Phylogeny of the genus Lotus (Leguminosae,
Loteae): evidence from nrITS sequences and
G.V. Degtjareva, T.E. Kramina, D.D. Sokoloff, T.H. Samigullin, C.M. Valiejo-Roman,
and A.S. Antonov
Abstract: Lotus (120–130 species) is the largest genus of the tribe Loteae. The taxonomy of Lotus is complicated, and a
comprehensive taxonomic revision of the genus is needed. We have conducted phylogenetic analyses of Lotus based on
nrITS data alone and combined with data on 46 morphological characters. Eighty-one ingroup nrITS accessions represent-
ing 71 Lotus species are studied; among them 47 accessions representing 40 species are new. Representatives of all other
genera of the tribe Loteae are included in the outgroup (for three genera, nrITS sequences are published for the first time).
Forty-two of 71 ingroup species were not included in previous morphological phylogenetic studies. The most important
conclusions of the present study are (1) addition of morphological data to the nrITS matrix produces a better resolved phy-
logeny of Lotus; (2) previous findings that Dorycnium and Tetragonolobus cannot be separated from Lotus at the generic
level are well supported; (3) Lotus creticus should be placed in section Pedrosia rather than in section Lotea; (4) a broad
treatment of section Ononidium is unnatural and the section should possibly not be recognized at all; (5) section Heineke-
nia is paraphyletic; (6) section Lotus should include Lotus conimbricensis; then the section is monophyletic; (7) a basic
chromosome number of x= 6 is an important synapomorphy for the expanded section Lotus; (8) the segregation of
Lotus schimperi and allies into section Chamaelotus is well supported; (9) there is an apparent functional correlation be-
tween stylodium and keel evolution in Lotus.
Key words: Leguminosae, Loteae, Lotus, nuclear ribosomal ITS sequences, morphology.
´:Le genre Lotus (120–130 espe
`ces) est le plus grand de la tribu des Loteae. La taxonomie des Lotus est compli-
´e, et une re
´vision taxonomique comple
`te du genre s’impose. Les auteurs ont conduit des analyses phyloge
Lotus, sur la base des donne
´es nrITS isole
´ment et combine
´es avec les donne
´es sur 46 caracte
`res morphologiques. Les au-
teurs ont e
´81 accessions nrITS d’un groupe interne repre
´sentant 71 espe
`ces de Lotus; parmi celle-ci, 47 accessions
´sentant 40 espe
`ces sont nouvelles. On retrouve des repre
´sentants de tous les autres genres de la tribu Loteae dans le
groupe externe (pour trois de ces genres, on publie les se
´quences nrITS pour la premie
`re fois). Des 71 espe
`ces du groupe
interne, 42 n’ont pas e
´incluses dans des e
´dentes. Les plus importantes conclusions
de cette e
´tude sont: (1) l’addition de donne
´es morphologiques a
`la matrice nrITS conduit a
`une meilleure re
´tique des Lotus; (2) on confirme les constats ante
`l’effet que les Dorycnium et Tetragonolobus ne peuvent
´s des Lotus au niveau du genre; (3) le L. creticus devrait e
´dans la section Pedrosia, pluto
ˆt que la
section Lotea; (4) le traitement ge
´ral de la section Ononidium n’est pas naturel et la section devrait possiblement ne pas
ˆtre reconnue du tout; (5) la section Heinekenia est paraphyle
´tique; (6) la section Lotus doit inclure le L. conimbricensis;
la section devient alors monophyle
´tique; (7) le nombre de base de chromosomes x= 6 est une importante synapomorphie
pour la section Lotus e
´tendue; (8) la se
´gation du L. schimperi et allie
´s dans la section Chamaelotus est bien supporte
(9) il y a une apparente corre
´lation fonctionnelle entre l’e
´volution du stylodium et de la care
`ne chez les Lotus.
´s:Leguminosae, Lotae, Lotus,se
´quences de l’ITS nucle
´ique ribosomal, morphologie.
[Traduit par la Re
There is little agreement in the literature regarding ge-
neric limits of Lotus (e.g., Greene 1890; Taubert 1894;
Brand 1898; Ottley 1944; Callen 1959; Gillett 1959; Hutch-
inson 1964; Polhill 1981, 1994; Isely 1981; Lassen 1986;
Kirkbride 1994, 1999; Kramina and Sokoloff 1997, 2001;
Talavera and Salgueiro 1999; Sokoloff 1999, 2000, 2003a,
2003b). The (lecto) type species, Lotus corniculatus, as well
as its closest relatives are native to the Old World. Many
species are confined to or common within the Mediterranean
Region. There are several Old World taxa that are either in-
cluded in Lotus or accepted as distinct genera by various
Received 30 August 2005. Published on the NRC Research
Press Web site at http://canjbot.nrc.ca on 30 June 2006.
G.V. Degtjareva. Faculty of Bioengineering and Bioinformatics,
Moscow State University, Moscow 119992, Russia.
T.E. Kramina, D.D. Sokoloff,1and A.S. Antonov. Higher
Plants Department, Biological Faculty, Moscow State
University, Moscow 119992, Russia.
T.H. Samigullin and C.M. Valiejo-Roman. Department of
Evolutionary Biochemistry, A.N. Belozersky Institute, Moscow
State University, Moscow 119992, Russia.
1Corresponding author (e-mail: sokoloff-V@yandex.ru).
Can. J. Bot. 84: 813–830 (2006) doi:10.1139/B06-035 #2006 NRC Canada
taxonomic authorities. Among them, the mostly Mediterra-
nean (also in other parts of Europe and western Asia) Dor-
ycnium Mill. (8–10 species) and Tetragonolobus Scop. (5–6
species) are most important (Rikli 1901; Dominguez and
Galiano 1979). Other problematic Old World genera vari-
ously included or excluded from Lotus are Podolotus Royle
(1 species found in India, Pakistan, Afghanistan, Iran, and
Oman; Rechinger 1984), Pseudolotus Rech.f. (1 species
found in Pakistan, Iran, and Oman; Rechinger 1984; Ali
and Sokoloff 2001), Kebirita Kramina & Sokoloff (1 species
in the Sahara, northwestern Africa; Kramina and Sokoloff
2001), and Benedictella Maire (1 species in Morocco; Maire
In the New World, species related to Lotus are most di-
verse in California. Recent studies based on nrITS sequences
(Allan and Porter 2000; Allan et al. 2003) and morphology
(Arambarri 2000a; Arambarri et al. 2005; Sokoloff 2006)
clearly show that New World species are not closely related
to Old World Lotus. According to nrITS data, Old World
Lotus is closer to the Old World genera Hammatolobium
and Tripodion than to New World Loteae (Allan et al.
2003; Degtjareva et al. 2003). Thus all New World species
should be excluded from the genus Lotus; in our opinion
(Sokoloff 1999, 2000; Sokoloff and Lock 2005), they form
four different genera (Hosackia Douglas ex Benth., Ottleya
D.D. Sokoloff, Acmispon Raf., and Syrmatium Vogel).
Phylogenetic studies of the tribe Loteae based on nrITS
sequences and morphology show a clade containing Doryc-
nium,Tetragonolobus, and Old World species of Lotus
studied so far (Allan and Porter 2000; Allan et al. 2003,
2004; Sokoloff 2003b, 2006). All analyses clearly show that
Tetragonolobus is derived from within Old World Lotus
(Allan and Porter 2000; Arambarri 2000b; Allan et al.
2003; Sokoloff 2006). It is logical to include Tetragonolobus
within Lotus. In the molecular phylogenetic study by Allan
et al. (2003), the four Dorycnium species analyzed did not
form a clade. In the morphological cladistic study of
Arambarri (2000b), Dorycnium is nested in the Old World
Lotus clade as a close relative of Lotus corniculatus and its
allies. Since morphological grounds for separation of Doryc-
nium from Lotus are equivocal, Sokoloff (2003a) has sug-
gested following Polhill (1981) in placing all Dorycnium
species in Lotus.
Of four monospecific and problematic Old World genera,
nrITS data have only been published for Kebirita (Allan et
al. 2003). Molecular and morphological data clearly show
that Kebirita is distinct from Old World Lotus and deserves
generic rank (Sokoloff 2006). Cladistic analyses based on
morphological characters suggest that Benedictella should
be included within Lotus (Sokoloff 2003b), but generic rank
is supported for Podolotus and Pseudolotus (Sokoloff 2006).
Although recent phylogenetic data have provided a much
better understanding of generic limits and relationships of
Lotus, the sectional classification of the genus remains prob-
lematic. Different authors accept very different classification
systems for Lotus species (e.g., Fig. 1). Only a few authors
discuss all species worldwide while many sectional systems
are introduced in regional Floras. Recent phylogenetic
(Allan and Porter 2000; Arambarri 2000b; Allan et al. 2003,
2004) and phenetic (Stenglein et al. 2004) studies clarified
some problems; however, many problematic species and
some sections were not included in these analyses. Phyloge-
netic trees based on morphology (Arambarri 2000b) and
nrITS data (Allan et al. 2003, 2004) differ significantly in
topology, but they also differ considerably in species sam-
The objectives of this paper are (1) to increase taxon sam-
pling in nrITS phylogenetic analyses of Loteae and (2) to
conduct, for the first time, a combined phylogenetic analysis
of Lotus based on morphological and nrITS data for the
same set of species. Our study should help to clarify sec-
tional limits in the genus Lotus and their phylogenetic rela-
Material and methods
Complete sequences of ITS1 and ITS2 were generated for
51 accessions representing 44 species of the genus Lotus and
related genera. In addition, GenBank data on the ITS region
in 49 taxa of Loteae are used (Table 1). In total, 81 ingroup
nrITS accessions representing 71 Lotus species were studied
(i.e., more than half of the total number of Lotus species,
which is estimated as 120–130). The taxon sampling covers
all sections of Lotus. However, we were able to produce
only ITS1 sequence of the rare endemic L. benoistii (Maire)
Lassen from Morocco (monospecific section Benedictella).
This sequence was not included in the main analyses. Ex-
cept for Lotus and Hammatolobium, each genus of the tribe
Loteae is represented by one species in the present study.
