Content uploaded by Thomas Marcussen
Author content
All content in this area was uploaded by Thomas Marcussen
Content may be subject to copyright.
Evolution, phylogeography, and taxonomy
within the Viola alba complex (Violaceae)
T. Marcussen
University of Oslo, Department of Biology, Division of Botany and Plant Physiology, Oslo, Norway
Received June 17, 2002; accepted November 27, 2002
Published online: March 20, 2003
Springer-Verlag 2003
Abstract. Taxa of the Viola alba complex were
investigated using alloz ymes and morphometry. A
taxonomic revision is presented. A wide delimita-
tion of V. alba with only three morphological and
geographical subspecies is suggested: (1) ssp.
dehnhardtii distributed in the Mediterranean east-
wards to Turkey; (2) ssp. alba flanking ssp.
dehnhardtii in the north and east; and (3) ssp.
cretica ende mic to Crete. Ssp. cretica, up to now
treated as a separate species, is particularly close to
ssp. dehnhardtii. Viola cadevallii (NW Mediterra-
nean) is included in the synonymy of ssp.
dehnhardtii. Ssp. scotophylla (S Europe), ssp. thes-
sala (Balkan), V. armena (Turkey), and V. besseri
(Caucasus) are reduced to synonyms of V. alba ssp.
alba. Viola pentelica (Greece) might represent
transitional forms between ssp. alba and ssp.
dehnhardtii. Glacial refugia for ssp. alba are
suggested from the eastern Mediterranean via
Turkey to the Caucasus, for ssp. dehnhardtii in
the Mediterranean area in general, and for ssp.
cretica in Crete. A key to the subspecies is
provided. Taxonomic recombination: Viola alba
Bess. ssp. cretica (Boiss. & Heldr.) Marcussen,
comb. nov.
Key words: Allozymes, Viola alba, Viola cretica,
phylogeography, glacial refugia.
Viola alba Bess. (von Besser 1809) is a
widespread and common woodland plant in
southern Europe and in the adjacent parts of
North Africa and West Asia. The species
expresses high variability on the regional,
local, and the individual scale, and within the
last two centuries this has brought about the
description of large numbers of taxa of various
taxonomic rank and distribution. Also, fre-
quent hybridisation with taxa outside the
complex has doubtlessly added to the confu-
sion (Marcussen and Borgen 2000).
Viola alba is a paleotetraploid (2n ¼20)
complex belonging to the Eurasian subsection
Viola (= Uncinatae Kupffer, Curvato-Pedun-
culatae Becker, Hypocarpea Godr.) with some
25 species (Okamoto et al. 1993, Marcussen
and Borgen 2000, Dinc¸ and Yıldırımlı 2002).
All of these are perennial acaulescent herbs
and the majority, including V. alba, have a
complex triple reproductive system. Showy
entomophilous flowers (potentially outcross-
ing) are produced in early spring and cleistog-
amous flowers (obligately self-pollinated) in
favourable periods during the rest of the
growing season. Additional clonal propaga-
tion by means of stolons also occurs regularly.
The main habitat of Viola alba is scrub and
woodland, but it is restricted to less-droughted
sites. It reaches high altitudes in the south
but becomes markedly thermophilous in the
Plant Syst. Evol. 237: 51–74 (2003)
DOI 10.1007/s00606-002-0254-5
northern parts and hardly reaches north of the
Alps and the Carpathians. It is almost contin-
uously distributed from northwestern Africa
and the eastern part of the Iberian Peninsula
eastwards to the Caucasus, with scattered
occurrences in the Middle East. It is probably
introduced in the most peripheral stations, e.g.
in Portugal (Mun
˜
oz Garmendia et al. 1993)
and Sweden (Sterner 1938). Also numerous
ornamental cultivars belong to this taxon
(M. Henry, J. Munzinger, T. Marcussen,
unpublished data).
With his monograph Violae Europaeae,
Becker (1910b) was the first author to unite a
high number of previously described taxa
under a single name, Viola alba, and this
system has served as a foundation for modi-
fication by later authors (Fig. 1). Becker (l. c)
characterised the complex primarily by its
narrow-lanceolate, fimbriate and usually ciliate
stipules, the slender aboveground stolons, the
fragrant flowers, and the cordate and usually
over-wintering leaves. Most earlier published
names were reduced to synonymy under either
of two widely-defined subspecies, ssp. alba and
ssp. dehnhardtii (Ten.) W. Becker. These two
subspecies are geographically as well as mor-
phologically defined (Table 1) but ecological
differentiation does not seem prevalent. Becker
(l. c.) included also a third subspecies, ssp.
sintenisii, distributed on the Caspian coast
of Azerbaijan, Iran and Turkmenistan. This
Fig. 1. Historical changes in the delimitation of taxa within the Viola alba complex. Only the most central
publications and taxa are included. Arrows from a taxon (header row) to a box indicates inclusion of this taxon
in the synonymy of the taxon in the box, as proposed by the current author(s) (left column). Grey boxes denote
species and white boxes subspecies of V. alba
52 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
taxon is at least partly sympatric with ssp. alba
and morphologically distinct, and is probably
better considered as a separate species
(T. Marcussen, L. Borgen and I. Nordal,
unpublished data).
Viola alba ssp. alba sensu Becker is the
most widespread and also the taxonomically
most disputed of Becker’s subspecies. It occurs
in a broad belt from the Caucasus in the east
(i.e. V. besseri Rupr.), via Turkey (i.e.
V. armena Boiss. & E
´
. Huet), the Middle East
and the Balkans (i.e. V. thessala Boiss. &
Spruner) to Central and South Europe (e. g.
V. alba s. str., V. scotophylla Jord.). This
subspecies, in the wide sense, is characterised
by the well-developed stolons, straight-
margined or slightly acuminate leaves, and
the pronounced pubescence (Becker 1910b,
Schmidt 1961, Espeut 1999). White-flowered
plants occur scattered throughout the range of
the subspecies (e.g. Ruprecht 1869) but are
predominant in the northwestern parts of its
distribution.
In Central Europe, a taxonomic distinction
has been made between white-flowered, antho-
cyan-free plants, i.e., ssp. alba s. str. (= V.
virescens Jord.), and white- or lilac-flowered
plants with distinctly purple-tinged leaves and
capsules, i.e. ssp. scotophylla. The former
occurs mainly north and west of the Alps (south
France to Poland), and the latter mainly south
and east of the Alps (Schmidt 1961), but they
are largely sympatric. This taxonomic treat-
ment was adopted in Flora Europaea (Valentine
et al. 1968) and, subsequently, by many later
European authors (Delipavlov 1979, Raus
1986, de Bolo
`
s and Vigo 1990).
Plants from the Balkans and the East
Mediterranean region, usually lilac-flowered
and with oblong-triangular, serrate (not cre-
nate) leaves, are by regional authors often
referred to as ssp. thessala (Boiss. & Spruner)
Hayek (von Hayek 1927; Post 1932; Rechinger
1961; Mouterde 1970; Davis et al. 1988;
Livaniou-Tiniakou 1991; Tiniakou 1991,
2000). Other authors have considered that
such plants fall within the morphological range
of Viola scotophylla (e.g. Becker 1910b,
Valentine et al. 1986, Raus 1986). This has
also been the case for the Turkish V. armena
Table 1. Morphological and geographical differences among the most central taxa within the Viola alba
complex. Viola cadevallii and V. pentelica are by most authors included in V. alba ssp. dehnhardtii
Taxon Leaf shape Stolons Indument Pigmentation
(e.g., petiole
and capsules)
Corolla
colour
Distribution
V. alba ssp.
alba s. str.
(=V. virescens)
cordate-
triangular;
crenate
long ± long green white SW Europe;
scattered eastwards
V. alba ssp.
scotophylla
cordate-
triangular,
crenate
long ± long purplish white
(or lilac)
SW Europe –
Caucasus
V. alba ssp.
thessala
oblong-
triangular,
serrate
long long purplish lilac Balkan – Middle East
V. alba ssp.
dehnhardtii
ovate,
crenate
± short,
or absent
± short purplish lilac Mediterranean
region
V. cadevallii ovate,
crenate
absent glabrous purplish lilac E Spain
V. pentelica triangular,
crenate
long glabrous purplish lilac Greece
V. cretica ovate,
crenate
very long very long purplish pale lilac Crete
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 53
(but see Coode and Cullen 1965) and the
Caucasian V. besseri.
Viola alba ssp. dehnhardtii has been accepted
by the vast majority of authors with minor
deviations only. This Mediterranean race oc-
curs from eastern Spain eastwards to Greece
and Turkey, and in Northwest Africa. It differs
from ssp. alba s. lat. by rounded, not acute,
leaves that are adpressed short-pubescent, and
lilac flowers. Stolons are often produced but
may be absent or reduced in many specimens
(often misidentified as V. collina Bess. or
V. hirta L.; Schmidt 1961, Espeut 1999).
Glabrous plants are not uncommon (Becker
1910b, Valentine et al. 1968, de Bolo
`
s and Vigo
1990, Tiniakou 1991, Mun
˜
oz Garmendia et al.
1993, Espeut 1999); these were usually given
specific rank by early authors (i.e. V. pentelica
Vierh. from Greece and V. cadevallii Pau from
Spain, France, and Corsica).
Becker (1910b) included in ssp. dehnhardtii
also the two island endemics V. jaubertiana
Mare
`
s & Vigin., from Majorca, and V. cretica
Boiss. & Heldr., from Crete. Whereas the
inclusion of V. jaubertiana was shown to be
erroneous (Tchourina 1909, Chodat 1924,
Schmidt 1961, Marcussen and Borgen 2000),
the situation for V. cretica is less clear.
Subsequent authors have treated it as a sepa-
rate species (von Hayek 1927, Valentine et al.
1968, Raus 1986, Livaniou-Tiniakou 1991,
Tiniakou 1991). Only Schmidt (1961) provided
a critical discussion, although inconclusive, on
its taxonomic rank. He stated that V. cretica
did fall within the general variability of V. alba
ssp. dehnhardtii while differing in other char-
acters (e. g. thinner and longer stolons, con-
spicuous hispid pubescence, light brown seed
coats, and smaller capsules), and concluded
that, at least, a geographical race of ssp.
dehnhardtii was at hand.
