultivation of this species. However, the lack of genetic knowledge, especially the levels and distribution of genetic diversity in Phoenix canariensis initially stalled the implementation of conservation management strategies for that species (González-Pérez et al. 2004).

In addition to the reduction of immediate population loss, species recovery programs may be aimed at increasing or maintaining the size of existing populations, increasing the genetic diversity of populations, creating or reintroducing new populations or enhancing connectivity among populations. An understanding of existing genetic diversity and differentiation among populations is therefore a significant component of recovery programmes to ensure success (McKay et al. 2005; Jones and Monaco 2007).

Study species

Diversity within Livistona is greatest in Australia and New Guinea, and secondarily in south-east Asia (Fig. 1). The closest extant Livistona species to L. carinensis occur in north-eastern India some 4,700 km to the east (Fig. 1). These species also appear to be phylogenetically closest to L. carinensis (Y. Isagi et al. in prep.). This exceptional disjunction is of considerable biogeographical and evolutionary interest, as it raises questions, of origin and dispersal that are difficult to reconcile (Uhl and Dransfield 1987). Fossils of Livistona affinity appear in India dated to the mid Eocene (Prakash and Ambwani 1980), although specimens of Livistona affinity have been excavated from European and North American sites dated to the late Cretaceous (Read and Hickey 1972) and the Eocene (Chandler 1978). Fossil fruits with affinity to Livistona have been reported from early Palaeocene deposits of Egypt (Gregor and Hagn 1982; Pan et al. 2006). Based on fossil pollen records, palms appear to have been much more widespread across Africa than they are at present (Jacobs 2004). The environmental history of north-east Africa and Arabia indicates that rapid changes occurred during the early Tertiary (Bayer 1984). Some areas, such as southern Yemen and Somalia, have continually remained above sea-level since that time, and include some of the oldest landforms on earth (Pilger and Rösler 1976). In the region, climate change during the Quaternary produced conditions tending to aridity, with a high rate of plant extinction and the creation of exceptionally high numbers of relictual species (Lovett 1993). It is possible that L. carinensis is thus a ‘survivor’ from an early establishment and has been able to persist through periods of climate change.

Ecology and natural history

Livistona carinensis most often occupies the flow lines of intermittent or permanent streams either in wadis or by oases on sandy soil (Bazara’a et al. 1990; Ford and Bealy 2004). The altitudinal ranges of the known localities are 210–875 m in Somalia, 200–450 m in Yemen and 600–975 m in Djibouti. The locations all fall within a semi-arid/arid climatic zone within 11°N–14°N latitude, and climate data for nearby locations (Djibouti, Djibouti; Aden,Yemen; Hargeisa, Somalia) indicate average annual rainfall of between 46 and 426 mm, with a strongly seasonal occurrence (Vose et al. 1992).

In Djibouti L. carinensis exists in small dense populations distributed amongst three valley (wadi) systems within the upland plateau of the Goda Massif (Fig. 2). The main wadi populations contain only mature individuals, and lack regeneration. However, in populations near springs there is some regeneration, but these populations are also in decline (Ford and Bealy 2004; Ford et al. 2008).
Fig. 2

The relative locations of the Livistona carinensis populations in Djibouti. Sampled populations indicated by ▲ unsampled populations by △. Population codes of sampled populations are indicated. The only allelic diversity in Djibouti populations was found at the Eg254 locus pie charts indicate the relative frequency of the alternative allele (176 bp) compared with the common allele (180 bp) and the location of the population in which it was found

Flowering has been observed in all populations, but the pollination vector is presently not known. Plants have been reported to produce abundant fruit, and birds have been observed consuming these. Birds have also been observed visiting flowering palms (Ford et al. 2008; Despite large numbers of seedlings being occasionally produced, there is a high mortality rate (Ford et al. 2008; There is some protection from grazing given by stands of Phoenix reclinata, where the two palm species occur together. Primary grazers such as fat-tailed sheep, goats, cattle, camels and donkeys are unable to penetrate the Phoenix thickets. Seedling survival is greatest in cultivated areas from where grazing stock are excluded. Conversely, palm stands have been extirpated to create gardens particularly near spring sites (Ford et al. 2008; Additionally, wadi systems are prone to irregular dynamic flood events which may dislodge seedlings and juveniles, floods may thus be implicated in seed dispersal.


