Wood anatomical relationships within Abies spp. from the Mediterranean area: a phyletic approach

An analysis was made of the wood anatomy of seven species, one subspecies and two varieties of the genus Abies from the Mediterranean area (A. alba, A. borisii-regis, A. cephalonica, A. cilicica, A. nordmanniana, A. nordmanniana subsp. equi-trojani, A. numidica, A. pinsapo, A. pinsapo var. marocana and A. pinsapo var. tazaotana) in order to find phyletically relevant features to help clarify and complement the evolutionary patterns based on molecular studies. The wood structure within the genus was qualitatively similar, except for specific features characteristic of certain provenances. However, the wood biometry allowed the different taxa to be grouped in accordance with their anatomical similarity. A. alba, A. pinsapo, A. pinsapo var. marocana and A. pinsapo var. tazaotana possess biometric features which distinguish them from the other Mediterranean firs. Furthermore, A. numidica showed biometric features which distinguish it from the eastern firs (A. borisii-regis, A. cephalonica, A. cilicica, A. nordmanniana and A. nordmanniana subsp. equi-trojani) and place it closer to certain Iberian populations of A. pinsapo. The maximum ray height in number of cells, frequency of rays with more than 30 cells and tracheid length can be regarded as specific patterns of the Mediterranean fir phylum.


Introduction
The Mediterranean firs, consisting of eight species, one subspecies and two varieties (Farjon, 2001), form a particularly homogeneous clade.They can be divided into two clearly distinct groups (Fig. 1 (Flous, 1936).Chloroplast DNA variation showed that the Mediterranean firs form a clade with closely related species which are very different from the Asian and North American species (Xiang et al., 2002;Kormutak et al., 2004).
Various hypotheses have been put forward as to how the genus Abies evolved, on the basis of several features such as morphological and anatomical (Dallimore and Jackson, 1923;Mattfeld, 1930;Flous, 1936Flous, , 1937;;Turill in Klaehn and Winieski, 1962), botanical (Pignatti, 1978;Quézel, 1978;Medus and Pons, 1980;Soto, 1998) and ecological (Arbez, 1969;Quézel et al., 1980;Mayer, 1981), in addition to pollen analysis (Aytug, 1959), phenolic composition (Granados and Rossell, 1966), terpene composition (Koedam, 1981;Fady, 1990), DNA analysis (Vicario et al., 1995;Parducci and Szmidt, 1999;Ziegenhagen et al., 2005), and allozyme analysis (Parducci et al., 2001).Fady et al. (1992) and Scaltsoyiannes et al. (1999) 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 speciation period of Abies spp.an adaptive projection of the ancient progenitor took place, resulting in the formation of A. alba and A. nordmanniana to the north and A. cephalonica to the south in response to the wide range of ecological features, while the central area remained occupied by the ancient progenitor.At the end of the Pliocene and during the Pleistocene, species such as A. cephalonica, A. alba, A. nordmanniana, A. cilicica and A. pinsapo became differentiated (Scaltsoyiannes et al., 1999).
Although it has been demonstrated that wood anatomy constitutes an important source of phylogenetic information which should be taken into consideration in systematics studies (Lens et al., 2007), the establishment of evolutionary patterns based on the wood structure of conifers has been studied very little (Jeffrey, 1905;Chamberlain, 1935;Flous, 1936;Greguss, 1955;Carlquist, 1975).The lack of defined evolutionary patterns may be due, among other reasons, to the homogeneity of conifer wood, which is particularly accentuated in the genus Abies.In fact, together with Pseudolarix, the genus Abies shows the simplest structure of the Pinaceae (Greguss, 1955), although at the same time it includes the largest number of primitive features of all the Abieteae (Abies, Cedrus, Tsuga, Pseudolarix) (Jeffrey, 1905).
The objective of this study was to define anatomical similarities and differences between the wood of the Mediterranean firs, in order to establish an evolutionary hypothesis related to the wood structure using the most accepted phylogenetic patterns.

