Introduction

“With the wealth of data available for Pinus, it is perhaps surprising that key phylogenetic relationships remain unresolved” (Syring et al. 2005).

The genus Pinus (Pinaceae) consists of around 110 species. It is the most abundant among the conifers, a group of 630 species (Farjon 1984, 2005). A century ago, following Koehne’s observations, Shaw (1914) proposed to split the genus into two subsets, Haploxylon (with only one fibrovascular bundle in the needles) and Diploxylon (with two fibrovascular bundles). The first subgenus (Strobus, “soft pines”) includes the Asiatic and North American pines with five needles per fascicle, while the second subgenus (Pinus, “hard pines”) possesses three or two needles with more or less persistent fascicle sheaths and populates the whole Paleartic region, including Middle Europe and the Mediterranean area.

This subgeneric differentiation was confirmed by Swedish chemists as early as the middle of the 20th century following structural analysis of phenolic substances isolated from the wood: “the consistency of the (phenolic) pattern appears typical of Diploxylon…. Subgenus Strobus has a more complex heartwood chemistry. The main feature is the occurrence of the stilbenes together with their hydrogenation products, the dibenzyls, and of the flavanones and their dehydrogenation products, the flavones.…It is possible that these chemical differences are due to a mutation in which case the Haploxylon pines could be regarded as being more ancestral than Diploxylon pines” (Norin 1972). Two decades later the advent of gas chromatography and high performance liquid chromatography allowed for the precise analyses of resins and oleoresins (terpenoids) and phenolics and other “micromolecules” respectively (Lebreton 1995). More recently relayed and even somewhat submerged by “macromolecular taxonomy” (proteins, DNA), chromatographic analyses have provided accurate and reproducible data which are very convenient for chemotaxonomical discussions at different levels. In addition phenolic compounds can act as dynamic markers depending on their positions in the biochemical pathway involved. Thus they can be ordered from “ancestral” (primitive) to “derived” (evolved).

Following four doctoral theses (Laracine (1984): Pinus sylvestris; Idrissi-Hassani (1985): Pinus pinaster; Lauranson-Broyer (1989): Pinus uncinata, mugo and nigra; Kaundun (1995): Pinus halepensis, brutia and eldarica) and several subsequent publications, the objective of this paper is to analyse flavonic data at the mesosystematic level (from species to genus). These are subsequently compared with macromolecular parameters (proteins, DNA) studied in the past decade. At the macrosystematic level (from genus to order), the abundance of proanthocyanidins (including prodelphinidin) and the absence of C-glycoflavones allows distinction of pines from the other Pinales; they are themselves characterised by the absence of biflavones, present in Cupressaceae and Araucariaceae (Lebreton 1990).

Materials and methods

Sampling

The present work is based on 45 taxa originating from 56 populations and a total of 620 individual plants. These include several species complex—for example Pinus grex nigra. Of the remaining 37 species 10 are from Strobus and 27 from Pinus subgenus. Morphometric and biochemical analyses were carried out on needles collected from forest trees in nature or in controlled plantations (French National Agricultural Research Institute; INRA) and botanical gardens (Les Barres Arboretum and Lyon Parc de la Tête d’Or, France).

Systematics

We have used the classification recently proposed by Gernandt et al. (2005), supported by the seminal works of Little and Critchfield (1969) and Price et al. (1998). Some considerations were given to classical works carried out by Gaussen (1960), Mirov (1967) and Krüssmann (1972). Number and size of needles were taken into consideration, as well as biogeographical localisation (New vs. Old World, Middle Europe vs. Mediterranean) and corresponding climates (hot vs. cold, dry vs. wet). In this way seven different biogeographical subsets were defined.

Biochemical analyses

Biochemical analyses were carried out on mature pine needles collected during winter and dried at room temperature in the absence of light. Following hot hydrochloric acid (2 N) hydrolysis in the presence of air, the anthocyanins generated from proanthocyanidins and the flavonol aglycones from native O-glycosides were extracted with relevant solvents and subsequently identified and measured with HPLC. Details of the biochemical analyses are included in similar works conducted in our laboratory (Idrissi-Hassani and Lebreton 1992; Lauranson-Broyer and Lebreton 1995; Kaundun et al. 1997).

