Biogeography and evolution of Abies (Pinaceae) in the Mediterranean Basin: the roles of long-term climatic change and glacial refugia

Authors


Juan Carlos Linares, Area de Ecología, Universidad Pablo de Olavide, Ctra. Utrera km. 1, 41002 Sevilla, Spain.
E-mail: jclincal@upo.es

Abstract

Aim  The genus Abies exemplifies plant diversification related to long-term climatic, geological and evolutionary changes. Today, the Mediterranean firs comprise nine species, one natural hybrid and several varieties. Here I summarize current knowledge concerning the origin and evolution of the genus Abies in the Mediterranean Basin and propose a comprehensive hypothesis to explain the isolation and speciation pattern of Mediterranean firs.

Location  The Mediterranean Basin.

Methods  The literature on Abies was reviewed, focusing on the morphology, fossil records, molecular ecology, phytosociology and biogeography of the genus in the Mediterranean Basin.

Results Abies fossils from the western Mediterranean indicate a wide Tertiary circum-Mediterranean distribution of the Abies ancestor. Palaeogeographical data also suggest a single eastern Mediterranean Tertiary ancestor. Following the Miocene to Pliocene climate crisis and marine transgressions, the ancestor of the northern Mediterranean firs is hypothesized to have separated into two eastern groups, one on the Balkan Peninsula and the other in Asia Minor. However, land bridges may have permitted gene flow at times. A southward migration of A. alba to refugia, where older fir species may have remained isolated since the Miocene, could explain recent findings indicating that morphologically distant species are more closely related than expected based on such morphological classification.

Main conclusions  The Abies genus appears to have undergone significant morphological differentiation that does not necessarily imply reproductive isolation. That is, long-term Mediterranean Basin dryness along a south-eastern to north-western gradient may have caused an initial Miocene–Pliocene speciation sequence. Pleistocene glacial cycles probably forced migrations to occur, leading to repeated contact between fir species in glacial refugia.

Introduction

The current distribution area of a given species results from the adaptation of the species as it shifts in space (Willis & McElwain, 2002). The presence of a species in an ecosystem requires not only an explanation based on the species’ current ecology, related to morphological and physiological adaptations to local conditions, but also an historical explanation, related to the sequence of events that allowed the presence of this species in this area and that drove evolution from an ancestral form (Hewitt, 1996; Svenning & Skov, 2005).

During the late Miocene and the Pliocene, the European Mediterranean flora underwent increasing differentiation (Palamarev, 1989; Kovar-Eder et al., 2008). The main selective pressure for Mediterranean plant communities in the late Tertiary seems to have been a long-term decline in summer precipitation (Suc, 1984). The drought associated with high temperatures during the growing season would have required specific adaptations and can be assumed to have led to the segregation of plant communities along elevational, latitudinal and topographic gradients, reflecting the ecophysiology of the constituent species. The orographic complexity derived from Alpine orogeny favoured this species segregation and the existence of local refugia around the Mediterranean Basin (Médail & Diadema, 2009). This sub-regional complexity appears to have been crucial in the evolution and persistence of the current Mediterranean flora (Taberlet & Cheddadi, 2002; Tzedakis et al., 2002; Carrión et al., 2003; Cheddadi et al., 2009).

The firs (genus Abies, Pinaceae) exemplify plant diversification related to long-term climatic, geological and evolutionary changes (Pignatti, 1978). Today, the Mediterranean firs comprise nine species, one natural hybrid and several varieties (Vidakovic, 1991). However, the current relict status of these species constitutes an example of an endangered ecosystem, highly vulnerable to global change. Various hypotheses have been proposed to explain how the genus Abies evolved (Table 1). Some proposals are contradictory, perhaps because the explanations are based generally on a single trait and we do not have enough macrofossils to validate the arguments (Gaussen, 1964; Liu, 1971; Xiang et al., 2009; Alba-Sánchez et al., 2010).

