Determination of taxa of the Achillea millefolium qroup and Achillea crithmifolia bv morpholoqical and ~ hvtochemical methods I . Characterisation of Central European taxa '

A survey of morphological and phytochemical data characteristic for several taxa of the Achillea millefoliurn group (A. aspleniifolia VENT., A. roseoalba EHREND., A. collina J. BECKER ex REICHENB., A. ceretanica SENNEN, A. setacea W. ET K., A. pratensis SAUKEL & LANGER, A. styriaca SAUKEL ined., A. pannonica SC H E ELE, A. distans W. ET K., A. millefolium s. I., A. millefolium ssp. sudetica OPIZ) and A. crithmifolia W. ET K. is presented. For each taxon a short morphological description and a guide for microscopic analysis is given as well as its sesquiterpene composition including the TLC characteristics. Based on GLC analyses of 1523 single plants collected in Central Europe the sums of sabinene + j3-pinene + j3-caryophyllene (SUM I ) , a-pinene + 1,8-cineole (SUM 2), camphene + camphor + borneol (SUM 3), camphene + camphor (SUM 4) and 1,8-cineole + borneol (SUM 5) were found to be highly significant for distinct taxa or groups of taxa.


Introduction
Yarrow is a widespread plant used in folk medicine due to its various effects.
The indications include gastric and intestinal disorders, inflammation of skin and mucosa as well as hemorrhages (Wichtl, 1997).The sesquiterpenes were shown to contribute to the antiphlogistic effects (Della Loggia et a/., 1992;Kastner et a/., 1993;Sosa et a/., 2001).The fact that the Achillea millefolium group consists of several different taxa (Ehrendorfer, 1953;Saukel et al., 1992Saukel et al., , 1992aSaukel et al., , 1992bSaukel et al., , 2002) ) and that in folk medicine herbal teas are prepared without regard to the species led to intensive morphological and phytochemical investigations.The latter were focused on sesquiterpenes and essential oil, both showing chemotaxonomic relevance (Kastner et a/., 1992;Kubelka et a/., 1999).The present paper gives an overview of the different taxa including brief morphological descriptions, pictures of typical leaflets and rayflorets, a summary of the sesquiterpenoids isolated up to now and the essential oil composition.Linear combination of the essential oil compounds is introduced as a new special feature which is highly significant for distinct taxa or groups of taxa.
Although A. crithmifolia does not belong to the A. millefolium group it is taken into consideration in the present paper.It was detected frequently in drug material available in Austrian pharmacies (Rehberger, 1996) and contains rupicolines which might trigger allergic reactions due to their exocyclic methylene group.The large amount of the obtained data is based on the analyses of more than 2600 single plants including plants from East Europe.This paper deals with those which were collected in Central Europe and which are of pharmaceutical relevance (1523 single plants).The combination of the presented morphological and phytochemical characteristics also allows the determination of unidentified drug material.

Flo wcytometry
Ploidy level was evaluated by cutting some leaflets of an upper stem-leaf.For higher ploidy levels the experiment was employed together with a diploid standard (cloned A. ceretanica 2x) and in case of diploid samples together with a tetraploid standard (cloned A. pratensis 4x) in a citric acid-buffer medium.After filtration, 4,6diamidino-2-phenyl-indole together with sulphorhodamine were added and 5 minutes later the sample was measured (for details see Wlach 2002).We used a Partec-CCA 1 (Cell Counter Analyser) equipped with a Hg-lamp as a source for UVradiation and a fluorescence-detector including a photomultiplier.

Sample preparation for GL C
100 mg flower heads were extracted with 1 ml dichloromethane for 10 minutes in an ultrasonic bath.1.5-2.0p1 of this solution were injected, the rest was evaporated under nitrogen, redissolved in 200 pl dichloromethane and used for TLC.