Morphometric analyses

Morphometric analyses were also carried out on mature pine needles dried at room temperature. The length and weight of the needles were measured to the nearest 0.5 mm and 0.1 mg respectively and averaged on a minimum of ten fascicles for each individual tree (20–50 depending on the species).

Results and taxonomic findings

The raw data for each of the 45 taxa are given in Tables 1 (geography and morphometry) and 2 (biochemistry). The chemical values are expressed as the percentage of the sum of absolute contents (mg/g dry weight) of homologous substances, proanthocyanidins (two molecules) or flavonols (six molecules). For a better readability, the values are sometimes expressed to the nearest whole digit even if calculations were performed with an additional decimal. Two indices were defined based on flavonic data: (1) an index of phenyl-O-methylation (Me-O), which is the sum of the relative amounts of the methylated flavonols isorhamnetin, larycitrin and syringetin; syringetin, i.e. di-O-methyl-3′,5′ myricetin, was attributed a coefficient of 2; and (2) an index of phenyl-tri-hydroxylation (Tri-OH), which is the sum of the relative amounts of prodelphinidin and myricetin; the former was attributed a coefficient of 1/3 due to the presence of two proanthocyanidins and six flavonols. The results are further summarized based on taxonomical (subgenera, sections, subsections) (Tables 3, 4), morphological (Table 5) or geographical (Table 6) subsets.

Table 1 Geographical and morphometric characteristics of 45 Pinus taxa studied here
Full size table
Table 2 Biochemical characteristics (flavonoids) of 45 Pinus taxa studied here
Full size table
Table 3 Morphometric characteristics of Pinus subsets recognised in this study
Full size table
Table 4 Biochemical characteristics of Pinus subsets recognised in this study
Full size table
Table 5 Morphometric and biochemical data according to the number of needles
Full size table
Table 6 Morphometric and biochemical data according to geographical origins
Full size table

Morphometric aspects

Pines with two, three or five needles show very distinct needle lengths and weights. Likewise, if an index of slimness is defined by length (L) divided by weight (W) of a needle, clearly distinct values are obtained for each type of fascicle. L/W values are 3.36, 4.29 and 8.75 for pines with two, three and five needles, respectively (Table 5). A significant difference is even obtained between the groups of pines with two and three needles (t = 1.88; P = 0.03). These results tend to show that the greater the number of needles per fascicle, the finer the needles. There is an obvious and positive correlation between individual needle length and weight (r = +0.694; P ≪ 0.001). There is also a highly negative correlation between weight and fineness (r = −0.612; P ≪ 0.001): the longer the needles, the heavier they are and the heavier they are, the stockier the needles.

If the calculation is based on the whole fascicle (global slimness index GSI = L/n × W, where n is the number of needles), no significant differences are observed between the two subgenera Pinus and Strobus or between the two sections Pinus and Trifoliae. The same is true for subsections Pinus and Pinaster (1.6 < GSI < 1.7). Taking into consideration the ratio L/W does in fact mean implicitly using the number of needles as a taxonomical character. At this level, it is thus better to use the number of needles directly rather than having recourse to the L/W index. Nevertheless, the GSI can be useful at lower taxonomical levels, for example when series “densiflorae” is compared to “sylvestres”. The equivalence of the GSI at the intrageneric level as well as its relatively low dispersion around the mean for the 45 taxa [L/n × W = 1.64 (0.52)] leads to the conclusion that needle ontogenesis proceeds from an initial structure common to the genus Pinus as a whole, which was then subjected to subsequent dichotomies by ontogenesis.

Geographical aspects

If the samples are grouped based on the three relevant continents (considered at the present geological time), clear differences are observed for some characters (Table 6 means and standard deviations). Significant differences are observed between Asia and Europe, and Asia and Europe + America for the number of needles (respectively, t = 2.40, P = 0.01 and t = 1.98, P = 0.03) but neither for length nor fineness. For O-methylation (but not for phenyl-tri-hydroxylation), significant differences exist between Asia and America (t = 2.64; P = 0.007), Asia and Europe (t = 5.74; P < 0.001) and even between Europe and America (t = 3.10; P = 0.002). There is thus a relationship between geographical location, number of needles and intensity of O-methylation. If climatic criteria (cold vs. hot, wet vs. dry) are taken into consideration, seven principal biogeographical regions can be identified: south-eastern USA, southern USA with Mexico, western and northern America, eastern and south-eastern Asia, continental Asia, continental Europe and Mediterranean region. The differences are clear as far as the proportion of the number of species of the two subgenus Strobus and Pinus and the methoxylation index MeO are concerned (Fig. 3).

