CHEMISTRY & BIODIVERSITY ± Vol.

3 (2006) 27

Infrared Spectroscopy as a Tool for Chemotaxonomic Investigations within the

Achillea millefolium Group
by Ingrid Werner*, Sabine Glasl, and Gottfried Reznicek
Department of Pharmacognosy, University of Vienna, PharmaCenter, Althanstrasse 14, A-1090 Vienna
(phone: ‡ 43-1-4277-55219; fax: ‡ 43-1-4277-9552; e-mail: ingrid.werner@univie.ac.at)

Dedicated to Prof. Wolfgang Kubelka on the occasion of his 70th birthday

An FT-IR-based method for the rapid characterization and identification of Achillea species was
developed. With the IR spectra of the CH2Cl2 extracts of A. aspleniifolia, A. collina, A. millefolium, A.
millefolium ssp. sudetica, A. pannonica, A. pratensis, A. roseoalba, A. setacea, and A. styriaca, a spectrum
library was created. This database allowed the identification of the respective taxa by comparison of IR
spectra. Moreover, information concerning dominant structural elements of sesquiterpene constituents
was obtained, providing hints about the composition of a plant sample of unknown origin.

Introduction. ± Yarrow has been used since ancient times for curing intestinal and
gynecological diseases due to its anti-inflammatory and spasmolytic effects [1].
Botanically, Achillea millefolium l. is an aggregate of a dozen taxa of different ploidies,
morphologies, and chemical properties. Unfortunately, differences in morphology as
the only parameter is insufficient for a reliable identification. Especially the
comminuted drug requires quick and easy chemical-screening methods for quality
control. In addition to the classical chromatographic methods, spectroscopic methods
like infrared (IR) spectroscopy have lately gained much influence in this field. The
time-consuming procedures prior to analysis such as sample preparation and
chromatographic separation are not necessary [2] [3]. In contrast to the most
commonly used near-infrared (NIR) methods, no chemometric algorithms are needed
for the interpretation of the data from Fourier-transform (FT-IR) methods, because
characteristic key bands can be observed here. These typical bands depend on the
composition of the extract and, consequently, on the polarity of the extracting agent.
In the case of Achillea, CH2Cl2 was chosen for the extraction, because this particular
solvent gives rise to an enrichment of sesquiterpenes. The sesquiterpenes of Achillea
are important with respect to both their pharmacological effects and chemotaxonomy.
For many sesquiterpene lactones such as the proazulenes (isolated from A.
aspleniifolia, A. collina, and A. roseoalba), anti-inflammatory activities have been
reported [4], while other compounds from this class such as the rupicolines (from A.
setacea) or the guaianolide endoperoxides (from A. millefolium) may cause allergic
reactions. Apart from these pharmacological effects, the chemotaxonomic relevance of
the sesquiterpenes [5] is important, as the differences in the sesquiterpene pattern are
the basis for the chemical distinction of the taxa.

¹ 2006 Verlag Helvetica Chimica Acta AG, Z¸rich

28 CHEMISTRY & BIODIVERSITY ± Vol. 3 (2006)

Results and Discussion. ± We investigated the taxa Achillea aspleniifolia, A. collina,

A. millefolium s. str., A. millefolium ssp. sudetica, A. pannonica, A. pratensis, A.
roseoalba, A. setacea, and A. styriaca.
In Fig. 1, the FT-IR spectra of the extracts of ten different single plants of A.
aspleniifolia are overlaid. As can be seen, there are only small differences. Thus, for
comparison of the different taxa, the averaged data of the single-plant spectra were
used. Interestingly, these averaged spectra show remarkable differences that allow an
IR-based differentiation between the single species (see Fig. 2 below).