Table 1.   Representative examples of features studied and hypotheses proposed to explain the evolution of the Mediterranean species of the genus Abies.
Species studiedStudied variablesMain hypothesis or conclusionsReferences
Mediterranean firsMorphological and anatomical traitsTwo groups based on the bracts of the cones, which are either longer or shorter than the scalesFarjon & Rushforth (1989); Dallimore & Jackson (1923); Liu (1971); Vidakovic (1991)
A. equi-trojaniPollen analysisHybrid characteristics for A. equi-trojaniAytug (1959)
Abies genusInter-specific hybridizationArtificial crosses suggest weak reproductive barriers among the Mediterranean Abies species as opposed to strict barriers between Mediterranean and North American fir species. Mediterranean firs are especially prone to hybridize, being closely related, suggesting recent speciationKlaehn & Winieski (1962)
A. pinsapoPhenolic compositionSpanish and Moroccan populations are very closeGranados & Rossell (1966)
Mediterranean firsPhytosociologyMonophyletic origin followed by geographical fragmentationArbez (1969); Pignatti (1978); Barbero & Quézel (1975)
Greek firsMorphology and monoterpene compositionIndividual trait variation shows a south-east to north-west clinal patternMitsopoulos & Panetsos (1987)
Eastern Mediterranean firsTerpene compositionSingle eastern Mediterranean Tertiary ancestor for present-day Aegean speciesFady et al. (1992)
A. borisii-regisSegregation and linkage of allozymesAbies borisii-regis populations are genetically very closeFady & Conkle (1992)
Mediterranean firsAllozyme differentiationGreat variation in heterozygosity (lowest in A. pinsapo, highest in A. cephalonica). Inter-population genetic diversity about 26% of the total genetic diversity. High affinity between the Calabrian fir population and the one from north-west Greece as well as between A. equi-trojani and southern Greek populations of A. cephalonica. Species-specific alleles in A. cilicicaScaltsoyiannes et al. (1999); Pascual et al. (1993)
Mediterranean firsAllozyme, chloroplast DNA and random amplified polymorphic DNAGenetic variation easily detectable using isozyme loci, but less evidence of genetic divergence by single-primer DNA amplification at the chloroplast DNA levelVicario et al. (1995); Ziegenhagen et al. (2005); Kaya et al. (2008)
A. albaPolymorphic chloroplast microsatellites, allozyme markersGenetic distances between most populations were high. Genetic distances increased with spatial distances. Present organizations of levels of allelic richness across the range of the species is likely have been shaped by the distribution of refugia during the last glaciation and the subsequent re-colonization processesVendramin et al. (1999); Parducci et al. (1996)
Abies genusRestriction fragment-length polymorphism analysis of chloroplast DNAThe chloroplast genome was highly variable in most of the taxa investigated. Species-diagnostic markers were found for four of ten species analysed, but intra-specific variation was also detected in four speciesParducci & Szmidt (1999)
A. nebrodensis, A. alba, A. numidica and A. cephalonicaChloroplast simple-sequence repeatsAbies nebrodensis differed from the other three Abies species investigated, supporting its classification as an independent taxonParducci et al. (2001a)
Mediterranean firsMitochondrial DNA markersFive haplotypes were defined. Two different haplotypes were found in each of A. alba and A. cephalonica, one haplotype being shared among the two species. A single species-specific haplotype was found in A. cilicica. Abies pinsapo and A. numidica shared one haplotype, while eastern Mediterranean firs (A. cephalonica, A. bornmuelleriana, A. equi-trojani and A. nordmanniana) also shared one haplotypeZiegenhagen et al. (2005)
Western Mediterranean firsChloroplast microsatellitesHigh levels of genetic diversity. There was a strong effect of the Strait of Gibraltar on western Mediterranean Abies differentiation, leading to two groups corresponding to Spanish and Moroccan populations. High diversity within populations and low differentiation among populations in Spain and Morocco suggest a certain level of pollen-mediated gene flow. Abies tazaotana is genetically very close to A. marocanaTerrab et al. (2007)
Abies genusSequence of the complete nuclear ribosomal internal transcribed spacer regionPhylogenetic analyses showed a close relationship between European and Asian species, but with weak supportXiang et al. (2009)
Mediterranean firsWood anatomyAbies alba, A. pinsapo, A. pinsapo var. marocana and A. pinsapo var. tazaotana possess biometric features that distinguish them from the other Mediterranean firs. Abies numidica shows biometric features close to A. pinsapo and differs from the eastern firs (A. borisii-regis, A. cephalonica, A. cilicica, A. nordmanniana and A. equi-trojani)Esteban et al. (2009)
A. alba and A. pinsapoPalaeopalynology and species distribution modelsAbies alba and A. pinsapo cannot be distinguished in palynological analyses; previous studies related to their glacial refugia or past distribution usually refer to the entire genus. Species-distribution models yielded a significant geographical and ecological segregation, suggesting that the two taxa remained isolated throughout the QuaternaryAlba-Sánchez et al. (2010)

Palaeoecological studies based on fossil pollen and plant macrofossils show that during the Pliocene (c. 5 Ma) the Mediterranean Basin was covered by vast forest ecosystems, presumably including a common ancestor of the current Mediterranean firs (Pignatti, 1978; Meyen, 1987; Palamarev, 1989). From this common ancestor, migrations and subsequent population fragmentation led to smaller, isolated fir forests around the current Mediterranean Basin (Farjon & Rushforth, 1989).