Materials and methods
The samples of A. alba and A. pinsapo were collected in their natural forests: A. alba in the Martinier-La Mascarina forest in the municipality of Forcayo El Cornato, in the province of Huesca; A. pinsapo in the three natural areas of the species in the south of Spain (Sierra del Pinar de Grazalema, in the province of Cádiz and Sierra Bermeja and Sierra de las Nieves, in the province of Málaga); and A. pinsapo var.marocana and var.tazaotana in Talassemtane National Park (Morocco), from mounts Magot and Tazaout, respectively.
Five trees more than 70 years old, representative of the forest, were felled in each zone.In the case of A. pinsapo the natural forests were located using the publications by Ceballos and Martín Bolaños (1930) for the province of Cádiz and by Ceballos and Vicioso (1933) for the province of Málaga.In the case of the Moroccan forests, the studies of Ceballos andMartín Bolaños (1928) andSánchez Cózar (1946) were used.
The samples of A. borisii-regis, A. cephalonica, A. cilicica, A. nordmanniana, A. nordmanniana  The wood samples were prepared for observation by means of light microscopy and scanning electron microscopy SEM mod.JEOL JSM-6380.The samples for light microscopy were prepared following the usual methods of softening, cutting and mounting, and were observed both without staining and after staining with safranine and Sudan 4 in order to make the resin red (Jane, 1970).The anatomical descriptions were made in accordance with the IAWA Committee (2004).
The measurements of tracheid diameter, tracheid length, tracheid pit diameter, height and frequency of rays and number and size of cross-field pits were made on the light microscopy slides using the WinCell image analysis program.The tracheid length was measured following Ladell's light method (Ladell, 1959).In the case of the samples collected from their natural areas, the biometry was done on 5 microslides from each tree, in all cases on mature wood from a basal disc between rings 70 and 100.On each slide 35 random measurements were made.The number of pits per cross-field was measured on the five slides from each tree, on ten different rays.The height and frequency of the rays were measured on all the rays contained in one square millimeter in the tangential section.
For the samples from collections, 35 random measurements for each feature were also made on a single slide, following the same criteria.
The most frequent average ray height (nº cells and µm) was obtained by fitting the distribution of the ray height frequency to a normal distribution curve, with the maximum of this distribution constituting the most frequent value.For the analysis of the other variables measured, the Lilliefors normality test and Bartlett's homogeneity of variance test were applied (α = 0.05).Because at least one of the conditions was not met in all cases, the Kruskal-Wallis test was used to compare the different Abies populations.The mean values were compared through LSD tests (α = 0.05).
A principal component analysis (PCA) was performed in order to observe possible groupings among the different Abies taxa according to their wood biometry.This method extracts a set of uncorrelated variables as linear combinations of the original variables, reducing the dimensionality of data while preserving most of the variance.The new variables, called components, are arranged in order of decreasing variance and in this way those with the highest variance are termed PCs.All the parameters measured (Table 1), including the maximum ray height (nº cells), were used as input variables.The PCs were displayed graphically as scatter plots of the scores, in order to observe any groupings in the dataset.In a PCA, coefficients by which the original variables must be multiplied to obtain the PCs are called loadings.The numerical value of a loading for a given variable in a PC shows how much the variable has in common with the component (Massart et al., 1988).In this way, loading values were used to ascertain the main anatomical 7.5 ± 1.2 (4.7-10.9) 3.8 ± 0.9 (1.7-6.8)2.6 ± 0.9 (1.1-5.2) 3.0 ± 0.9 (0.9-6.0) 1 Nº pits per cross-field 2.0 ± 0.8 (1-4) 1.9 ± 0.6 (1-4) 1 Nº rays/mm 2 > 30 cells 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.025 ± 0.041 0.002 ± 0.009 0.021 ± 0.047 0.0 ± 0.0 0.004 ± 0.017 1.8 ± 0.9 (1-4) 2.3 ± 0.9 (1-4) 1 Nº rays/mm 2 > 30 cells 0.040 ± 0.054 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 variables related to the localization of the samples with respect to the PCs in the scatter plot of the scores.The statistical calculations were done with the MAT-LAB V.6.5 Release 13 program.The ray height frequency distribution curves were fitted using Curve Fitting Toolbox Version 1.1 (R13) from the same program, for an interval of confidence of 95%.