Efforts to conserve L. carinensis have been initiated and this study forms an essential component of an international team effort to assist with the conservation of this species. This study aimed to determine how genetically variable are the Djibouti populations? How much genetic differentiation is there among the Djibouti populations and how distinctive are they compared with representatives from Yemen and Somali. We aimed to incorporate the demographic and ecological observations with the genetic results to evaluate the species and population viability and identify what key actions could increase the species viability. The study also aimed to use the results to identify strategies or approaches that need to be incorporated into any future reintroduction, population enhancement or ex-situ conservation programmes for this species.


Plant collections

A survey of the extant populations of the Bankoualé palm, Livistona carinensis in Djibouti was undertaken and sampled for genetic analysis in 2006 by H. Ford and colleagues. The palm is distributed amongst three valley systems; Randa containing four populations; Bankoualé with seven populations and Ditilou with two populations (Ford et al. 2008). As seedlings and rosette stage plants were observed to be under strong grazing pressure only juvenile and adult plants from populations within each of the three valley systems were sampled. Forty-two individuals from seven populations were thus sampled and locations were recorded as GPS co-ordinates. If necessary, trees to be sampled were climbed to reach the leaf material and a sample cut (approximately 5 × 5 cm) from the leaf ( Individuals from a population from the Yemen have been successfully grown from seed in the Townsville Palmetum, Queensland, Australia, and five plants were thus sampled for this study. Similarly, two L. carinensis plants originally collected from Somalia as seed and now growing in the Fairchild Tropical Gardens, Florida USA (73398A; 89514B) were sampled for this study. This enabled a small number of individuals from the Yemen and Somalia to be compared with the Djibouti samples. Samples were cut into smaller pieces and were placed in silica gel to dry after collection and sent to the University of the Sunshine Coast for processing.