Mediterranean firs have been classically divided into two groups, based on the bracts of the cones, which are either longer or shorter than the scales (Dallimore & Jackson, 1923; Liu, 1971; Farjon & Rushforth, 1989; Vidakovic, 1991): a pinsapo fir-type archaic group and a silver fir-type modern group. The archaic group, which presents a southern Mediterranean distribution, is morphologically characterized by enclosed bracts. This apparently ancient group comprises the species A. pinsapo Boissier, A. pinsapo var. marocana (Trabut) Ceballos & Bolaños, A. pinsapo var. tazaotana (Cózar ex Huguet del Villar) Pourtet, A. numidica de Lannoy ex Carrière and A. cilicica (Antoine & Kotschy) Carrière. The modern group has a northern Mediterranean distribution and is morphologically characterized by protruding bracts. This group includes the species A. alba Miller, A. borisii-regis Mattfeld, A. cephalonica Loudon, A. nebrodensis (Lojac.) Mattei, A. nordmanniana (Steven) Spach, A. equi-trojani (Asch. & Sint. ex Boiss.) Coode & Cullen and A. bornmuelleriana (Steven) Spach.

However, recent evidence based on allozyme differentiation, chloroplast and mitochondrial DNA variation, as well as on other DNA markers, has shown that the speciation sequence and the Pleistocene overlap among different Mediterranean firs is not well understood (Vicario et al., 1995; Scaltsoyiannes et al., 1999; Vendramin et al., 1999; Parducci et al., 2001a; Ziegenhagen et al., 2005; Xiang et al., 2009). For instance, an unexpectedly close relationship between A. pinsapo and A. alba (two of the morphologically most distant Mediterranean fir species) was reported by Scaltsoyiannes et al. (1999). In addition to this molecular evidence, recent papers on wood anatomy (see Esteban et al., 2009) and palaeopalynology and species distribution models (see Alba-Sánchez et al., 2010) support these surprising results and underscore biogeographic issues that remain unsolved.

The present paper reviews the morphology, molecular ecology and biogeography of Abies, with a focus on the Mediterranean Basin, in order to summarize current knowledge concerning the origin and evolution of the genus Abies in this area. A comprehensive hypothesis to explain the isolation and speciation pattern of the Mediterranean fir is developed and discussed.

Materials and methods

This paper is based on a review of the literature on Abies fossil records, morphology, molecular ecology and biogeography. Science Citation Index (SCI) papers were used as far as possible, avoiding excessive use of references in German, Spanish and French in order to invite a wide readership to examine the presented hypothesis by means of the references used. The keywords used in the review were: Abies, biogeography, evolution, firs, fossil pollen, Mediterranean, Miocene, palaeogeography, phylogeny, plant macrofossils, Quaternary refugia.

Results and discussion

Palaeogeography, palaeoclimate and plant evolution in the Mediterranean Basin

The Mediterranean Basin represents the remnant of the ancient Tethys Sea, a Mesozoic ocean that underwent profound changes during the Alpine orogeny, beginning in the Middle Cretaceous, about 100 Ma (Gradstein et al., 2004), and extending to the late Miocene, about 7 Ma. Regional trends towards decreasing rainfall and, to a lesser extent, towards a temperature decrease have been inferred at the end of the Oligocene (35–23 Ma), and mainly during several stages of the Miocene (23–5.2 Ma; Bruch et al., 2006; Jost et al., 2009). As a result, the Mediterranean Basin would have acquired the pattern of high atmospheric pressure that caused a water deficit in the Mediterranean summer (Pignatti, 1978; Suc, 1984; Mai, 1989; Fauquette et al., 1999).

The Miocene (23–5.3 Ma) would have included periods during which there were large emerged areas, alternating with others during which, for instance, the current south-western edge of Europe (in the Iberian Peninsula) was a sea belt dotted by small islands (Popov et al., 2006). This configuration led to the geographical isolation of several plant populations and eventually allowed the speciation of some taxa. Palaeoclimatic researchers studying the late Miocene (Messinian, 7.2–5.3 Ma) have identified an event of regional aridity, during which much of the Mediterranean Sea became a marsh (Kovar-Eder et al., 2008). Thereafter, during the Pliocene (5.3–1.8 Ma), the Mediterranean seasonality and the regional cooling accentuated around 4.5, 3.6, 3.2, 2.8 and 2.4 Ma (Suc, 1984; Kovar-Eder et al., 2008; Jost et al., 2009). These climatic oscillations, ending with the Pleistocene glaciations, resulted in the cumulative loss of several thermophilic species from the European continent, as well as in episodic expansions of xerophytic communities (Pignatti, 1978; Suc, 1984; Palamarev, 1989). Palaeoclimatic analysis suggests the establishment of the current Mediterranean climate seasonality, characterized by two intra-annual stress periods for plant growth, in summer and in winter, during at least three climatic crises dated to 3.2, 2.8 and 2.4 Ma (Suc, 1984; Fauquette et al., 1999, 2007; Bruch et al., 2006).

