Genetic Structure of Atriplex halimus Populations in the Mediterranean Basin

Abstract

• Background and Aims The saltbush Atriplex halimus is a chenopodiaceous plant well adapted to dry saline habitats and widely distributed in the Mediterranean Basin. A study was carried out to analyse the genetic diversity of A. halimus at the level of the Mediterranean Basin.

• Methods To assess the intra- and interpopulational variation of A. halimus a total of 51 populations and six plants per populations was analysed with the RAPD-PCR technique. For the study of the phylogeny of the populations, 21 samples of A. halimus and seven samples of other species of Atriplex were analysed by the sequencing of the ITS (internal transcribed spacer) region of the ribosomal DNA.

• Key Results The AMOVA analysis of the RAPD results showed that populations were divided into two discrete genetic groups, as the variation among groups accounted for 54·36 % of the total variance of the collection. At the same time, the intrapopulational diversity was high, as 301 out of 306 plants analysed constituted an individual RAPD haplotype. The sequencing of the ITS region also showed a significant separation of the two genetic groups, with a genetic distance of 0·023 nucleotide substitutions per site. Using A. breweri, A. canescens, A. glauca and A. prostrata as outgroups in the phylogenetic analysis, A. breweri and A. canescens are the species closest to A. halimus from this group, while A. prostrata is the most distant.

• Conclusions The present work indicates that two genetic groups of A. halimus can be distinguished after analysing the genetic diversity of 51 populations from ten countries in the Mediterranean Basin.

INTRODUCTION

Atriplex species (saltbushes) are dominant in many arid and semi-arid regions of the world, particularly in habitats that combine relatively high soil salinity with aridity (Osmond et al., 1980; McArthur and Sanderson, 1984). Most saltbush species prosper in areas with annual rainfall ranging from 200 to 400 mm. Saltbushes have been used as a resource for domestic livestock and wildlife, and for rehabilitation of degraded lands (sand dunes, saline/alkaline soils, mine waste, badlands, shallow soils, etc.). Over 400 species of Atriplex have been identified on all continents. The Mediterranean Basin, with 40–50 Atriplex species, mostly in its southern and eastern bordering areas, is a region where saltbushes have been extensively used as fodder reserves during periods of scarcity (e.g. drought and cold periods), and as a supplementary forage resource in arid and semi-arid countries.

Atriplex halimus (Mediterranean saltbush) is a perennial native shrub of the Mediterranean Basin with an excellent tolerance to drought and salinity. It ramifies almost from the base, can grow 1–3 m high and may reach 3 m in diameter. It is monoecious and inflorescences are in dense spikes. Male flowers are at the top of the spike and female flowers at the base. It blooms between May and December. Native populations are found in loam or clay depressions containing moderately saline soils and a temporary water-table, and also on outcrops of gypso-saline marls.

Genetic research on shrubs has received little attention because, for many shrub species, breeding systems and genetic control mechanisms are not yet known, but also because of their low economic value compared with the high cost of breeding programmes (Stringi et al., 1992). Despite this, saltbushes exhibit outstanding features, such as drought hardiness, ability to grow in saline and disturbed environments and tolerance to metal toxicity, that allow them to occupy a wide ecological range (Cibils et al., 1998), and might justify the implementation of shrub improvement programmes to domesticate relevant fodder shrub species such as A. halimus. It seems that there is a large variability and heterozygosity in A. halimus populations of different Mediterranean origins (Stringi et al., 1992; Le Houèrou, 2000; Talamali et al., 2003), and such heterogeneity could be exploited to select clones or develop synthetic populations with a combination of good traits such as high palatability, high edible biomass production, rapid re-growth capacity and good adaptability to environmental limiting factors in semi-arid Mediterranean environments, such as summer drought, high salinity and cold winters. Prior to this work, a genetic study of intraspecific variation in A. halimus is needed.