Correlation between variables (45 taxa × 12 traits)

The number of needles per fascicle (two, three or five) is significantly correlated (negatively) to the level of isorhamnetin (r = −0.482; P = 0.001) (Table 7). This relationship is accentuated with the O-methylation (=methoxylation) index MeO (r = −0.527; P < 0.001): the greater the number of needles, the lower the level of methylated flavonols. In fact, the methoxylation index differs more significantly when the difference in the number of needles is higher: between the pines with three and five needles, t = 1.92, P = 0.035; between two and three needles, t = 2.12, P = 0.020; between two and five needles, t = 6.59, P ≪ 0.001. Quercetin and isorhamnetin are negatively correlated (r = −0.418; P = 0.004); isorhamnetin derives from quercetin by O-methylation of the lateral phenyl group of this flavonoid.

Table 7 Correlations between morphometric and/or biochemical variables
Full size table

The level of prodelphinidin, a phenyl-trihydroxylated pro-anthocyanidin, is correlated to the relative amount of myricetin, a phenyl-trihydroxylated flavonol (r = +0.345), larycitrin = methyl-3′ myricetin (r = +0.274) and syringetin = dimethyl-3′,5′ myricetin (r = +0.289; P = 0.02–0.06), thus reflecting the existence of the phenyl-trihydroxylation pathway. There is indeed a highly positive correlation between myricetin and its two mono- and di-methylated derivatives (larycitrin, r = +0.684, P ≪ 0.001; syringetin, r = +0.475, P = 0.001) and between the latter two (r = +0.824, P ≪ 0.001). However, the number of needles is neither correlated to the tri-O-substitution index (r = +0.120, NS), nor is the latter to the methoxylation index MeO (r = −0.085, NS).

Differentiation between subgenera Pinus and Strobus

The two subgenera Strobus (=Haploxylon) and Pinus (=Diploxylon) taken globally differ from each other by their average number (t = 11.90, P ≪ 0.001), length (t = 2.44, P = 0.009) and individual weight of the needles (t = 4.12, P < 0.001). The needles of subgenus Strobus are shorter and lighter than those of the subgenus Pinus. Also, based on flavonoid data, subgenus Pinus is significantly richer in isorhamnetin than subgenus Strobus: 16 versus 4% (t = 5.55, P < 0.001). This complete differentiation between subgenus Strobus and Pinus has been recognized since long ago by many authors and disciplines (Bergmann and Gillet 1997; Kutil and Williams 2001) and thus validates our methodology based on flavonoid and needle characteristics.

Based on the species analysed here (which neglects section Parrya, including subsection Cembroides of Mexican pines), we observe the homogeneity of subsection Strobus Loud., which groups species formerly distinguished as Strobi and Cembrae by some authors (Klaus 1989; Liston et al. 1999). With the exception of Pinus bungeana, all other species analysed belong to the section Strobus (= Quinquefoliae). Gernandt et al. (2005), Wang et al. (1999) or Liston et al. (1999) did not differentiate among them either based on chloroplastic or ribosomal DNA analyses though P. strobus, P. peuce or P. parviflora were detached from the lot as subsection Strobi. Moreover, Pinus bungeana (subsection Gerardianae) is not biochemically distinguishable from the other species of the same subgenus, although it has fewer (n = 3) and heavier needles (average 23 vs. 11 mg). Thus, our discussion will be focused mainly on the morphometric and flavonic taxonomy of species of subgenus Pinus and its diverse sections and subsections, better represented in this study.