The Pleistocene (1.8 Ma to the present) was characterized by cycles of fragmentation and re-expansion in the distribution range of several species (Thompson, 2005). This period has been analysed theoretically under three main evolutionary scenarios (Schluter, 2001). The first could be called species persistence. Under this approach, the species would have undergone contractions and re-expansions in their distribution area as a consequence of the glacial–interglacial cycles (Willis & McElwain, 2002). These biome shifts would have taken place without genetic differentiation or leading to a loss in local differentiation following the re-expansion and overlap of populations previously isolated. This hypothesis results in evolutionary stasis (Hewitt, 1999; Willis & Niklas, 2004). The second scenario involves species extinction. The species distribution area would have become reduced and progressively fragmented, whereupon subsequent population isolation would have led to extinctions (Svenning & Skov, 2007). The third possibility is that of allopatric speciation, whereby geographic isolation would have promoted local adaptations and genetic differentiation among populations (Turelli et al., 2001). As a consequence, the new species would have shown reproductive isolation during re-expansion and possible geographic overlaps (Hewitt, 1996; Willis & McElwain, 2002; Willis & Niklas, 2004; Thompson, 2005). These three evolutionary processes appear to have operated during the Mediterranean fir history (Pignatti, 1978).

It has been shown that each species has responded independently to climate changes. The species patterns of differentiation, persistence in a refugium, migration rates and the expansion–retraction routes cannot easily be generalized (Willis & Whittaker, 2000; Willis & Niklas, 2004). Moreover, the genus Abies appears to have acquired significant morphological differentiation, which does not suggest reproductive isolation (Klaehn & Winieski, 1962). Introgression between the Mediterranean Abies species seems to have been common (Tzedakis, 1993; Scaltsoyiannes et al., 1999; Xiang et al., 2009; Alba-Sánchez et al., 2010).

Evolution of the archaic Mediterranean firs

The discovery of Abies fossils in different parts of the Iberian Peninsula gives rise to the hypothesis that the arrival of Abies in the southern Iberian Peninsula as well as in northern Morocco and Algeria occurred across Tethys bridges from the northern Iberian Peninsula (Fig. 1). This evidence also supports the idea of a wide Tertiary circum-Mediterranean distribution of the Abies ancestor, prior to its range fragmentation and subsequent allopatric speciation (Liu, 1971). On the other hand, the above-mentioned fossil records suggest an apparent evolutionary stasis in the morphology of the western Mediterranean firs since the Miocene (Areitio, 1874; Depape, 1928; Dubois-Ladurantie, 1941; Gaussen, 1964; Menéndez & Florschütz, 1964; Pons & Reille, 1988; Carrión et al., 2001).

Figure 1.

 Hypothesized speciation sequence of Mediterranean fir (Abies) species during the Miocene–Pliocene. 1, A. alba ancestor; 2, A. cilicica ancestor; 3, A. pinsapo ancestor; 4, A. numidica ancestor; 5, A. cephalonica ancestor. Hypothesized expansion routes based on molecular data and fossil records appear as arrows, and uplift areas are indicated by the other lines. Letters denote Tertiary fossil findings. a, Abietites latisquamosus Ludw. Tertiary. Shrine of Saint Eulalia, Totana (Spain; Areitio, 1874). b, Abies ramesii Saporta. Early Miocene 20–16 Ma. Majorca Island (Spain; Depape, 1928); France (Saporta, 1865). c, Abies saportana Rerolle. Late Miocene. Santa Eugenia, Coll del Saig, Cerdaña (Spain; Dubois-Ladurantie, 1941; Kovar-Eder et al., 2006); Poland; and Rhodope region, Bulgaria (Palamarev, 1989).

Southward migration of A. alba to southern Iberian Peninsula refugia (Fig. 2a), where A. pinsapo would have remained isolated since the Miocene (Pignatti, 1978; Palamarev, 1989), could explain recent findings that establish closer than expected relationships between A. alba and A. pinsapo, paradoxically the most distant species among the Mediterranean firs, based on morphological classification (Dallimore & Jackson, 1923; Liu, 1971; Farjon & Rushforth, 1989; Vidakovic, 1991). According to allozyme differentiation, A. pinsapo shows greater genetic distance from A. cilicica and A. numidica than from A. alba populations from the Pyrenees, Alps or Apennines (Scaltsoyiannes et al., 1999). Recent wood biometry studies (Esteban et al., 2009) have shown high anatomical proximity between A. alba from the Pyrenees and A. pinsapo, despite the fact that they belong to two different morphological groups, suggesting that contact occurred between the two species in the past (Fig. 2a). Moreover, similar results were found in the genetic distance of different populations of the fir-root pathogen Heterobasidion abietinum (Basidomycota) isolated from A. pinsapo, which showed close relationships with H. abietinum populations isolated from A. alba in France and Italy (Luchi et al., 2008).

Figure 2.

 Hypothesized speciation sequence and migration routes of the Mediterranean fir (Abies) species (a) during the late Pliocene to the early Pleistocene and (b) during the late Pleistocene. 1, A. alba ancestor; 2, A. cilicica ancestor; 3, A. pinsapo ancestor; 4, A. numidica ancestor; 5, A. cephalonica ancestor; 6, A. bornmuelleriana ancestor. Migration routes appear as arrows; introgression areas appear as dotted lines; isolation areas appear as solid lines.