Laboratory methods

DNA was extracted by first cutting up approximately 0.3 g of leaf material, freezing in liquid nitrogen and grinding using an automated bead shaker (Retsch MM200 grinding mill; Haan, Germany) then extracting and purifying the DNA using the DNeasy Plant Mini-kit (QIAGEN) and their associated methodology. The DNA extracted was quantified by comparing the intensity of the test sample using quantification standards on agarose gels (Lambda HindIII; Axygen Biosciences). The DNA extracted generally fell within the 50–100 ng/μl concentration range. Following this 0.5 μl of DNA was used in reactions to ensure at least 25 ng DNA per reaction as indicated in the original papers. A selection of 24 microsatellite primers developed for several palm species, shown to exhibit cross generic transferability, were trialled using the methods in the original papers (Perera et al. 1999; Billotte et al. 2001, 2004a; Gaiotto et al. 2001; Montufar et al. 2007). This set included five primer pairs developed for Phoenix dactylifera (Billotte et al. 2004a), five primer pairs developed for Elaeis guineensis (Billotte et al. 2001), five primer pairs developed for Bactris gasipaes (Billotte et al. 2004b), five primer pairs developed for Cocos nucifera (Perera et al. 1999), three primer pairs developed for Oenocarpus bataua (Montufar et al. 2007) and two primer pairs developed for Euterpe edulis (Gaiotto et al. 2001). Primer combinations were tested for transferability by separating the products on a 1.5% agarose gel containing ethidium bromide comparing against a 100 bp DNA size standard (Axygen Biosciences) to check for selective amplification of product within the expected fragment size range. These were visualised under UV light. To reduce the possibility of selecting for artefacts (Selkoe and Toonen 2006) only primers which gave single bands close to the expected size range given in the original publications were considered. Blank controls were used to test for random PCR amplification products. While samples of all the original test species were not available to confirm that the PCR reactions were able to be repeated in our laboratory, during the initial screening for primers that may transfer to Livistona carinensis we also ran a sample of Phoenix dactylifera obtained from a cultivated specimen and samples of Lepidorrhachis mooreana and Hedyscepe canterburyana which were being concurrently trialled for analysis. The Phoenix primers were successfully amplified in the Phoenix sample with products within the expected size ranges and different primers were successfully amplified producing a band in the expected size range, in different species. Several primers were successfully amplified in Bankoualé palm and these were further optimised to reduce additional random PCR amplification products for this species. Nine microsatellite loci primers and conditions were thus used for this study: three primer loci developed by Billotte et al.(2004a) for Phoenix dactylifera mPdCIR015, mPDCIR016, mPDCIR090; three developed for Elaeis guineensis by Billotte et al. (2001) mEgCIR0254, mEgCIR0353, mEgCIR0476 and two developed by Billotte et al. (2004b) for Bactris gasipaes mBgCIR042, mBgCIR062. The final PCR mix used was made in 10 μl with 0.2 mM dNTP’s, 0.5 U Taq polymerase (Taq F1, Fisher Biotech), 1.5 mM MgCl2, 1× PCR buffer (Fisher Biotech). The forward primer was end tagged to enable a fluorescently labelled (Hex) tag to be added in the PCR reaction following the Universal fluorescent labelling method of Shimizu et al. (2002). In the PCR mix 0.1 μM of forward primer, 0.8 μM reverse primer and 0.05 μM of tag end label was used. Amplification was carried out on an Eppendorf Mastercycler (Hamburg), the PCR cycling conditions were as per Billotte et al. (2004a): Denaturation 95°C for 1 min; 35 cycles of 94°C for 30 s, 52°C for 1 min, 72°C for 2 min; Final elongation step at 72°C for 8 min. In addition to these, one primer loci originally developed for Oenocarpus bataua by Montufar et al. (2007) Ob15 was also successfully amplified in L. carinensis. The PCR mix was modified in that only 0.5 Units of Taq and 1.5 mM of MgCl2 were used and PCR cycling conditions were as per the original publication (Montufar et al. 2007).

The resultant fragments were then run on a fragment analyser in this case a Gelscan 2000 (Corbett Research). PCR products (5 μl) were diluted with equal volumes of loading dye and denatured at 95°C for approximately 15 min prior to separation on an acrylamide gel made with TBE buffer. A Gene Scan ROX 400HD size standard (Applied Biosystems) used to calibrate the gels was inserted at four points in a 24 well mould, across the gel to enable uneven migration in the gel to be corrected for. The CR Gelscan program v 7.2.7 (Corbett Research 2001) was then used to run and visualise the results. The resultant bands that were scored were within the size range reported in the original paper or differed in multiples of a minimum of two base pairs. The bands behaved in the manner expected of microsatellites with stutter bands sometimes present. The bands were scored using the oneD Scan v 2.05 program (Scanalytics Inc).

Statistical analysis

The scored fragments were assigned to alleles at each locus based on base pair size using the oneD Scan v 2.05 program (Scanalytics Inc). The multilocus genotype of each individual then documented and compared between individuals and populations. The GenAlEx V6 program (Peakall and Smouse 2005) was used to calculate population and species level allelic frequencies, and measures of genetic diversity (A, P, He, Ho) were calculated. Measures of gene flow (Nm) and partitioning of genetic diversity among populations (FST), within populations (FIS), and across the whole species population (FIT), were also calculated using Wrights F statistics. The significance of genetic partitioning among populations within the species was statistically tested by AMOVA using the GenAlEx V6 program (Peakall and Smouse 2005). Nei’s genetic distance was calculated between pairs of populations and between all individuals sampled. The genetic distance between populations was used in a hierarchical cluster analysis to construct a dendrogram indicating the relationships among populations (using the Primer 5 statistical package PRIMER-E Ltd). The genetic relationships among all individuals sampled was used in a Principal Coordinates Analysis (PCoA) to assess the distinctiveness of populations where genetic distance measures were standardised and 999 bootstrap permutations used, to determine if there were distinct genetic groupings among populations, valley systems or between countries using the GenAlEx V6 program (Peakall and Smouse 2005). Allelic frequencies at individual loci were plotted as pie charts and overlaid onto maps to investigate specific geographic relationships among alleles.