Structure of subgenus Pinus

The two sections Pinus (n = 23) and Trifoliae (n = 12) are completely distinguished by their average number of needles (t = 4.19; P < 0.001) but not by morphometry; the first is significantly richer in phenyl-trihydroxylated flavonoids (LD: t = 2.65, P = 0.006; Myr: t = 3.93, P < 0.001; Tri-OH index: t = 4.39, P < 0.001) and poorer in methylated flavonols (IRh: t = 3.56, P < 0.001; MeO index t = 2.19, P = 0.017). Thus, the number of needles and flavonic biochemistry validate the two sections recently explained by Gernandt et al. (2005).

The Nearctic section Trifoliae Duhamel is heterogeneous, with complete individualisation of subsection Contortae (enriched with P. muricata) according to various criteria: number of needles (n) (2.0 vs. 3.0, P ≪ 0.001), length (L) (161 vs. 50 mm, t = 4.60, P < 0.001), weight (W) (45 vs. 12 mg, t = 3.27, P = 0.005) and GSI (1.29 vs. 2.24, t = 4.95, P < 0.001). Myricetin content (8 vs. 13%, t = 1.87, P = 0.046) and Tri-OH index (38 vs. 45, t = 2.02, P = 0.036) are weaker, as are isorhamnetin (4 vs. 11%, t = 4.09, P = 0.001) and MeO index (9 vs. 21, t = 3.39, P = 0.004). Separated and amended, the two subsections with three needles, Australes (hereby reduced to three species of south-eastern America: P. palustris, rigida and taeda) and Ponderosae, share a common foliar biometry (L = 160 and 162 mg, GSI = 1.29 and 1.30) and a close combination of trihydroxylated flavonoids (Tri-OH index = 36 and 40). This conclusion can justify a posteriori the recognition by Gaussen (1960) of the subclass Taedoponderosoides which today is obsolete. Nevertheless, the methylated flavonol content allows individualisation of the two subsections: IRh = 2 versus 6% (t = 4.00, P = 0.005) and MeO = 3 versus 13 (t = 3.01, P = 0.015).

There remains the case of Pinus patula, a California pine with three needles per fascicle. The needles are of identical length to those of the two precedent similar subsections; this species is in fact positioned (but with P. muricata) by Gernandt et al. (2005) in the Southern subsection discussed earlier. However, the heavy and slim needles (GSI = 2.3 vs. 1.3; cf. subsection Contortae) do not favour this insertion, nor does a very low trihydroxylated flavonoid content (Tri-OH = 32) and a very high isorhamnetin content (IRh = 31%), the only methyl-flavonol present. For many authors (Little and Critchfield 1969; Price et al. 1998; Liston et al. 1999), Pinus patula should be placed in an Oocarpeae subsection, which was unfortunately not analysed here. In this context, this species is considered as incertae sedis within the Trifoliae section.

Contrary to the Trifoliae, section Pinus is heterogeneous from a geographical standpoint. It comprises some Nearctic representatives (such as P. resinosa) together with a few Palearctic members ranging from the Canary Islands (P. canariensis) to Japan (P. thunbergi). The first dichotomy proposed by our biometric and biochemical data properly separates the two subsections, Pinus and Pinaster, distinguished by Gernandt et al. (2005). The needles of the first subsection are relatively shorter on average (88 vs. 126 mm, t = 1.63, P = 0.057) and have lower levels of phenyl-trihydroxylated flavonoids (LD: 83 versus 94%, t = 3.62, P < 0.001; Myr: 1 versus 4%, t = 2.09, P = 0.023; Tri-OH: 29 vs. 35, t = 3.85, P < 0.001), and also low methylated flavonol content (IRh: 17 vs 28%, t = 2.71, P = 0.006; MeO: 17 vs. 36, t = 5.06, P < 0.001). This dichotomy is equally accepted by Wang et al. (1999) based on chloroplastic DNA data. Their study recognizes two clades in subgenus Pinus including Mediterranean pines on the one side and species from subsection Sylvestres on the other side.