Several pieces of evidence indicate that this hypothetical connection between A. alba and A. pinsapo would have been disrupted during the Pleistocene, with the insertion of a semi-arid wedge in the south-eastern Iberian Peninsula (Fig. 2b; Konnert & Bergmann, 1995; Combourieu-Nebout et al., 1999; Burjachs & Allue, 2003; Carrión et al., 2003). However, the fossil records are scant and do not validate the hypothesis of Pleistocene contacts between silver and pinsapo firs. In addition, a strong niche differentiation between the two Iberian firs has recently been found by Alba-Sánchez et al. (2010), suggesting an early ecological divergence of A. alba and A. pinsapo, and possibly indicating that the two species have followed independent biogeographic dynamics (Fig. 2b; see also Muller et al., 2007).

In summary, for A. pinsapo, the findings suggest range overlaps (Fig. 2a), at the end of Pliocene or during early Pleistocene glacial periods, with the ancestors of A. alba (see Alba-Sánchez et al., 2010, and references therein). However, for A. cilicica (in Turkey) and A. numidica (in Algeria), the Tertiary isolation process would not have been broken during the Pleistocene (Fig. 2).

Evolution of the modern Mediterranean firs

Palaeogeographical data support the possible existence of a single eastern Mediterranean Tertiary ancestor for present-day Aegean species (Mitsopoulos & Panetsos, 1987). Following the Messinian climatic crisis (Suc, 1984; Palamarev, 1989) and the Pliocene marine transgression, the ancestor of the northern Mediterranean firs would have split into two groups, one on the Balkan Peninsula and the other in Asia Minor.

As the current eastern populations of A. alba are more similar to the Pontic fir (A. bornmuelleriana) than they are to A. cephalonica, it has been suggested that A. alba may have originated in the Pontic region during the Pliocene (see Fig. 2a; Fady et al., 1992). This dynamic new taxon would have then spread throughout central and western Europe during the Pliocene (Fig. 2a). Indeed, Pliocene macrofossils attributed to A. alba have been discovered in western Europe (Gaussen, 1964; Liu, 1971). However, the occasional land bridges probably maintained gene flow between the Pontic populations of the A. bornmuelleriana ancestor and the eastern Balkan populations of the A. alba ancestor (see Fady et al., 1992).

After the Pliocene marine transgression of the Aegean basin, A. alba would have appeared in the northern part of the Aegean and then colonized all of Europe, while A. bornmuelleriana would have appeared in the eastern part (Fig. 2a; Pignatti, 1978; Mitsopoulos & Panetsos, 1987). Frequent Pleistocene contacts between the two species (Fig. 2b) could have contributed to the individualization (the occurrence of geographic isolation, local adaptations and genetic differentiation of a new taxon) of A. nordmanniana, and thereafter to the individualization of A. equi-trojani, and furthermore could have given rise to the particular morphology of the south-eastern ecotypes of A. alba (Kaya et al., 2008).

In the western Aegean basin, A. cephalonica populations would also have appeared after the Pliocene (Fig. 2a). Southern migrations of A. alba during the ice ages and contact with A. cephalonica in glacial refugia probably led to the development of A. borisii-regis hybrid populations (see Fig. 3). Indeed, evidence for gene flow with A. alba has been suggested for the southernmost A. cephalonica populations by Fady et al. (1992). Moreover, in northern Greece and the Balkans, hybrids between A. cephalonica and A. alba appeared (A. borisii-regis) during the glaciation period, when the two taxa came into contact (Fady & Conkle, 1992; Scaltsoyiannes et al., 1999).

Figure 3.

 Current distribution and diversity of Mediterranean fir (Abies) species. 1, A. alba; 2, A. cilicica; 3, A. pinsapo; 4, A. numidica; 5, A. cephalonica; 6, A. bornmuelleriana; 7, A. nordmanniana; 8, A. equi-trojani; 9, A. borisii-regis (A. × borisii-regis = A. alba × A. cephalonica); 10, A. nebrodensis; 11, A. pinsapo var. marocana; and 12, A. pinsapo var. tazaotana. Post-glacial expansion routes hypothesized for A. alba based on molecular data and fossil records appear as arrows.

Based on allozyme differentiation, Scaltsoyiannes et al. (1999) also proposed a monophyletic evolutionary hypothesis for the genus Abies in the Mediterranean, according to which an ancient progenitor fir existed at the beginning of the Miocene in the Balkan Peninsula, in the Aegean, between the Ionian Sea and Turkey. During the evolution of Abies species, an ancient progenitor gave rise to A. alba and A. nordmanniana to the north and to A. cephalonica to the south in response to the wide range of ecological features, while the central area remained occupied by the ancient progenitor (Fig. 2). Under this hypothesis, at the end of the Pliocene and during the Pleistocene, species such as A. cephalonica, A. alba, A. nordmanniana, A. cilicica and A. pinsapo would have become differentiated (Scaltsoyiannes et al., 1999; see also Fig. 2a).