It had been pointed out already by Becker
(1910b) that the distinction between ssp. alba
and ssp. dehnhardtii appears to break down in
certain geographical regions. A latitudinal
transition zone connecting ssp. alba s. lat.
and ssp. dehnhardtii extends across South
Europe from northern Spain to Greece (Becker
1910b, Mun
˜
oz Garmendia et al. 1993, Espeut
1999). In addition, the two subspecies have
been reported to be sympatric and linked by
transitional forms in some areas, i.e. sspp.
dehnhardtii and thessala in Greece (Tiniakou
1991, 2000), sspp. alba and dehnhardtii in
Romania (Grint¸ escu et al. 1955), and sspp.
dehnhardtii and scotophylla in Spain (de Bolo
`
s
and Vigo 1990). Some recent authors have
therefore preferred not to subdivide V. alba at
all and recognise only one morphologically
variable species (Meikle 1977, 1980; Mun
˜
oz
Garmendia et al. 1993; Mus et al. 2000).
The main objectives of this study were (i)to
characterise the taxa within the Viola alba
complex genetically and to determine their
relationships, (ii) to elucidate the phylogeo-
graphic history of the complex, and finally (iii)
to provide a discussion and a re-evaluation of
their taxonomic validity and status. Specific
taxonomic questions were whether ssp. thes-
sala and ssp. scotophylla can be distinguished
from ssp. alba, and whether V. cretica is
distinct from ssp. dehnhardtii. For the genetic
characterisation allozyme electrophoresis was
applied; this method has proved a powerful
tool in resolving systematic relationships with-
in subsection Viola, particularly at the species
level (Marcussen 1998, Marcussen and Nordal
1998, Marcussen and Borgen 2000, Marcussen
et al. 2001).
Materials and methods
Materials. A total of 281 individuals from 62 sites
were collected in the period 1995–2001 and grown
in a greenhouse for subsequent genetic and mor-
phological analyses (Table 2). A subset of 273
individuals from all sites was analysed using
allozyme electrophoresis and a subset of 240
individuals from 53 of the sites was included in
the morphometric analyses. Germination was ob-
tained after 2–8 weeks of combined stra tification
and treatment with gibberellins (Blaxland 1996).
Morphometric and allozymic data are available
from the author upon request.
In order to investigate geographical and taxo-
nomic variation mate rial was classified into 14
54 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
Table 2. Materials included in allozymic (a) and morphometric (m) analyses. Geographical distribution is indicated for each race. Asterisks
denote Greek populations bridging ssp. dehnhardtii and ssp. thessala and treated as ‘‘ungrouped’’ in the Canonical Discriminant Analyses. All
populations were collected by the author except where indicated: (GK) K.-G. Knoche; (IN) I. Nordal; (KB) K. Blaxland; (MD) M. Dinc¸ ; (MR)
M. Ronikier; (PS) P. Scho
¨
nswetter; (WK) W. Klettring
Taxon Ref. code Group Collection site (collector) Number of
individuals
am
Viola alba ssp. alba s.str. TM-22 1 France, Ain: Bouis (with ssp. scotophylla). 4 5
(=V. virescens; Central- TM-39 1 France, Ain: Bourg-en-Bresse, Rue d’Ypres. 7 6
west Europe) TM-54 1 France, Ain: Btw. Veronnat & Reinat. 4 3
TM-52 1 France, Ain: Les Conches (with ssp. scotophylla). 1 1
TM-53 1 France, Ain: Meillonas. 6 5
TM-111 1 France, Ise
`
re: Btw. Uriage-les-Bains &
Chamrousse, 900 m (with ssp. scotophylla).
1–
TM-389 1 Switzerland, Sarnen canton: Alpnachstad,
below Mt. Pilatus, 450 m (with ssp. scotophylla).
1–
V. alba ssp. scotophylla TM-22 2 France, Ain: Bouis (with ssp. alba). 6 6
(Central-west Europe) TM-52 2 France, Ain: Les Conches (with ssp. alba). 1 1
TM-111 2 France, Ise
`
re: Btw. Uriage-les-Bains &
Chamrousse, 900 m (with ssp. alba).
21
TM-389 2 Switzerland, Sarnen canton: Alpnachstad,
below Mt. Pilatus, 450 m (with ssp. alba).
32
V. alba ssp. thessala
(E Mediterranean)
TM-352 3 Greece, Evia: Mt Dirphys, above the
refuge, 1170–1200 m.
57
TM-357 3 Greece, Evia: Mt Dirphys, along the
dirtroad S of the pass of the
Steni-Stropones road, 1040 m.
44
TM-353 3 Greece, Evia: Mt Dirphys, btw. the refuge
and the summit, 1140 m.
22
TM-351 3 Greece, Evia: Mt Dirphys, close to the
refuge, 1160 m.
11
TM-320 3 Greece, Peloponnisos, Pa
´
rnon Mts:
Arkadı
´
a, Agios Petros, 850 m.
21
TM-323 3 Greece, Peloponnisos, Pa
´
rnon Mts: Arkadı
´
a/Lakonı
´
a,
btw. Agios Petros & Agios Vassilios, 855 m.
2–
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 55
Table 2 (continued)
Taxon Ref. code Group Collection site (collector) Number of
individuals
am
TM-496 3 Greece, Koza
´
nti: Euipeus Gorge at
base of Mt Olympos. (KB)
3–
TM-343 3 Greece, Pindhos Mts: Kardı
´
tsa,
hill above Kastania, 900 m.
44
TM-346 3 Greece, Pindhos Mts: Kardı
´
tsa,
N side of Mt Itamos, btw. Karoplesi &
Rahou´ la, 1040 m.
55
TM-344 3 Greece, Pindhos Mts: Kardı
´
tsa, roadside
btw. Mucha & Lake Tavropou´ weir, 890 m.
22
V. alba ssp. alba s. lat.
(race not specified in
TM-469 4 Cyprus, Troo
¨
dos Massif: nr. Saı
¨
ttas, along
Mesapotamos creek.
66
the literature) TM-474 4 Cyprus: nr. B9, btw. Kato Amiantos & Spilı
´
a. 4 4
TM-458 5 Turkey, Erzurum: Yusufeli btw. Artvin &
Erzurum, 1950 m (=V. armena). (WK)
11
TM-237 6 Azerbaijan,
_
IIsmayilli: 1 km S TMzMkMnd, 800 m. 3 3
TM-223 6 Azerbaijan,
_
IIsmayilli: btw. Basqal & Bizlan, 800 m. 6 6
TM-234 6 Azerbaijan,
_
IIsmayilli: nr. Qalac¸ iq, 500 m. 2 3
TM-224 6 Azerbaijan,
_
IIsmayilli:
MrMkit nr. Lahic¸ , 1400 m. 8 8
TM-233 6 Azerbaijan, O
gguz: Das¸ agil River bed, 500 m. 3 1
TM-221 6 Azerbaijan, Quba: Btw. Quba & QMc¸ rMs, 800 m. 5 5
TM-222 6 Azerbaijan, Quba: QMc¸ rMs, 1000 m. 5 6
TM-236 6 Azerbaijan, Qusar: C¸ ılMgir, 800 m. 7 5
TM-226 6 Azerbaijan, QMbMlM:VMndMm, 800 m. 4 4
TM-227 6 Azerbaijan, S¸ Mki: Hill btw. S¸ Mki & Bas¸ ZMyzit, 1600 m. 5 5
TM-228 6 Azerbaijan, S¸ Mki: Hill btw. S¸ Mki & Bas¸ ZMyzit, 1650 m. 3 2
TM-230 6 Azerbaijan, S¸ Mki: Hill btw. S¸ Mki & Bas¸ ZMyzit, 1750 m. 5 5
V. alba ssp. dehnhardti i
(Mediterranean region)
TM-455 7 Morocco, Mekne
`
s: Street to Aı
¨
n Leuh, 25 km N Azrou,
1420 m. (GK)
68
TM-494 8 Spain, Castello
´
n: Tirig N Castello
´
n de la Plana,
500 m (=V. cadevallii). (GK)
3–
TM-81 9 France, Alpes-Maritimes: Btw. Le Bar & Gourdon,
800 m.
8–
56 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
Table 2 (continued)
TM-72 9 France, Alpes-Maritimes: Le Bayou de St. Jeannet,
nr. Vence. (IN)
88
TM-85 9 France, Alpes-Maritimes: Les Barres nr. La Colle-sur-Loup. 2 –
TM-86 9 France, Alpes-Maritimes: Parc St. Donat nr. La Colle-sur-Loup. 4 –
TM-384 9 France, Aveyron: c. 1 km
N St.-Rome-de-Cernon, 600 m.
54
TM-380 9 France, Aveyron: close to D277 btw. Ste.
Eulalie-de-Cernon & Cavale
´
rie, 750 m.
98
TM-65 9 France, Bouches-du-Rhoˆ ne: Cassis, Vallon d’En-Vau. 3 3
TM-37 9 France, Bouches-du-Rhoˆ ne: Parc St. Pons nr. Ge
´
me
´
nos. 6 5
TM-84 9 France, Var: Esterel, Gorges Blavet. 3 –
TM-397 9 France, Var: Les Lecques-sur-Mer, Alle
´
e des Palmiers,
E La Ciotat.
22
TM-69 9 France, Corsica: Bocca di Bonasa nr. Bonifato, 1000 m. (IN) 1 1
TM-70 9 France, Corsica: Bridge of Figarella nr. Calenzana,
400 m. (IN)
11
TM-68 9 France, Corsica: Nr. Calenzana, 300 m. (IN) 2 2
TM-400 10 Croatia, Zadar, Starigrad Paklenica: 0-80 m. (PS) 7 7
TM-401 10 Croatia, Zadar, Velebit: Paklenica National Park,
500–720 m. (PS)
14 14
TM-51 11 Italy, Sicily: Nr. Linguaglossa. 1 1
TM-358 12* Greece, Evia: Mt Dirphys, S of the pass of the
Agios Athanasios-Glifada road, 1000 m. (=V. pentelica)
23
TM-342 12 Greece, Peloponnisos: Achaı
´
a, 2 km up the road to
Rogio & Kerpini.
34
TM-335 12 Greece, Peloponnisos: Arkadı
´
a (Menalon), btw.
Kardaras & N. Kardaras, 990 m.
34
TM-331 12 Greece, Peloponnisos: Arkadı
´
a, btw. Liko
´
hia &
Hrissovitsi, 1150 m.
34
TM-332 12 Greece, Peloponnisos: Arkadı
´
a, descent towards
Hrissovitsi, along road from Liko
´
hia, 1150 m.
22
TM-340 12 Greece, Peloponnisos: Korinthı
´
a, main road btw.