Fig. 1. Overlaid IR spectra of ten different samples of Achillea aspleniifolia

The proazulene-containing taxa A. aspleniifolia, A. collina, and A. roseoalba are

very similar in their sesquiterpene patterns, which leads to similar FT-IR spectra. These
taxa are characterized by azulenogenic guaianolides like the artabsines (1 ± 3) and the
non-azulenogenic guaianolides such as the matricarines (4, 5) [5]. All of these
compounds possess a saturated g-lactone moiety, no a-methylidene g-lactones being
known in these taxa so far. Thus, the spectra are dominated by CˆO and C O
absorptions at 1770 ± 1785 and 1155 ± 1175 cm 1, respectively, due to saturated g-
lactones, and by the absorptions of CˆO functions of acetyl (Ac; 1740), angeloyl (Ang;
1715), and tigloyl (Tig; 1705 cm 1) esters, as well as their corresponding C O
vibrations at 1255 (Ac), 1235 (Ang), and 1260 cm 1 (Tig).
 CHEMISTRY & BIODIVERSITY ± Vol. 3 (2006) 29

Compared to the above resonances, the spectra of A. millefolium s. str., A.

millefolium ssp. sudetica, and A. setacea show a shift of the lactone bands (CˆO at
1750 ± 1770, C O at 1135 ± 1150 cm 1). This shift can be rationalized by the dominating
sesquiterpene lactones with a-methylidene g-lactone moieties such as the rupicolines
(6, 7) from A. setacea and the guaianolide endoperoxides (8 ± 13) in A. millefolium s.
str. and A. millefolium ssp. sudetica [5]. As these types of lactones are known to have
allergenic potential, this difference in the IR spectra is very important. Even though the
difference is small (D~ nZ ˆ 20 cm 1), the type of the dominating lactone (saturated or
unsaturated) can be readily determined.
The single-plant FT-IR spectra of the extracts of the three taxa A. pratensis, A.
pannonica, and A. styriaca show larger differences due to a greater heterogeneity
within these species, and no similarities to other taxa. A. pratensis is characterized by
the occurrence of eudesmanolides with both types of lactones; the corresponding
spectra show also characteristics of both lactone types, i.e., the CˆO bands typical for
saturated lactones and the C O bands of a-methylidene g-lactones. In the spectra of
A. pannonica, no remarkable lactone band at all can be observed (see Fig. 2). This has
to do with the observation that this species contains sesquiterpenes of different
skeletons mostly lacking a lactone moiety. A. styriaca contains longipinenes without
lactone moiety, beside minor amounts of eudesmanolides. Thus, the lactone band is
significantly weaker than in the spectra of the other Achillea taxa, except A. pannonica
[5].
For electronic comparison of the IR spectra, an IR library was established. For each
species, the spectra of ten single plants per location were collected, and, in most cases,
plants were collected on more than one location. Half of the experimentally recorded
spectra were added to the library, the other half were used as test set. When a spectrum
of an Achillea extract was compared with this library, a hit list was created, indicating
similarity by a correlation coefficient (R). Hence, a perfect match would correspond to
a coefficient of 1.0. In Table 1, the average correlation coefficients of the spectra from
30 CHEMISTRY & BIODIVERSITY ± Vol. 3 (2006)

Fig. 2. Average IR spectra of different taxa of

Achillea millefolium. a) A. aspleniifolia, b) A.
collina, c) A. roseoalba, d) A. millefolium s. str., e)
A. millefolium ssp. sudetica, f) A. setacea, g) A.
pannonica, h) A. pratensis, i) A. styriaca.
 CHEMISTRY & BIODIVERSITY ± Vol. 3 (2006) 31

the test set are compared to the ones of the spectra in the library. The obtained factors
show again the great similarity between the three proazulene-containing taxa on the
one hand, and between the three taxa dominated by a-methylidene g-lactones on the
other hand: A. millefolium, A. millefolium ssp. sudetica, and A. setacea. The electronic
comparison of the spectra in the test set with the those from the library revealed the
correct identification in all cases, mostly with a correlation factor higher than 0.9900 for
the first hits, and a great number of spectra of the same species as top hits.

Table 1. Average Correlation Factors ( R) for Comparison of IR Test Spectra (columns) with Library
Spectra (lines) for Extracts of Various Achillea sp. For details, see Exper. Part.