A suitable tool for assessing the genetic structure of the populations is the technique RAPD (random amplified polymorphic DNA)-PCR (Williams et al., 1990). In recent years, a number of authors have shown the potential of this approach for studying genetic variation in populations (for a review, see Hadrys et al., 1992). Analyses carried out with RAPD markers have mostly been directed at the estimation of genetic relationships among haplotypes, and little attention has been paid to analysis of genetic differentiation among populations. However, a first attempt at quantifying RAPD variation, and partitioning it within and among populations, was made by Huff et al. (1993) as an extension of the analysis of molecular variance (AMOVA) (Excoffier et al., 1992; Excoffier and Smouse, 1994). In the present study, this analytical approach was applied to a set of RAPD-PCR data to examine the genetic diversity of a group of A. halimus populations from the Mediterranean Basin. The study of the phylogeny of A. halimus in relation to four other species of the genus was also addressed by the sequencing of the ITS region of the ribosomal DNA (White et al., 1990).

MATERIALS AND METHODS

Plant material used and experimental design

A total of 51 populations of Atriplex halimus L. from ten countries from the Mediterranean Basin were analysed (Table 1 and Fig. 1). After collection or receipt of seeds, these populations were planted and maintained in the collection of the IMIDA, at the coastal location of Mazarrón (Murcia, south-east Spain). All the populations were collected in natural environments, except one (FAO), that was obtained from the germplasm collection of INRF (Tunis), its origin being unknown.

Table 1.

List of the populations of Atriplex halimus used in the study, with their origin and intrapopulational diversity calculated by the average number of pairwise differences derived from the RAPD-PCR experiment

Sample	Origin	Country*	ITS accession no.	Average no. of pairwise differences ± s.d.
E.1	La Coruña	Spain (NW)		0·533 ± 0·508
E.2	Aranjuez	Spain (C)		3·133 ± 1·884
E.3	Perales	Spain (C)		4·800 ± 2·728
E.4	Alhama	Spain (SE)	AY873932	5·933 ± 3299
E.5	Cieza	Spain (SE)		4·800 ± 2·728
E.6	Librilla	Spain (SE)		4·533 ± 2·593
E.7	Lorca	Spain (SE)		4·800 ± 2·728
E.8	Mazarrón	Spain (SE)		8·533 ± 4·603
E.9	Mula	Spain (SE)		3·933 ± 2·290
E.10	Caspe	Spain (C)		2·266 ± 1·441
E.11	Almudévar	Spain (C)	AY873933	3·933 ± 2·290
E.12	Córdoba	Spain (S)	AY873934	2·600 ± 1·612
E.13	Ses Figueretes	Spain (E)		3·333 ± 1·986
E.14	Cala Tarida	Spain (E)	AY873935	2·933 ± 1·783
E.15	Playa Oliva	Spain (S)	AY873936	3·200 ± 1·918
F.1	Marseille	France (S)	AY873937	2·333 ± 1·476
FAO	FAO-INRF-Tunis	Unknown	AY873938	1·200 ± 0·883
I.1	Trapani	Italy (Sicily)		8·600 ± 4·637
I.2	Pietranera	Italy (Sicily)		8·466 ± 4·570
I.3	Eraclea	Italy (Sicily)		8·333 ± 4·503
I.4	P.di Montechiaro	Italy (Sicily)	AY873939	9·133 ± 4·904
I.5	Butera	Italy (Sicily)		4·666 ± 2·660
G.1	Heraklion	Greece (Crete)	AY873940	10·533 ± 5·606
S.1	ICARDA	Syria (C)		11·533 ± 6·107
S.2	Maragha	Syria (C)	AY873941	12·133 ± 6·407
S.3	Wadi El Azib	Syria (C)	AY873942	11·867 ± 6·274
P.1	Beer Sheva	Israel (C)		10·666 ± 5·673
Z.1	Marsa Matrouth	Egypt (N)	AY873943	12·933 ± 6·808
T.1	Cekhira	Tunisia (C)		8·067 ± 4·370
T.2	Rohia	Tunisia (C)	AY873944	8·867 ± 4·771
T.3	Sbikha-Kairouan	Tunisia (C)	AY873945	14·267 ± 7·475
T.4	Medenine	Tunisia (S)		13·600 ± 7·141
T.5	Sfax	Tunisia (C)		10·200 ± 5·439
T.6	Sidi Bousaid	Tunisia (N)		9·267 ± 4·971
T.7	Sousse	Tunisia (C)	AY873946	10·933 ± 5·806
T.8	Tajerouine	Tunisia (C)	AY873947	10·733 ± 5·706
T.9	Tatahouine	Tunisia (S)		14·200 ± 7·442
T.10	Amilcar	Tunisia (N)		10·400 ± 5·539
T.11	El Alam	Tunisia (C)		8·533 ± 4·603
T.12	El Grin	Tunisia (C)		6·067 ± 3·365
T.13	Hergla	Tunisia (C)		10·467 ± 5·572
T.14	El Kef	Tunisia (C)		13·467 ± 7·075
T.15	Sebka Kelbia	Tunisia (C)		16·600 ± 8·643
A.1	Djelfa	Algeria (N)	AY873948	9·600 ± 5·138
A.2	Hoggar	Algeria (S)	AY873949	5·533 ± 3·097
M.1	Tensift	Morocco (C)	AY873950	8·733 ± 4·704
M.2	Moulay Hafid	Morocco (C)	AY873951	10·666 ± 5·673
M.3	Riff	Morocco (N)	AY873952	11·200 ± 5·940
M.4	Tamarakis	Morocco (S)		10·933 ± 5·806
M.5	Ziz	Morocco (S)		9·400 ± 5·038
M.6	Aouffouss	Morocco (S)		8·333 ± 4·503