Louzio & Lykouria, 920 m.
22
TM-338 12* Greece, Peloponnisos: Korinthı
´
a, Mt Kyllini, btw.
Trikala & Agios Nikolaos above Go
´
ura, 1550 m.
32
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 57
geographical categories (Fig. 2) prior to data
analysis, 13 of which were used in the discriminant
analysis. Groups 1–6 correspond to ssp. alba s. lat.:
1. SW Europe (ssp. alba s. str.), 2. SW Europe (ssp.
scotophylla), 3. Greece (ssp. thessala), 4. Cyprus, 5.
Turkey: Erzurum, 6. Azerbaijan. Groups 7–13
correspond to ssp. dehnhardtii: 7. Morocco,
8. Spain (V. cadevallii, not included in the mor-
phological analyses), 9. SW Europe, 10. Croatia,
11. Sicily, 12. Greece, 13. Turkey: Adana; Group 14
corresponds to Crete (V. cretica).
Allozyme electrophoresis. Six isozyme systems
were analysed, AAT (aspartate aminotransferase),
AMP/LAP ([leucine] aminopeptidase), GPI (glu-
cose-6-phosphate isomerase), IDH (isocitrate de-
hydrogenase), PGM (phosphoglucomutase), and
SKD (Shikimate dehydrogenase), yielding a total
of 14 interpretable and variable loci. In order to
avoid the problems associated with sliming often
encountered in Viola the use of very young and still
cigar-shaped leaves was imperative for successful
analysis. Leaves were crushed in 4 drops of
grinding buffer (Morden et al. 1987) and the
homogenates centrifuged and stored at )80 C.
The supernatant was loaded in 0.7 · 3mm(25 ll)
wells in the gel (62 g hydrolysed starch, 14 g sucrose
per 500 ml gel-buffer). Gels were run at 4 C for 5–6
hours using a modification of Wendel and Wee-
den’s (1989) buffer system 6 (pH 8.3; electrode-
buffer: Li-Borate, 12.4 g B(OH)
3
/1 H
2
O; gel-buffer:
Tris-Citrate, 6.1 g Tris/l H
2
O) for gels stained for
AAT, AMP/LAP, and GPI, and a modification of
buffer system 1 (Histidine-Citrate, 10.1 g Histidine /
lH
2
O; pH 6.5) for gels stained for IDH, PGM, and
SKD. Allelic and homomeric bands (heterodimers
excluded) were labelled alphabetically from the
most anodal position and scored as anonymous
(0, 1) markers. Within the 14 geographical catego-
ries, band frequencies and diversity indices (Pielou
1969) were calcul ated. The full data set was
analysed in a PCO (Gower 1966) using the
NTSYSpc-2.02h package (Rohlf 1997).
Morphometry. Due to relatively high mortality
only a subset of the plants was included for
morphometry (Table 2). No substantial chasmog-
amous flowering was obtained under greenhouse
conditions and floral characters were therefore not
included in this analysis. Nine morphological
characters were scored, eight on summer leaves
and one on stolons. Out of 3–10 leaves sampled per
individual one median leaf was scored for each
Table 2 (continued)
Taxon Ref. code Group Collection site (collector)
Number of
individuals
am
TM-325 12 Greece, Peloponnisos: Messinı
´
a (Tay
¨
getos Mts),
Langada Gorge, 910 m.
21
TM-330 12 Greece, Peloponnisos: Messinı
´
a, along main
road S Megalopolis, 400 m.
66
TM-478 12 Greece, Pindhos Mts: Ioa
´
nina (Timfi Massif),
btw. Aristi & Papigo. (MR)
12 11
TM-484 13 Turkey, C5 Adana: Pozantı, above Akc¸ atehir,
1500 m. (MD)
66
V. cretica (Crete) TM-360 14 Greece, Crete, Rethymno: River gorge near the
Arkadiou monaster, 500 m.
11 11
58 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
individual; for the two cases where single plants
constituted a discriminant group (5 Turkey: Erzu-
rum, 11 Sicily) three leaves were scored. Stolon
length was estimated by measuring (1) maximal
Internode length; missing values were replaced by
population means. Aspects of leaf blade shape was
estimated by calculating (2) Lamina breadth/length
ratio (measured from basal lobes to apex), (3)
Lamina length to greatest breadth/lamina length
ratio (measured from basal lobes to where the blade
was at its broadest); sinus shape by (4) Sinus depth/
lamina length ratio and (5) Sinus width/lamina
length ratio; (6) Apex angle was estimated by leaf
breadth 10% length below apex/lamina length
ratio; (7) Number of crenations along one blade
margin from basal lobe to apex was counted; hair
cover on leaf surface was estimated by measuring
(8) maximal Hair length and (9) Hair density as
number of ha irs per random 2·2mm
2
(avoiding
veins where pubescence was more developed). The
resulting morphological data were examined using
both univariate (box-plots and ANOVA with
Contrast tests) and multivariate methods
(Canonical Discriminant Analysis) using the SPSS
11.0.0 statistical package.
Results
Allozyme electrophoresis. Number of isozyme
loci and their compartmentalisation was con-
sistent with expectations for a tetraploid Viola
species, i.e. all cytosolic loci were duplicated
except two (‘Amp-1’ and ‘Pgm-1’) (Kim et al.
1991, Marcussen and Nordal 1998, Nordal
and Jonsell 1998, Culley and Wolfe 2001).
Banding patterns in AAT, GPI and IDH
were consistent with dimeric enzymes
expressed in two subcellular compartments,
one ‘fast’ compartment which was not scored
(due to invariability in IDH, and indistinctness
in AAT and GPI) and a ‘slow’ one with two
loci. AMP/LAP, PGM and SKD were
interpreted as monomeric. AMP (low substrate
specificity) and LAP (high substrate specificity)
Fig. 2. Distribution of the Viola alba complex. Sampled populations are superimposed and groups applied in
analysis indicated. 1–6 V. alba ssp. alba s. lat. (1 V. alba ssp. alba s. str., 2 ssp. scotophylla ,3ssp.thessala), 7–13,
ssp. de hnhard tii (8 V. cade vallii), 14 V. cretica
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 59
were stained on the same gel-slice and inter-
preted as two ‘fast’ LAP loci and one ‘slow’
AMP locus. Banding patterns were consistent
with three interpretable PGM loci, one ‘fast’
and two ‘slow’ ones, and two SKD loci.
Variation was detected in 14 of the 15
interpretable loci. Most variation could be
interpreted as allelic but also apparent lack of
gene expression (‘null alleles’) occurred in either
of the duplicated loci occasionally. In polyp-
loids, the absence of allelic band may result
either from concerted evolution among dupli-
cated genes or from ‘true’ silencing of one gene;
besides assigning alleles to loci is usually difficult
(Werth 1989). In consequence, polyploid geno-
types should not be interpreted from pheno-
types without preceding crossing experiments
and analysis of progeny. In the present analyses,
allozyme markers (allelic and homomeric
bands) were therefore treated as anonymous.
Band frequencies and diversity indices
(Pielou 1969) obtained within the 14 geographic-
taxonomic groups are presented in Table 3.
The allozyme data showed a distinct, bimodal
pattern coinciding with Becker’s delimitation
of the complex. Differentiation between ssp.
alba s. lat. (i.e. including ssp. scotophylla and
ssp. thessala) and ssp. dehnhardtii s. lat.
(including Viola cretica) was found in AAT,
GPI. IDH, and SKD: Aat
b
, Gpi
c
, Idh
c
, and
Skd
b
were predominant within the former and
Aat
c
, Gpi
e
, Idh
a
and Skd
a
in the latter. For
AAT and IDH the difference was close to
absolute: only a very few population possessed
the ‘wrong’ allele, most notably populations of
ssp. dehnhardtii from Greece (TM-338, 342,
358 with Aat
b
or Idh
c
), Croatia (Aat
b
), and
Turkey (Aat
b
). Within ssp. alba s. lat. none of
ssp. alba s. str., ssp. scotophylla, and ssp.
thessala had private bands and they could not
be recognised based on allozymes. Private
markers were seen only in Azerbaijani (Amp
c
and Pgm
c
) and Cypriot (Pgm
g
) populations at
low frequency; the locus of Lap
d
was appa-
Table 3. Diversity pa rameters and allozyme marker frequencies within taxonomic and geographical groups
within the Viola alba complex, as defined in Fig. 2. Geography is indicated for ssp. alba s. lat. and ssp.
dehnhardtii unless the taxon has a restricted distribut ion, e.g. V. alba s.str. and ssp. scotophylla (both
western Europe), ssp. thessala (Greece), and V. cretica (Crete). N = number of individuals; N
p
= number
of populati ons; M = number of multilocus phenotypes; D
f
= the complement of Simpson’s Diversity
Index (D) corrected for finite sample size (Pielou 1969), calculated as D
f
=1) {[Rn
i
(n
i
) 1)]/[N (N ) 1)]},
where n
i
is the number of individuals with multilocus phenotype i and N is as above
Taxon/region/
group
Diversity
parameters
Frequency of allozyme markers
NN
p
MD
f
Aat
a
Aat
b
Aat
c
Amp
a
Amp
b
Amp
c
Gpi
a
Gpi
b
Gpi
c
Gpi
d
Gpi
e
alba ssp. alba 1 alba s.str 25 7 3 0.50 1 1 – – 1 – 1 – 0.76 – 0.24
s. lat. 2 scotophylla 11 4 2 0.29 1 1 – 0.27 1 – 1–1––
3 thessala 30 10 15 0.91 1 0.83 – – 1 – 1 – 0.83 – 0.23
4 Cyprus 10 2 6 0.78 11–– 1 – 1–1––
5 Turkey 1 1 1 11–– 1 – 1–1––
6 Azerbaijan 56 12 31 0.95 1 1 – – 1 0.04 1 – 0.95 – 0.20
alba ssp. 7 Morocco 6 1 1 01–1– 1 – –1––1
dehnhardtii 8 Spain 3 1 1 01–1– 1 – 1–––1
9 W Europe 54 13 27 0.95 0.57 – 1 0.59 0.44 – 0.96 0.06 – – 1
10 Croatia 21 2 4 0.74 1 0.24 0.76 – 1 – 1–––1
11 Sicily 1 1 1 1–11 – – –1––1
12 Greece 38 10 14 0.86 1 – 1 0.21 0.79 – 0.89 – 0.24 0.13 0.63
13 Turkey 6 1 2 0.33 1 1 – 0.17 0.83 – 1–––1
cretica 14 Crete 11 1 1 01–1– 1 – 1–––1
60 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
rently silenced in SW European populations.