The subsection Pinus in turn reveals a dichotomy based on geography. The holartic subgroup have very slim needles (GSI: 1.96 vs. 1.43, t = 4.00, P < 0.001), and the European subgroup are significantly richer in isorhamnetin (IRh: 21 vs. 8%, t = 3.04, P = 0.004; MeO index: 21 vs. 9, t = 2.80, P = 0.007). Dividing subsection Pinus into segments would make things clearer but would further complicate the phylogenetic scheme. We prefer not to question this subsection, but at the same time distinguish within it two series designated as “densiflorae” and “sylvestres”. Yet, an argument of coherence resides in the fact that subsection Pinus is the only one with species containing taxifolin (=dihydro-2,3 quercetin) (one European P. sylvestris, one North American P. tabuliformis, one Asian P. massoniana). The synthesis of taxifolin is genetically controlled in mountainous and meadow populations of Pinus sylvestris (Laracine-Pittet and Lebreton 1988; Lebreton et al. 1990; Yazdani and Lebreton 1991). Actually, on the basis of chloroplastic DNA, Krupkin et al. (1996) recognized that “the North American species P. resinosa is closely allied to P. sylvestris” (following the same analysis, P. canariensis is close to the Mediterranean pines P. pinea and brutia).

In fact, attention must now be focused on the distinction between the two European subgroups, “sylvestres” (series) and Pinaster (subsection), which questions the autonomy of the Middle European and Mediterranean pines, recognised by numerous authors (Schirone et al. 1991; Piovesan et al. 1993) following seed enzyme analysis. The morphometric arguments are inefficient due to the presence, in both cases, of long and/or heavy needles (salzmanii vs. pinaster) and short and/or light needles (uncinata vs. halepensis). Flavonic characteristics are however different; methylation of flavonols is certainly high in “sylvestres”, but lower than that of Pinaster: MeO = 21 versus 36 (t = 3.74, P < 0.001). This is mainly due to the almost complete absence of larycitrin (0.1 vs. 5%) in series “sylvestres”. The path of phenyl-trihydroxylation is equally less active in the latter series as compared to subsection Pinaster. The Tri-OH index is respectively 29 versus 35 (t = 4.26, P < 0.001), as a result of lower levels of prodelphinidin (LD: 85 vs. 94%, t = 4.83, P < 0.001) and myricetin (Myr: 1 vs. 4%, P < 0.001) in series “sylvestres”.

Our biochemical analyses therefore favour the recognition of a group of Mediterranean pines sensu lato, previously suggested by various authors including Klaus (1989) who clearly differentiates the Mediterranean pines from the “diploxyl mountain pines from the area surrounding the Mediterranean”. This is typically the case in the species complex Pinus nigra (laricio + salzmanii vs. nigricans + pallasiana), as mentioned below. On the other hand, we refuse to agree with same author on the fact that the Mediterranean pines could be in “close relationships with Central American and Caribbean pines”, which is rejected mainly based on MeO index or the abundance of two-needled pines in the Atlanto-Mediterranean area. Referring to these pines, Krupkin et al. (1996) made a close connection between P. canariensis and the Mediterranean pines P. pinea and brutia. Also Liston and co-workers (1999) group the canariensis and pinaster species with pinea and halepensis. The same year, Wang and co-workers connected P. halepensis, brutia and pinaster to pinea and canariensis. Moreover, the close relationship among the three species P. halepensis, P. brutia and P. eldarica has been documented by means of flavonols (Kaundun et al. 1997). The correlation between drought (summer) and O-methylation will be discussed below.

As a conclusion to this section devoted to the comparison of our biometric and biochemical results from a taxonomic (=static) point of view, we can acknowledge, on one hand, the boldness of the new data, and on the other hand, their general adequacy as compared to more recent propositions made by different authors (Lopez et al. 2002), even if certain modifications have been suggested and uncertainties exist here and there, which are discussed later on from a phylogenetic perspective.

Discussion and systematic aspects

Flavonic chemotaxonomy

The concept of “flavonic evolution” is based on two principles (Lebreton 1990). The first and of higher importance here is that the methylation of the hydroxyl of the lateral phenyl group is a derived character. This is founded upon the irreversible nature of phenyl-O-methyl-transferase activity and on its macrosystematic expression in the phanerogams (increasing from ligneous to herbaceous plants). In this respect, isorhamnetin is more “evolved” than quercetin from which it is derived; similarly larycitrin (monomethylated) and syringetin (dimethylated) are more advanced than myricetin.

The second principle consists of tri-hydroxylation of the lateral phenyl group of flavonoids being a primitive (ancestral) character, founded upon the primordial nature of the shikimate biosynthesis pathway and its macrosystematic expression in the phanerogams (decreasing from the gymnosperms to the monocotyledons). In this respect, prodelphinidin and myricetin are more “ancestral = primitive” than procyanidin and quercetin respectively.