In summary, within the silver fir group, the eastern A. alba populations appear to be closely related to A. nordmanniana and A. bornmuelleriana (Scaltsoyiannes et al., 1999; Ziegenhagen et al., 2005). A relationship between A. alba and A. bornmuelleriana was also found by Arbez (1969), while the Calabrian silver firs show great differentiation from the rest of the A. alba populations (Konnert & Bergmann, 1995; Parducci et al., 1996; Scaltsoyiannes et al., 1999). On the other hand, central Apennines silver fir populations appear to be intermediate between the typical A. alba population and the Calabrian one (Scaltsoyiannes et al., 1999). The Pleistocene glacial cycles probably forced the eastern populations of A. alba to migrate towards Asia Minor and, in glacial refugia, to come into repeated contact with the populations derived from the Pontic fir (A. bornmuelleriana). This was made possible by a general lowering of the sea level (100–120 m below the current sea level). Gene exchange probably occurred and could be responsible for the present biochemical and morphological differences between the eastern and western populations of A. alba (Konnert & Bergmann, 1995; Hewitt, 2004). For instance, eastern European A. alba specimens have brush-shaped needles on current-year branches comparable to those of A. equi-trojani and A. bornmuelleriana (Fady et al., 1992).

On the other hand, the Greek A. cephalonica appears to be closely related to A. equi-trojani and A. borisii-regis (Kaya et al., 2008). The great distinction within A. cephalonica populations (Scaltsoyiannes et al., 1999) led to the proposal of two varieties (A. cephalonica var. cephalonica and A. cephalonica var. graeca, with A. equi-trojani belonging to A. cephalonica var. graeca; see Liu, 1971).

An alternative explanation to the above-mentioned relationships between A. cephalonica var. graeca and A. equi-trojani is provided by Fady et al. (1992), who hypothesized that gene exchanges between A. nordmanniana and A. alba could have been responsible for the emergence of A. equi-trojani (see Fig. 3), which has the characteristics of a relatively new species (Arbez, 1969) and strong similarities in its terpene composition to the A. alba populations of Romanian provenance (Fady et al., 1992). On the other hand, southern Pleistocene migrations of A. alba must have led to gene exchange with A. cephalonica (which had hypothetically evolved from the Balkan firs) in glacial refugia (Figs 2 & 3). These exchanges probably contributed to the origin of A. borisii-regis, in which the present natural distribution is the result of multiple introgressions with its parent species (Fady & Conkle, 1992).

Although an eastern Mediterranean Tertiary ancestor for present-day Mediterranean fir species seems likely, the archaic characteristic of the western Mediterranean fir species (A. pinsapo and A. numidica; see Xiang et al., 2009) is widely recognized, supporting the hypothesis that this common ancestor had a broad circum-Mediterranean distribution prior to the successive isolation and speciation of distinct populations (Figs 2 & 3).

The south-east to north-west speciation-sequence hypothesis

Probably since the middle Miocene, and mainly during the Pliocene (Fig. 1), increasing aridity led to a contraction in the range of the ancestral fir species (Kovar-Eder et al., 2008). However, the establishment of this regional dryness trend would not have been spatially simultaneous throughout the Mediterranean Basin. Recent findings suggest that the long-term Mediterranean Basin dryness could have spread from the south-east to the north-west (Fauquette et al., 1999, 2007; Bruch et al., 2006; Jost et al., 2009). Studies such as those associated with the ‘Environments and Ecosystem Dynamics of the Eurasian Neogene’ (EEDEN) project of the European Science Foundation have focused on the palaeoclimatic reconstruction of the period from the mid-Miocene (c. 14 Ma) to the mid-Pliocene (c. 3.6 Ma). These studies have shown that this period was characterized by deep palaeogeographical changes and climatic shifts (Bruch et al., 2006; Popov et al., 2006; Kovar-Eder et al., 2008; Jost et al., 2009) related to plate movements (e.g. Alpine uplifts, closing and opening of marine gateways). The results show that the climate was warmer than it is today throughout the Neogene, that the transition from a weak latitudinal thermal gradient (from around 0.48 °C per degree in latitude to 0.6 °C per degree in latitude) took place at the end of the Miocene, and that the latitudinal precipitation gradient was, from the mid-Miocene to the mid-Pliocene, more accentuated than it is today, with higher precipitation than today in the north-western Mediterranean but conditions drier than or equivalent to today in the south to south-eastern Mediterranean region (Fauquette et al., 1999; Jost et al., 2009). In this sense, some of the above-discussed phylogenetic studies, based on allozyme differentiation, support a south-east to north-west Abies speciation sequence in the Mediterranean Basin (Scaltsoyiannes et al., 1999; Fig. 1). At a smaller spatial scale, based on tree morphology and monoterpene composition, Mitsopoulos & Panetsos (1987) also found a south-east to north-west speciation pattern for Greek firs in the Peloponnesian peninsula.

