Pine species are usually classified to morphologically distinct and taxonomically typified taxa. Nevertheless, a large number of hybrids have evolved as a result of introgressive hybridisation. The individual morpho-anatomical characteristics of these hybrids in several cases can be deceiving and can lead to wrong assignment to typified taxa.
The high degree of genotypic variability observed in a number of Pinus species is reflected in the biochemical variability, which is usually studied at the levels of terpene composition and isoenzyme variation.
In the framework of our chemical and biological investigations on the volatile metabolites of Greek endemic and Mediterranean conifers (Roussis et al., 1999; Petrakis and Roussis, 1997) we had the opportunity to study the composition of the volatile metabolites from the needle essential oils of 178 Pine trees belonging to the 10 most abundant Pinus species (Pinus halepensis, P. brutia, P. canariensis, P. pinea, P. nigra, P. pinaster, P. peuce, P. radiata, P. attenuata and P. heldreichii) and growing in Greece in natural habitats.
The essential oils were analysed by GC and GC-MS and the identification of the chemical constituents was based on comparison of their Rts and mass spectra with those obtained from authentic samples and/or libraries spectra. Forty-nine metabolites, constituting the majority of the volatile metabolites, were identified and quantified. The majority of the identified metabolites were found to be monoterpenes and sesquiterpenes.
On the basis of the major terpene metabolites, distinct and characteristic chemotypes were assigned for the 10 investigated Pine species. Among the metabolites with the highest discriminant value were a-, b-pinene, limonene, myrcene, E-caryophyllene and germacrene D. Statistical analyses of the chemical and morphoanatomical characters of the investigated trees produced broadly similar classifications for the studied Pinus species.
Pine species are usually classified to morphologically distinct and taxonomically typified taxa. Nevertheless, a large number of hybrids have evolved as a result of introgressive hybridisation. The individual morpho-anatomical characteristics of these hybrids in several cases can be deceiving and can lead to wrong assignment to typified taxa. The high degree of genotypic variability observed in a number of Pinus species is reflected in the biochemical variability, which is usually studied at the levels of terpene composition and isoenzyme variation.
In the framework of our chemical and biological investigations on the volatile metabolites of Greek endemic and Mediterranean conifers (Roussis et al., 1999; Petrakis and Roussis, 1997) we had the opportunity to study the composition of the volatile metabolites from the needle essential oils of 178 Pine trees belonging to the 10 most abundant Pinus species (Pinus halepensis, P. brutia, P. canariensis, P. pinea, P. nigra, P. pinaster, P. peuce, P. radiata, P. attenuata and P. heldreichii) and growing in Greece in natural habitats. The essential oils were analysed by GC and GC-MS and the identification of the chemical constituents was based on comparison of their Rt's and mass spectra with those obtained from authentic samples and/or libraries spectra. Forty-nine metabolites, constituting the majority of the volatile metabolites, were identified and quantified. The majority of the identified metabolites were found to be monoterpenes and sesquiterpenes.
The essential oils of 148 Pine trees belonging to the 10 most abundant Pinus species growing in Greece were analysed. Forty-nine metabolites, constituting the majority of the volatile metabolites, were identified and quantified on the basis of their chromatographic characteristics and mass spectra. The morphoanatomical characteristics of the investigated trees were also recorded.
Both chemical and morphoanatomical data sets were used in separate cluster analyses of the ten studied Pinus species. Although morpho-anatomically based classification is not shown here it is reflected in the terminology used to characterise trees.
UPGMA was the method to produce the dendrogram shown in Fig. 1.
Canonical discriminant analysis was used to characterize the groups at each clustering level and estimate the contribution of individual terpenes in shaping the chemotypes. Canonical discriminant analysis is particularly suited in sorting statistical differences in parametric variables and for this it serves as a evolutionary probe. The algorithm CONTML (Felsenstein, 1984) was used to produce the phylogenetic network in the discriminant space shown in Fig.2 on the basis of group centroid data with the continuous character option. This was done because the underlying evolutionary model is the multivariate evolution model (Lande, 1979).
On the basis of all terpene metabolites, ten distinct chemotypes are recognized. Although chemotypes broadly correspond to the ten investigated pine species (Fig. 1 & 2) the congruence is not perfect. Some species span to more than one chemotype while some chemotypes include more than one species.
Among the metabolites with the highest discriminant value were a-, b-pinene, limonene, myrcene, E-caryophyllene and germacrene-D. Statistical analyses of chemical and morphoanatomical characters of the investigated trees produced broadly similar classifications for the studied Pinus species.
From the dendrograms it is evident that morphoanatomical and chemical characters produce broadly similar classifications. The discrepancies are due to the differential degree of correlation between morphological and terpenoid characters. In particular monoterpenes can produce the same classification scheme of the ten Pine species examined as the entire set of all volatile secondary metabolites. This is because the biosynthesis of sesquiterpenes is based on the availability of monoterpenes and expectedly sesquiterpene based characters are correlated to monoterpene based ones.