At the species level Livistona carinensis contained low genetic diversity with 16 alleles recorded from across the nine loci tested (Table 1). Of the variation found in the species, a significant 36% was distributed among populations with moderate genetic partitioning among populations as measured by FST (Table 1). Most of this variation among populations was due to the variation between the samples from Yemen and Somalia compared with those in Djibouti (Fig. 3). Perhaps the most interesting result was that despite very small sample sizes, the individuals representing the Yemen (LC13), and Somalia (LC14) were more genetically diverse than Djibouti populations which were almost monomorphic (Fig. 3; Table 2).
Fig. 3

The genetic relationships among Livistona carinensis populations in Djibouti (LC 1, 2, 3, 4, 5, 7, 9) in relation to populations in Somalia (LC13) and Yemen (LC14) as indicated by a cluster analysis of population genetic distances across 9 microsatellite loci (a). The genetic relationship between individuals sampled is indicated in the Principal Coordinates analysis output where symbols (see legend) indicate the population of each individual (b)

Table 1

Summary of species level diversity and partitioning of variation across nine microsatellite loci and among nine Livistona carinensis populations














Variation among populations


Variation within populations






Wrights F statistics are given (FIS, FIT, FST), past gene flow among populations (Nm), the mean number of alleles per locus within the species (As). In addition the results of an AMOVA analysis are given where the % of variation within and between populations is given and the test statistic PhiPT and its significance probability

Table 2

Summary of genetic diversity measures for Livistona carinensis


Location (valley)




P (%)





LC1 (5)

Bankoualé Cascade (2)









LC2 (12)

Satabou (2)









LC3 (2)

Wer (3)









LC4 (8,9,10)

Bankoualé Side (2)









LC5 (3)

Disay (2)









LC7 (4)

Aggorogouba (2)










Randa (1)







































N, number of samples; A, mean number alleles per locus; Ae, mean number effective alleles per locus; P, percentage of loci polymorphic; I, information index; Ho, observed heterozygosity; He, expected heterozygosity; Ap, number of private alleles values are given for each population. The population codes (corresponding site numbers oné), location and valley system (1, 2, 3) is indicated

While there were only a small number of individuals sampled for this study from each population, the species was sampled from seven populations across its distribution in Djibouti. This study found extremely low genetic diversity in the L. carinensis samples analysed (Table 2). In all seven populations from Djibouti all samples were monomorphic for the same allele at each of the nine loci tested. These results indicate that there is no significant genetic difference between the populations within each of the three valley systems or within each system. There was a single variant individual at one locus (Eg254) found within the largest sampled population (LC1; 23 samples; Fig. 2) that was homozygous for an alternative allele at this locus. This private allele was not found in any other population within the species sampled (Table 2; Fig. 2). This population was also one of the largest known for the species in Djibouti and in the centre of the species distribution within Djibouti (Ford et al. 2008). This makes this population and individual distinctive within the Djibouti populations as well as the species (Fig. 2). Given not every plant was sampled, the fact that some diversity was found where the population and sample size was larger is hopeful that there may be some low level of variation present within the remaining plants within the Djibouti populations.