Viola cretica differed from ssp. dehnhardtii
only by the private Lap
b
. Ssp. dehnhardtii
produced a higher number of private markers
than ssp. alba s. lat.: Lap
a
was private to and
fixed in the Moroccan plants, Gpi
b
was
restricted to the West Mediterranean (Moroc-
co, SW Europe, Sicily), Gpi
d
to Greece, and
Idh
b
to SW Europe. Of considerable interest
were the markers that displayed the same
geographical distribution within both ssp. alba
s. lat. and ssp. dehnhardtii. Pgm
a
(Azerbaijan,
Cyprus, Turkey) and Skd
d
(Cyprus, Turkey)
were restricted to the eastern parts. Skd
e
(SW
Europe) was restricted to the western parts at
low frequency.
There were considerable differences in esti-
mated genetic diversity (D
f
) among the 14
geographical categories (Table 3), D
f
values
ranging from 0 to almost 1. However, for six of
the categories (1, 7, 8, 11, 13, 14) the estimate
was based on one population alone and might
be inaccurate. Among the remaining groups,
D
f
proved particularly low in ssp. alba s. str.
(D
f
¼.50) and in ssp. scotophylla (D
f
¼.29) in
SW Europe (pooled D
f
¼.46) while the other
six groups had much higher D
f
values ranging
between .74 and .95.
A scatter plot of the two first PCO axes
(Fig. 3) visualises some of the points made
above. PCO axis 1 revealed a bimodal distri-
bution of individuals, separating ssp.
dehnhardtii (left), including Viola cretica, from
ssp. alba s. lat. (right), including ssp. scotophy-
lla and ssp. thessala. PCO axis 2 separated
V. cretica (top) from ssp. dehnhardtii. Geo-
graphical resolution was rather poor within
ssp. dehnhardtii and ssp. alba s. lat. but a few,
important points should be made: Firstly, ssp.
alba s. str. and ssp. scotophylla (SW Europe),
as well as the Cypriot plants, expressed only
small subsets of the total variation within the
ssp. alba s. lat. cluster. Secondly, the Greek
materials tended to bridge the two main
Idh
a
Idh
b
Idh
c
Idh
d
Lap
a
Lap
b
Lap
c
Lap
d
Lap
e
Pgm
a
Pgm
b
Pgm
c
Pgm
d
Pgm
e
Pgm
f
Pgm
g
Skd
a
Skd
b
Skd
c
Skd
d
Skd
e
––11– –1–––1––11––10.96 – 0.04
––11––1–––1––11––11––
––11––0.90 0.50 0.03 – 1 – 0.23 0.63 0.93 – 0.03 0.97 1 – –
––11––11–0.30 0.70 ––110.20 – 0.60 1 0.30 –
––11––11––1––11––11––
0.02 –11––0.96 0.70 0.39 0.16 0.93 0.05 0.21 0.70 0.91 – 0.02 0.98 1 – –
1––11–––––1––11––11––
1––1––11––1––11–1–1––
1 0.19 – 0.83 – – 0.96 0.65 0.35 – 1 – 0.24 0.78 0.81 – 0.87 0.19 0.98 – 0.02
1––1––0.14 1––1––0.81 1–1–1––
1––1––11––1––11––11––
0.79 – 0.21 1––10.84 – – 1 – 0.71 0.89 0.29 – 0.66 0.37 0.95 – –
1––1––11–1–––11–––11–
1––1–11–––1–––1––11––
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 61
clusters by transitional forms: i.e. three popu-
lations with ‘recombinant’ AAT/IDH pheno-
types, indicated with population numbers in
the plot (TM-338, 342, and 358), fell well
within ssp. alba s. lat. in this analysis (but see
the morphometric results).
Morphometry. Extensive variability was
documented also in the nine morphological
characters scored. Boxplots of six of the
characters varying among the 14 categories
are show in Fig. 4. For several characters a
correlation between variability and number of
populations was evident, indicating that quite a
lot of variation was expressed among popula-
tions within each category. Within ssp. alba s.
lat., ssp. alba s. str. was characterised by
slightly higher numbers of crenulae (the only
character by which it differed from ssp. scoto-
phylla), Whereas ssp. thessala had generally low
scores for Number of crenulae, Sinus depth/
lamina length ratio, and Apex angle. Within
ssp. dehnhardtii, variation was also high; most
noteworthy were perhaps the broad leaves of
the Moroccan plants (Lamina breadth/length
ratio) and the long pubescence of the Sicilian
plant (Hair length). Two of the ‘recombinant’
Greek populations, intermediate between ssp.
thessala and ssp. dehnhardtii, generally pointed
in the direction of the latter, but took extreme
Hair length values: TM-338 having long hairs
(similar to the Sicilian plant) and TM-358 being
completely glabrous. Also Viola cretica gener-
Fig. 3. Scatter plot of the two first axes in a PCO analysis of Jaccard’s Similarity between 273 individual
allozyme multilocus phenotypes in the Viola alba complex, using Squared Euclidean Distance. Axes 1, 2, and 3
(last not shown) extracted 39.5%, 7.8%, and 7.1% of the variation, respectively. Subdivision of materials into
geographical and taxonomical groups, indicated by numbers, is according to Fig. 2: 1–6 V. alba ssp. alba s. lat.
(1 ssp. alba s. str., 2 ssp. scotophylla, 3 ssp. thessala), 7–13 V. alba ssp. dehnhardtii (8 V. cadevallii), 14 V. cretica
(Crete). Fences, according to clusters obtained in UPGMAs (not shown), are superimposed to highlight the
bimodal distribution along the first PCO axis. Thr ee Greek populations, here grouped within ssp. alba s. lat.
And indicated with population numbers (TM-338, 342, 358), showed clear morphological affinities to ssp.
dehnhardtii
62 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
ally fell within the range of ssp. dehnhardtii but
differed significantly by high values for both
Hair length and Internode length.
The placement of three populations display-
ing ‘recombinant’ AAT or IDH phenotypes
TM-338, TM-342 and TM-358, was attempted
resolved using a Canonical Discriminant Anal-
ysis of the Greek material (Fig. 5). Discrimi-
nant groups corresponded to populations (15)
and the three target populations were left
Fig. 4. Boxplots of six morphological characters discriminating among taxa and populations within the Viola
alba complex (cf. Table 4). Groups are according to Fig. 2. alb ¼ssp. alba s. str. (SW Europe); sco ¼ssp.
scotophylla (SW Europe); the ¼ssp. thessala (Greece); Cy ¼Cyprus; Tu ¼Turkey; Az ¼Azerbaijan; Mo ¼
Morocco; WE ¼SW Europe; Cro ¼Croatia; Si ¼Sicily; Gr ¼Greece; Cre ¼Crete; intermed. ¼Greek
populations deemed intermediate between ssp. alba s. lat. and ssp. dehnhardtii (TM-338, 358)
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 63
ungrouped. There was a remarkably distinct
separation of the Greek plants into three
clusters, (i) ssp. thessala from Evia (left), (ii)
ssp. thessala from Peloponnisos and Pindhos
(top), and (iii) ssp. dehnhardtii (right). The
regional differentiation of ssp. thessala was,
however, not repeated by allozymes. The
ungrouped populations TM-338 (Peloponn-
isos) and TM-358 (Evia) fell out as intermedi-
ate between thessala and dehnhardtii in
morphology whereas TM-342 clustered within
dehnhardtii. Apex angle (.456), Internode
length ().410) and Hair density (.294) corre-
lated strongly with Discriminant axis 1, Num-
ber of crenulae (.632) with axis 2, and Hair
length ().264, .525) with both.
The overall structure in the full morpho-
metric data set was resolved using a Canonical
Discriminant Analysis (Fig. 6). The centroids
of the 13 discriminant groups, defined by
geography and taxon in combination (cf.
Table 2, Fig. 2), are indicated in bold figures.
The two Greek populations deemed interme-
diate between ssp. dehnhardtii and ssp. alba s.
lat. (Fig. 5), were left ungrouped in the ana-
lysis; their centroids are indicated with popu-
lation numbers (TM-338 and 358). Although
there was considerable overlap in variation
among individuals of the different discriminant
groups, the two first axes in combination
separated well among the group centroids of
ssp. dehnhardtii, ssp. alba s. lat., and V. cretica.
Ssp. alba s. lat. (1–6) obtained high scores
along both axes and fell out in the upper right
part of the plot. Ssp. dehnhardtii (7–13) and
V. cretica (14) fell out in a diagonal belt from
upper left to lower right, with dehnhardtii to
the left and cretica to the right, and with the
single individual of dehnhardtii group 11
(Sicily) intermediate between them and close
to ssp. alba s. lat. The two somewhat interme-
diate ungrouped Greek populations here fell
within ssp. dehnhardtii, but one (TM-338) near
the Sicilian plant in close proximity of ssp. alba
s. lat.; both were nested within ssp. alba s. lat.
in the allozyme PCO. Internode length (.569)
and the Apex angle ().305) were both corre-
lated with axis 1 and thus discriminated
between ssp. dehnhardtii on the one hand and
ssp. alba s. lat. and V. cretica on the other,
whereas Lamina breadth/length ratio (.320)
and Sinus breadth/lamina length ratio (.194)
Fig. 5. Canonical Discriminant Analysis (two first axes) of Greek Viola alba ssp. dehnhardtii and ssp. thessala,
based on nine morphological characters and 64 individuals in 15 discriminating groups (populations). The three
ungrouped populations (TM-338, 342, 358) nested within ssp. alba s. lat. in the allozymic analysis, fell here
within ssp. dehnhardtii (TM-342) or were intermediate (TM-338, 35 8)
64 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
were correlated with axis 2, discriminating
V. cretica from the other two. Number of
crenulae (.501, .661) was strongly positively
correlated with both axes, thus separating ssp.
alba s. lat. from ssp. dehnhardtii and V. cretica.
Hair length (.524, ).461) was strongly
positively correlated with axis 1, but negatively
with axis 2, and thus delimited all three main
groups, ssp. dehnhardtii (minus group 11), ssp.
alba s. lat., and V. cretica.
Distinctions among the three main groups,
ssp. alba s. lat., ssp. dehnhardtii, and V. cretica
were resolved using ANOVAs with correspon-
dent Contrast tests for each of the nine
morphological characters. Character descrip-
tives (mean, standard deviation, range, and
interquartile range) and ANOVA results with
individual contrasts are shown in Table 4.