Members of Robinieae (Robinia) and Sesbanieae (Sesbania)
are used as outgroups because higher level molecular phylo-
genetic studies of legumes strongly support a close relation-
ship of these two tribes to the Loteae (e.g., Wojciechowski
et al. 2000; Lewis et al. 2005). In the Results and Discussion
sections, the taxonomy of Kramina and Sokoloff (2003) and
Sokoloff (2003a, 2003b) is used (see Table 2 for details) be-
cause it is the only recent system of Lotus that assigns each
species worldwide to a particular section.
DNA was isolated from leaf tissue using the CTAB
method of Doyle and Doyle (1987). PCR reactions were per-
formed with universal primers (White et al. 1990). Both
spacer regions were sequenced in their entirety for both
strands. The sequences obtained were aligned manually us-
ing the SED editor of the VOSTORG package (Zharkikh et
A morphological data matrix was produced for the same
set of species (Appendix A and supplementary data.2). A to-
tal of 46 characters were obtained mostly from original mor-
phological observations. Literature data on chromosome
numbers were also used (Grant 1965, 1995; Fedorov 1969;
Goldblatt and Johnson 1996, 1998). Three multistate mor-
phological characters were coded as additive while others
were binary or multistate nonadditive. The following charac-
ters were coded as additive: flower number per partial in-
florescence (18), flower size (22), and basic chromosome
2Supplementary data for this article are available on the journal Web site (http://canjbot.nrc.ca) or may be purchased from the Depository of
Unpublished Data, Document Delivery, CISTI, National Research Council Canada, Building M-55, 1200 Montreal Road, Ottawa, ON
K1A 0R6, Canada. DUD 5039. For more information on obtaining material refer to http://cisti-icist.nrc-cnrc.gc.ca/i rm/unpub_e.shtml.
814 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada
number (46). We decided to use the additive coding because
of the nature of the characters was such that some character
states are intermediate between the others. For example,
there are reasons to hypothesize that evolutionary transitions
between basic chromosome numbers x= 8 and x= 6 were
most likely performed through the intermediate number x=
7 (see also Grant 1991). It is reasonable to suppose that evo-
lutionary transitions between large and small flowers oc-
curred via mid-size flowers. All characters were not a priori
polarized in our analyses. Maximum parsimony and Baye-
sian analyses were performed for a combined molecular–
morphological data set as well as for the molecular data
alone. No phylogenetic analysis of morphological data alone
was performed because there are insufficient characters to
produce a resolved phylogeny. Some characters, such as
those of pollen morphology (Crompton and Grant 1993;
´ez and Ferguson 1994) are relatively uniform among Old
World Lotus species and offer little phylogenetic informa-
tion at this level. Seed morphology (Arambarri 1999) and
leaf epidermal microcharacters (Stenglein et al. 2004) offer
significant and useful characters, but many species included
in the present analyses have not yet been studied for these
Bayesian inference of phylogeny was explored using the
MrBayes program (version 3.1; Ronquist and Huelsenbeck
2003). The evolutionary model implemented in MrBayes
for morphological data are analogous to a Jukes–Cantor
model with a variable number of states. For the analyses of
molecular data, the GTR+I+Gmodel of nucleotide substitu-
tions was selected by the Akaike Information Criterion in
Modeltest (Posada and Crandall 1998). A total of 3 000 000
generations were performed and trees from first 2 200 000
generations were discarded. The number of generations to
be discarded was determined using a convergence diagnos-
tic. Parsimony analysis involved a heuristic search con-
ducted with PAUP* (version 4.0b8; Swofford 2000) using
tree bisection–reconnection (TBR) branch swapping with
character states specified as equally weighted. One hundred
replicates with random addition of sequences were per-
formed and all shortest trees were saved. Bootstrap (Felsen-
stein 1985) analysis was performed to assess the degree of
support for particular branches on the tree. Bootstrap values
were calculated from 100 replicate analyses with TBR
branch swapping and random addition sequence of taxa.
One thousand most parsimonious trees from each replicate
were saved. In the parsimony analyses all gaps were treated
as missing data.
Analyses of nrITS sequences (Fig. 2)
The length of the ITS region (ITS1, 5.8S, and ITS2)
ranged from 587 to 617 bp for the 99 accessions of the in-
group and two outgroup taxa studied. The length of the ITS1
region varied from 210 to 239 bp and the ITS2 region from
194 to 229 bp. The 5.8S gene was 163–164 bp in length.
The alignment of 101 ITS sequences resulted in matrix of
646 nucleotide positions after excluding 332 ambiguous po-
sitions. A total of 261 characters were parsimony-informa-
tive, 299 characters were constant, and 86 variable
characters were parsimony-uninformative. Our study re-
vealed a length polymorphism of the ITS1 spacer for two
species, a 4 bp duplication in Lotus cytisoides and a 1 bp
duplication in Lotus preslii.
In the maximum parsimony analysis, 20 004 shortest trees
(1461 steps) were found, with a consistency index of 0.411
Fig. 1. Historical changes in sectional classification of Old World Lotus.
Degtjareva et al. 815
#2006 NRC Canada
Table 1. GenBank accession numbers and sources of nrITS sequences used in this paper.
Species GenBank No. First publication of the sequence or voucher data
Acmispon americanus (Nutt.) Rydb. [=Lotus unifoliolatus
AF450183 Allan et al. (2003)
Anthyllis onobrychioides Cav. AF450210 Allan et al. (2003)
Antopetitia abyssinica A. Rich. DQ166212 This paper; Auquier 2598 (BE)
Coronilla viminalis Salisb. DQ166213 This paper; Morocco, Podlech 53755 (M)
Cytisopsis pseudocytisus (Boiss.) Fertig AY325282 Degtjareva et. al. (2003)
Dorycnopsis abyssinica (A. Rich.) V.N. Tikhom. &
AF450235 Allan et al. (2003)
Hammatolobium lotoides Fenzl AY325279 Degtjareva et. al. (2003)
Hippocrepis emerus (L.) Lassen AF218531 Allan and Porter (2000)
Hosackia crassifolia Benth. [=Lotus crassifolius (Benth.)
AF218523 Allan and Porter (2000)
Kebirita roudairei (Bonnet) Kramina & D.D. Sokoloff
(=Lotus roudairei Bonnet)
AF450200 Allan et al. (2003)
Ornithopus micranthus (Benth.) Arechav. AY325277 Degtjareva et. al. (2003)
Ottleya oroboides (Kunth) D.D. Sokoloff [=Lotus oro-
boides (Kunth) Ottley]
AF218510 Allan and Porter (2000)
Podolotus hosackioides Benth. DQ166214 This paper; Afghanistan, 13 Apr. 1967, Freitag s.n.
Pseudolotus villosus (Blatter & Hallb.) Ali &
DQ166215 This paper; Oman, Redcliffe-Smith 3901 (K)
Robinia pseudoacacia L. AF218538 Allan and Porter (2000)
Scorpiurus vermiculatus L. AF218536 Allan and Porter (2000)
Sesbania vesicaria (Jacq.) Elliott AF398761 Lavin et al. (2001)
Syrmatium glabrum Vogel [=Lotus scoparius (Nutt.) Ott-
AF218521 Allan and Porter (2000)
Tripodion tetraphyllum (L.) Fourr. AF218498 Allan and Porter (2000)
Lotus sect. Benedictella (Maire) Kramina & D.D. Sokoloff (1/1)
L. benoisstii (Maire) Lassen DQ372916 This paper; Morocco, 31 Mar. 1934, Maire & Wilczek
Lotus sect. Bonjeanea (Rchb.) D.D. Sokoloff (3/3)
L. hirsutus L. [=Dorycnium hirsutum (L.) Ser.] AY294292 Allan and Porter (2000)
L. rectus L. [=Dorycnium rectum (L.) Ser.] AF218503 Allan and Porter (2000)
L. strictus Fisch. & C.A. Mey. DQ160286 This paper; Asiatic Russia, 18 Sep. 2003, Korolyuk s.n.
Lotus sect. Canaria (Rikli) D.D. Sokoloff (3/1)
L. broussonetii Choisy ex Ser. [=Dorycnium broussonetii
(Choisy ex Ser.) Webb et Berth.]
DQ160278 This paper; plant cultivated at Royal Botanic Gardens,
Kew, introduced from Canary Is., Chase 16057 (K)
Lotus sect. Chamaelotus Kramina & D.D. Sokoloff (3/2)
L. glinoides Del. (1) DQ160282 This paper; Spain, Canary Is., Nydegger 26086 (MHA)
L. glinoides Del. (2) DQ166220 This paper; Egypt, 7 May 1962, Bochantsev s.n. (LE)
L. schimperi Steud. ex Boiss. DQ166218 This paper; Oman, McLeish 3458 (E)
Lotus sect. Dorycnium (Mill.) D.D. Sokoloff (5/2)
L. dorycnium L. s.l. [=Dorycnium herbaceum Vill.] AF218501 Allan and Porter (2000)
L. graecus L. [=Dorycnium graecum (L.) Ser.] AF218500 Allan and Porter (2000)
Lotus sect. Erythrolotus Brand (1/1)
L. conimbricensis Brot. AF450186 Allan et al. (2003)
Lotus sect. Heinekenia Webb & Berth. (23/14)
Lotus arabicus group
L. arabicus L. AF450176 Allan et al. (2003)
L. lalambensis Schweinf. DQ166216 This paper; Saudi Arabia, Collenette 7908 (E)
L. lanuginosus Vent. DQ166221 This paper; Jordan, Townsend 65/22 (LE)
L. laricus Rech.f., Aellen & Esfand. DQ166233 This paper; Abu Dhabi, Western 275 (E)
L. quinatus (Forssk.) J.B. Gillett DQ166217 This paper; Yemen, Thulin et al. 9374 (E)
Lotus australis group
L. australis Andrews AF450187 Allan et al. (2003)
816 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada
Table 1 (continued).