Anatomical description
The wood anatomy of the different taxa studied was in general qualitatively similar.Therefore, the following description refers to all the Mediterranean firs except in the case of particular anatomical features observed in certain provenances.
Cross section.The sapwood and heartwood are indistinct.The growth ring boundaries are distinct (Fig. 2a).In some species the fresh wood is fetid, e.g. A. pinsapo.Tracheids are rectangular or hexagonal in outline in earlywood and rectangular in latewood.Tracheid pitting in tangential wall, generally in tracheids near the growth ring boundary (Fig. 2b).Axial parenchyma present but sparse, generally in marginal position, in isolated cells with high cellular content (Fig. 2c).Resin canals absent.Occasional presence of traumatic resin canals (Fig. 2d.1), surrounded by subsidiary cells with high cellular content (Fig. 2d.2) and crystals.Organic deposits in axial tracheids adjacent to rays in the Grazalema provenance of A. pinsapo (Fig. 2e).Tori well defined (Fig. 2f), not scalloped, but extensions present.
Tangential section.Maximum ray height (nº cells) high in the western-most firs, particularly in A. alba and A. pinsapo (Fig. 2g).Rays almost exclusively uniseriate, as partially biseriate rays constitute less than 10% of the total (Fig. 2h).

Biometric study
The biometric characterization of the different fir groups is shown in Table 1.
For all the variables analyzed with the Kruskal-Wallis test, the provenance factor was significant (P < 0.01).For the tracheid diameter, the Spanish and Moroccan pinsapo firs and A. alba showed values significantly higher than the other Mediterranean firs (P < 0.05).Similarly, the Moroccan pinsapo firs showed larger tracheid diameters than their Spanish counterparts (P < 0.05).In terms of the tracheid pit diameter, A. pinsapo from Sierra Bermeja, the Moroccan firs, A. alba and A. borisii-regis form a distinct group from the others, with significantly higher values (P < 0.05) than A. numidica, the Grazalema and Sierra de las Nieves pinsapo firs, A. nordmanniana, A. nordmanniana subsp.equi-trojani, A. cephalonica and A. cilicica.In relation to the tracheid length, A. numidica showed significantly lower values (P < 0.05) than the other provenances, with the exception of the Grazalema provenance of A. pinsapo, and the Grazalema and Sierra de las Nieves pinsapo firs showed shorter tracheids than the Moroccan firs (P < 0.05), while the Sierra Bermeja firs occupy an intermediate position between the two groups.The most frequent average ray height (nº cells) showed maximum values in A. cilicica and A. numidica.There is a very obvious difference between the most frequent ray height in the Grazalema provenance of A. pinsapo (8 cells) and the Sierra Bermeja and Sierra de las Nieves pinsapo firs (4 and 5 cells, respectively).In the case of the maximum ray height (nº cells), two different groups of A. pinsapo can be distinguished: on the one hand, the Grazalema and Sierra de las Nieves pinsapo firs, with 40 and 50 cells respectively, and on the other, the Sierra Bermeja pinsapo firs, with 36 cells, and the two Moroccan varieties, with 25 in var.marocana and 30 in var.tazaotana.These values place the Grazalema and Sierra de las Nieves wood closer to the biometry of the rays of A. alba (48 cells) than to that of the Moroccan firs.With regard to the most frequent average ray height (µm), the high values observed in A. cilicica, A. numidica, A. pinsapo from Grazalema and A. alba are notable in comparison with the other populations.The number of rays per mm 2 in A. numidica and A. alba was significantly higher (P < 0.05) than the values for the Spanish and Moroccan A. pinsapo and for A. borisii-regis and A. cephalonica.In terms of the largest cross-field pit diameter, the Spanish and Moroccan A. pinsapo, in addition to A. alba and A. numidica, showed higher values (P < 0.05) than the other populations studied.The smallest crossfield pit diameter in A. numidica was larger than in the other taxa (P < 0.05), except for A. borisii-regis, while in the case of the Moroccan varieties of A. pinsapo it was smaller than in the other taxa (P < 0.05), with the exception of A. nordmanniana.As regards the number of pits per cross-field, the Grazalema A. pinsapo showed significantly lower values than the other Spanish A. pinsapo and the Moroccan varieties (P < 0.05), but similar values to A. numidica.Lastly, all the eastern firs, including A. alba, with the exception of A. pinsapo var.marocana, had rays with more than 30 cells.