Highly significant differences (p < .0005)
among taxa were found for Hair length,
Number of crenulae, Internode length, Apex
angle and Hair density. Contrasts yielded
significant differences between all taxon pairs
both for Hair length and for Internode length,
significant differences only between alba and
dehnhardtii and between alba and cretica for
Number of crenulae, significant differences
between alba and dehnhardtii and between
dehnhardtii and cretica for Apex angle, and
only between alba and dehnhardtii for Hair
density. Statistically significant, although not
necessarily biologically significant, differences
were detected within other characters as well,
due to high N’s and in spite of discouragingly
large overlaps in range.
Discussion
Systematic considerations. Ssp. alba. Neither
allozymes nor morphometry discriminated
consistently among the three taxa ssp. alba s.
str., ssp. scotophylla, and ssp. thessala. Wher-
ever differing, they fell within the general range
of variability of ssp. alba in the wide sense.
Ssp. alba s. str. and ssp. scotophylla in
Europe are distinguished solely on the basis of
pigmentation pattern, as noted already by
Jordan (1849, p. 11) in the protologue of Viola
Fig. 6. Canonical Discriminant Analysis (two first axes) of the Viola alba complex, based on nine
morphological characters and 240 individuals in 13 discriminant groups (group 8 excluded). The three first
axes extracted 83.4% of the total variance, 46.6%, 28.4%, and 8.4%, respectively. Discriminant group centroids
are indicated by numbers and taxa delimited by fences, e.g. 1–6 V. alba ssp. alba s. lat., 7–13 ssp. dehnhardtii,14
V. cretica (cf. Fig. 2). Two Greek populations, indicated with centroids (TM-338, 358) and intermediate
between ssp. dehnhardtii and ssp. thessala in Fig. 5, were treated as ungrouped in this analysis
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 65
scotophylla. Schmidt’s (1961, p. 63) statement
that they differ markedly in their Central
European distributions seems to be incorrect.
The two colour morphs are in fact largely
sympatric, also outside Europe (Ruprecht
1869), and often co-occur in mixed
populations (Becker 1910a, Gams 1925, Hess
et al. 1970). This is evident also from herbarium
Table 4. Analysis of nine morphological characters of Viola alba ssp. alba s. lat., ssp. dehnhardtii, and
V. cretica. Character descriptives, ANOVA results, and individual Contrast Test results are indicated.
Contrast
1
=1· [alba] ) 1 · [dehnhardtii]+0· [cretica]; contrast
2
=1· [alba]+0· [dehnhardtii] )
1 · [cretica]; contrast
3
=0· [alba]+1· [dehnhardtii] ) 1 · [ cretica]. Signific ant values (p < .05) are
indicated in bold
Character
Descriptives
Taxon N Mean Std.
Dev.
Percentile
25
Percentile
75
Range
Hair length (mm) alba 122 0.685 0.196 0.550 0.800 1.275
cretica 11 1.141 0.155 1.000 1.250 0.450
dehnhardtii 111 0.503 0.204 0.375 0.625 1.000
Total 244 0.623 0.244 1.525
Number of crenulae alba 122 22.07 4.89 19 26 22
cretica 11 14.73 2.20 13 17 7
dehnhardtii 111 16.68 2.78 14 18 14
Total 244 19.28 4.85 23
Internode length (mm) alba 104 30.58 10.99 23 36 63
cretica 11 37.73 21.00 21 55 72
dehnhardtii 77 16.13 9.32 7.5 22 47
Total 192 25.19 13.44 79
Apex angle alba 122 0.212 0.043 0.185 0.239 0.243
cretica 11 0.228 0.049 0.197 0.275 0.152
dehnhardtii 111 0.253 0.051 0.212 0.286 0.296
Total 244 0.231 0.051 0.311
Hair density alba 122 8.39 4.03 6 10 22
(hairs / 4 mm
2
) cretica 11 8.73 1.56 7 10 4
dehnhardtii 111 11.14 5.50 7 15 29
Total 244 9.66 4.87 29
Lamina length alba 122 0.321 0.0414 0.296 0.347 0.232
to greatest breadth : cretica 11 0.289 0.0471 0.262 0.300 0.188
length ratio dehnhardtii 111 0.336 0.0446 0.302 0.370 0.243
Total 244 0.326 0.0443 0.278
Sinus depth : alba 122 0.262 0.0406 0.241 0.290 0.236
lamina length ratio cretica 11 0.291 0.0247 0.283 0.313 0.084
dehnhardtii 111 0.269 0.0501 0.235 0.303 0.246
Total 244 0.267 0.0450 0.246
Lamina breadth : alba 122 0.765 0.0676 0.718 0.815 0.324
length ratio cretica 11 0.733 0.0869 0.677 0.758 0.320
dehnhardtii 111 0.775 0.0796 0.714 0.826 0.356
Total 244 0.768 0.0744 0.356
Sinus width : alba 122 0.340 0.0854 0.262 0.408 0.487
lamina length ratio cretica 11 0.350 0.0641 0.296 0.374 0.238
dehnhardtii 111 0.347 0.1078 0.259 0.419 0.603
Total 244 0.344 0.0953 0.603
66 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
records, and was the case in all four sites where
ssp. scotophylla was sampled for this study.
Equivalent pigmentation morphs occur
within other species in the subsection as well,
e.g. in V. odorata (Walters 1946) and V. collina
(Gerstlauer 1943). The genetic basis for such
polymorphism is probably simple and may be
explained by bi-allelic variation in two loci,
Analysis of nine morphological characters of Viola alba ssp. alba s. lat., ssp. dehnhardtii,andV. cretica.
Character descriptives, ANOVA results, and individual Contrast Test results are indicated.
Contrast
1
=1· [alba] ) 1 · [dehnhardtii]+0· [cretica]; contrast
2
=1· [alba]+0· [dehnhardtii] )
1 · [cretica]; contrast
3
=0· [alba]+1· [dehnhardtii] ) 1 · [ cretica]. Signific ant values (p < .05) are
indicated in bold
ANOVA Contrast tests
Sum of
Squares
df Mean
Square
F p Contrast Value of
Contrast
Std.
Error
tdfp
Between Groups 200.338 2 100.169 63.525 0.000 1 0.1817 0.0260 6.975 241 0.000
Within Groups 380.021 241 1.577 2 )0.4559 0.0625 )7.293 241 0.000
Total 580.359 243 3 )0.6375 0.0628 )10.158 241 0.000
Between Groups 1933.076 2 966.538 61.544 0.000 1 5.3981 0.5198 10.385 241 0.000
Within Groups 3784.842 241 15.705 2 7.3465 1.2476 5.889 241 0.000
Total 5717.918 243 3 1.9484 1.2527 1.555 241 0.121
Between Groups 11067.602 2 5533.801 44.604 0.000 1 14.4471 1.6746 8.627 189 0.000
Within Groups 23448.268 189 124.065 2 )7.1503 3.5315 )2.025 189 0.044
Total 34515.870 191 3 )21.5974 3.5902 )6.016 189 0.000
Between Groups 0.098 2 0.049 22.297 0.000 1 )0.0411 0.0062 )6.673 241 0.000
Within Groups 0.532 241 0.002 2 )0.0160 0.0148 )1.081 241 0.281
Total 0.630 243 3 0.0251 0.0149 1.693 241 0.092
Between Groups 446.812 2 223.406 10.120 0.000 1 )2.7417 0.6163 )4.449 241 0.000
Within Groups 5320.270 241 22.076 2 )0.3338 1.4791 )0.226 241 0.822
Total 5767.082 243 3 2.4079 1.4852 1.621 241 0.106
Between Groups 0.028 2 0.014 7.624 0.001 1 )0.0144 0.0057 )2.548 241 0.011
Within Groups 0.449 241 0.002 2 0.0325 0.0136 2.395 241 0.017
Total 0.477 243 3 0.0470 0.0136 3.442 241 0.001
Between Groups 0.009 2 0.005 2.363 0.096 1 )0.0066 0.0059 )1.130 241 0.260
Within Groups 0.482 241 0.002 2 )0.0288 0.0141 )2.045 241 0.042
Total 0.491 243 3 )0.0222 0.0141 )1.567 241 0.118
Between Groups 0.020 2 0.010 1.803 0.167 1 )0.0102 0.0097 )1.048 241 0.296
Within Groups 1.326 241 0.006 2 0.0314 0.0234 1.344 241 0.180
Total 1.346 243 3 0.0416 0.0235 1.773 241 0.077
Between Groups 0.003 2 0.002 0.180 0.836 1 )0.0070 0.0125 )0.560 241 0.576
Within Groups 2.203 241 0.009 2 )0.0097 0.0301 )0.321 241 0.748
Total 2.207 243 3 )0.0026 0.0302 )0.088 241 0.930
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 67
one coding for anthocyan production, and
another for its expression in the corolla (thus,
giving rise to three possible morphotypes: the
alba morphotype, and the white- and lilac-
flowered scotophylla morphotypes). The
long-held assumption that ssp. scotophylla
originated through introgression of anthocyan
genes from V. hirta into ssp. alba (e.g. Murr
1924, Gerstlauer 1943, Scho
¨
fer 1954) is not
supported by allozymes nor karyotypes (Sch-
midt 1961).
In summary, the differentiation between
ssp. scotophylla and ssp. alba seems too tenu-
ous to deserve taxonomic recognition; hence,
scotophylla should be included in the synonymy
of ssp. alba.
It is also clear that the Greek plants,
referred to as ssp. thessala in recent taxonomic
publications (Livaniou-Tiniakou 1991; Tini-
akou 1991, 2000), are not particularly deviant
within ssp. alba s. lat. Like ssp. scotophylla,
ssp. thessala is probably better included in ssp.
alba s. lat. However, plants from the island of
Evia, off the southeastern coast of the Greek
mainland, are morphologically distinct from
plants from Pindhos and Peloponnisos, e.g. in
having more triangular, acute leaves with an
open, shallow sinus and shorter indument (cf.
Fig. 3). This may be interpreted as incipient
geographical race formation but should not be
given too much taxonomic weight.