ITS accession number column indicates the populations used in the phylogenetic analysis.

^*N, north; C, central; S, south; E, east; W, west.

Fig. 1.

Map showing the distribution of A. halimus in the Mediterranean Basin as well as the sampling sites analysed and the distribution of the two genetic groups detected by RAPD-PCR in this work.

To assess the intra- and interpopulational diversity, six individuals (12-week-old seedlings) from each population were analysed by RAPD-PCR, with a total of 306 plants. For the phylogenetic analysis, the entire ITS1 and ITS2 regions, together with the 5·8S gene sequence between them, were sequenced, by using the universal primers its-4 and its-5 described by White et al. (1990). This analysis was performed using a single plant of each one of 21 populations (ITS column in Table 1). Four species of Atriplex other than A. halimus were used as outgroups in the phylogenetic analysis, namely A. breweri, A. canescens, A. prostrata and A. glauca. One plant was used for each species except for A. glauca, for which four plants were analysed.

DNA extraction and PCR reactions

The plant DNA was extracted from young leaves, according to the method proposed by Torres et al. (1993), and based on the extraction with CTAB buffer after grinding the material in liquid nitrogen. RAPD-PCR was carried out according to the protocol of Williams et al. (1990) with some modifications. The reaction volume was 12 µL, composed of reaction buffer (50 mM KCl, 50 mM Tris–HCl, pH 9, 20 mM (NH₄)₂SO₄, 2·5 mM MgCl₂, 0·005 % BSA), 0·2 mM dNTPs, 0·5 units of DNA-polymerase Replitherm, (Epicentre Technologies, Madison, WI, USA), 7·5 ng of primer and 30 ng of plant DNA. The amplification was performed in a Gene Amp PCR System 9600 (Perkin-Elmer, Norwalk, CT, USA) through 40 cycles of denaturation (1 min at 94 °C), annealing (1 min at 34 °C) and extension (1 min at 72 °C). The reaction products were separated by electrophoresis, run at 5 V cm⁻¹ for 2 h in a 1·5 % agarose gel, and visualized by staining with ethidium bromide. Primers used were 10-mer oligonucleotides supplied by Operon Technologies (Alameda, CA, USA). A total of 100 primers of sets A, B, C, E and F was tested with a reduced group of four plants in order to select the ones giving the best amplification for further analysis. Finally, a group of 12 primers was used in the experiment: OPA-02, OPA-05, OPB-01, OPB-03, OPB-06, OPC-07, OPC-08, OPC-15, OPD-08, OPD-11, OPD-15 and OPE-12.