Essential oilsum parameters and calculation
Computer guided classification and comparison of the different taxa based on the essential oil required the determination of distinct parameters.According to oc-currence and reproducibility nine monoterpenes and O-caryophyllene were chosen and used for all computations, as their peak areas showed linear correspondence with concentrations (Rauchensteiner et al., 2002a ined.).The peak areas of the chosen substances were summarised, set 100% and recalculated, which permits a comparison of the single plants independent of the injected amount.However, within all the taxa of the genus Achillea a great variability of the data is observed.In none of the investigated species one of the compounds shows stable values.E. g. in A. setacea the values of 1,8-cineol, camphor and borneol vary between 0 and 80% (fig. 1, ai, j).The analyses of the correlation coefficients show for A. setacea -0,82 (camphor against borneol) in the case of A. collina -0,84 (sabinene against O-pinene) and for A. pannonica -0,74 (a-pinene against 1,8-cineol).An extensive examination of the data material led to the conclusion that especially the sums of sabinene + P-pinene + P-caryophyllene (SUMI), a-pinene + 1,8-cineole (SUM2) and camphene + camphor + borneol (SUMS) are very characteristic for distinct taxa or groups of taxa: Fig. 2 shows a scatterplot for the sums of the proazulene containing species (proaz), A. setacea (set), A. pannonica (pan) and A. styriaca (styr), respectively.It is clearly visible that the oils of A. pannonica and the proazulene group are fully described by SUM 1 and SUM 2, whereas A. setacea is characterised by small values of SUM 1 and small to high values of SUM 2. Additional linear combinations were found for camphene + camphor (SUM 4) and 1,8-cineole + borneol (SUM 5) shown in fig. 3 with the same species as in fig. 2. A. setacea is represented with high significance by SUM 4 and SUM 5. A. styriaca shows some similarities but the different position is due to additional compounds in the oil.
The sum parameters 1, 2, 3 were therefore plotted in ternary graphs (tab.2a -I).Within these three coordinates each single plant is determined by one distinct data point revealing characteristic values for the respective taxon.(Dequeker, 1964) and heated to visualise the non-proazulenes.

Quick-Testing for proazulenes and other sesquiterpene lactones
One capitulum was placed on a slide and heated over a flame after addition of a small amount of CP-reagent (60 % chloralhydrate -85 % phosphoric acid (2+1); Saukel, 1993).This procedure makes it easier to prepare the rayflorets.The reaction of the glands was obselved in the microscopea blue or black colour indicated the presence of proazulenes whereas any other or missing colour indicated a different composition within the sesquiterpene lactones (tab.2a -I).An important requirement is to use only dried plant materialfresh material gives in most cases a wrong positive reaction, even though proazulenes lack.