Species GenBank No. First publication of the sequence or voucher data
L. cruentus Court AF450182 Allan et al. (2003)
Lotus discolor group
L. discolor E. Mey. DQ160288 This paper, Lisocuski B-3330 (BE)
L. goetzei Harms DQ166235 This paper; Kenya, Gillett 16179 (LE)
L. mlanjeanus J.B. Gillett DQ166232 This paper; Malawi, J.D. & E.G. Chapman 8807 (E)
L. wildii J.B. Gillett DQ160287 This paper; Zimbabwe, Bayliss 10166 (E)
Lotus gebelia group
L. aegaeus (Griseb.) Nym. DQ160276 This paper; Turkey, Khokhryakov & Mazurenko 1135
L. gebelia Vent. AF450188 Allan et al. (2003)
L. michauxianus Ser. AF450206 Allan et al. (2003)
Lotus sect. Krokeria (Moench) Ser. (1/1)
L. edulis L. AF450184 Allan et al. (2003)
Lotus sect. Lotea (Medik.) DC. (10/8)
L. cytisoides L. (A) DQ160280 This paper; Cyprus, Seregin & Sokoloff 280 (MW)
L. cytisoides L. (B) DQ166241 This paper; Cyprus, Seregin & Sokoloff 280 (MW)
L. halophilus Boiss. & Spruner DQ160283 This paper; Greece, Raus 9307 (MHA)
L. longisiliquosus R. Roem. AF218526 Allan & Porter (2000)
L. ornithopodioides L. AF450205 Allan et al. (2003)
L. peregrinus L. AF450177 Allan et al. (2003)
L. polyphyllus Clarke DQ160289 This paper; Egypt, 06 Apr. 1962, Bochantsev s.n. (LE)
L. weilleri Maire AF450180 Allan et al. (2003)
Lotus sect. Lotus (30/19)
Lotus angustissimus group
L. angustissimus L. DQ166243 This paper; Australia, Norfolk Island, introduced, 14
Oct. 1999, Waterhouse 5510 (NSW)
L. castellanus Boiss. & Reut. (1) DQ160272 This paper; Portugal, Malato-Beliz & Guerra 13585
L. castellanus Boiss. & Reut. (2) DQ166223 This paper; Spain, Segura Zubizarreta 15112 (LE)
L. castellanus Boiss. & Reut. (3) DQ166238 This paper; Spain, Segura Zubizarreta 38111 (MHA)
L. cf. castellanus (4) DQ160275 This paper; Turkey, 17 Oct. 1999, Majorov s.n. (MW)
L. parviflorus Desf. (1) DQ166230 This paper; Spain, Segura Zubizarreta 34567 (MHA)
L. parviflorus Desf. (2) AF450194 Allan et al. (2003)
L. praetermissus Kuprian. (1) DQ166227 This paper; European Russia, 20 July 1993, Kramina s.n.
L. praetermissus Kuprian. (2) DQ168370 This paper; Ukraine, Tzvelev et al. 1630 (LE)
L. subbiflorus Lag. (syn. L.suaveolens Pers.) (1) DQ166239 This paper; cultivated at the Botanic Garden of Moscow
University, 1998 Kramina s.n. (MW)
L. subbiflorus Lag. (2) DQ166237 This paper; Australia, Kodela et al. 163 (NSW)
L. subbiflorus Lag. (3) DQ166231 This paper; Italy, Iberite 15222 (MHA)
L. subbiflorus Lag. (4) DQ168369 This paper; France, Dutartre 570 (MHA)
Lotus corniculatus group
L. alpinus (DC.) Schleicher ex Ramond DQ160274 This paper; Spain, Segura Zubizarreta 43694 (MHA)
L. borbasii Ujhelyi DQ166226 This paper; Czech Republic, 14 May 1961, Smejkal 1441
L. corniculatus L. AF218527 Allan and Porter (2000)
L. delortii Timb.-Lagr. ex F.W. Schultz DQ166228 This paper; Spain, Sandwith 4772 (LE)
L. glaber Mill. DQ166225 This paper; Slovakia, 16 July 1974, Chrtek & Kr
L. japonicus (Regel) K. Larsen ‘Gifu’ AJ512882
Nanni et al. (2004)
L. japonicus (Regel) K. Larsen ‘Miyakojima’ AJ512881
Nanni et al. (2004)
Degtjareva et al. 817
#2006 NRC Canada
and a retention index of 0.745. A strict consensus of all
shortest trees is shown in Fig. 2. The Bayesian tree (not
shown) is generally similar to the strict consensus.
The genus Lotus (including Tetragonolobus and Doryc-
nium) is revealed as a clade both in the Bayesian and parsi-
mony analyses. A group containing Hammatolobium,
Tripodion, plus Cytisopsis is well supported as a clade sister
to Lotus. The problematic genera Podolotus and Pseudolotus
do not group with the Lotus clade. In the Bayesian analysis,
Pseudolotus is sister to another monospecific Old World ge-
nus, Antopetitia (tree not shown), while in the parsimony
analysis, the position of Pseudolotus is unresolved. On the
Bayesian tree, Podolotus is poorly supported as sister to a
large clade comprising all New World taxa plus Old World
Dorycnopsis,Antopetitia,Pseudolotus, and Kebirita (tree
not shown). In the strict consensus of shortest trees, Podolo-
tus is sister to Hippocrepis plus Scorpiurus, but this group-
ing has a bootstrap support of less than 50%.
Basally branching nodes within the Lotus clade are poorly
supported in both the Bayesian and parsimony trees. In the
Bayesian tree, as well as in the strict consensus of shortest
trees, members of section Chamaelotus (L. schimperi and
L. glinoides) are sister to the rest of Lotus, but posterior
probability and bootstrap support for this grouping are very
Species of section Lotus fall into two clades. Clade A is
highly supported but relationships are unresolved. Clade A
includes three species of annuals, namely L. parviflorus and
L. subbiflorus of section Lotus plus L. conimbricensis (sect.
Erythrolotus). Clade B comprises the rest of the sampled spe-
cies of section Lotus. Within this clade, members of the L.
corniculatus group form a strongly supported subclade. Rela-
tionships within the L. corniculatus group are well resolved.
The second subclade of clade B contains the perennials
L. uliginosus and L. pedunculatus plus the annuals (biennials)
L. angustissimus,L. praetermissus, and L. castellanus, and a
Table 1 (concluded).
Species GenBank No. First publication of the sequence or voucher data
L. krylovii Schischk. & Serg. AF450209 Allan et al. (2003)
L. palustris Willd. AF450195 Allan et al. (2003)
L. peczoricus Miniaev et Ulle AF450191 Allan et al. (2003)
L. preslii Ten. (A) DQ166229 This paper; Algeria, 22 July 1968, Bochantsev s.n. (LE)
L. preslii Ten. (B) DQ166236 This paper; Algeria, 22 July 1968, Bochantsev s.n. (LE)
L. schoeleri Schweinf. DQ166224 This paper; cultivated at the Botanic Garden of Moscow
University, 16 Sep. 1994 Kramina s.n. (MW)
L. stepposus Kramina DQ166242 This paper; Ukraine, 28 June 1989, Kramina 14-4 (MW)
Lotus pedunculatus group
L. pedunculatus Cav. DQ166222 This paper; Spain, 18 July 1972, Segura Zubizarreta s.n.
L. uliginosus Schkuhr (1) DQ160273 This paper; Denmark, Larsen 29349 (LE)
L. uliginosus Schkuhr (2) AF450197 Allan et al. (2003)
Lotus sect. Ononidium Boiss. (4/3)
L. garcinii DC. DQ166234 This paper; Iran, Leonard 5899 (LE)
L. ononopsis Balf.f. DQ166219 This paper; Yemen, Miller et al. 10097 (E)
L. simonae Maire, Weiller & Wilczek DQ160285 This paper; Morocco, Podlech 49444 (M)
Lotus sect. Pedrosia (Lowe) Christ (29/11)
L. arenarius Brot. AF218528 Allan and Porter (2000)
L. assakensis Brand DQ160277 This paper, Morocco, Podlech 40448 (M)
L. azoricus P.W. Ball AY294293 Allan et al. (2004)
L. campylocladus Webb & Berth. AF450196 Allan et al. (2003)
L. creticus L. DQ160279 This paper; Portugal, June 2001, Severova s.n. (MW)
L. emeroides R.P. Murray AY294295 Allan et al. (2004)
L. eriosolen (Maire) Mader & Podlech DQ160281 This paper; Morocco, Podlech 52619 (M)
L. jacobaeus L. AY294299 Allan et al. (2004)
L. jolyi Battand. DQ166240 This paper; Morocco, Lewalle 11581 (LE)
L. lancerottensis Webb & Berth. AY294300 Allan et al. (2004)
L. maroccanus Ball AF450181 Allan et al. (2003)
L. pseudocreticus Maire, Weiller & Wilczek DQ160284 This paper; Morocco, Podlech 52358 (M)
Lotus sect. Rhyncholotus (Monod) D.D. Sokoloff (3/2)
L. berthelotii Masf. AY294306 Allan et al. (2004)
L. maculatus Breitf. AY294308 Allan et al. (2004)
Lotus sect. Tetragonolobus (Scop.) Benth. & Hook.f. (5/2)
L. maritimus L. [=Tetragonolobus maritimus (L.) Roth.] AF218505 Allan and Porter (2000)
L. tetragonolobus L. (=Teteragonolobus purpureus
AF218506 Allan and Porter (2000)
Note: Sections of Lotus are indicated. Numbers after sectional names show total number of species in a section / number of species studied here.
818 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada
putative new taxon labeled ‘‘L. cf. castellanus’’. The sister
group relationship between clade A and clade B is not sup-
ported by analyses of nrITS sequences, which is unexpected
given that these clades contain members of section Lotus,a
group that was traditionally thought to be natural on the basis
of morphological evidence (e.g., Kramina 1999; Valde
2000). In the tree inferred from the Bayesian analysis, clades
A and B group together with species traditionally included in
Dorycnium sections Bonjeanea and (Eu)Dorycnium
(L. rectus, L. hirsutus,L. dorycnium,L. graecus). Lotus stric-
tus which was only recently classified as Dorycnium (Lassen
1986) also falls here. In the parsimony analysis, clades A and
B, former members of Dorycnium plus section Chamaelotus
form an unresolved polytomy at the base of the Lotus clade.