For the phylogenetic analysis, PCR reactions were set up in 50 µL of the same reaction mix described before, but with 0·2 µm of the primers its-4 (5′-TCCTCCGCTTATTGATATGC-3′) and its-5 (5′-GGAAGTAAAAGTCGTAACAAGG-3′) (White et al., 1990). The amplification was performed by 40 cycles as described for RAPDs, but with the annealing temperature set to 56 °C. The fragments amplified by PCR were directly sequenced in an ABI Prism 310 sequencer (Applied Biosystems, Foster City, CA, USA) of the University of Valencia (Burjassot, Valencia, Spain).

Genetic analysis of data

The RAPD bands obtained with all the primers were aligned and scored as present or absent. A measure of genetic distance was established by the number of differences (Nei and Kumar, 2000) and the relationships among the individuals analysed were represented in a neighbour joining dendrogram. All of this was done with the program MEGA version 2·0 (Kumar et al., 2001).

Partitioning of the observed genetic variation and calculation of the corresponding F-values were carried out by means of AMOVA (Excoffier et al., 1992), which gave three levels of partitioning: (1) differences between genetic groups (F_ct); (2) differences among populations within groups (F_sc); (3) differences within populations (F_st). As a measure of genetic distance, the number of fragment differences between haplotypes was used. The contribution of the three partitions to the total variance and the three F-statistics were statistically tested by randomization tests based on 2000 permutations. As a measure of intrapopulational diversity, the average number of pairwise differences (ANPD) was used. All the analyses were performed using the package ARLEQUIN ver. 2000 (Schneider et al., 2000).

The phylogenetic analysis was made by applying the model of genetic distance of Kimura-2 parameters (Kimura, 1980). Significance of nodes was estimated by 2000 bootstrap replicates. Phylogenetic analyses were conducted using MEGA version 2·0 (Kumar et al., 2001).

RESULTS

The RAPD-PCR amplification of the 306 samples (six individuals of 51 populations), with the 12 primers indicated, produced a total of 114 bands of good quality. The RAPD-PCR patterns obtained showed differences between populations and between individuals of some populations (Fig. 2). Genetic similarity between individuals was calculated from the rectangular matrix of presence/absence of bands. However, given the big size of the resulting dendrogram, only a reduced set of 12 representative populations is shown (Fig. 3). These populations were: S.1 (Syria), P.1 (Israel), Z.1 (Egypt), G.1 (Greece), I.1, I.5 (Sicily, Italy), T.7 (Tunisia), M.3 (Morocco), A.2 (Algeria), F.1 (France), E.1, E.8 (Spain). The dendrogram reveals considerable intrapopulational diversity, with a clear separation of individual plants. In fact, the 306 plants analysed were distributed in 301 RAPD-PCR haplotypes. The dendrogram also shows the grouping of all samples into two separate clusters of nine (group G2) and three populations (group G1), respectively. This genetic separation into two groups was also clear when working with the complete set of 306 plants. The 15 samples from Spain and the samples F.1 (France) and FAO grouped together in group G1, while the rest of the populations (34) belonged to the larger group G2 (not shown). All the populations from the southern and eastern Mediterranean Basin countries belong to the G2 group. Figure 1 shows the geographical distribution of the two genetic groups (G1 and G2) deduced from the present data, combined with the colonization area of A. halimus derived from previous floristic distribution studies (Post, 1932; Máire, 1952–1987; Tutin et al., 1964; Pignatti, 1982; Greuter et al., 1984; and several other floras not quoted).

Fig. 2.

Example of an agarose gel showing the amplified DNA patterns obtained with a RAPD-PCR reaction with primer OPA-9 and six plants of six populations (reference to populations in Table 1 is given in parentheses). Top: Observe the separation between two populations of A. halimus group G2 (Djelfa and Hoggar) and one population A. halimus group G1 (FAO). Bottom: Observe the differences among individuals within three Tunisian populations from A. halimus G2. SM, Size marker, 123-bp ladder.