Under this south-east to north-west speciation-sequence hypothesis (Fig. 1), the first isolated group would have been the ancestor of the current Turkish A. cilicica. Indeed, this species was identified as the oldest Mediterranean fir based on allozyme differentiation (Scaltsoyiannes et al., 1999) and has enclosed bracts, characteristic of the morphological archaic group. However, it has been reported that A. cilicica, A. cephalonica and A. nordmanniana appear very close anatomically (see Esteban et al., 2009), and have significant genetic proximity according to chloroplast DNA markers (see Parducci & Szmidt, 1999) and sequences of the complete nuclear ribosomal internal transcribed spacer (see Xiang et al., 2009).

The second isolated group would have been the ancestor of the Algerian A. numidica, and finally, the Spanish and Moroccan A. pinsapo (Fig. 1; Xiang et al., 2009). Analysis of chloroplast DNA markers (Parducci & Szmidt, 1999) and mitochondrial markers (Ziegenhagen et al., 2005) have placed A. numidica genetically close to A. pinsapo, as these two species share haplotypes. The early isolation of A. cilicica, A. numidica and A. pinsapo would account for features such as their low heterozygosity and high gene differentiation.

Although A. cilicica, A. pinsapo and A. numidica belong to the same taxonomic subsection (Pinsapones Franco) and are characterized by enclosed cone-bract scales, great genetic distances between them were found by Scaltsoyiannes et al. (1999). These contrasting results could be attributed partially to the prolonged isolation of the species and also to their restricted distribution and the subsequent genetic drift (Parducci & Szmidt, 1999).

Abies cilicica, A. numidica and A. pinsapo grow in the same regions as Cedrus species (A. cilicica with Cedrus libani; A. numidica and A. pinsapo with C. atlantica), suggesting that firs and cedars may have a common biogeographic history in Africa, southern Turkey and the Middle East (Scaltsoyiannes, 1999; Scaltsoyiannes et al., 1999; Qiao et al., 2007). For instance, fossil records from the Miocene and Pliocene (see also Fig. 1) have identified Cedrus sp. (Cedrus vivariensis N. Boul. = C. miocenica Lauby) and Abies saportana Rér. in the eastern Iberian Peninsula (Alvarez Ramis & Golpe-Posse, 1981), France, Poland and Bulgaria (see Palamarev, 1989). Moreover, Abies species from the Himalayas (e.g. A. pindrow, which grows in mixed stands with Cedrus deodara) could possibly be related to the ancestral Mediterranean fir species. Indeed, A. pindrow has enclosed bract scales (Vidakovic, 1991), as in the pinsapo fir-type archaic group, and, with A. pinsapo, it forms an interspecific hybrid (A. vasconcellosiana Franco; see Klaehn & Winieski, 1962), despite the fact that both species (A. pinsapo and A. pindrow) have probably remained geographically isolated since their origin (Xiang et al., 2009).

Phylogenetic studies show that Abies and Cedrus originated simultaneously (Magallón & Sanderson, 2005), although it is not possible to infer a common biogeographic history. The present revision suggests a northern Holocene expansion of the Abies genus over the entire Mediterranean Basin (Fig. 3), while Cedrus appears to have expanded throughout the southern Mediterranean Basin from glacial refugia, such as those located in the Rif Mountains of northern Morocco (Cheddadi et al., 2009), where remnant A. pinsapo populations are located (see Fig. 4). However, Cedrus is not present today in Spain, Italy, or the Peloponnesian peninsula.

Figure 4.

 The Rif Mountains (northern Morocco) experience high precipitation, essentially as a result of the orographic uplift of air charged with moisture from the nearby coast. These conditions probably also occurred during the glacial period, buffering the impact of regional aridity on tree populations. At present, these refugia may aid the persistence of relict species (Abies pinsapo, Cedrus atlantica, Acer granatense, Sorbus aria, Taxus baccata, Ilex aquifolium, Pinus nigra, etc.). Photograph: J. C. Linares.

Based on the spatial climate patterns and gradients reported for the Neogene (Fauquette et al., 1999; Bruch et al., 2006), the early south-eastern to north-western isolation of some Mediterranean firs could also have affected the ancestor of the Greek A. cephalonica (Mitsopoulos & Panetsos, 1987) and even of the Sicilian A. nebrodensis (Parducci et al., 2001a,b). However, repeated overlaps with northern species or subspecies that migrated following long-term climatic oscillations, from the Balkan range (to the southern Peloponnesian peninsula) and from the Alps (to the southern Italian peninsula), could explain modern features in these species (Mitsopoulos & Panetsos, 1987; Bennett et al., 1991; Tzedakis, 1993).