Clade C comprises members of sections Canaria,Heine-
sia, and Rhyncholotus. The only sampled member of the
section Canaria is sister to the rest of this large clade in the
tree inferred from the Bayesian analysis, however, bootstrap
support for this grouping in the parsimony analysis is poor,
and the grouping is also not present in strict consensus. Sec-
tion Heinekenia is not monophyletic according to analyses
of nrITS sequences. Its members fall into two clades (D and
E) forming a grade within clade C. Clade D is composed en-
tirely of members of section Heinekenia. Clade E contains
seven species of section Heinekenia plus two of section
Ononidium (L. garcinii and L. ononopsis). Relationships of
L. garcinii and L. ononopsis within clade E are not resolved;
more data are needed to determine if they are sister taxa.
Clade F includes members of sections Tetragonolobus,
Krokeria, and Lotea plus L. simonae. The two GenBank ac-
cessions of the sect. Tetragonolobus group together with
very low support. A well-supported subclade of clade F in-
cludes members of section Lotea plus the rare endemic
L. simonae from south Morocco, which was originally
placed in section Stipulati (Maire et al. 1935) and subse-
quently transferred to section Ononidium (Sokoloff 2003b).
All sampled members of sections Pedrosia and Rhyncho-
lotus plus a problematic species Lotus creticus (that has
been placed in either sect. Lotea or Pedrosia) form a well-
supported clade (clade G) with posterior probability of 1.00
and bootstrap support of 100%. Section Pedrosia is paraphy-
letic with section Rhyncholotus embedded within it. Lotus
creticus is supported (posterior probability 1.00; bootstrap
support 83%) as a member of a clade that includes members
of section Pedrosia (L. campylocladus,L. lancerottensis,
and L. assakensis).
Sister-group relationship between clades F and G is only
supported in the tree inferred from the Bayesian analysis
(posterior probability 0.85; not shown in Fig. 2). These two
clades form a polytomy with clade E in the strict consensus.
The ITS1 sequence of Lotus benoistii (sect. Benedictella),
according to our preliminary data (tree not shown), groups
with sequences of L. glinoides and L. schimperii, but boot-
strap support of this grouping is low.
Analyses of the combined matrix (nrITS sequences plus
morphology) (Fig. 3)
In the maximum parsimony analysis, 34 000 shortest
trees (1762 steps) were found, with a consistency index of
0.374 and a retention index of 0.720. A strict consensus of
all shortest trees is shown in Fig. 3. The Bayesian tree (not
shown) is generally similar to the strict consensus. In both
analyses, the genus Lotus is a well-supported clade sister to
Tripodion,Hammatolobium, and Cytisopsis.
Only a few well-supported clades in the molecular analy-
ses are unresolved in trees inferred from analyses of the
combined matrix. For example, Lotus maroccanus and
L. eriosolen group in the molecular analyses (bootstrap sup-
port 69%, posterior probability 0.86), but this is not sup-
ported in the combined analyses.
Some clades receiving low support in the molecular anal-
yses are well supported in the combined analyses. For exam-
ple, in the latter, the Tetragonolobus clade has a posterior
probability of 0.91 and a bootstrap support of 78%; the
Rhyncholotus clade has a posterior probability of 1.00 and a
bootstrap support of 100%. These two groups are morpho-
logically well defined by apomorphic character states.
In contrast to the analyses of nrITS data alone, analyses
of the combined data set show clades A and B grouping to-
gether with bootstrap support of 57% and posterior probabil-
ity of 0.97. In the combined analyses, the strict consensus of
shortest trees shows a clade comprising all sampled mem-
bers of sections Dorycnium and Bonjeanea (i.e., former ge-
nus Dorycnium). This clade is sister to clades A +
B. However, the Dorycnium+Bonjeanea clade and its sister
group relationship with clades A + B received very low
bootstrap support and posterior probabilities.
Monophyly of the genus Lotus
The present analyses support the segregation from Lotus
of the Old World monospecific genera Podolotus,Pseudolo-
tus, and Kebirita, as well as the New World genera Ho-
sackia s.str., Ottleya,Acmispon, and Syrmatium. These
genera were previously included in Lotus by various authors
(e.g., Polhill 1981). In the trees obtained in this study, the
genera Hosackia,Ottleya,Acmispon, and Syrmatium are rep-
resented by one species each. We have also performed anal-
yses with more extensive sampling of these American
genera. Each segregate genus is monophyletic in these anal-
yses (data not included).
The present study supports monophyly of the genus Lotus
within the limits suggested by Sokoloff (2003a, 2003b), that
is, including the segregate genera Tetragonolobus and Dor-
ycnium. The current circumscription of the genus Lotus is
restricted only to Old World species. The monophyly of
this group was also revealed in the molecular phylogenetic
studies of Allan and Porter (2000) and Allan et al. (2003,
2004). Previous studies, however, did not include material
for all genera of the tribe. Some problematic taxa within Lo-
tus were also previously not sampled for DNA, for example,
sections Canaria and Ononidium.
Morphological synapomorphies of major clades within
Lotus are summarized in Table 2. Of 30 major clades recog-
nized (Table 2), 14 clades have no obvious morphological
synapomorphies. Six clades have a single synapomorphy
each, four clades two synapomorphies each, two clades three
synapomorphies, three clades four synapomorphies, and one
clade (sect. Tetragonolobus) has five synapomorphies. Nine
putative uniquely derived synapomorphies within Lotus are
Degtjareva et al. 819
#2006 NRC Canada
found. There is no obvious correlation between number of
morphological synapomorphies and node support in molecu-
lar phylogenetic analyses. For example, clade E2-1 (Fig. 3),
with four morphological synapomorphies, including one
uniquely derived synapomorphy, does not appear in a strict
consensus of trees inferred from molecular analysis (Fig. 2).
Alternatively, we found no obvious morphological synapo-
morphy for Lotus corniculatus group (clade B2, Fig. 3),
although it has bootstrap support of 100% and posterior
probability of 1.00 in molecular analyses (Fig. 2). A lot of
authors have proposed the same limits of the Lotus cornicu-
latus group solely on the basis of morphological data (see
below). This shows that a search of synapomorphies does
not represent a panacea in analyses of morphological data.
Most morphological characters show significant level of
homoplasy in the genus Lotus. Nevertheless, adding of mor-
phological data set to nrITS data allows the resolution of re-
lationships for some critical nodes.
Section Chamaelotus and section Benedictella
The section Chamaelotus was described by Kramina and
Sokoloff (2003) to segregate three closely related species of
desert annuals having sessile umbels and very small flowers.
In the majority of Lotus species, the umbels are pedunculate.
Molecular data support segregation of section Chamaelotus
(although only two of three species have so far been
sampled). Members of section Chamaelotus were tradition-
ally associated with Lotus arabicus (sect. Heinekenia). Our
data on ITS1 of L. benoistii (sect. Benedictella) clearly
shows that this species belongs to the genus Lotus. Lassen
(1986) suggested that L. benoistii should be placed in the
same section with species that we classified as section Cha-
maelotus. The ITS1 sequence in L. benoistii does not allow
testing of this hypothesis.
Section Lotus and section Erythrolotus
Section Lotus is not revealed as monophyletic in all clad-
istic analyses since L. conimbricensis (sect. Erythrolotus)is
resolved together with members of section Lotus (Arambarri
2000b; Allan et al. 2003; Sokoloff 2006; this study).
Brand (1898) accepted two species-rich sections of Lotus,
sect. Erythrolotus Brand and sect. Xantholotus Brand. These
sections were considered to share such characters as the sty-
lodium lacking a tooth, leaves with five leaflets, and fruit
dehiscent by two valves. According to Brand, members of
the section Erythrolotus have red (or pink) flowers while
members of section Xantholotus have yellow (or white)
flowers. The name Xantholotus is illegitimate because the
lectotype of the genus, L. corniculatus, belongs here, and
thus the section should be called sect. Lotus (although
Brand’s section Xantholotus also includes many species that
are now excluded from section Lotus). Chrtkova
(1984) selected L. conimbricensis as a lectotype of sect. Ery-
throlotus, and Kramina and Sokoloff (2003) postulated that
this species alone should be included in sect. Erythrolotus.
The most important difference between L. conimbricensis
and section Lotus is petal color (red vs. yellow), although
this character is much more variable in the genus than was
considered by Brand (1898). In particular, some species that
undoubtedly belong to section Lotus such as L. krylovii and
L. schoelleri often have red petals (e.g., Schweinfurth 1896;
Schischkin and Sergievskaja 1932). Given the phylogenetic
data, it is clear that L. conimbricensis should be placed in
the section Lotus.
Although it is clear that section Lotus is not monophyletic
if L. conimbricensis is excluded, it remains to be ascertained
whether it is monophyletic even with L. conimbricensis in-
cluded. The present study splits this group into two clades
(clade A and clade B). Each clade is strongly supported in
all analyses, but their sister relationship is not supported in
the analyses of molecular data alone and has high support
only in the Bayesian analysis of the combined data set. It is
important that all members of section Lotus (including
L. conimbricensis) studied so far share basic chromosome
number x= 6, and this may represent a uniquely derived
synapomorphy within the genus Lotus. The number x=6
has been reported for some species of other lineages (e.g.,
L. aegaeus,L. arabicus,L. polyphyllus). However, x=7
was also reported for these species (Grant 1995), and they
merit future cytological studies.
We hesitate to further subdivide section Lotus (e.g., into
two sections corresponding to clades A and B) until strong
phylogenetic evidence for doing so can be demonstrated,
for example, by using different DNA markers.
The Lotus corniculatus group (sect. Lotus)
The present phylogenetic data allow discussion of the lim-
its of the Lotus corniculatus species group. There are two
principal questions regarding the limits of this group.