Actually, the situation can be more complex: not only the second character can be masked by methylation of the phenolic group but the first is consecutive to the second one (syringetin, for instance, cannot appear if the myricetin pathway is closed), and both can be uncoupled. This can be illustrated with the species complex Pinus nigra Arn. of which the Calabrian taxa (LD = 84%, IRh = 12%) of the Mediterranean region can be considered as the strain of two (postglacial) phyla, respectively occidental/meridional (laricio and salzmanii: LD = 85 ± 4%, IRh = 21 ± 2%) and oriental/continental (nigricans and pallasiana: LD = 76 ± 1%, IRh = 11 ± 1%). On the first phylum, prodelphinidin remained constant whilst isorhamnetin took an upward trend; on the second, isorhamnetin remained stable and prodelphinidin took a downward trend (Lauranson-Broyer and Lebreton 1995). Overall based on the relative contents of myricetin, isorhamnetin, larycitrin, syringetin and prodelphinidin, two global indices were defined: an index of methoxylation, MeO, and an index of phenyl-tri-hydroxylation, Tri-OH (cf. section “Material and methods”). It is noteworthy that the methoxylation index is of greater amplitude than that of hydroxylation. The means and standard deviations are respectively 16 (12) and 34 (6) with variation coefficients 74 and 19%. Thus methylation generates more information than hydroxylation, the latter being of minor importance for the present discussion.

Phylogenetic relationships

Statistical analysis of the whole genus Pinus, coupled with that of its taxonomic subgroups (statistical comparison of means), shows a net negative correlation between the number of needles and amounts of O-methylated flavonols that are “derived”. It can therefore be safely concluded that a large number of needles, that is five needles as compared with three or two needles per fascicle, is an ancestral character. The same is true for subgenus Strobus when compared to subgenus Pinus. This proposition is in contradiction with Gaussen (1960), but concurs with Norin (1972). This being said, it would be unwise to conclude without any other considerations that there is a direct derivation of subgenus Pinus from Strobus. On the one side the phenyl-tri-O-substitution level does not differentiate between the two subgenera (Table 1b), and on the other side subgenus Pinus is manifestly heterogeneous in its two flavonic characteristics. In particular, section Trifoliae shows a significantly higher Tri-OH index than subgenus Strobus: 40 versus 35 (t = 2.57, P = 0.009). In this section, myricetin and its two methylated derivatives together represent not less than 12% of total flavonols, thus testifying to the persistence of the archaic shikimate pathway in contrast with the subgenus Strobus (only 5% for the total of the tri-O-substituted flavonols). The same holds for the subsection Pinaster (15%), making it most unlikely for subgenus Pinus to be directly derived from Strobus.

Under these circumstances a common original strain for the different species of Pinus and its subsets can be considered to have the following archaic characteristics: a large number of needles per fascicle (originally five), low level of O-methylation (<2) and high level of phenyl-tri-hydroxylation (around 45). As indicated in Table 7 the first two characters are highly correlated. From a phylogenetic point of view, a first dichotomy could have generated two groups, the first leading to subgenus Strobus, the second to subgenus Pinus, with subsequent subdivision into Neartic Trifoliae and Holartic Pinus sections (Fig. 1a, b). In the first case, the MeO index progresses less than in the second (14 vs. 23) and the Tri-OH index remains high (40 vs. 31); thus, Eurasiatic pines are globally more “evolved = derived” than the American pines. However, the situation is not homogeneous at other levels. In North America, two subsets appear, the first composed of subsections Australes + Ponderosae, with similar number and morphometry of needles, the second, Contortae, with two short and slim needles per fascicle (Fig. 1a). Taking into consideration the number of needles and MeO index, subsection Contortae is the more “evolved” of the group of American pines (Fig. 1b). This scheme agrees with the Dollo phylogenetic tree proposed by Govindaraju et al. (1992) on the basis of ribosomal DNA but differs from the Wagner tree, where the Ponderosae group is separated from Australes and Contortae. In the classification presented by Syring et al. (2005), P. contortae is detached from P. ponderosa + P. taeda . According to Krupkin et al. (1996), the “subsection Contortae emerged as a sister group to all the other North American pines”. Contrary to Syring et al. (2005), we do not believe that “the position of (sub)sect. Contortae remains equivocal”, and its evolution in the New World can be compared to that of subsection Pinaster in the Mediterranean with maximal methoxylation within these two subsections (Fig. 2).