A. asplen. A. collin. A. roseo. A. setac. A. millef. A. pann. A. prat. A. sudet. A. styr.

A. asplen. 0.9974 0.9907 0.9894 0.9626 0.9657 0.8902 0.9747 0.9599 0.9566
A. collin. 0.9902 0.9978 0.9936 0.9810 0.9785 0.9173 0.9680 0.9737 0.9699
A. roseo. 0.9890 0.9935 0.9977 0.9756 0.9821 0.9172 0.9607 0.9786 0.9689
A. setac. 0.9613 0.9779 0.9751 0.9984 0.9925 0.9364 0.9550 0.9845 0.9734
A. millef. 0.9657 0.9816 0.9828 0.9937 0.9972 0.9414 0.9519 0.9915 0.9771
A. pann. 0.8884 0.9166 0.9167 0.9361 0.9403 0.9987 0.8759 0.9562 0.9292
A. prat. 0.9622 0.9680 0.9595 0.9499 0.9450 0.8771 0.9831 0.9306 0.9563
A. sudet. 0.9589 0.9736 0.9785 0.9849 0.9909 0.9570 0.9358 0.9977 0.9710
A. styr. 0.9660 0.9803 0.9794 0.9847 0.9872 0.9403 0.9688 0.9815 0.9868

As commercial drugs often contain mixtures of different taxa or hybrids, mixtures

of A. aspleniifolia and A. setacea in ratios of 1 : 1, 1 : 2, and 1 : 3 were prepared. When the
IR spectra of these mixtures were compared with those of the library, the correlation
factor was significantly lower (< 0.9), and the hit list showed spectra of different taxa as
top hits. These effects can be observed within commercial drug materials as well, even
though the differences are smaller, depending on the ratio of components in the
mixture.

Table 2. Hit List in Terms of Correlation Factor ( R) for a Library Search of a Commercial Drug and a
Pure Sample of Achillea aspleniifolia. Collection sites are indicated in parentheses (see Exper. Part).

Hit No. R Pure sample Hit No. R Commercial drug

1 0.9985 A. aspleniifolia ( Rust) 1 0.9925 A. collina (Liechtenstein)
2 0.9982 A. aspleniifolia ( Rust) 2 0.9913 A. roseoalba ( K‰rnten)
3 0.9977 A. aspleniifolia ( Retz) 3 0.9912 A. collina (Liechtenstein)
4 0.9976 A. aspleniifolia ( Rust) 4 0.9909 A. millefolium ( Bruchsdorf)
5 0.9976 A. aspleniifolia ( Rust) 5 0.9909 A. collina (Liechtenstein)
6 0.9972 A. aspleniifolia ( Retz) 6 0.9908 A. collina (Voesendorf)
7 0.9968 A. aspleniifolia ( Retz) 7 0.9902 A. collina (Voesendorf)
8 0.9965 A. aspleniifolia ( Rust) 8 0.9901 A. millefolium ( Lungau)
9 0.9965 A. aspleniifolia ( Retz) 9 0.9900 A. collina (Liechtenstein)
10 0.9962 A. aspleniifolia ( Rust) 10 0.9895 A. collina (Liechtenstein)
11 0.9958 A. aspleniifolia ( Retz) 11 0.9895 A. collina (Liechtenstein)
12 0.9953 A. aspleniifolia ( Rust) 12 0.9895 A. collina (Liechtenstein)
13 0.9953 A. aspleniifolia ( Retz) 13 0.9892 A. collina (Liechtenstein)
14 0.9948 A. aspleniifolia ( Retz) 14 0.9891 A. millefolium ( Lungau)
15 0.9939 A. aspleniifolia ( Rust) 15 0.9887 A. millefolium ( Bruchsdorf)
32 CHEMISTRY & BIODIVERSITY ± Vol. 3 (2006)

Table 2 shows the hit list of a commercial and of a pure sample of A. aspleniifolia.
The pure sample produced a hit list with the correct identification result on the first 15
places, whereas the hit list of the commercial drug contained three different taxa within
the first ten hits. To get this information from the hit list, it is important to collect the
spectra of at least five different plant samples and to use their individual spectra rather
than the averaged spectra of all measured spectra of one taxon in the library.