(1) Lotus palustris is either included in the L. corniculatus
group (Ball and Chrtkova
´1968) or allied with
L. angustissimus (Brand 1898; Heyn 1970a) by different
taxonomic authorities. Lotus palustris is similar to the spe-
cies of the L.angustissimus group by the indumentum type,
leaf rachis usually prolonged above the insertion of upper
lateral leaflets, comparatively small flowers (ca. 6—
10 mm), keel shape (similar to that in L. castellanus), but it
differs from them by predominantly perennial life form and
larger dimensions of vegetative organs. Both nrITS and the
combined molecular and morphological data, however,
show that L. palustris belongs to the L. corniculatus species
group. Allan et al. (2003) when they first published the ITS
sequence of L. palustris also revealed its grouping with
L. corniculatus and its allies, but with bootstrap support less
than 50%. However, a marked seasonal polymorphism in L.
palustris noted by Heyn (1970a) and Zohary (1987) as well
as varying chromosome numbers in this species (2n= 12,
14, and 24; Grant 1965, 1995) may bear evidence to a con-
siderable variability of this taxon. Its limits and relationships
Fig. 2. Strict consensus of 20 004 trees (1461 steps) derived from a maximum parsimony analysis of ITS sequence data. Numbers above
branches are bootstrap support values obtained by maximum parsimony analysis with bootstrap resampling and posterior probabilities found
in Bayesian analysis. Only bootstrap values above 50% are shown. Terminal groups represented by new nrITS sequences are underlined.
Lotus species are attributed to sections according to the classification of Kramina and Sokoloff (2003) and Sokoloff (2003a, 2003b).
820 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada
Degtjareva et al. 821
#2006 NRC Canada
with other species need to be tested by additional analyses
using different DNA markers.
(2) Lotus pedunculatus and its allies (L. uliginosus and
L. granadensis Z
ˇertova) are either treated as members of
the L. corniculatus group (Ball and Chrtkova
or as more isolated members of section Lotus (see Grant and
Zandstra 1968; Raelson and Grant 1988, 1989). Lotus pe-
dunculatus and allies share a perennial life form with the
L. corniculatus group but differ in the presence of stolons
and other characters. Lotus pedunculatus resembles mem-
bers of L. angustissimus group in the type of hairs on the
stems and leaves and leaf rachis often prolonged above the
insertion of upper lateral leaflets (e.g., Chrtkova
1966). The present data show that L. pedunculatus and
L. uliginosus should not be included in the L. corniculatus
species group (L. granatensis has not been studied to date).
In the above-defined limits (i.e., including L. palustris and
excluding L. pedunculatus s.l.), the Lotus corniculatus group
is revealed as a monophyletic group. Relationships within
the group are relatively well resolved. Among the species
included here, diploids L. schoelleri,L. glaber,
L. stepposus,L. peczoricus,L. borbasii,L. krylovii, and
L. japonicus are closest to tetraploid L. corniculatus. These
species should be taken into account when discussing the al-
lotetraploid origin of L. corniculatus (for a review, see Grant
and Small 1996). Two GenBank nrITS accessions of
L. japonicus are closest to each other despite of obvious
morphological (Kawaguchi et al. 2001; Barykina and Kra-
mina 2005) and genomic (Hayashi et al. 2001) differences
between these plants (Barykina and Kramina (2005) even
suggest that L. japonicus ‘Miyakojima’ could be accepted
as a distinct species, Lotus miyakojimae Kramina nom. nov.
The Lotus angustissimus group (sect. Lotus)
This group, as traditionally circumscribed, is clearly not
monophyletic in the analyses presented here. It is subdivided
into two subgroups.
The first subgroup includes L. castellanus,
L. praetermissus, and L. angustissimus. The studied acces-
sions of L. angustissimus and L. praetermissus do not form
a clade. Lotus praetermissus was segregated by Kuprijanova
(1937) on the basis of complex of characters including an
indumentum of long but sparse patent hairs, solitary erect
stems with spreading branches, wider and shorter legumes
(16–20 mm long, not 20–30 mm as in L. angustissimus),
and dark brown seeds. However, many authors consider
these characters as not decisive and prefer to treat
L. praetermissus as a synonym of L. angustissimus. The
present data suggest that further studies should be conducted
prior to accepting the synonymy of these two names.
The second subgroup of the traditional L. angustissimus
group includes L. parviflorus and L. subbiflorus. These spe-
cies are closely associated with L. conimbricensis. Differen-
ces between nrITS sequences of L. parviflorus,
L. subbiflorus, and L. conimbricensis are surprisingly low
(morphologically, the three species are clearly distinguish-
able by several characters, especially by fruit shape and
size). In contrast, we found significant differences between
nrITS sequences of L. subbiflorus and L. castellanus. Mor-
phologically, L. subbiflorus and L. castellanus are rather
closely related species differing from each other mainly by
keel shape (with a long straight-tipped beak in L.subbiflorus
Lag., syn. L.suaveolens Pers.; long-beaked with incurved tip
in L.castellanus Boiss. & Reut., syn. L. subbiflorus sensu
Heyn, non Lag.) (Heyn 1970a). However, this character is
variable to some extent. Some authors accepted
L. castellanus as a subspecies of L. subbiflorus (Ball and
´1968). Kramina (in preparation) found
other morphological differences justifying the specific rank
of L. castellanus. One of the most important characters is
the presence of hairs along the ventral suture of the ovary
and fruit in L. castellanus and absence of such hairs in
L. subbiflorus. Except for L. castellanus, and some speci-
mens of L. palustris all other studied members of section
Lotus have glabrous pods. Lotus castellanus is mostly re-
stricted to Western Mediterranean (Kramina, in preparation).
A specimen from Turkey (listed in Table 1 as L. cf. castel-
lanus) fits traditionally used morphological features of L.
castellanus. However, it has completely glabrous fruits.
This specimen may represent an undescribed species.
The non-monophyletic nature of the L. angustissimus
group is an unexpected finding of the present study. Mor-
phologically members of this group are alike. The unex-
pected tree topology in this region is unlikely to result from
low species sampling. We have sampled all members of the
L. angustissimus group (as accepted by Heyn 1970a), with
exception of the rare endemic of Turkey, L. macrotrichus
Boiss. It is also unlikely that members of section Lotus ex-
hibit high infraspecific polymorphism in nrITS sequences.
To test this hypothesis, we have studied several accessions
of L. castellanus and several accessions of L. subbiflorus.
We have revealed only very low infraspecific variation in
Former members of the genus Dorycnium
Rikli (1901) accepted three sections of the genus Doryc-
nium, namely Canaria,Bonjeanea, and (Eu)Dorycnium.
This study analysed members of all three sections. Section
Canaria includes three closely related species endemic to
the Canary Islands. It is represented by L. broussonettii in
our analyses. Our phylogenetic data clearly show that sec-
tion Canaria is not closely related to sections Bonjeanea
and Dorycnium. This supports previous findings by Gillett
(1959). Morphologically, section Canaria differs from sec-
tions Bonjeanea and Dorycnium by large leaves, long petal
claws, pronouncedly rostrate keel, and by presence of some
papillae on stylodium. In addition, the geographical distribu-
Fig. 3. Strict consensus of 34 000 trees (1762 steps) derived from a maximum parsimony analysis of the combined matrix (ITS sequence
data plus morphology). Numbers above branches are bootstrap support values obtained by maximum parsimony analysis with bootstrap
resampling and posterior probabilities found in Bayesian analysis. Only bootstrap values above 50% are shown. Terminal groups repre-
sented by new nrITS sequences are underlined. Lotus species are attributed to sections according to the classification of Kramina and
Sokoloff (2003) and Sokoloff (2003a, 2003b).
822 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada
Degtjareva et al. 823
#2006 NRC Canada
Table 2. Putative morphological synapomorphies of major clades in the genus Lotus.
Clade name and correspond-
ing sectional namesaMorphological synapomorphies Apomorphy status and comments
L. glinoides +L. schimperi
Habit: annuals Also in clades A, B1–2, in some members of DEFG clade,
and in some outgroups
Peduncle shortened Also in F2–1 and in some outgroups
Flowers less than 7 mm long Also in L. garcinii, L. rectus, L. dorycnium, L. graecus, and
in some outgroups
Clade X Basal leaflets with maximum width near
Almost unique, 8 reversals within X clade
Clade Y None
Clade Z (=sect. Dorycnium +
Elongate internode between the sterile
bract and umbel
Almost unique within Lotus (present as an unstable feature
in few Lotus taxa, e.g., sect. Canaria), occurs also in
some distantly related outgroups; absent in L. strictus (re-
Umbels typically with more than 8 flow-
Stylodium smooth (not papillose) A uniquely derived synapomorphy within Lotus but present
in many outgroups (including those closest to Lotus)
L. dorycnium +L. graecus
Rachis shortened (leaves palmate) Also in clade E2–1 and some species of clade G,
L. simonae, L. polyphyllus, and some outgroups
Flowers less than 7 mm long Homoplastic, see above
Keel obtuse (not rostrate) Also in L. rectus and in some outgroups
Fruit twice as long as the calyx or shorter Also in L. garcinii, L. polyphyllus, L. parviflorus, and some
Clade A+B (=sect. Lotus +
Flowers yellow With a reversal in L. conimbricensis, also in many other
Basic chromosome number x= 6 Possibly a uniquely derived synapomorphy within Lotus but
present also in some outgroups
Clade A Habit: annuals (biennials) Homoplastic, see above
Clade B None
Clade B1 None
Clade B1–1 Flowers more than 10 mm long Also in other clades
Clade B1–2 Habit: annuals (biennials) Homoplastic, see above
Clade B2 (=L. corniculatus
None We found no synapomorphies also for B2–1 and B2–2
Clade C Flowers more than 10 mm long Very homoplastic
Clade D+E+F+G None
Clade D (=part of sect. Hei-
None We found no synapomorphies also for D1 and D2
Clade E+F+G None
Clade E Flowers less than 10 mm long Also in other clades; not always so in two species of E
Clade E1 (=part of sect. Hei-
Clade E2 Leaflet number variable Also in most outgroups, in clade G1, in L. graecus,
L. australis, L. cruentus; not so in L. arabicus and
Basal leaflets of a leaf with maximum
width near the middle or at the apex of
Homoplastic, see under clade X.
Clade E2-1 (=part of sect.