Fig. 1
figure 1

a Phylogeny of genus Pinus and morphometrical characteristics. n Number of needles, L length of needle, GSI global slimness index. b Phylogeny of genus Pinus and biochemical characteristics. Myr Myricetin %, Tri-OH phenyl-tri-hydroxylation index, IRh isorhamnetin %, MeO phenyl-O-methylation index

Full size image
Fig. 2
figure 2

Phylogeny of genus Pinus and biological characteristics: number of needles/global slimness index//MeO index

Full size image

In the Holarctic section Pinus, MeO index and taxifolin distinguish clearly the two subsections Pinus and Pinaster, as seen in the former taxonomical discussion (cf. Fig. 1b). Thus, Pinus is more ancestral than Pinaster, in spite of a higher level of prodelphinidin and trihydroxylation index in subsection Pinaster. This is confirmed by the near absence of larycitrin and syringetin in subsection Silvestres on the one hand, and their substantial presence (together around 9%) in Pinaster on the other. In fact, despite the common weakness in substituted phenyl-tri-O flavonols and the presence of taxifolin, the subsection Pinus as proposed by Gernandt et al. (2005) is heterogeneous for two reasons: slimness of the needles (GSI = 1.96 vs. 1.43) and low levels of methylated flavonol (MeO index = 9 vs. 21), hence the acknowledgment of the two “series”, densiflorae and sylvestres, the latter being more evolved (=derived) than the former (Fig. 2). However, instead of concluding on a direct relationship (prohibited by geography), it seems more appropriate to propose two separate groups with distinct levels of evolution.

Biogeographical and ecophysiological phenomena

Over and above the few distinctions based on the strictly geographically oriented continental classification, we can also compare the influence of climate to the genetic/environmental interface, whereby biochemistry can be the eco-physiological mediator. In this respect, North America is not homogeneous with a north-western set which is cold, and two warm subsets, south-central and south-eastern, which are dry and humid respectively. Taking biogeography and the methoxylation index together, it can be emphasised that methylation is a chemophysiological character associated with lipophily and water balances for plants living in hot and dry regions. At equal masses, pines with five needles show a greater surface (thus more evapotranspiration) than those with two or three needles. This can explain the adaptive correlation between number of needles and MeO index. Yet, if both temperature and dryness are taken into consideration, seven “bioclimatic regions” A to G can be distinguished according to Critchfield and Little (1966). These include three regions in North America and four in Eurasia (Table 8 and map, Fig. 3). As for subsection Pinaster and the Mediterranean region, there is often a close correspondence between taxonomy and ecogeography: subsection Australes with south-eastern USA, subsection Ponderosae with southern USA and Mexico, and series “sylvestres” of the Pinus subsection with Middle Europe. Nevertheless, the collective species Pinus nigra is split between Middle and southern Europe.

Table 8 Bioclimatic dependence of biological characteristics (number of needles, MeO index)
Full size table
Fig. 3
figure 3

Bioclimatic regions and biological characteristics (MeO index/number of needles) within the genus Pinus

Full size image

In continental Asia and northern America, with simultaneously cold and wet climates, the number of needles remains high: 3.7 and 3.6, and the MeO index is low or medium: 5 and 14; for the Mexican region compared to northern USA, the number of needles and MeO index are similar. In Middle Europe, the specific characteristics are very different compared to the Mediterranean ones: 2.6 versus 2.1 and 20 versus 31, respectively. Actually, bringing together the different species of the halepensis and pinaster groups, as well as Pinus canariensis and Pinus pinea, the subsection Pinaster constitutes an atlanto-mediterranean subset particularly evolved (=derived) within the Eurasian category. At the opposite, when the climate is hot but wet, as in south-eastern USA and Asia, the number of needles decreases (to 2.6 and 2.7 respectively) whereas the MeO index remains low (7 and 7 respectively), contrasting with the situation in other latitudes.