Fig. 3.

Neighbour-joining dendrogram showing the genetic similarity of six plants of each of three populations from A. halimus group G1 and nine populations from A. halimus group G2 (for populations refer to Table 1). Support values from 2000 bootstrap replicates are indicated beside each node. Data derived from the RAPD-PCR experiment.

To quantify the significance of the genetic separation of the two groups, AMOVA was carried out. The results (Table 2) indicate that 54·36 % of the total variation is due to differences between the two groups, G1 and G2 (P < 0·001). The variation between populations within groups accounts for 29·18 % of the total variation and the variation within populations for 16·46 % of the total. This indicates that, together with a high intrapopulational variation, there is also a significant structuring and separation of populations.

Table 2.

AMOVA analysis of the RAPD-PCR variation of the 51 populations of Atriplex halimus

	Among groups	Among populations within groups	Within populations	Total
G1 vs. G2
d.f.	1	49	255	305
Sum of squares	1795·191	2221·613	993·833	5010·637
Variance components	12·867	6·907	3·897	23·671
Percentage of variation	54·36	29·18	16·46
P-value	<0·001	<0·001	<0·001
Italy vs. the rest of G2
d.f.	1	32	170	203
Sum of squares	382·068	1581·820	836·833	2800·721
Variance components	6·500	7·418	4·923	18·841
Percentage of variation	34·50	39·37	26·13
P-value	<0·001	<0·001	<0·001

Partitioning of variance contrasting the two genetic groups: G1 (17 populations) against G2 (34 populations) and also the Italian (5) against the rest (29) of the G2 populations.

The dendrogram derived from RAPD-PCR data shows that the populations clustered following a pattern tentatively related with geographic distance (Fig. 3). The quantification of the genetic distances between populations is provided by the pairwise F_st values (Table 3). The lowest values in the larger group (G2) are observed between the populations from Syria, Israel and Egypt. By contrast, the highest distances in the group G2 are observed between the Italian populations and the rest. To assess the significance of this separation, an AMOVA analysis was performed grouping the five Italian populations against the other 29 populations of the group G2 (Table 2). The results indicate that 34·50 % of the total variation (P < 0·001) is due to the separation between groups, which suggests a certain genetic separation of the Italian populations, all of them collected in Sicily. Another AMOVA analysis was performed to see if there is any genetic difference between G2 populations from the west and the east of the Mediterranean Basin (not shown). The groups consisted of populations from Morocco, Algeria and Tunisia in the west and populations from Greece, Syria, Israel and Egypt in the east. The percentage of variance due to the group component was 11·35 %, and the among-populations component was 52·25 %, suggesting that there is not a significant genetic differentiation of populations located on opposite sides of the Mediterranean Basin.

Table 3.

Pairwise F_st values among the 12 populations of Atriplex halimus depicted in the dendrogram of Fig. 3, derived from the RAPD-PCR data

	Greece G.1	Syria S.1	Israel P.1	Egypt Z.1	Moroc. M.3	Algeria A.2	Italy I.1	Italy I.5	Tunisia T.7	Spain E.1	Spain E.8
Syria-S.1	0·537
Israel-P.1	0·510	0·442
Egypt-Z.1	0·540	0·463	0·461
Morocco-M.3	0·655	0·625	0·652	0·555
Algeria-A.2	0·757	0·733	0·731	0·670	0·707
Italy-I.1	0·708	0·679	0·690	0·705	0·770	0·841
Italy-I.5	0·761	0·752	0·768	0·762	0·816	0·862	0·644
Tunisia-T.7	0·583	0·591	0·598	0·538	0·622	0·757	0·680	0·764
Spain-E.1	0·881	0·864	0·869	0·827	0·866	0·931	0·917	0·947	0·871
Spain-E.8	0·797	0·789	0·781	0·731	0·772	0·843	0·840	0·865	0·792	0·477
France-F.1	0·855	0·84	0·846	0·789	0·838	0·905	0·894	0·923	0·852	0·803	0·454