The present investigation has shed some
light on the infraspecific identity of Viola alba
in the east Mediterranean region, which has
long been unclear. Firstly, the Cypriot plants
belong to ssp. alba s. lat. (cf. Meikle 1977). In
Turkey, ssp. alba s. lat. (i.e. V. armena) was
shown to occur in the north and ssp.
dehnhardtii in the south. This contrasts with
earlier treatments, in which Turkish plants
were referred collectively to ssp. alba (Becker
1910b, 1918) or Anatolian plants, including
V. armena, to ssp. dehnhardtii and plants from
Turkey-in-Europe to ssp. thessala (Coode and
Cullen 1965, Davis et al. 1988). However, there
are indications that the distinction between
ssp. alba s. lat. and ssp. dehnhardtii becomes
unclear in the east Mediterranean region.
Traits characteristic of the other subspecies
are expressed in Cypriot ssp. alba (e.g. short
pubescence) as well as in Turkish ssp. de-
hnhardtii (Aat
b
), not counting the numerous
intermediate plants from Greece.
Ssp. dehnhardtii, although shown here to be
as variable as ssp. alba s. lat., has not been
subdivided to the same extent into geograph-
ical and morphological races. Where taxa have
been delimitated it has mainly been glabrous
or glabrescent forms, which occur scattered
throughout the range of ssp. dehnhardtii and
have been included in its synonymy by the
majority of recent authors. Not surprisingly,
allozyme analysis confirmed the inclusion of
the glabrous V. cadevallii (TM-494) from
Spain in ssp. dehnhardtii (morphometry was
not performed). On the other hand, the situ-
ation for the glabrous plants from Evia
(TM-358), referred to as V. pentelica, is more
complex. These individuals are morphological-
ly and enzymatically somewhat intermediate
between ssp. dehnhardtii and ssp. alba and may
constitute parts of the transitional zone
between ssp. alba in the north and ssp.
dehnhardtii in the south.
Viola cretica appears to fall within the
range of allozymic variation of V. alba. The
same conclusion may be drawn from
karyotype analyses on the two taxa (Liva-
niou-Tiniakou 1991, Tiniakou 1992) and from
preliminary results obtained from ITS seque-
nces (M. Henry, J. Munzinger, T. Marcussen,
unpublished data).
The present data indicate a particularly
close relationship between V. cretica and ssp.
dehnhardtii. Indeed, Becker (1910b) included
V. cretica in its synonymy, while later authors
generally treated it as a separate species.
Characters that have been considered diagnos-
tic for V. cretica are the long and slender
stolons, the hispid pubescence, and the smaller
flowers lacking trichomes inside the lateral
petals (‘‘petal bearding’’); their diagnostic value
against ssp. dehnhardtii is, however, debatable.
Firstly, both pubescence and flower size
are characters particularly prone to vary
between mainland and island populations of
68 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
the same species and should not be given too
much taxonomic importance (Polunin 1980).
Furthermore, both hispid and glabrescent
forms of Viola cretica occur throughout its
range in Crete (Levka Ori, Psiloritis, Dhikti)
(Livaniou-Tiniakou 1991). Thus, for this
character, V. cretica falls within the general
range of variability of V. alba ssp.
dehnhardtii although hispid plants are rare
in dehnhardtii (Espeut 1999). This appears to
be the case for corolla size and petal
bearding as well, as both small-flowered (cf.
Livaniou-Tiniakou 1991, Marcussen 1998)
and unbearded forms of V. alba occur
(e.g. TM-401). Apparently, petal bearding is
closely associated with pollination syndrome
in Viola (Beattie 1974). It is often lost in
isolated populations, such as island endemics
(V. cretica, V. jaubertiana), and in presum-
ably inbreeding taxa. The only investigated
character that could be considered diagnostic
for V. cretica is the stolons, which are
usually longer than in V. alba.
Based on the evidence above, Viola cretica
seems to belong naturally in V. alba.If
included in ssp. dehnhardtii, to which it is the
most closely related, it would be as a quite
untypical morph. Therefore, it seems most
appropriate to treat this taxon as a separate
subspecies, i.e., V. alba ssp. cretica (Boiss. &
Heldr.). In this way, the close relationship
between V. cretica and V. alba is pointed out
while also underlining that a deviant geo-
graphical race is at hand.
Phylogeography. The differentiation on
the population level observed in this material
of Viola alba is probably attributable to its
inbreeding reproductive system where cleistog-
amous selfing and clonal propagation by
means of stolons seem common. Apparently,
introgression is not rare among species within
subsection Viola (Scho
¨
fer 1954, Schmidt 1961,
Marcussen and Borgen 2000) but any recent
impact of importance was not confirmed in
this study.
Obviously, the major regional trends of
variation in the Viola alba complex have been
shaped by historical events. Most recently,
these comprise the size and location of glacial
refugia and the course of immigration follow-
ing the climatic improvement towards the end
of the Wu
¨
rm glaciation, c. 10,000 years ago.
Evidence from palynological and phyloge-
ographic studies (e.g. Polunin and Walters
1985, Bennett et al. 1991, Fukarek et al. 1995,
Brewer et al. 2002, Denk et al. 2002) suggests
that deciduous forests, which is the prime
habitat for Viola alba, existed in refugia in a
narrow belt around the Mediterranean basin,
primarily in the southwest and in the north-
east, as well as in the Ponto-Caucasian region.
A narrow channel in the Aegean region may
have interconnected these two regions.
In the light of this evidence it is plausible to
assume that ssp. alba s. lat. survived the Wu
¨
rm
glaciation in the eastern, Ponto-Caucasian ref-
ugium and, hence, colonised Europe from the
east. The low diversity detected in the western-
most populations is in accordance with this
hypothesis (cf. Hewitt 1999). Ssp. dehnhardtii
seems to have been recruited from a Mediter-
ranean refugium, and to have recolonised
Europe from the south. An in situ refugium
for ssp. cretica in Crete seems probable consid-
ering the isolated location of this island; the
mountains of Crete are also high enough
(2,400 m) to have ensured enough precipitation
and allowed for vertical migration.
This phylogeographic scenario is further
supported by the presence of a transitional zone
between ssp. alba s. lat. and ssp. dehnhardtii.
This zone, stretching across southern Europe
from northeast Spain to Greece (Becker 1910b,
Merxmu
¨
ller 1982, Mun
˜
oz Garmendia et al.
1993, Espeut 1999), may be interpreted as a
persistent hybrid zone where dehnhardtii and
alba met during recolonisation. Its localisation,
running more or less parallel to the Mediterra-
nean coast and close to the watershed, indicates
that the mountains of southern Europe were
major obstacles in the postglacial recolonisa-
tion of Viola alba into Europe, particularly for
ssp. dehnhardtii.
The phylogeographic scenario of Viola alba,
where most of Europe was recolonised from the
east, coincides rather well with the ‘‘grasshop-
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 69
per paradigm’’, one of three general postglacial
recolonisation patterns envisioned by Hewitt
(1999, 2001) for European biota. This pattern
of recolonisation was apparently shared by a
wide range of plant and animal species, includ-
ing the ecologically important tree species
beech (Demesure et al. 1996, Denk et al.
2002) and black alder (King and Ferris 1998).
The differentiation between ssp. alba and
ssp. dehnhardtii probably reflects long-time
isolation in two refugia, probably repeatedly
during more than one glaciation cycle. On the
other hand, a more recent origin of ssp. cretica
is suggested, probably through peripatric der-
ivation from ssp. dehnhardtii.
Within ssp. alba, the increased frequency of
white-flowered and anthocyanine-free plants in
west Europe, i.e. the classical delimitation of
ssp. alba s. str. and ssp. scotophylla, is prob-
ably caused by recurrent founder events,
resulting in skewed allele frequencies, during
recolonisation from the Pontic-Caucasian
refugium. Equivalent east-west clinical varia-
tion patterns have been reported also in beech
(Denk et al. 2002), and due to the long
migration distances such variation patterns
are probably only to be expected among taxa
sharing the phylogeographic ‘‘grasshopper
paradigm’’.
The practically allopatric distributions of
ssp. alba and ssp. dehnhardtii call into question
reports of the latter from outside the Mediter-
ranean basin. Particularly, records from north-
ern Greece (Livaniou-Tiniakou 1991, Tiniakou
1991), Bulgaria (Delipavlov 1979), Romania
(Grint¸ escu et al. 1955), and most of Turkey
(Coode and Cullen 1965) probably refer to
broad-leaved forms of ssp. alba (cf. illustra-
tions in Grint¸ escu et al. 1955). However, it
should be pointed out that populations in the
Aegean area, geographically located between
the two main putative refugia, might have
been connected to either of them, at least
temporarily, during history. Primary contact
between alba and dehnhardtii probably ex-
plains the presence of intermediates in Greece,
and to some extent also the slightly deviating
plants from southern Turkey (TM-484) and
Cyprus (TM-469). Indeed, the restricted oc-
currence of two alleles, Pgm
a
and Skd
d
, within
both ssp. alba and ssp. dehnhardtii in the
eastern parts of their distributions might be
indicative of some gene flow between the two
subspecies in this region. Future research
should therefore include plants from the east-
ern Mediterranean region, where the delimita-
tion and distribution of ssp. alba and ssp.
dehnhardtii is still uncertain.
Taxonomic treatment and nomenclature
Key to the subspecies of Viola alba Besser,
Prim. fl. Galiciae austriac. 1: 171 (1809). Char-
acters are based on the literature (Becker
1910a, b; Schmidt 1961) and on observations
made in this study. Character ranges, where
indicated, correspond to interquartile ranges
(the middle 50% of data points) of data
presented in this paper (Table 4). Characters
should therefore be used in combination for a
correct determination.
1. Leaves ± acute with straight margins (i.e.
triangular-cordate), usually with 19–26
crenulae along one margin. Stolons long;
maximum internode length 23–36 mm.
Hairs 0.5–0.8 mm long. Flowers lilac or
white (usually white in W and C Europe).
Capsule purplish or light green. South
Europe to Caucasus, Mediterranean only
in the east...................................1. ssp. alba
– Leaves ± obtuse with convex margins (i.e.,
ovate-cordate), usually with 13–18 crenulae
along one margin. Flowers lilac. Capsule
purplish. Mediterranean........................... 2.
2. Stolons usually short; maximum internode
length usually 7.5–22 mm. Plant short-
pubescent or occasionally glabrous; hairs
usually 0.4–0.6 mm. ........2. ssp. dehnhardtii
– Stolons very long; maximum internode
length usually 21–55 mm. Plant hispid; hairs
usually 1.0–1.3 mm long. Endemic to
Crete.......................................3. ssp. cretica
1. Viola alba Bess. ssp. alba.
Type: Poland, ‘‘In montosis et sylvaticis versus
Duklam E. C.’’ Besser (KW). — Topotype:
70 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
‘‘Culta in horto botan. Lemberg e loco
classico Besseriano Galiciae 1889 ’’ Rehman
(LY-Gandoger!).