Conclusions. ± FT-IR Analysis allows a quick and cheap identification of Achillea

taxa present in CH2Cl2 extracts. Additionally, the method provides information about
dominating structural elements of the pertinent sesquiterpenes. Unknown taxa can be
characterized, and mixtures or hybrids can be differentiated from the pure taxa. In
combination with other methods such as thin-layer chromatography (TLC) and high-
performance liquid chromatography (HPLC) [6], IR spectroscopy represents a useful
tool for quality control.

Experimental Part
Materials. Plant material was collected from the following locations in Austria and Italy: A.
aspleniifolia Vent., Rust, Burgenland, and Retz, Lower Austria; A. collina J. Becker ex Reichenb.,
Liechtenstein and Vˆsendorf, Lower Austria; A. millefolium L. s. str., Bruckdorf and Lungau, Salzburg;
A. millefolium ssp. sudetica Opiz, Gro˚eck, Salzburg; A. pannonica Scheele, Neusiedl, Burgenland, and
Falkenstein, Lower Austria; A. pratensis Saukel & LÌnger, Tamsweg, Salzburg, and ÷densee, Styria; A.
roseoalba Ehrend., Rosental, Carinthia, and Studena bassa, Italy; A. setacea W. et K., St. Margarethen,
Burgenland, and Retz, Lower Austria; A. styriaca Saukel & Danihelka ined., Einach, Styria.
Specimens were deposited at the herbarium of the Institute of Pharmacognosy, University of Vienna.
From each location, ten single plants were investigated, five of the resulting spectra were stored in the
library as single spectra, the other five spectra were used as test sets.
Sample Preparation and IR Spectroscopy. The IR analyses were carried out on a Perkin Elmer
Spectrum-2000 FT-IR spectrometer. Flower heads (100 mg) were extracted with CH2Cl2 (1 ml) by
ultrasonication (10 min). A part (10 ml) of the resulting extract was placed on a silicon plate (13  1 mm;
Korth Kristalle GmbH, Altenholz, Germany). After evaporation of the solvent, the IR spectra were
recorded. The following parameters were used: resolution, 4 cm 1; J-stop resolution, 7.77 cm 1;
apodization, strong; gain, 1; OPD velocity, 2 cm/s; interferogram, bidirectional double-sided; number
of scans, 1; scan range, 5200 ± 370 cm 1; interval, 1.0 cm 1.
Parameters for spectrum processing: interpolation, from 2000 to 800 cm 1; data interval, 1 cm 1;
baseline correction, automatic; normalization, ordinate limit 0.1 A, zero-point auto zero. The spectra
were compared by intensity weighting and noise weighting, with the following correlation factor R:

P
wi Ai Bi
Rˆ P 1=2 P 1=2
… wi Ai Ai †  … wi Bi Bi †

where Ai and Bi are the absorbance values in spectra A and B at frequency i, and where wi is the product
of the intensity and noise weighting. For intensity weighting, a black-body filter was used to reduce the
influence of data in the regions at the ends of the spectrum where the signal-to-noise ratio is lower. For
noise weighting, a black band filter was used to reduce the influence of regions where the transmission is
low due to high noise in absorbance.
 CHEMISTRY & BIODIVERSITY ± Vol. 3 (2006) 33

REFERENCES
[1] M. Wichtl, −Teedrogen und Phytopharmaka×, Wissenschaftliche Verlagsgesellschaft, Stuttgart, 2002,
p. 399.
[2] H. Schulz, B. Schrader, R. Quilitzsch, S. Pfeffer, H. Kr¸ger, J. Agric. Food Chem. 2003, 51, 2475.
[3] H. Schulz, M. Baranska, H.-H. Belz, P. Rˆsch, M. Strehle, J. Popp, Vib. Spectrosc. 2004, 35, 81.
[4] U. Kastner, S. Sosa, A. Tubaro, R. Della Loggia, J. Jurenitsch, Planta Med. 1993, 59, A669.
[5] W. Kubelka, U. Kastner, S. Glasl, J. Saukel, J. Jurenitsch, Biochem. Syst. Ecol. 1999, 27, 437.
[6] S. Glasl, U. Kastner, J. Jurenitsch, W. Kubelka, J. Chromatogr., B 1999, 729, 361.

Received September 16, 2005