The quantification of the within-population variation is provided by the average number of pairwise differences within each population (ANPD, Table 1; note that RAPD primers were pre-selected for polymorphism). In the group G1 of populations from Spain and France, the data of intrapopulational variability indicates that the value of ANPD is lowest (0·533) in the population E.1 (La Coruña), collected in Galicia (north-west Spain). This is not unexpected given that this population, located in a garden, is clearly a recent introduction to the area. By contrast, the highest variability (8·533) was detected in the population E.8 (Mazarrón) collected in Murcia (south-east Spain). The values of the intrapopulational diversity of the G2 group range from a minimum of 4·666 at Butera (Sicily, Italy) to a maximum of 16·600 at Sebka Kelbia (Tunisia) (Table 1). A population located in Hoggar (Algeria), in the Sahara desert, also shows a relatively low variability (5·533) and a clear morphological differentiation (small, wavy leaves). In general, the intrapopulational variability of the G2 group is higher than that of the G1 group.

The phylogenetic analysis based on the ITS sequences (Fig. 4) is fully consistent with the RAPD data. The two genetic groups of A. halimus appear clearly separated, in a node supported by a bootstrap value of 100. There is also a considerable intragroup diversity. The mean genetic distances between G1 and G2 are 0·023 nucleotide substitutions per site (Table 4). This is about half the distance from both groups to one of the outgroup species, A. glauca.

Fig. 4.

Neighbour-joining dendrogram representing the genetic distance of 14 plants of A. halimus group G2, seven plants of A. halimus group G1 and four Atriplex species: A. glauca (GenBank accessions AY873925–28), A. breweri (GenBank accession AY873929), A. canescens (GenBank accession AY873930) and A. prostrata (GenBank accession AY873931). Genetic distance was calculated by the Kimura two-parameters model from data of a DNA sequence in the ITS region of the rDNA. Support values from 2000 bootstrap replicates are indicated beside each node.

Table 4.

Genetic distances (nucleotide substitutions per site) (under diagonal) and standard error (above diagonal) calculated from the sequences in the ITS region of rDNA of Atriplex halimus groups G2 and G1, as well as four outgroup species: A. glauca, A. breweri A. canescens and A. prostrata

	A. halimus-G2	A. glauca	A. breweri	A. canescens	A. prostrata	A. halimus-G1
A. halimus-G2		0·009	0·007	0·008	0·010	0·004
A. glauca	0·059		0·008	0·009	0·010	0·008
A. breweri	0·038	0·047		0·006	0·009	0·007
A. canescens	0·054	0·055	0·023		0·010	0·008
A. prostrata	0·061	0·065	0·047	0·058		0·010
A. halimus-G1	0·023	0·055	0·037	0·052	0·059

DISCUSSION

The study of genetic diversity of Atriplex halimus has received little attention until now. The first study was made by Haddioui and Baaziz (2001), analysing the isoenzyme polymorphisms of nine populations of A. halimus from several locations in Morocco. This experiment showed a very high intrapopulational diversity. Through the analysis of three enzyme systems (esterases, acid phosphatases and glutamate oxaloacetate transaminase), these authors found that genetic diversity of their collection was explained mainly by the within population component. Only 8 % of the whole diversity was explained by the between-populations differentiation. This study is expanded in the present work with two different types of genetic markers at the level of the Mediterranean Basin.

The present results indicate that about half of the genetic variation of A. halimus in the Mediterranean Basin is explained by differentiation between two clear groups. Moreover, further work done on chromosome counting and genome size by flow cytometry (Walker et al., 2004) also confirms such separation: populations of group G2 are tetraploid (4x) and contain twice as much DNA as populations of group G1, which are diploid (2x).

The genetic differentiation between populations is moderate at the overall level of the entire collection, but some populations, like those from Sicily, are clearly separated from the rest. This genetic separation is consistent with morphological and ecological traits of the Sicilian populations (Pignatti, 1982). They have larger leaves, a higher leaf : shoot ratio and produce more edible biomass than other G2 populations. This could be the result of adaptation to a different biotope as the mean rainfall in Sicily—over 500 mm—is higher than in southern and eastern Mediterranean areas where most group G2 populations are present.