Leaf rachis shortened Homoplastic, see above
Peduncle shortened Not always so in L. ononopsis; see also above
Sterile bract scale-like Almost unique within Lotus (present as an unstable feature
in sect. Canaria); present in some outgroups
Umbels always one-flowered Homoplastic
Clade E2-2 (=part of sect.
Clade F+G (‘‘Zygocalyx
Flowers yellow With a lot of reversals; present also in other clades
Calyx monosymmetric With a reversal in L. edulis; present also in few other Lotus
spp. and in some outgroups.
Clade F None
824 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada
tion of Canaria is distinctive, because the sections Bonjea-
nea and Dorycnium are absent from the Canary Islands.
Section Dorycnium is represented here by two species,
L. dorycnium (=D. pentaphyllum) and L. graecus
(=D. graecum). The latter species was previously placed in
section Bonjeanea (e.g., Rikli 1901; Demiriz 1970) but So-
koloff (2003a, 2003b) recently suggested its placement in
section Dorycnium. A clade comprising L. dorycnium and
L. graecus has been found by Allan and Porter (2000). This
finding was repeated in the present analysis. Other members
of section Dorycnium are morphologically very close to
L. dorycnium (see Demiriz 1970; Lassen 1979; Diaz Lifante
2000; Sokoloff 2003a).
According to Sokoloff (2003a, 2003b), section Bonjeanea
includes three species, L. strictus (D. strictum), L. hirsutus
(D. hirsutum), and L. rectus (D. rectum). All species were
included in the present analyses. The combined analysis sug-
gests that section Bonjeanea is paraphyletic but that Bonjea-
nea+Dorycnium may be monophyletic. The paraphyly of
section Bonjeanea was found earlier by Sokoloff (2003b)in
a parsimony analysis of morphological data, the two sections
differing only by plesiomorphic characters. It may be rea-
sonable to combine sections Bonjeanea and Dorycnium.
However, prior to making any taxonomic decisions, it is im-
portant to produce a well-supported molecular phylogeny for
Section Heinekenia and section Ononidium
According to Kramina and Sokoloff (2003), section Hei-
nekenia includes most species that were placed by Brand
(1898) into his broadly defined section Erythrolotus. Brand
characterized this section by 5-foliolate leaves and red
(pink) flowers. The lectotype of the section Erythrolotus is
L. conimbricensis. Since morphologically and phylogeneti-
cally it is not close to other members of Brand’s section
(see above), another name, Heinekenia, must be used for
the rest of the section (see also Lassen 1986). In addition to
L. conimbricensis, we have excluded from this section
small-flowered desert annuals (section Chamaelotus, see
above). Finally, Kramina and Sokoloff (2003) placed in sec-
tion Heinekenia some species that Brand included in his sec-
tion Xantholotus (L. discolor,L. namulensis Brand,
L. aegaeus). Section Heinekenia is unusual among sections
of Lotus (except sect. Ononidium) in having its diversity
centers outside the Mediterranean region. Kramina and
Sokoloff (2003, see also Sokoloff 2001, Kramina and Sokol-
off 2004) accepted four informal groups within section Hei-
nekenia (Table 1). The present study does not support the
monophyly of section Heinekenia. Also, of the four informal
groups, only two are monophyletic (L. australis group and
L. discolor group). Members of section Heinekenia fall in
two clades (one of them includes also two species of section
Ononidium,L. garcinii and L. ononopsis). These two clades
(clade D and clade E) are close to each other in our phylo-
genetic trees, but never group together. It is almost impossi-
ble to distinguish clades D and E by using of morphological
characters. However these clades show a good correspond-
ence with geographic distribution.
Species of clade E occur in Africa plus in western and
southwestern parts of the Arabian Peninsula, and in Socotra.
The only exception is L. garcinii, which has a wide distribu-
tion extending from Somalia eastwards to Pakistan and west
India. Two subclades of clade E are also well defined in
terms of ecology and geography. The first subclade includes
L. wildii,L. discolor,L. mlanjeanus, and L. goetzei, and cor-
responds to the L. discolor species group that occurs in
mountains of tropical Africa and is morphologically well de-
fined. Its sister group (L. lalambensis,L. quinatus,
Table 2 (concluded).
Clade name and correspond-
ing sectional namesaMorphological synapomorphies Apomorphy status and comments
Clade F1 (=sect. Tetragono-
Foliage leaves encircle their nodes A uniquely derived synapomorphy within Lotus; present in
Basal leaflets of a leaf fused to rachis A uniquely derived synapomorphy
Flowers more than 15 mm long Homoplastic
Dorsal stylodium outgrowth present A uniquely derived synapomorphy
Paired fruit wings present A uniquely derived synapomorphy within Lotus
Clade F2 (=sect. Lotea +
L.simonae of sect. Ononi-
Clade G (=sect. Pedrosia +
Ventral stylodium tooth present A uniquely derived synapomorphy
Hairs along the ventral slit on ovary and
Also in Lotus castellanus and some outgroups (plus in
L. hebecarpus of sect. Heinekenia that is not covered by
Clade G1 (=sect. Rhyncholo-
Leaflet number variable Also in some other Lotus (outside clade F) and in most out-
Flowers more than 25 mm long Also in very few species (L. maritimus in our matrix)
Standard indumentum present Also in very few Lotus spp. not included in our matrix and
in few outgroups
Standard strongly bent backward Autapomorphy
Clade G2 (=part of sect. Ped-
Note: For clade names, see Fig. 3.
aAccording to classification by Kramina and Sokoloff (2003) and Sokoloff (2003a, 2003b).
Degtjareva et al. 825
#2006 NRC Canada
L. arabicus,L. ononopsis, and L. garcinii) has a center of
diversity in Ethiopia, around the Red Sea and in Socotra.
Clade D includes Asian and Australian species. Those
species that are present in the Arabian peninsula occur in its
northern (L. lanuginosus) or eastern (L. laricus) part. Lotus
gebelia is the only member of clade D that was recorded
from Africa (north Libya); however, most of its wide distri-
bution area lies in Asia, and Libya is far away from centers
of diversity of clade E in Africa. The Australian species
L. australis and L. cruentus are sister to L. laricus, which
has the most eastern distribution among species included in
clade D except to the Australian species (eastwards to Paki-
stan). Morphologically, L. laricus has no obvious synapo-
morphies with Australian Lotus species.
Relationships between L. gebelia,L. michauxianus, and
L. aegaeus are of particular interest. Lotus aegaeus is the
only species of section Heinekenia that has yellow petals.
Other species of the section are red-, pink- or white-flow-
ered. Our results support earlier conclusions by Heyn
(1970b) and Chrtkova
´(1967) that L. aegaeus is
closest to L. gebelia. In some cases it is difficult to distin-
guish between L. aegaeus and L. gebelia if petal color is
lost on herbarium material. The position of L. michauxianus
on some distance from L. gebelia is intriguing. Morphologi-
cally, these two species are closest to each other. The main
difference is flower size (Chrtkova
´1984). Some au-
thors consider L. michauxianus as a synonym of L. gebelia
(e.g., Heyn 1970b). More material should be studied to
understand if we have indeed an obvious conflict between
morphological and nrITS data in this case.
Section Ononidium was traditionally circumscribed as a
small group of species occurring in southwestern Asia and
east Africa. These species differ from most representatives
of the genus in having leaves with typically three (not five)
leaflets. Lotus simonae also has leaves with three leaflets but
occurs in Morocco, that is, far away from members of sect.
Ononidium. Sokoloff (2003b) suggested including
L. simonae in sect. Ononidium. Present data does not sup-
port this idea. It is not clear if two other species of section
Ononidium studied here form a single clade. Grouping be-
tween these species (L. ononopsis and L. garcinii) is well
supported only in the combined analysis. The only species
of section Ononopsis not studied here is L. mollis Balf.f., a
rare endemic of Socotra. Morphologically it is similar to
It is obvious, on the basis of present phylogenetic data,
that trifoliolate leaves represent a derived condition that ap-
peared many times in the evolution of Lotus. This condition
is also characteristic for L. robsonii E.S. Martins &
D.D. Sokoloff that is morphologically close to L. goetzei
(Martins and Sokoloff 2003). Leaflet number is variable in
some Lotus species (Sokoloff 2003b; Kramina and Sokoloff
2004). It seems that this character cannot be used to segre-
gate taxa of sectional rank. It may be possible to include
L. garcinii and L. ononopsis (plus L. mollis) into section
Paraphyletic nature of section Heinekenia creates a very
difficult taxonomic problem because of a lack of obvious
morphological differences between clades D and E. Even if
paraphyletic nature of the section will be confirmed by fu-
ture studies, it might be possible to keep it as paraphyletic
one at least until morphological evidence will be found to
characterize segregated sections.
‘‘Zygocalyx’’ clade (=clade F+G)
Members of this clade usually have a zygomorphic
(monosymmetric) calyx while most other Lotus species
have an actinomorphic (polysymmetric) calyx. Thus we
could also call this lineage ‘‘Zygocalyx’’ clade. This moder-
ately supported clade includes members of sections Lotea,
Tetragonolobus,Krokeria,Pedrosia, and Rhyncholotus.A
similar clade, although with lower taxon sampling, was ear-
lier found by Allan et al. (2003, 2004). Although the ten-
dency to have a monosymmetric calyx is very prominent
and characteristic for this clade, this feature occurs rarely
also in some other species of Lotus, for example in L.bor-
basii,L.delortii (Ujhelyi 1960), and L.peczoricus (Miniaev
and Ulle 1977). Besides, a few members of the ‘‘Zygo-
calyx’’ clade have polysymmetric calyx, for example,
L. edulis (monospecific section Krokeria).
The present phylogenetic data support the presence of two
major clades (F and G) within the ‘‘Zygocalyx’’ clade. Clade
F includes sections Tetragonolobus,Krokeria, and Lotea.