Whilst the samples from Yemen (LC14) and Somalia (LC13) shared the same monomorphic allele at seven loci, at two loci (Ob15 and Pd15) they also contained distinctive alternative alleles (Fig. 4). The Yemen samples had two private alleles present in this small sample (Table 2; Fig. 4). Despite small sample sizes heterozygotes were recorded at the variable loci in both the Yemen and Somali samples (Table 2). Given the extremely low variation found within and among the Djibouti populations, in contrast to the variation found from just a few samples from the populations in Somalia and Yemen, there is some potential for significant variation in the remaining populations in these countries and especially the population in Yemen.
Fig. 4

Map showing the relative locations of Djibouti, Somalia and Yemen with Pie chart overlay indicating the allelic composition of the Djibouti Livistona carinensis populations at the Ob15 and Pd15 microsatellite loci in comparison with the Yemen and Somalia allelic compositions. Where Ob15 allele 1 (246 bp); allele 2 (250 bp); allele 3 (248 bp); allele 4 (254 bp) and Pd15 allele 1 (162 bp) allele 2 (160 bp) are indicated


Genetic diversity

This study found extremely low genetic diversity within Livistona carinensis across its range. Particularly there was almost no genetic variation found within or among any of the Djibouti populations sampled. Low genetic diversity has been reported in several other endangered palm species. For example, Dowe et al. (1997) reported very low genetic diversity within the remaining wild populations of the endangered palm Carpoxylon macrospermum from Vanuatu. Extremely low genetic diversity both within and among populations, similar to the levels documented in this study, was also documented for the regionally endangered palm Ptychospermamacarthurii (as P. bleeseri Shapcott 1998a).

Genetic diversity within populations is predicted to be lost over generations in small populations due to genetic drift as well as sampling effects during population crashes (Ellstrand and Elam 1993; Ouborg et al. 2006). Correlations between decreasing population size and diversity have been documented in a range of species (e.g. Gaudeul et al. 2000; Culley and Grub 2003; Hensen and Oberprieler 2005; Armstrong and DeLange 2005; Godt et al. 2005; Leimu et al. 2006). In particular it has been reported that rare alleles are lost from small populations when compared to retention in larger populations (White et al. 1999; Ohara et al. 2006).

The L. carinensis population sizes in Djibouti are typically small and they have been shown to be rapidly declining (Ford et al. 2008). The only allelic diversity was found within the largest population studied which also had the largest number of plants sampled. The near absent genetic diversity within and among the Djibouti populations contrasts with the samples from Somalia and Yemen. Despite small sample sizes these few samples contained more genetic diversity than found in all the Djibouti samples. The records indicate that the L. carinensis population sizes in both the Yemen and Somalia were formerly much larger than those recorded from Djibouti (Bazara’a et al. 1990; Welch and Welch 1998; Ford et al. 2008). The genetic results would support the historical presence of larger population sizes that contained a larger genetic allelic pool in those countries. The critically endangered palm from Madagascar Beccariophoenix madagascariensis appears to have recently undergone population reductions and much higher levels of genetic variation than expected for such small population sizes were found in that species (Shapcott et al. 2007). The results suggest that the remaining populations in Yemen and Somalia are likely to contain higher levels of genetic diversity than found in Djibouti populations.

The Goda Massif of Djibouti is deeply dissected, and the populations of L. carinensis there are fragmented and subpopulations are somewhat isolated from each other (Ford et al. 2008). Whilst the sample sizes were small their genetic uniformity indicates that the remaining populations in Djibouti contain overall low diversity. However, they are not randomly different from each other, as one would have expected if they had more recently lost genetic diversity because of drift. The similarity among all populations in Djibouti seems more indicative of a species that has been dispersed to the area and subsequently expanded from a low genetic base. This situation could be the result of either the contraction of a formerly much larger population to a single population and subsequent expansion, or that the populations expanded from a small genetically depauperate founder population.