Results and Discussion
Twelve Central European taxa were investigated with regard to the morphology, the essential oil composition and the sesquiterpene pattern.For the morphological characterisation the morphology of the leaflets and rayflorets turned out to be most suitable (Saukel et a/., 1992(Saukel et a/., , 1992a(Saukel et a/., , 1992b(Saukel et a/., , 1994(Saukel et a/., , 2002)).
The essential oils of 1523 single plants collected in Central Europe were examined by GLC.Each tested individual showed a stable composition (Kastner et a/., 1992) whereas plants of one species differed in the content of their main compounds (fig.1).By summing up two or three out of ten of these compounds we got stable values, the sums (see experimental), which allowed the assignment of the samples to distinct species or to aggregates of species.Extensive investigations of the whole data material revealed significant negative correlations of the terpenes used for the sum parameters.In the diagrams the single datapoints are located within a triangle, each of the three corners represents 100% of one of the five sum parameters.As a result the data of a large number of single plants can be shown in one diagram allowing a comparison between the different taxa.The plots with the sum parameters 1, 2 and 3 (tab.2a -I) showed the same phenomenons and tendencies as those with the sum parameters 1, 4 and 5. Remarkably seems that no datapoint was located on the axis between sum parameter 1 and 3.
As result the proazulene containing species A. aspleniifolia, A. roseoalba, A.
collina and A. ceretanica clearly show homogeneity of the essential oil with a high content of sabinene + R-pinene + R-caryophyllene (SUM 1).This is confirmed by the sesquiterpene composition which shows the typical proazulene pattern (tab.3, fig.5).In contrast, A. setacea as well as A. crithmifolia are located along the axis between SUM 2 and SUM 3 lacking sabinene + The scatterplot of A. millefolium s.1.(tab.2k) includes a large number of single plants (320) and gives an impression of the heterogeneity within that group, as the individuals are spread almost over the whole area of the triangle.89 individuals from Northwest Europe show a high value of SUM 1 and therefore resemble to proazulene containing species.For differentiation a check of the sesquiterpene pattern by TLC andlor morphological investigations (the value of the diameter of pollengrains and the dimension of the rayflorets are very high) have to be performed.In accordance to its ploidy level (6x), A. millefolium s.1.can be described as a polyphyletic species with many phenotypes depending on the geographical origin.
From the more or less distinct A. millefolium ssp.sudetica in mountainous and alpine habitats also several phenotypes exist (Saukel in: Adler et al. 1994).As a consequence, A. styriaca, A. pannonica, A. distans, A. millefolium ssp.sudetica and A.
millefolium s.1.can only be determined by the combination of morphology, sesquiterpene pattern and essential oil composition.

Conclusion
The essential oil represents a helpful tool to distinguish between proazulene and non-proazulene containing species of taxa of the A. millefolium group and A.
crithmifolia.Sum parameters of the essential oil components correspond to morphological features in most cases.A discriminant analysis of the di-, tetra-and octoploid species (all proazulene containing taxa were concentrated in one group) was performed with all five sum parameters and the contents of p-cymene and bornylacetate.The analysis revealed a correct assignment for the proazulene containing group with 99,7%, A. pannonica 97,1%, A. setacea 94,3%, A. styriaca 93,2% and A. prafensis 87,8%.Consequently the presented results can also be combined to identify unknown drug material.A key for determination is given in tab.
1. Up to now this key is valid only for uncomminuted drugs, as leaflets and flowers from one plant are required to obtain correct results.The identification of cut drug requires a modification of the key which is in progress.

Colour of glands:
uncoloured or pink.

Colour of glands:
uncoloured or light green.

Figure 4 .
Figure 4. Blue colour of the glands indicating proazulenes.
The essential oil characteristics of A. styriaca resemble to those of A. setacea but the shape of the leaflets is different.The surprising number of sesquiterpene skeletons (eudesmane, germacrane and longipinane) could partly be explained by a hybridisation process with A. pratensis.The remaining taxa A. pratensis, A. pannonica, A. distans, A. millefolium ssp.sudetica and A. millefolium s.1.show high variability and overlappings so that based on the essential oil only, the assignment to a certain taxon will fail.Only A. pratensis is characterised by one sesquiterpene type (eudesmanolides).The hexaploid A.

Table 1 .
Determination key for uncomminuted drugs of taxa of the A. millefolium group and A. crithmifolia.

Table 2a .
Short morphological, cytological and chemical characteristics of A. aspleniifolia. S

Table 2c .
Short morphological, cytological and chemical characteristics of A. collina.

Table 2e .
Short morphological, cytological and chemical characteristics of A. setacea.

Table 2f .
Short morphological, cytological and chemical characteristics of A. crithmifolia.

Table 29 .
Short morphological, cytological and chemical characteristics of A

Table 2i .
Short morphological, cytological and chemical characteristics of A. distans.

Table 2j .
Short morphological, cytological and chemical characteristics of A. pannonica.

Table 2k .
Short morphological, cytological and chemical characteristics of A. millefolium s.I.

Table 2 .
List of taxa of the A. millefolium group and A. crithmifolia including sesquiterpenes, R, values, colours with spraying reagents and references.