Within each of the two genetic groups, about half of the genetic variation is explained by the within-population component, making it more difficult to differentiate populations. The intrapopulational genetic variability is larger in group G2 than in G1. The allogamic mode of reproduction of A. halimus could explain high levels of gene flow. The high levels of variability observed may be required to maintain plasticity in a highly fluctuating and diverse environment like the Mediterranean Basin. Another reason for the higher intrapopulational variation of group G2 could be its polyploid character. According to Soltis and Soltis (2000), polyploids, both individuals and populations, maintain higher levels of heterozygosity than do their diploid progenitors. Moreover, most polyploids are polyphyletic, incorporating genetic diversity from multiple progenitor populations. In this way, polyploids may have a much better adaptability to diverse ecosystems, which may contribute to their success in nature. This is illustrated in the case of A. halimus group G2 by its much bigger distribution area than A. halimus group G1 and by its presence in very contrasting biotopes (Sahara Desert vs. Sicily).

Little was known previously about the phylogenetic relationships within the genus Atriplex. Only one ITS sequence of A. spongiosa can currently be found in GenBank (Pyankov et al., 2001). The data presented here indicate that the phylogenetic analysis of a group of five species of Atriplex also sustains the separation between the two A. halimus genetic groups. It also appears that A. breweri and A. canescens are relatively close to A. halimus; in fact, A. breweri can be confused morphologically with A. halimus in its non-fruiting stage (Le Houèrou, 2000).

It is interesting to remark that, according to previous reports, A. halimus includes two quite different groups in terms of habitat and morphology. Le Houèrou (1992, 2000) described the characteristics of these two groups, that he named subspecies: (1) subspecies halimus, which is present on the northern shores of the Mediterranean Basin and on the shores of the Atlantic and the North Sea, and can be identified by its short and leafy fruiting branches and by its erect habit; and (2) subspecies schweinfurthii (Boiss.) Le Houèrou, common on the southern shores of the Mediterranean Basin, North Africa and Near East, and which is characterized by long (>50 cm) and leafless, somewhat reddish, fruiting branches and an intricate bushy habit. However, these subspecies described by Le Houèrou are not accepted taxonomical units, as they have not been formally characterized yet under the rules of botanical taxonomy.

According to Le Houèrou (2000), both subspecies are extremely heterogeneous in terms of their morphology, ecology, productivity and palatability to herbivores. However, subsp. halimus predominates in semi-arid to subhumid areas and has a higher leaf : shoot ratio than subsp. schweinfurthii, which is better adapted to arid environments but is less productive in terms of browsing biomass. Contrasting the description of Le Houèrou (1992, 2000) with our observations on the material used in their study, it was observed that the populations included in group G1 are rather coincident with the characters described for A. halimus sub. halimus, while populations from group G2 present features of A. halimus sub. schweinfurthii. Despite these observations, the connection between our two genetic groups and the two subspecies described by Le Houèrou is tentative, as we have not analysed directly the material used by this author. So, all these aspects are pending further clarification.

Practical consequences of the present results for genetic resources management could be: (a) that a small number of A. halimus populations would be sufficient to preserve most of its genetic diversity, and (b) that an efficient strategy to select plant material (e.g. with a high edible biomass yield, good palatability, tolerance to cold etc.) would be to concentrate on the most variable populations and exploit the intrapopulation variability for these traits.

The analysis presented here is intended mainly to quantify the genetic distance between two A. halimus groups, by two methods: RAPD-PCR marker and sequencing of the ITS region of ribosomal DNA. According to the authors' field experience, populations of both groups exhibit a large variability in morphological and adaptive traits, G2 being the more variable. Great variability also exists among individual shrubs within any population. Thus, the present analysis confirms that these differences between groups, populations and individuals have a genetic basis and provides genetic markers to separate two types of morphologies which sometimes are obscured by the existence of intermediate morphotypes.