Principal component analysis
The PCA of the anatomical variables was used to obtain the scatter plot of the scores from the first two PCs (Fig. 4).PC1 groups the Iberian Peninsula firs, including A. alba, in the positive semi-axis with the Moroccan firs, and groups the other Mediterranean firs in the negative semi-axis.PC2 separates A. numidica and the Grazalema provenance of A. pinsapo, which have the most negative scores, from the other Mediterranean firs, placing A. alba in an intermediate position between the two groups.
The loadings of PC1 and PC2 are shown in Table 2.The parameters selected because of their high influence in PC1 according to their loadings (absolute value >  0.70) were: tracheid pit diameter, maximum ray height (nº cells), tracheid diameter, most frequent average ray height (nº cells) and number of rays per mm 2 with more than 30 cells.The parameters selected because of their high influence in PC2 according to their loadings (absolute value > 0.70) were: most frequent average ray height (µm), tracheid length and number of pits per cross-field.

Discussion
This study shows that the biometry of Abies wood from the Mediterranean Arc allows the different taxa to be grouped and distinguished according to their anatomical similarity.The grouping observed in the positive semi-axis of PC1 between the Spanish and Moroccan pinsapo firs (Fig. 4) coincides with the molecular similarity seen between these taxa in earlier studies with allozyme markers (Scaltsoyiannes et al., 1999) and chloroplast DNA markers (Parducci and Szmidt, 1999).The position of A. alba in relation to the PC1 axis, grouped with the Spanish and Moroccan provenances of A. pinsapo, coincides with the clade observed using allozyme markers (Scaltsoyiannes et al., 1999), but not with the results observed using DNA markers, in which A. alba is placed closer to the eastern firs (Parducci and Szmidt, 1999;Ziegenhagen et al., 2005).However, in A. alba high intraspecific variability has been observed, with unique haplotypes which distinguish it from the other eastern firs (Vendramin et al., 1999;Parducci et al., 2001;Ziegenhagen et al., 2005).The distribution of A. alba in Europe is very wide, encompassing the principle mountain areas of central and southern Europe (Fig. 1).The current geographical distribution of forest species in Europe is a result of post-glacial recolonization, which began around 13,000 ago years from the glacier refuges in the south of Europe (Huntley, 1990).The Pyrenees were likely to have been a glacier refuge of A. alba.After the glacial period, the populations in this area were isolated from the others, as suggested by the major genetic differences with neighboring populations (Vendramin et al., 1999).In view of this, phyletic studies which relate A. alba to other species of the genus should take into account the provenance of the population studied.Given the anatomical proximity observed between A. alba from the Pyrenees and A. pinsapo, particularly the Grazalema population, it would be interesting to study whether common genetic factors exist between them.For exam-ple, a common haplotype has been found to exist between A. alba and A. numidica (Parducci et al., 2001), despite the fact that they belong to two different subsections, suggesting contact between the two species in the past.
The species A. nordmanniana, A. cephalonica and A. cilicica appear very close anatomically on the scatter plot of the PCA scores (Fig. 4), corresponding to their genetic proximity observed using chloroplast DNA markers (Parducci and Szmidt, 1999).Near these species is A. borisii-regis, a taxon regarded as a natural hybrid between A. alba and A. cephalonica (Scaltsoyiannes et al., 1999).The genetic isolation of the Pyrenees populations of A. alba could explain why A. borisii-regis is positioned closer to A. cephalonica than to A. alba. A. numidica is far from this group according to PC2, in which it is located near the Grazalema provenance of A. pinsapo.Chloroplast DNA markers (Parducci and Szmidt, 1999) and mitochondrial markers (Ziegenhagen et al., 2005) place A. numidica genetically close to A. pinsapo, as they share haplotypes.It is possible that the Grazalema population, which in PC2 is positioned far from the other A. pinsapo provenances, has a closer phyletic relation to A. numidica than the other populations of A. pinsapo and its Moroccan varieties do.The closeness between the two groups is seen in the similar values for most frequent average ray height, tracheid length and largest cross-field pit diameter.
The finding of Abies fossils in different parts of the Iberian Peninsula has enabled some hypotheses on the phylogeny of Abies in Western Europe to be validated, in particular that established between the Pyrenees and the pinsapo fir enclaves.The fact that some of the fossil findings are similar to the firs now existing in the same region implies a certain degree of stability in the past and present geographical distribution of the genus in the Mediterranean since the Miocene (Viguié and Gaussen, 1928).Liu (1971) and Parducci (2000) proposed that the arrival of Abies in southern Spain and north Africa occurred through the Pyrenees.The route followed the Thetic bridges from the southeast of France → Majorca → Granada → Sierra de las Nieves → Sierra de Grazalema → Sierra Bermeja, reaching as far as Morocco, to the mountainous regions of Tazaout and Talassemtane, and the Babor mountains in Algeria (Fig. 5).On allowing this route and the direction of the southern phylum as that proposed by Dubois-Ladurantie (1941), which establishes from most primitive to most evolved as A. pinsapo → A. pinsapo var.marocana → A. pinsapo var.tazaotana → A. numidica → A. cilicica, it can be accepted that the speciation of Abies occurred after fragmentation of its territory.The fragmentation of the area of A. pinsapo into isolated populations must have caused reproductive isolation between them due to the already limited dispersal of pinsapo fir pollen (Arista and Talavera, 1994), and this situation may have been the start of the speciation of these populations and their subsequent morphological (Arista and Talavera, 1995) and genetic differentiation (Pascual et al., 1993).This in turn gave rise in the different species to a specific wood structure related to their antiquity rather than geographical proximity, as suggested by the fact that A. pinsapo var.tazaotana, although geographically much closer to A. pinsapo var.marocana, is nonetheless botanically closer to A. numidica.(Esteban et al., 2007).