Major monoterpenes are in that case the most significant characters at various nodes of the hierarchical structure. Importantly, certain terpenes are responsible for the split at several levels. Germacrene D, b-caryophyllene, a-pinene, and b-pinene exhibit such a behaviour.
Figure 1. UPGMA based hierarchical clustering of all 148 pine trees on the basis of terpenoid characters. The arrangement of taxa at the top of the classification is not arbitrary. It is based on more resolving clustering not shown in this presentation, where population / forest spot means are used. CDF stands for Canonical Discriminant Functions.
The fact that no single terpene can be used to account for the separation at a certain node is most probably due to the fact that the evolution in the genus Pinus proceeds through gene elimination rather than addition or mutation. However, it is important to note that in the biogenetic scheme of terpenes the existence of backward reactions of products with substrates blurs the overall picture.
Figure 2. Discriminant analysis of all 148 pine trees at the ten clusters level of the hierarchical phenogram shown in Fig.1. Standardised coefficients of the discriminant functions are used as estimators of the contribution of individual terpenes to the separation of groups. A negative or positive sign indicates whether the separation on a specific discriminant axis is caused by a higher or lower content of a pine group member in terms of the terpenoid character. It is important to note that at each clustering level the contribution of various characters varies. These character loads are specific for this clustering level. A phylogenetic network is shown on the basis of CONTML (PHYLIP) algorithm. No rooting of the network was attempted since we polarity of terpenoid characters is largely unknown.
The same situation holds for insect pests -other than bark beetles- that exhibit a co-evolutionary relationship with their pine host. Thaumetopoea pityocampa exhibits a feeding preference behaviour at a chemotype level. Even if no single terpene can account for this behaviour, some terpenes are present at higher concentrations in the less preferred or avoided host pines. Among those terpenes a- and b-pinene, E-caryophyllene and myrcene are considered repellent to the pine moth caterpillars.
Terpenoid characters alike morphological ones, suffer from the same correlation "disease" that blurs any phylogenetic pattern.
The adoption of canonical discriminant analysis capable in partitioning the overall covariance matrix of a set of taxa sharing a common ancestor, as an evolutionary probe revealed a basic phylogenetic scheme for pines, though not all species were included in the data set.
Since the pine needle essential oil can be altered from a variety of environmental factors this procedure was suited to the scope of the study under the multivariate evolutionary model as underpinning mechanism. The model states that
Phenotypic variance = Genotypic variance X Environmentally induced variance
Canonical discriminant analysis partitions the observed phenotypic variance successfully in genotypic variance, which is phylogenetically useful and the environmental noise.
While terpene based chemosystematics do not confirm with higher subgeneric classification it is particularly useful in revealing specific and subspecific natural groups and hybridisation complexes.
Canonical discriminant analysis coupled with numerical cladistic algorithms is very successful in incorporating apomorphic variance (non-communality-based).
On practical grounds in terms of integrated pest management techniques of major pine pests such studies can serve as a theoretical framework for decision making tasks. They can be used as action probes since they provide evolutionary argumentation for practical decision making. Typical questions that can be asked in this context are:
Does it make any sense to search for single compounds for feeding deterrence of pine moth caterpillars or it would be more efficient to submit in a bioassay screening program the set of foliar chemical profiles available in the existing chemotype gamut?
Is it economically feasible to invent a spray blend of repellent terpenes and formulate it for slow release action or it would be better to suggest a certain specific pine synthesis of managed forests in afforestation projects?
Pest resistance to behaviour modifying chemicals is more probable to be evolved if the semiochemical blend involved substantially diverges from the existing chemotype profiles, which were tested in evolutionary time by means of natural selection acting on the plant.
Adams R.P. (1982): A comparison of multivariate methods for the detection of hybridization, Taxon, 31, 646-661.
Lande R.(1979): Quantitative genetic analysis of multivariate evolution applied to brain: body size allometry. Evolution, 33,402-416.
Petrakis P.V. & Roussis V. (1997): Evolution in Mediterranean climate regions. TREE, 12, 113.
Roussis V., CouladisM.,Tzakou O.& LoukisA.:A comparative studyonthe needlevolatile constituents of three Abies species grown in South Balkans. J.Ess.Oil Res.(in press).
von Rudloff E. (1975): Volatile oil analysis in chemosystematic studies of North American conifers. Biochemical Systematics and Ecology, 2, 131-168.
Sorensen J.T. & Foottit R. (1992): Ordination In The Study Of Morphology, Evolution And Systematics Of Insects: Applications And Quantitative Rationals. Amsterdam: Elsevier.