Synonyms:
– V. scotophylla Jord., Observ. pl. nouv. 7: 9–
11 (1849); V. alba ssp. scotophylla (Jord.)
Nyman, Consp. Fl. Eur. 1: 78 (1878).
Type: ‘‘flore albo, [France:] Lyon frequens
primovere’’ Jordan (Lectotypes. –selected
here–LYJB-Sagot!); ‘‘flore albo, e
´
peron or-
dinairement violace
´
[note added later], Lyon
a
`
Francheville etc.’’ Jordan (Paratypes
LYJB-Sagot!); ‘‘flore violaceo, Lyon a
`
Francheville’’ Jordan (Paratypes LYJB-Sa-
got!); ‘‘Lyon, Maj. 1849’’ Jordan (Para. K).
Jordan selected (at least) four syntypes,
annotated with an exclamation mark after
the author’s name, among a large number of
specimens collected by him from the district
of Lyon. Jordan included both white- and
lilac-flowered individuals in V. scotophylla
but stated in the protologue that the white-
flowered morph was most common; a lecto-
type was therefore selected among these.
– V. thessala Boiss. & Spruner, Diagn, Pl.
Orient., se
´
r. 1, 8: 51–52 (1849); V. alba ssp.
thessala (Boiss. & Spruner) Hayek, Prodr.
Fl. Penins. Balc. 1: 502 (1925).
Type: Greece, ‘‘in monte Oeta Thessaliae,
1837’’ Spruner (Lectotypes. –selected here–
G-BOIS!).
This voucher consists of two specimens of
which the lower was selected as a lectotype.
This effectuates the lectotypification per-
formed by Livaniou-Tiniakou based on the
same specimen(s) and published ineffective-
ly in her Ph.D. thesis (Livaniou-Tiniakou
1991: 75).
– V. armena Boiss. & E
´
. Huet, Diagn. Pl.
Orient., se
´
r. 2, 5: 48 (1856).
Type: Turkey, ‘‘Erzeroum, Jun. 1853’’ Huet
du Pavillon (Holotypes. G-BOIS!).
Of the two specimens mounted on this sheet
only the left specimen corresponds with the
protologue and is to be considered the
holotype of Viola armena. The right speci-
men, with ‘‘Viola armena Boiss. & Huet.’’
written on a separate tag, lacks all the
necessary information about collection site,
date and collector and should be discarded.
– V. virescens Jord. ex Boreau, Fl. Centre
France ed. 3, 2: 77 (1857).
Type: France, ‘‘Tassin (Rhoˆ ne): bois, haies.
28. mars, 30. avril’’ Boullu 704 (Lectotypes
–selected here– Herb. Jordan!, deposited in
ISARA, in Faculte
´
Catholique de Lyon;
Isotypes LY-Bonaparte!).
Apparently, Jordan assigned no collec-
tions to this taxon himself. Boullu 704 is
the only specimen of this taxon in his
herbarium.
– V. besseri Rupr., Fl. Caucasi I: 151–153
(1869).
Described from Georgia, type not traced.
Probable syntype: Georgia: ‘‘Iberia
caucasia pr. Tiflis, 09.04.1861’’ Ruprecht
(G-BOIS!).
2. Viola alba ssp. dehnhardtii (Ten.) W. Becker,
Ber. Bayer. Got. Ges. 8 (2): 257 (1902).
Basionym: V. dehnhardtii Ten., Ind. Sem.
Hort. Neap.: 12 (1830); ex. ej. Syll: 117–
119 (1831).
Type: Italy, Naples, ‘‘Camaldoli’’ Tenore
(Lectotypes –selected here– NAP!). This
effectuates the unpublished lectotypification
performed by Schmidt (11.04.1961) on the
same specimens. – Topotype: ‘‘Loc. Cam-
pania. – In silvis circa Neapolim (Napoli),
loco dicto Camaldoli (loc. class.), alt. 400 m.,
solo siliceo. Apr. 1907’’ Guadagno (LY-
Bonaparte!).
Plants from the locus classicus range from
completely glabrous (the type collection) to
hispid ones (Guadagno’s collection).
Synonym:
– V. cadevallii Pau, Mem, Acad. Ci. Artes
Barcelona, ser. 3, 2; 62–63 (1895).
Described from Spain, Catalonia. Type not
traced.
3. Viola alba ssp. cretica (Boiss. & Heldr.)
Marcussen, comb. nov.
Basionym: V. cretica Boiss. & Heldr., Diagn.
Pl. Orient., se
´
r. 1, 8: 51 (1849).
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 71
Type: Crete, ‘‘Montagnes de Lassiti, 5000’ u
¨
. m.,
20.05.1846’’ Heldreich-1667 (G-BOIS!).
Dubia:
– Viola pentelica Vierh., Verh. Zool. -Bot. Ges.
Wien 64: 266 (1914), described from Greece
and often included in ssp. dehnhardtii
(Valentine et al. 1968, Raus 1986, Tiniakou
1991), may in fact be transitional between
ssp. alba and ssp. dehnhardtii.
This paper is based on parts of a M.Sc. Thesis
(Marcussen 1998). Professors I. Nordal and
L. Borgen are acknowledged as supervisors of
the project and for reading and commenting on
this manuscript. Thanks are also due to N. A.
Sivertson, Y. Adigozalov, F. Fedorov and the
other staff at the Norwegian Refugee Council in
Baku, without whose help and contribution field-
work in Azerbaijan would not have been possible;
to Dr. A. Tiniakou and Professor A. Strid for
translations from Greek; to A.-C. Scheen for
general comments; to the collectors of additional
plant material; and to the curators of the herbaria
of G, Jordant, K, LY and LYJB. The project has
received economical support through the Norwe-
gian Research Council (project 135144/410 ‘‘Hy-
bridisation and evolutionary patterns in Viola sect.
Viola’’), the Nansen Foundation of the Norwegian
Academy of Science (projects 215/1994 ‘‘Studies of
evolution and population biology in the genus
Viola’’ and 50/2001 ‘‘Evolution and geographical
structure within Viola secti on Viola’’), and the
Rygh Legacy.
References
Beattie A. J. (1974) Floral evolution Viola. Ann.
Missouri Bot. Gard. 61: 781–793.
Becker W. (1910a) Die Violen der Schweiz. Neue
Denkschr. Naturforsch. Ges. 65: 1–82.
Becker W. (1910b) Violae europaeae. Dresden-N,
Verlag von C. Heinrich.
Becker W. (1918) Violae asiaticae et australenses
III. Beihefte Bot. Centralbl. 36: 15–59.
Bennett K. D., Tzedakis P. C., Willis K. J. (1991)
Quartenary refugia of north European trees.
J. Biogeogr. 18: 103–115.
von Besser W. S. J. G. (1809) Viola alba . In:
Primitiae florae Galiciae austriacae utriusque,
Vol. 1. Vienna, pp. 171.
Blaxland K. (1996) Violet germination. Bull.
Alpine Gard. Soc. 64: 323–325.
de Bolo
`
s O., Vigo J. (1990) Viola L. In: Flora dels
Paı
¨
sos Catalans, Vol. 2. Barcelona, Editorial
Barcino, pp. 230–245.
Brewer S., Cheddadi R., de Beaulieu J. L., Rei lle
M. (2002) The spread of deciduous Quercus
throughout Europe since the last glacial period.
Forest Ecol. Manag. 156: 27–48.
Chodat L. (1924) Contributions a
`
la ge
´
o-botanique
de Majorque. Thesis 734, Geneva.
Coode M. J. E., Cullen J. (1965) Subgen. Viola. In:
Davis P. H. (ed.) Flora of Turkey and the East
Aegean islands, Vol. 1. Edinburgh, University
Press, pp. 525–529.
Culley T. M., Wolfe A. D. (2001) Population
genetic structure of the cleistogamous plant
species Viola pubescens Aiton (Violaceae), as
indicated by allozyme and ISSRs molecular
markers. Heredity 86: 545–556.
Davis P. H., Mill R. R., Tan K. (1988) Flora of
Turkey and the East Aegean islands, Vol. 10.
(Supplement). Edinburgh, University Press.
Delipavlov D. (1979) Viola L. In: Velcev V.,
Jordanov D., Kuzmanov B. (eds.) Flora na
narodna republika Bulgaria/Flora Reipublicae
popularis Bulgaricae, Vol. 2. Sofija Izdatelstvo
na Bulgarska akademkija na naukite, pp. 339–
395.
Demesure B., Comps B., Petit R. J. (1996) Chlo-
roplast DNA phylogeography of the common
beech (Fagus sylvatica L.) in Europe. Evolution
50: 2515–2520.
Denk T., Grimm G., Sto
¨
gerer K., Langer M.,
Hemleben V. (2002) The evolutionary history of
Fagus in western Eurasia: Evidence from genes,
morphology and the fossil record. Plant Syst.
Evol. 232: 213–236.
Dinc¸ M., Yıldırımlı S. (2002) A new species of
Viola (Violaceae) from Turkey. Bot. J. Linn.
Soc. 138: 483–487.
Espeut M. (1999) Approche du genre Viola dans le
Midi mediterrane
´
en franc¸ ais. Monde Plant. 94:
15–38.
Fukarek F., Hu
¨
bel H., Ko
¨
nig P., Mu
¨
ller G. K.,
Schuster R., Succow M. (1995) Die grosse
farbige Enzyklopa
¨
die Urania-Pflanzenreich,
Vol. 5. Vegetation. Leipzig, Urania-Verlag.
Gams H. (1925) Viola L. In: Hegi G. (ed.)
Illustrierte Flora von Mittel-Europa, Vol. 5(1).
Mu
¨
nchen, J. F. Lehmanns Verlag, pp. 586–657.
72 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
Gerstlauer L. (1943) Vorschla
¨
ge zur Systematik der
einheimischen Veilchen. Ber. Bayer. Bot. Ges.
26: 12–55.
Gower J. C, (1996) Some distance properties of
latent root and vector methods used in multi-
variate analysis. Biometrica 53: 325–338.