Although there is no obvious relation between phyletic status and tracheid length, with the exception of Araucariaceae in comparison with the other conifer genera (Carlquist, 1975), tracheid length could be used as a phyletic pattern within the group of western firs.The fact that the tracheids are longer in the Moroccan firs (Table 1) than in the Spanish pinsapo firs was expected, as there is a correlation between plant size and tracheid length (Carlquist, 1975), and the Moroccan firs are larger.Only the pinsapo firs from Sierra Bermeja, although not located in their ecological optimum, are close to the Moroccan firs in terms of tracheid length.Thus it appears that tracheid length could be used as a phyletic pattern in the pinsapo firs and that the Sierra Bermeja provenance therefore occupies an intermediate phyletic position between the Spanish pinsapo firs and the Moroccan firs.This situation shows that, within certain limits, the geographical location did not influence the morphological speciation of the Moroccan firs.Similarly, the tracheid length places A. numidica closer to the Spanish pinsapo firs than to the Moroccan varieties.Some researchers maintain that the populations of A. pinsapo, A. pinsapo var.marocana and A. numidica could even have formed a single species before the final separation of the Iberian Peninsula from the north of Africa at the end of the Tertiary (Sánchez Cózar, 1946;Bocquet et al., 1978;Jeanmonod and Bocquet, 1981;Blanca, 1993).This shared past prior to speciation may have been the common vector of the southern phylum.
The maximum height of the rays and their frequency can be used to characterize the Abies species (Castellarnau, 1880;Wiesehuegel, 1932;Phillips, 1948;Greguss, 1955;Peraza, 1964).The biometry of the rays in Abies appears to provide evolutionary patterns.The maximum ray height (nº cells) allows two clades to be distinguished in A. pinsapo, coinciding with those established on the basis of tracheid length; firstly that of Sierra de Grazalema-Sierra de las Nieves, and secondly that of the Moroccan firs, while the Sierra Bermeja firs occupy an intermediate position.The wood of the samples from Sierra de las Nieves-Sierra de Grazalema presents a very similar structure to A. alba in terms of the maximum ray height in number of cells and the frequency of rays with more than 30 cells, although in A. alba rays with more than 30 cells are more frequent.The Moroccan provenances do not have this very high ray structure, despite their location in the ecological optimums of pinsapo fir forests, and therefore a relation between maximum ray height and ecological conditions is ruled out.
In general, the wood structure of the two Moroccan varieties is similar to the Iberian Peninsula firs, particularly the Sierra Bermeja provenance.Once again, this provenance could well present an intermediate situation, in phyletic terms, between the Grazalema and Sierra de las Nieves provenances, on the one hand, and the Moroccan firs on the other, in the same way as A. nebrodensis does between A. alba and A. numidica (Ducci et al., 1999;Parducci et al., 2001;Conte and Cristofolini, 2003).
The wood structure of the Moroccan firs is different from that of A. numidica, as seen in the separate position of the latter in the scatter plot of the PCA scores and in the values shown in Table 1.These results coincide with the distance observed between the two groups in their allozyme composition (Scaltsoyiannes et al., 1999).This distance may be imputed, in part, to extended isolation of the species and also to its restricted distribution and subsequent genetic displacement (Flous, 1936;Scaltsoyiannes et al., 1999).The same genetic divergence has also been observed between the populations of Abies pinsapo and var.marocana through the study of the inheritance of isozyme variations in their seed tissues (Pascual et al., 1993).
The wood samples of Abies spp.studied were in general qualitatively similar, except for particular features of certain provenances, as in the case of the organic deposits observed in the axial tracheids adjacent to the rays in the Grazalema provenances of A. pinsapo.The deposits, similar to those described in Araucariaceae (Jones, 1912;Patton, 1927;Heady et al., 2002), may be associated with a large amount of resin in the ray parenchyma which then passes into the interior of the tracheid lumen through the cross-field pitting (Heady et al., 2002;Esteban et al. 2005).The presence of resin canals should not be regarded as diagnostic, as they appear in Abies as a response to wounding (Anderson, 1897;Jeffrey, 1905;Vierhapfer, 1910;Rhoads, 1923;Chamberlain, 1935;Jane, 1970).However, the calcium oxalate crystals observed in the ray parenchyma of all the taxa, in both the marginal and submarginal rows, should be regarded as characteristic of Abies wood (Greguss, 1955;Jane, 1970;Core et al., 1979;IAWA Committee, 2004).All the species studied lack ray tracheids.Although they are recorded by some authors in A. balsamea (Penhallow, 1907), A. homolepis (Thompson, 1912) and A. veitchii (Jane, 1970), the presence of ray tracheids in Abies spp. is associated with wounding (Jeffrey, 1917;Chamberlain, 1935;Phillips, 1948).The degenerate cells in marginal position observed in A. pinsapo from Grazalema and Sierra de las Nieves have been described as a characteristic element of the genus Abies (Thompson, 1912).