Section Tetragonolobus is well defined by four uniquely
derived synapomorphies (Table 2). As mentioned above, it is
not reasonable to accept generic rank for this taxon. Our phy-
logenetic data suggest that L. simonae should be re-classified
as a member of section Lotea. Although this species differs
from other members of the section in trifoliolate leaves, it
has a monosymmetric calyx, a key morphological character
of this group. Morphologically, this species differs from
members of other sections of the ‘‘Zygocalyx’’ clade in fruit
and stylodium morphology. Clade G includes sections Pedro-
sia and Rhyncholotus. Our data support previous findings by
Allan et al. (2004) on the paraphyly of section Pedrosia.Itis
paraphyletic because members of the section Rhyncholotus
are embedded within it. It may be possible to combine both
sections under the name Pedrosia, although more extensive
morphological and molecular data should be collected to
make formal taxonomic decisions. A clear morphological
synapomorphy of clade G is presence of a ventral tooth on
the stylodium. In clade F, the ventral stylodium tooth is al-
ways absent, although species of Tetragonolobus have a dor-
sal tooth or outgrowth. The presence of dorsal stylodium
structures in Tetragonolobus and similar ventral structures in
Pedrosia/Rhyncholotus is an obvious example of evolution-
ary parallelism. Similar structures are rare in the family Le-
guminosae. They are definitely absent among other members
of Loteae and their closest relatives, Robinieae and Sesba-
¨nch 1910; Lavin and Delgado 1990; Lavin and
Sousa 1995; Kramina and Sokoloff 1999).
There is an apparent functional correlation between stylo-
dium and keel evolution in Lotus. In sections Dorycnium
and Bonjeanea, the keel is often obtuse and the stylodium
is always smooth, while in other members of Lotus the keel
is beaked and the stylodium is papillose. The beak is espe-
cially long in some members of the ‘‘Zygocalyx’’ clade. An
elaborated stylodium surface has a functional significance
for secondary pollen presentation. In Loteae, during the visit
of a pollinator, the pollen is pushed through an opening at
the top of the keel aided by the dilated stamen filaments
(Faegri and van der Pijl 1979). The stylodium may also act
826 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada
in this process. A papillose stylodium surface may help to
push the pollen, and the ventral or dorsal tooth may be
even more effective mean for achieving this. We could spec-
ulate that, in Lotus, the longer the keel beak the more im-
portant is the contribution of the stylodium to secondary
pollen presentation. All Lotus species with obtuse keel have
a least elaborated stylodium. Species with an exceedingly
long keel beak have the ventral tooth on the stylodium.
Phylogenetic relationships of Lotus creticus are of special
interest because this species has a very small ventral stylo-
dium tooth (Kramina and Sokoloff 1999; Valde
´s 2000). The
tooth in L. creticus is smaller than in most members of Pe-
drosia and Rhyncholotus. Sometimes, the tooth is almost ab-
sent. Traditionally, L. creticus was placed in section Lotea
(Ball and Chrtkova
´s 2000). Kramina
and Sokoloff (1999) have suggested moving L. creticus to
section Pedrosia. Apart from the presence of the stylodium
tooth, they noted similarity between this species and
L. pseudocreticus (sect. Pedrosia) in general habit, certain
floral features, and ecology. However, Allan et al. (2003,
2004), on the basis of nrITS molecular phylogenetic data
suggested placement of L. creticus in section Lotea. That re-
sult implies a double origin of the ventral stylodium tooth in
the genus Lotus. We have produced a new ITS sequence
based on another voucher specimen. Our data strongly sup-
port placement of L. creticus in the section Pedrosia, close
to L. pseudocreticus and L. campylocladus. Our results sug-
gest a single origin of the ventral stylodium tooth in the ge-
nus Lotus. Detailed studies should be undertaken to
determine if the ITS region is variable in L. creticus.
The work is supported by grants from the President of
Russia (MD-1200.2005.4) and RFBR (03-04-48831 and 06-
04-48113). We are very grateful to Prof. M.W. Chase and
L. Csiba (Royal Botanic Gardens, Kew) for extracted DNA
of Lotus brousonnettii,Podolotus,Pseudolotus, and Antope-
titia, to Prof. D. Podlech (M), Prof. H. Freitag (KAS),
I.D. Illarionova (LE), and Herbaria BE, E, K, LE, MHA,
MW, Z for leaf fragments for DNA extraction, and to
E. Severova for collecting Lotus creticus material. We are
indebted to L. Gillespie, P.K. Endress, G. Sandral, and two
anonymous reviewers for helpful suggestions and criticism.
Ali, S.I., and Sokoloff, D.D. 2001. A new combination in Pseudo-
lotus Rech.f. (Leguminosae: Loteae). Kew Bull. 56: 721–723.
Allan, G.J., and Porter, J.M. 2000. Tribal delimitation and phyloge-
netic relationships of Loteae and Coronilleae (Faboideae: Faba-
ceae) with special reference to Lotus: evidence from nuclear
ribosomal ITS sequences. Am. J. Bot. 87: 1871–1881. PMID:
Allan, G.J., Zimmer, E.A., Wagner, W.L., and Sokoloff, D.D.
2003. Molecular phylogenetic analyses of tribe Loteae (Legumi-
nosae): implications for classification and biogeography. In Ad-
vances in legume systematics. Part 10. Higher level systematics.
Edited by B.B. Klitgaard and A. Bruneau. Royal Botanic Gar-
dens, Kew. pp. 371–393.
Allan, G.J., Francisco-Ortega, J., Santos-Guerra, A., Boerner, E.,
and Zimmer, E.A. 2004. Molecular phylogenetic evidence for
the geographic origin and classification of Canary Island Lotus
(Fabaceae: Loteae). Mol. Phylogenet. Evol. 32: 123–138.
Arambarri, A.M. 1999. Illustrated catalogue of Lotus L. seeds (Fa-
baceae). In Trefoil: the science and technology of Lotus.Edited
by P.R. Beuselinck. CSSA Special Publication No. 28, Crop
Science Society of America, Madison, Wis. pp. 21–41.
Arambarri, A.M. 2000a. A cladistic analysis of the New World
species of Lotus L. (Fabaceae, Loteae). Cladistics, 16: 283–297.
Arambarri, A.M. 2000b. A cladistic analysis of the Old World spe-
cies of Lotus (Fabaceae: Loteae). Can. J. Bot. 78: 351–360.
Arambarri, A.M., Stenglein, S.A., Colares, M.N., and Novoa, M.C.
2005. Taxonomy of the New World species of Lotus (Legumi-
nosae: Loteae). Aust. J. Bot. 53: 797–812. doi:10.1071/
Ball, P.W., and Chrtkova
´, A. 1968. Lotus L. In Flora Euro-
paea. Vol. 2. Edited by T.G. Tutin, V.H. Heywood,
N.A. Burges, D.M. Moore, S.M. Valentine, and D.A. Webb.
Cambridge University Press, Cambridge, UK. pp. 173–176.
Barykina, R.P., and Kramina, T.E. 2005. A comparative anatomical
study of Lotus japonicus and related species. Bull. Moscow Soc.
Naturalists Biol. Ser. 110(5): 36–51. [In Russian.]
Brand, A. 1898. Monographie der Gattung Lotus. Bot. Jahrb. 25:
Callen, E.O. 1959. Studies in the genus Lotus (Leguminosae).
I. Limits and subdivisions of the genus. Can. J. Bot. 37: 157–
´, A. 1966. Bemerkungen zur Taxonomie von Lo-
tus uliginosus Schkuhr und L. pedunculatus Cav. Folia Geobot.
Phytotax. 1: 78–81.
´, A. 1967. The variability of critical species of
Lotus in Iran and in the neighbouring countries. I. Folia Geobot.
Phytotax. 2: 283–310.
´, A. 1984. Lotus.In Flora Iranica. No. 157. Edited
by K.H. Rechinger. Akademische Druck und Verlagsanstalt,
Graz, Austria. pp. 327–342.
Crompton, C.W., and Grant, W.F. 1993. Pollen morphology in Lo-
teae (Leguminosae) with particular reference to the genus Lotus
L. Grana, 32: 129–153.
Degtjareva, G.V., Valiejo-Roman, C.M., Kramina, T.E., Mironov,
E.M., Samigullin, T.H., and Sokoloff, D.D. 2003. Taxonomic
and phylogenetic relationships between Old World and New
World members of the tribe Loteae (Leguminosae): new insights
from molecular and morphological data, with special emphasis
on Ornithopus. Wulfenia, 10: 15–50.
Demiriz, H. 1970. Dorycnium Miller. In Flora of Turkey and the
east Aegean Islands. Vol. 3. Edited by P.H. Davis. Edinburgh.
Diaz Lifante, Z. 2000. Dorycnium Mill. In Flora Iberica. Vol. 7.
Edited by S. Talavera, C. Aedo, S. Castroviejo, A. Herrero, C.
Romero Zarco, F.J. Salgueiro, and M. Velayos. Real Jardin
Botanico, Madrid. pp. 812–823.
´ez, M.J., and Ferguson, I.K. 1994. The pollen morphology of the
tribes Loteae and Coronilleae (Papilionoideae: Loteae). 2. Lotus
L. and related genera. Rev. Palaeobot. Palynol. 81: 233–255.
Dominguez, E., and Galiano, E.F. 1979. Revision del genero Tetra-
gonolobus Scop. (Fabaceae). Lagascalia, 8: 189–214.
Doyle, J.J., and Doyle, J.L. 1987. A rapid DNA isolation procedure for
small quantities of fresh leaf tissue. Phytochem. Bull. 19: 11–15.
Endress, P.K. 1994. Diversity and evolutionary biology of tropical
flowers. Cambridge University Press, Cambridge, UK.
Faegri, K., and van der Pijl, L. 1979. The principles of pollination
ecology. 3rd ed. Pergamon Press, Oxford, UK.
Degtjareva et al. 827
#2006 NRC Canada
Felsenstein, J. 1985. Confidence limits on phylogenetics: an ap-
proach using the bootstrap. Evolution, 39: 783–791. doi:10.
Fedorov, A. A. (Editor). 1969. Chromosome numbers of flowering
plants. Nauka, Leningrad.
Gillett, J.B. 1959. Lotus in Africa south of the Sahara (excluding
the Cape Verde islands and Socotra) and its distinction from
Dorycnium. Kew Bull. 3: 361–381.