The origin and the phylogenetic relationships between A. nebrodensis and A. alba have not been definitively identified. Some authors have hypothesized that A. nebrodensis originated from the southern silver fir (A. alba) populations as a consequence of isolation during post-glacial periods (Fig. 3; see Parducci et al., 2001b, and references therein). Results based on allozyme, chloroplast DNA and random amplified polymorphic DNA show a clear differentiation among A. nebrodensis and A. alba populations and provide support for their classification into two distinct taxonomic groups (Vicario et al., 1995). Nevertheless, A. nebrodensis showed a closer affinity to A. alba populations from southern Italy, suggesting that this species may have originated through a past hybridization event (Parducci et al., 2001b) or through genetic drift resulting from the prolonged isolation of southern populations.

Past and present roles of glacial refugia

The current natural distribution of Abies species around the Aegean and central Europe was inherited from Holocene colonization patterns from the last glacial-age refugia (see Fig. 3). The Mediterranean area may have been the starting point for a wave of Abies colonization towards central Europe (Konnert & Bergmann, 1995; Hewitt, 1999; Tzedakis et al., 2002; Muller et al., 2007; see also Fig. 3). The strong human pressure (agriculture and grazing), c. 7500–8000 yr bp (see Pons & Quézel, 1985), caused the decline and geographical disjunction of temperate fir forests.

As suggested by palaeopalynological and palaeoclimatic studies (see Tzedakis et al., 2002, and references therein), the factors that lead to the currently high precipitation in some mountainous areas of the Mediterranean Basin (essentially, orographic uplift of air charged with moisture from the nearby coasts, see Fig. 4) also operated during the glacial period, moderating the impact of regional aridity on tree populations (Jost et al., 2005). At present, the role that these refugia may play in allowing the persistence of relict species and the possibility of further migration in response to future climate change, as has happened repeatedly in the past, is subject to a number of uncertainties (Taberlet & Cheddadi, 2002; Terhürne-Berson et al., 2004; Cheddadi et al., 2006, 2009). The intense habitat fragmentation and degradation as a result of human action, in addition to such ecological factors as low genetic variability and low population sizes, increase the extinction risk of some of these Mediterranean firs, for example A. numidica and A. pinsapo (Linares & Carreira, 2009; Linares et al., 2009; Rodríguez-Sánchez et al., 2010). Further interdisciplinary research integrating climate models, biogeographic studies and palaeoecological as well as genetic data are needed for a better understanding of the role of these geographic areas, which have acted as refugia in both the past and present (Médail & Diadema, 2009; Alba-Sánchez et al., 2010; Rodríguez-Sánchez et al., 2010).

Concluding remarks

Since the Miocene, and mainly during the Pliocene, increasing aridity has led to increasing fragmentation and contraction of the range of ancestral Mediterranean fir species, probably following a south-eastern to north-western sequence. The speciation of the ancestor of the current Turkish A. cilicica, the ancestor of the Algerian A. numidica, and the ancestor of the Spanish and Moroccan A. pinsapo has been related to this regional dryness. The selective pressures over these early isolated species might be responsible for the current low heterozygosity and high gene differentiation among them, and also between them and the other Mediterranean fir species.

Based on palaeoclimate studies, the early isolation could also have affected the ancestor of the Greek A. cephalonica and even the Sicilian A. nebrodensis. However, repeated geographic overlaps with northern species or subspecies that migrated following long-term climatic oscillations, from the Balkans range to the southern Peloponnesian peninsula and from the Alps to the southern Italian peninsula, could explain the modern morphological features and complex genetic relationships for these species.

In the north-eastern Mediterranean Basin, following the Miocene to Pliocene climate crisis and marine transgressions, the ancestor of the northern Mediterranean firs would have been separated into two eastern groups, one on the Balkan Peninsula and the other in Asia Minor, despite the occasional land bridges that probably maintained gene flow.

During the Pleistocene glacial cycles, the southward migration of A. alba to glacial refugia, where older fir species may have remained isolated, could explain recent findings that establish closer than expected relationships between distant species based on morphological classification.

The orographic complexity of the Mediterranean Basin played a key role in the evolution and the persistence of the Mediterranean firs. Factors that currently lead to high precipitation in some coastal mountain areas also operated during the glacial period, moderating the impact of regional aridity on tree populations. Today, however, the role of these refugia in the persistence of relict species could be limited by rapid climatic change and human-induced land degradation.

Acknowledgements

I would like to thank David Nesbitt for English revision, and F. Steininger and J. Eder for palaeoclimatic information. I also thank M. Ebach, R.J. Whittaker and A. Whittaker, and two anonymous referees for their very constructive and thorough comments on the manuscript.

Biosketch

Juan Carlos Linares is currently an assistant professor in ecology at the University Pablo de Olavide (Seville, Spain). His research is focused on topics related to climate variability as well as land use changes and their impact on relict Mediterranean mountain forests (mainly Abies, Cedrus and Pinus). The main challenge of his research has been to disentangle the interacting effects of long-term climate trends, climate variability and land use on forest dynamics from the landscape scale to the individual (ecophysiology).

Editor: Malte Ebach

Ancillary