Grint¸ escu G., Gus¸ uleac M., Nya
´
ra
´
dy E. I. (1955)
Viola L. In: Sa
ˇ
vulescu, T. (ed.) Flora Republicii
Populare Romıˆ ne, Vol. 3. Editura Academiei
Republicii Populare Romıˆ ne, pp. 553–625.
von Hayek A. (1927) Viola L. In: Prodromus florae
peninsulae Balcanicae. Dahlem, pp. 500–517.
Hess H. E., Landolt E., Hirtzel R. (1970) Viola L.
In: Flora der Schweiz und angrenzender Gebie-
te, Vol. 2. Basel, Birkha
¨
user Verlag, pp. 729–
750.
Hewitt G. M. (1999) Post-glacial re-colonization
of European biota. Biol. J. Linn. Soc. 68: 87–
112.
Hewitt G. M. (2001) Speciation, hybrid zones and
phylogeography – or seeing genes in space and
time. Molec. Ecol. 10: 537–549.
Jordan A. (1849) Observations sur plusieurs plantes
nouvelles rares ou critiques de la France, Vol. 7.
Paris, J.-B. Ballie
`
re.
Kim K. S., Sun B. Y., Whang S. S., Chung G. H.
(1991) Biosystematic study on the genus Viola in
Korea – Comparative study of the Viola albida
complex. Korean J. Bot. 34: 229–238.
King R. A., Ferris C. (1998) Chloroplast DNA
phylogeography of Alnus glutinosa (L.) Gaertn.
Molec. Ecol. 7: 1157–1161.
Livaniou-Tiniakou A. (1991) Viosystimatiki meleti
tou genous Viola section Viola (Violaceae) stin
Ellada [In Greek, with English summary]. Ph. D.
thesis, University of Patras, Patras.
Marcussen T. (1998) Viola L. subsection Viola:
Variation, hyb ridization, and relationships
among European taxa. Master thesis, University
of Oslo, Oslo.
Marcussen T., Borgen L. (2000) Allozymic varia-
tion and relationships within Viola subsection
Viola (Violaceae). Plant Syst. Evol. 223: 29–57.
Marcussen T., Borgen L., Nordal I. (2001) Viola
hirta (Violaceae) and its relatives in Norway.
Nord. J. Bot. 21: 5–17.
Marcussen T., Nordal I. (1998) Viola suavis, a new
species in the Nordic flora, with analyses of the
relation to other species in the subsection Viola
(Violaceae). Nord. J. Bot. 18: 221–237.
Meikle R. D. (1977) Flora of Cyprus. Kew, The
Bentham-Moxon Trust, Royal Botanic Gardens.
Meikle R. D. (1980) Violaceae. In: Townsend C. C.,
Guest E. (eds.) Flora of Iraq, Vol. 4. Cornaceae to
Resedaceae, Baghdad, Ministry of Agriculture
and Agrarian Reform, pp. 172–181.
Merxmu
¨
ller H. (1982) Viola L. In: Pignatti S. (ed.)
Flora d’Italia, Vol. 2. Bologna, Edagricole,
pp. 102–118.
Morden C. W., Doebley J., Schertz K. (1987) A
manual of techniques for startch gel electropho-
resis of Sorghum isozymes, Texas, The Texas
Agricultural Experimental Station, The Texas A
and M University System.
Mouterde S. J. P. (1970) Viola L. In: Nouvelle flore
du Liban et de la Syrie, Vol. 2. Beyruth, Dar
El-Machreq E
´
diteurs, pp. 531–535.
Mun
˜
oz Garmendia F., Montserrat P., Laı
´
nz M.,
Aldasoro J. J. (1993) Viola L. sect. Viola.
In: Castroviejo, S., Aedo, C., Cirujano, S., Laı
´
nz,
M., Montserrat, P., Morales, R., Mun
˜
oz Gar-
mendia, F., Feliner N., Rico E., Talavera S.,
Villar L. (eds.) Flora iberica, Vol. 3. Madrid, Real
Jardı
´
n Bota
´
nico, C.S.I.C., pp. 276–298.
Murr J. (1924) Neue U
¨
bersicht u
¨
ber die Farn- und
Blu
¨
tenpflanzen von Vorarlberg und Liechten-
stein. Bregenz, 1923–1926.
Mus M., Sa
´
ez L., Rossello
´
J. A. (2000) Viola barceloi
Nyman, V. barceloi L. Chodat and V. suavis var.
barceloi O. Bolo
´
s & Vigo: too many eponyms too
many taxa? Taxon 49: 785–788.
Nordal I., Jonsell B. (1998) A phylogeographic
analysis of Viola rupestris: Three postglacial
immigration routes into the Nordic area? Bot.
J. Linn. Soc. 128: 105–122.
Okamoto M., Okada H., Ueda K. (1993)
Morphology and chromo some number of Viola
pilosa, and its systematic position. Taxon 42:
781–787.
Pielou E. C. (1969) An introduction to mathemat-
ical ecology. New York, Wiley-Interscience.
Polunin O. (1980) Flowers of Greece and the
Balkans: a field guide. Oxford, Oxford Univer-
sity Press.
Polunin O., Walters M. (1985) A guide to the
vegetation of Britain and Europe. Oxford , Ox-
ford University Press.
Post G. E. (1932) Viola L. In: Flora of Syria,
Palestine and Sinai, Vol. 1. Beirut, American
Press, pp. 149–152.
Raus T. (1986) Viola L. In: Strid A. (ed.) Mountain
Flora of Greece, Vol. 1. Cambridge, Cambridge
University Press, pp. 608–640.
T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba 73
Rechinger K. H. (1961) Die Flora von Euboea.
Bot. Jahrb. 80: 294–382.
Rohlf F. J. (1997) NTSYS-pc: Numerical taxono-
my and multivariate analys system, version 2.0
Setauket, New York, Exeter Software.
Ruprecht F. J. (1869) Me
´
moires de l’acade
´
mie
impe
´
riale de St. Pe
´
tersbourg, 7. se
´
rie, Vol. 12.
Flora Caucasii. St. Petersburg.
Schmidt A. (1961) Zytotaxonomische Untersuch-
ungen an europa
¨
ischen Viola-Arten der Sektion
Nomimium.O
¨
sterr. Bot. Zeitschr. 108: 20–88.
Scho
¨
fer G. (1954) Untersuchungen u
¨
ber die Poly-
morphie einheimischer Veilchen. Planta 43: 537–
565.
Sterner K. R. (1938) Flora der Insel O
¨
land. Acta
Phytogeogr. Suec. 9: 121.
Tchourina O. (1909) Note sur le Viola Jaubertiana
Mare
`
s [sic!]. Bull. Soc. Bot. Gene
´
ve Ser. 21: 204–
207.
Tiniakou A. (1991) On some species of Viola sect.
Viola from Greece. Bot. Chron. 10: 661–665.
Tiniakou A. (1992) Reports (56–60). In: Kamari
G., Felber F., Garbari F. (eds.) Mediterranean
chromosome number reports, Vol. 2. pp. 232–
235.
Tiniakou A. (2000) Distribution of the genus Viola
(Violaceae) in Peloponnisos. Bot. Chron. 13:
357–368.
Valentine D. H., Marxmu
¨
ller H., Schmidt A.
(1968) Viola L. In: Tutin T. G., Heywood V.
H., Burges N. A., Moore D. M., Valentine
D. H., Walters S. M., Webb D. A. (eds.) Flora
europaea, Vol. 2. Cambridge, University Press,
pp. 270–282.
Walters S. M. (1946) Observations on varieties of
Viola odorat a L. Bot. Exchange Club 12: 834–
839.
Wendel J. F., Weeden N. F. (1989) Chapter 1.
Visualization and interpretation of plant iso-
zymes. In: Soltis D. E., Soltis P. S. (eds.)
Advances in plant sciences series, Vol. 4.
Isozymes in plant biology. Portland, Oregon,
Dioscorides Press, pp. 6–45.
Werth C. R. (1989) The use of isozyme data for
inferring ancestry of polyploid pteridophy tes.
Biochem. Syst. Ecol. 17: 117–130.
Address of the author: Cand. Scient. Thomas
Marcussen, University of Oslo, Department of
Biology, Division of Botany and Plant Physiology.
P. O. Box 1045, Blindern, N-0316 Oslo, Norway
(e-mail: thomas.marcussen@bio.uio.no).
74 T. Marcussen: Evol ution, phylogeography, and taxonomy within Viola alba
Plant
Syst,
Evol . 239:
169
(2003)
DOI
1
0,
I 007/s00606-003-0039-5
Erratum
In the
paper "Evolution,
phylogeography,
and
taxonomy
within
the Viola alba
complex
(Violaceae)"
by
T. Marcussen
fPlant
Syst.
Evol .
237:
5I*7
4
(2003)l
the following
correc-
tions
have to be
made:
Page 71,
left
column,
lines
9*15
(under
o'V.
scotophylla"),
should
read:
Type:
"flore
albo,
[France:]
Lyon
frequens
primovere'
'
Jordan
(Lectotype
-
selected
here
*
LYJB-Sagot!);
"fl,ore
qlbo,
6p6ron ordinaire-
ment violace
fnote
added
later], Lyon å
Fran-
cheville
etc," Jordan
(Paratype
LYJB-sagotl);
"flore violaceo, Lyon
åL
Francheville"
Jordan
(Paratype
LYJB-Sagot!);
"
Lyon,
Mai.
I849"
Jordan
(Paratype
K),
Page
7I, left column,
lines
29*31
(under
"V.
thessala"),
should
read:
Type:
Greece,
"
in
monte
Oeta
Thessaliae,
1837" Spruner
(Lectotype
selected
here
G-BOIS!),
Plnnf
$ystematies
ftmd
Evtlution
Prirr t*d
irr A urstritr
Page 7I,
left column,
lines
4l-42
(under
"
V, armena"),
should
read:
Type:
Turkey,
"Erzeroum,
Jun,
1853"
Huet du
Pavillon
(Holotype
G-BOIS!).
Page
JI, right
column,
lines
6-10
(under
"V,
virescens"),
should
read:
Type:
France,
"Tassin
(Rhone):
bois, haies.
28.
mars, 30.
avril"
Boullu
704
(Lectotype
selected
here
Herb. Jordan!,
deposited
in
ISARA,
in
Facultå Catholique
de
Lyon;
Isotypes
LY-Bonaparte!).
Page 7I, right
column,
lines 26-27
(under
"V,
dehnhardtii"),
should
read:
Type:
Italy, Naples,
"Cqmaldoli"
Tenore
(Lectotype
-
selected
here
*
NAP!).
t