Conclusions
The results obtained in this study show the potential of wood biometry in Abies, in conjunction with earlier molecular studies, for studying phyletic relations between taxa.
In the same way that features such as the stomata position, general cone shape, bract morphology, bud shape etc. can determine evolution into a phylum or clade and be constant in another phylum or clade (Flous, 1936;Dubois-Ladurantie, 1941), the features which determine evolution in wood can also establish a clade.The correlation between wood structure and climate conditions can make it difficult to use certain anatomical features (e.g.tracheid diameter) in studies on phyletic evolution.However, it was seen that geographical proximity was not a determining factor in the values of certain anatomical variables (e.g.ray height, tracheid length and number of pits per cross-field).Three main conclusions can be drawn from this study: i) the Mediterranean firs can be classified according to their wood anatomy into two large clades: the Spanish populations (including A. alba) and the Moroccan populations, on the one hand, and the rest of the Mediterranean populations on the other; ii) the most frequent average ray height, tracheid length and largest cross-field pit diameter place the populations of A. pinsapo from Grazalema and A. numidica close together; and iii) the maximum ray height in number of cells, frequency of rays with more than 30 cells, and tracheid length can be regarded as specific patterns of the southern phylum of the western Mediterranean firs.

Table 1 .
Biometry of Abies spp. in the Mediterranean

Table 1 .
Biometry of Abies spp. in the Mediterranean (continued)

Table 2 .
Loadings of the factors with the greatest influence in the principal component analysis