The whole L. carinensis distribution is disjunct from the distribution of the genus (Fig. 1). The geological records indicate the centre of the L. carinensis distribution has remained above sea level (Pilger and Rösler 1976) and it seems possible that the species has a relict distribution. The low diversity found within the species generally would support this, as it would be expected that the species would have progressively lost diversity due to drift over time. However, while the sample sizes for Somalia and Yemen populations are very small, the higher levels of genetic variation found in these regions suggest they are more likely to have been the core of the species distribution historically. The fact that there is regional differentiation documented from this limited sample is also indicative of genetic differentiation between the three regions. González-Pérez et al. (2004) found Phoenix canariensis contained a subset of the alleles found in the more widespread congener P. dactylifera and suggested that P. canariensis is recently evolved from a common ancestor to P. dactylifera. Likewise Henderson et al. (2006) suggested that geographic isolation lead to the divergence of P. atlantica from P. dactylifera.

Gene flow

The L. carinensis populations in Djibouti are distributed in at least three catchments. However, this study has shown that there is no genetic differentiation among these catchments. This also indicates that there has been significant gene flow. Together with the observation of bird visitations (pers. obs. H. Ford), this result suggests that in addition to seed dispersal downstream during flood events, seed dispersal by birds is highly likely to have occurred between catchments. Gaiotto et al. (2003) found high levels of gene flow in the heart-of-palm species Euterpe edulis and that seeds had been dispersed by birds and become established up to 22 km away. Shapcott et al. (2007) found evidence of considerable gene flow between Beccariophoenix madagascariensis populations that were within 3 km of each other and hypothesized this was due to seed dispersal. Other palms, such as Carpentaria acuminata, have been shown to have effective seed dispersal by birds and other mobile frugivores of many kilometres which enabled small populations to remain viable within a naturally fragmented landscape (Shapcott 1998b). Galetti et al. (2006) reported that many species were now documented to have been affected by loss of seed dispersers resulting from forest fragmentation and hunting.

Conversely, Adin et al. (2004) found that in the peach palm Bactris gasipaes bird dispersal was only local (100–200 m) and instead found that gene flow by human trading was likely to have lead to long distance dispersal and reduced genetic differentiation between populations from two adjacent rivers. Dowe et al. (1997) found that many of the Carpoxylon macrospermum palm plants remaining on the Vanuatu islands had been cultivated by local villagers as they had some cultural significance. However, these appear to have come from very limited genetic stock as little or no variation was found in these samples.

Dispersal by humans of palms in this region could explain the higher altitude location and low genetic diversity of the Djibouti populations in comparison to Somalia and Yemen if humans had a history of culturally utilising this species. However, this species does not appear to have cultural significance or utilitarian uses to the present inhabitants reducing this as a potential explanation (H. Ford pers. obs.).

Population viability

The reason why there is such a disproportionately large number of endangered palms is possibly explained by their unique morphology, this being the possession of a single apical meristem (Bernal and Galeano 2006), a situation that occurs in about half of all palms (Henderson 2002). Single stemmed palms are killed if their meristem is damaged and hence they are particularly sensitive to grazing and harvesting (Bernal and Galeano 2006). Grazing of seedlings has been documented in many palm species to have lead to endangerment or to be preventing population regeneration (Henderson et al. 2006). For example, Liddle et al. (2006) found that introduced animals had reduced the survival of Ptychosperma macarthurii (as P. bleeseri) seedlings. They found that in sites where introduced animals had been excluded the seedlings had survived and populations had a more even size structure. In Hawaii, Chapin et al. (2007) found that many species of Pritchardia were endangered and that the wild populations had poor to absent regeneration. Their study recorded that seed predation, goat grazing, pig damage and human harvesting were contributing to the lack of regeneration (Chapin et al. 2007). Maunder et al. (2002) reported that high levels of seed predation of endangered Mascarene Island palms were preventing natural regeneration, and that active enhancement programs were needed to overcome this problem. Where there was grazing, Ford et al. (2008) observed a lack of recruitment of seedlings to larger class sizes in L. carinensis. But where populations were protected from grazing other class sizes were present. As L. carinensis is a single stemmed palm, there is a considerable threat from grazing, a situation that can otherwise potentially be managed.