On other hand, other latitudinal clines can be observed concerning biological segregation in the genus Pinus. First, the number of species increases, for example from one to seven from northern to southern Europe (calculated from Atlas Florae Europaeae, Jalas and Suominen 1973) and from 18 to 30 in North America (based on 100 species including the 37 taxa studied here). Second, the importance of the subgenus Strobus and, correlatively, of the pines with five needles, decreases in the same way: if “cold” (continental Eurasia and America) and “hot” (Atlantic and Central America, south-eastern Asia, Mediterranean area) pines are compared, there are 17 and 11 species of the Strobus subgenus versus 22 and 50 of the Pinus subgenus, respectively. The difference between the two is highly significant (χ2 = 7.71; P = 0.006), the first subgenus being twice as rich for “cold pines” as the second (61 and 31% of species in each subgenus).

Conclusion

At the present time, the systematics of the genus Pinus cannot be understood without simultaneous recourse to classical morphological characters and geography, and those factors proposed more recently by biochemistry and molecular biology at two levels: macromolecular (proteins, DNA) and micromolecular (“chemotaxonomy”, here polyphenolic markers). Beyond their length and weight (thus their fineness), the number of needles per bundle appears negatively correlated to the content of O-methylated foliar flavonols. Given that this biochemical character is directed biogenetically speaking because methylation is generally irreversible, it can be concluded that a low level of methylation (almost nil) and a large number of needles (up to five) are both ancestral characters, as observed for subgenus Strobus. Inversely, pines with two needles and a methoxylation index exceeding 30 are to be considered as evolved (=derived), as observed in subgenus Pinus. This is particularly the case for the two subsections Contortae (section Trifoliae) and Pinaster (section Pinus). On other hand, several characters show coherence with the two former subsections relative to taxonomy (balance between the two subgenus Strobus and Pinus), biogeography (northern vs. southern area) or ecophysiology (cold vs. hot, wet or dry regions).

From a dynamic point of view, there is a strong correspondence between the taxonomical subsets that have been recognised and the evolutionary stages of the genus defined by morphometric and biochemical characters given that the first generally shows a good biogeographical and climatic homogeneity; the Mediterranean case (subsection Pinaster) is a symptomatic one. In this respect, we postulate the existence of a common ancestor (or “Archeopinus”) of present Pinus species of continental/septentrional location in the present geographical sense, allowing consideration of the geological phenomena which gave birth, about 180 millions years ago (during the Mesozoic era between the Cretaceous/Jurassic), to the two major blocks, Gondwana and Laurasia from Pangea. Our proposal of an Archeopinus taxon with a Laurasian origin nullifies the present distinction between Palearctic and Neartic domains, which would have previously populated the same continuous territory.

Starting from this hypothetical primary Laurasian origin, the migration and then the segregation following separation of present Eurasia and America could have thus allowed the divergence of two phyla: the first, corresponding to the subgenus Strobus, remains near the “Archeopinus” stem group; the second, corresponding to the subgenus Pinus, has generated two sub-phyla, the first mainly Palearctic (section Pinus, with some species of the Nearctic “series” densiflorae) and North American (section Trifoliae). Considering the low number of needles and the high MeO index, subsection Contortae is the more derived subset of the North American pines. In Eurasia, the two Asian subsets, continental and eastern, share a low methylation index. There is a clear difference compared to the European bloc which in turn, also comprises two sub-classes, Middle European and Mediterranean. The latter with a low number of needles and high methylation index could thus be derived from the first. In the Old World, continental Asia (northern, cold and humid) and the Mediterranean area (southern, hot and dry) show extreme characteristics which are highly significant: 54 versus 0% for the presence of the subgenus Strobus, 3.7 versus 2.1 for the average number of needles per bundle, 5 versus 31 for the O-methylation index. Thus Pinus is a genus “that came from the cold”!

To conclude, it is hoped that our multi-disciplinary contribution, which adds a morphometric, biogeographical and micromolecular dimension to the many recent works of macromolecular nature, may partly be considered as an answer to the pertinent comment recently made by Syring et al. (2005): “Despite ca. 30 published studies over the past two decades, a well-resolved phylogeny of Pinus remains a work in progress.”