Ford et al. (2008) also suggested that the frequency of extreme flooding events was likely to have consequences for demographic viability of L. carinensis populations in Djibouti as plants are washed away during flood events. Thus any active recovery program for L. carinensis in situ would need to consider both protection from grazing as well as targeting sites less likely to be flood prone for active programs.


A multi-disciplined approach is the most effective method of conserving endangered species. For example, in a case study of threatened Pritchardia species in Hawaii, Chapin et al. (2004) advocate conservation of existing populations, establishing ex situ populations and reintroducing plants into the wild. On the Mascarene Islands seven palm taxa have populations of 100 or fewer mature individuals in the wild and an active recovery involving both in-situ and ex-situ approaches has been recommended to prevent extinction (Maunder et al. 2002). Dowe et al. (1997) also indicated that recovery of Carpoxylon macrospermum required active conservation programs to establish new populations containing the range of genetic diversity found in the species.

In Djibouti, in situ demographic enhancement is one of the options to protect L. carinensis from continuing decline caused by grazing. The cultivation of populations of L. carinensis in areas protected from grazing will enhance the species survival demographically. This may take the form of fencing, supplementing existing populations with seedlings propagated ex-situ, or the creation or expansion of new populations within the species distribution range.

Many plant species are locally adapted over a range of geographic scales (McKay et al. 2005). Population enhancement can have either beneficial or harmful impacts on the viability and fitness of populations and needs to be undertaken with consideration of the genetic impacts (Fischer and Matthies 1997; Montalvo and Ellstrand 2001; McKay et al. 2005; Jones and Monaco 2007). The importance of an environmental match and the use of local sources for translocations and restoration activities have been identified and stressed by some authors (e.g. Keller et al. 2000; Gustafson et al. 2004; O’Brian et al. 2007). Given the lack of genetic diversity found both within and among populations of L. carinensis in Djibouti, plants could be cultivated for in-situ population enhancement from any seed that is available from within Djibouti with no significant genetic impacts of provenance mismatch. Seed collected from the largest populations/population clusters is the most likely to contain diversity as shown by this study and is also those most likely to form the sources for seed collections.

Some authors strongly advocate the exclusive use of local provenance in restoration activities while others advocate that this may leave populations at evolutionary dead ends (McKay et al. 2005; Harris et al. 2006). Clearly the populations from Djibouti, Somalia and Yemen are different genetic provenances. This raises some issues about the recovery and conservation of L. carinensis. (1) Should some of the variation found in individuals from Somalia and Yemen be introduced into the Djibouti populations? (2) Should we attempt to conserve the populations or increase the population size within Yemen and Somalia? This study strongly indicates that the populations in Yemen and Somalia are highly significant for the conservation of the diversity of the species. (3) Given that there is a greater opportunity to conserve the species in Djibouti than in either Yemen or Somalia, should individuals from there be introduced into the Djibouti populations as a means to conserve what genetic diversity remains across the entire distribution of the species? However, as the populations represent distinct genotypes, care must be taken that any introductions do not interfere with the genetic integrity of the local genotypes, and that selective pressures, such as the differing altitudes in which the populations occur, must be considered.


We would like to acknowledge contributions to this project of Gavin Conochie, Phil Trathan, and Harriet Gillet, as well as William Baker and John Dransfield (RBG KEW UK), and Larry Noblick (Montogomery Botanical Center, Miami USA). We would also like to acknowledge the help and assistance of Houssein Abdillahi Rayaleh, Ministry of Housing, Urban Affairs, Environment and Land Management in Djibouti and Secretary of Djibouti Nature which was essential for the project success and Houmed Ali, Bankoualé for his knowledge and abilities in the field. We would like to extend a special thanks to Nur Ali, who was indispensable in the field. Rhonda Stokoe (USC) provided technical advice in the laboratory. The funding for the project was provided by grants from the International Palm Society and the University of the Sunshine Coast with in-kind support from The Royal Botanic Gardens Kew.

Copyright information

© Springer Science+Business Media B.V. 2008