Goldblatt, P., and Johnson, D. 1996. Index to plant chromosome
numbers for 1992–1993. Monogr. Syst. Bot. Mo. Bot. Gard. 58:
Goldblatt, P., and Johnson, D. 1998. Index to plant chromosome
numbers for 1994–1995. Monogr. Syst. Bot. Mo. Bot. Gard. 69:
Grant, W.F. 1965. A chromosome atlas and interspecific hybridiza-
tion index for genus Lotus (Leguminosae). Can. J. Genet. Cytol.
Grant, W.F. 1991. Chromosomal evolution and aneuploidy in Lo-
tus. In Chromosome engineering in plants: genetics, breeding,
evolution. Part B. Edited by T. Tsuchiya and P.K. Gupta.
Elsevier Science Publishers, Amsterdam. pp. 429–447.
Grant, W.F. 1995. A chromosome atlas and interspecific-intergene-
ric index for Lotus and Tetragonolobus (Fabaceae). Can. J. Bot.
Grant, W.F., and Small, E. 1996. The origin of the Lotus cornicu-
latus (Fabaceae) complex: a synthesis of diverse evidence. Can.
J. Bot. 74: 975–989.
Grant, W.F., and Zandstra, I.I. 1968. The biosystematics of the
genus Lotus (Leguminosae) in Canada. II. Numerical chemotax-
onomy. Can. J. Bot. 46: 585–589.
Greene, E.L. 1890. Enumeration of the North American Loti. Pitto-
nia, 2: 133–150.
Hayashi, M., Miyahara, A., Sato, S., Kato, T., Yoshikawa, M., Ta-
keta, M., Hayashi, M., Pedrosa, A., Onda, R., Imaizumi-Anraku,
H., Bachmair, A., Sandal, N., Stougaard, J., Murooka, Y., Ta-
bata, S., Kawasaki, S., Kawaguchi, M., and Harada, K. 2001.
Construction of a genetic linkage map of the model legume Lo-
tus japonicus using an intraspecific F2 population. DNA Res. 8:
301–310. doi:10.1093/dnares/8.6.301. PMID: 11853317.
Heyn, C.C. 1970a. Studies in Lotus. III. The L. angustissimus
group. Isr. J. Bot. 19: 271–292.
Heyn, C.C. 1970b.Lotus L. In Flora of Turkey and the east Aegean
Islands. Vol. 3. Edited by P.H. Davis. Edinburgh. pp. 518–531.
Hutchinson, J. 1964. The genera of flowering plants (Angios-
permae). Vol. 1. Clarendon Press, Oxford, UK.
Isely, D. 1981. Leguminosae of the United States. III. Subfamily
Papilionoieae: Tribes Sophoreae, Podalyrieae, Loteae. Mem.
New York Bot. Gard. 25: 1–264.
Kawaguchi, M., Motomura, T., Imaizumi-Anraku, H., Akao, S.,
and Kawasaki, S. 2001. Providing the basis for genomics in Lo-
tus japonicus: the accessions Miyakojima and Gifu are appropri-
ate crossing partners for genetic analysis. Mol. Genet.
Genomics, 266: 157–166. PMID: 11683256.
Kirkbride, J.H., Jr. 1994. Taxonomic circumscription of the genus
Lotus Linnaeus (Fabaceae, Loteae), its tribal position, and its
species. In Proceedings of the 1st International Lotus Sympo-
sium, St. Louis, Missouri, 22–24 March 1994. Edited by
P.R. Beuselinck and C.A. Roberts. Univ. Missouri Ext. Publ.
MX 411, Columbia, Mo. pp. 11–15.
Kirkbride, J.H., Jr. 1999. Lotus systematics and distribution. In Tre-
foil: the science and technology of Lotus.Edited by
P.R. Beuselinck. CSSA Special Publication No. 28, Crop
Science Society of America, Madison, Wis. pp. 1–20.
Kramina, T.E. 1992. A study of geographical variability of some
morphological characters of Lotus corniculatus L. s.l. (Legumi-
nosae Juss.) in European Russia and adjacent territories. Bull.
Moscow Soc. Nat. Biol. Ser. 97: 108–119. [In Russian.]
Kramina, T.E. 1999. Taxonomic revision of Lotus L. section Lotus
(Leguminosae) in the territory of European Russia and adjacent
countries. Ph.D. thesis, Higher Plants Department, Moscow State
University, Moscow. [In Russian.]
Kramina, T.E., and Sokoloff, D.D., 1997. Lotus roudairei Bonnet
and taxonomic relationships between African and North Ameri-
can species of the tribe Loteae (Papilionaceae). Adansonia Ser.
3, 19(2): 321–328.
Kramina, T.E., and Sokoloff, D.D. 1999. Taxonomic bearing of
stylodium tooth in the genus Lotus (Papilionaceae) with special
reference to Lotus creticus L. Feddes Repert. 110: 527–533.
Kramina, T.E., and Sokoloff, D.D. 2001. Kebirita, a new genus of
Leguminosae-Loteae from NW Africa. Bull. Moscow Soc. Nat.
Biol. Ser. 106: 58–63. [In Russian.]
Kramina, T.E., and Sokoloff, D.D. 2003. On Lotus sect. Erythrolo-
tus and related taxa (Leguminosae). Bull. Moscow Soc. Nat.
Biol. Ser. 108: 59–62. [In Russian.].
Kramina, T.E., and Sokoloff, D.D., 2004. A taxonomic study of
Lotus australis complex (Leguminosae), with special emphasis
on plants from Pacific Ocean islands. Adansonia Ser. 3, 26(2):
Kramina, T.E., and Tikhomirov, V.N. 1991. Within-population
variability of some morphological characters of Lotus cornicula-
tus L. s.l. (Leguminosae) on the territory of European part of
USSR. Bull. Moscow Soc. Nat. Biol. Ser. 96: 117–126. [In Rus-
Kuprijanova, L.A. 1937. A new annual species of Lotus. Bot. Mat.
(Leningrad), 7: 37–39. [In Russian.]
Lassen, P. 1979. Dorycnium fulgurans, a neglected species from
the Balearic Islands. Bot. Notis. 132: 357–358.
Lassen, P. 1986. Acmispon sect. Simpeteria,Acmispon roudairei,
Dorycnium strictum,Lotus benoistii.In Med-Checklist Notulae,
13. Edited by W. Greuter and T. Raus. Willdenowia 16(1): 107–
Lavin, M., and Delgado, A. 1990. Pollen brush of Papilionoideae
(Leguminosae): morphological variation and systematic utility.
Am. J. Bot. 77: 1294–1312.
Lavin, M., and Sousa, S.M. 1995. Pylogenetic systematics and bio-
geography of the tribe Robinieae (Leguminosae). Syst. Bot.
Monogr. 45: 1–165.
Lavin, M., Wojciechowski, M.F., Richman, A., Rotella, J., Sander-
son, M.J., and Beyra-Matos, A. 2001. Identifying tertiary radia-
tions of Fabaceae in the Greater Antilles: alternatives to cladistic
vicariance analysis. Int. J. Plant Sci. 162(Suppl. 6): S53–S76.
Lewis, G., Schrire, B., Mackinder, B., and Lock, M. 2005. Le-
gumes of the world. Royal Botanic Gardens, Kew, UK.
Maire, R. 1924. Contributions a l’etude de la Flore de l’Afrique du
Nord (9efascicule). Bull. Soc. Hist. Nat. Afrique N. 15: 380–395.
Maire, R., Weiller, M., and Wilczek, E. 1935. Sertulum austro-
maroccanum. Bull. Soc. Hist. Nat. Afr. N. 26: 120–122.
Martins, E.S., and Sokoloff, D.D. 2003. Lotus robsonii
E.S. Martins & D.D. Sokoloff, sp. nov. In Flora Zambesiaca.
Vol. 3. Part 7. Edited by G. Pope R.M. Polhill, and E.S. Martins.
Royal Botanic Gardens, Kew, UK. pp. 3–4.
Miniaev, N.A., and Ulle, Z.G. 1977. De speciebus generis Lotus
L. in regionibus orientali-septentrionalibus partis Europaeae
URSS crescentibus. Nov. Syst. Plant. Vasc. 14: 153–161. [In
¨nch, C. 1910. U
¨ber Griffel und Narbe einiger Papilionaceae.
Beih. Bot. Centralbl. 27: 83–126.
828 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada
Nanni, L., Ferradini, N., Taffetani, F., and Papa, R. 2004. Molecu-
lar phylogeny of Anthyllis spp. Plant Biology (Stuttgart), 6: 454–
Ottley, A.M. 1944. The American Loti with special consideration
of a proposed new section, Simpeteria. Brittonia, 5: 81–123.
Polhill, R.M. 1981. Loteae, Coronilleae. In Advances in legume
systematics. Edited by R.M. Polhill and P.H. Raven. Royal Bo-
tanic Gardens, Kew, UK. pp. 371–375.
Polhill, R.M. 1994. Complete synopsis of legume genera. In Phyto-
chemical dictionary of the Leguminosae. Vol. 1. Edited by
F.A. Bisby, J. Buckingham, and J.B. Harborne. Chapman &
Hall, London. pp. xlix–lvii.
Posada, D., and Crandall, K.A. 1998. Modeltest: testing the model
of DNA substitution. Bioinformatics, 14: 817–818. doi:10.1093/
bioinformatics/14.9.817. PMID: 9918953.
Raelson, J.V., and Grant, W.F. 1988. Evaluation of hypotheses con-
cerning the origin of L. corniculatus using isoenzyme data.
Theor. Appl. Genet. 76: 267–276.
Raelson, J.V., and Grant, W.F. 1989. An isoenzyme study in the
genus Lotus (Fabaceae). Experimental protocols and genetic ba-
sis of electrophoretic phenotype. Theor. Appl. Genet. 77: 595–
Rechinger, K.H. 1984. Pseudolotus,Podolotus.In Flora Iranica.
No. 157. Edited by K.H. Rechinger. Akademische Druck und
Verlagsanstalt, Graz, Austria. pp. 346–350.
Rikli, M. 1901. Die Gattung Dorycnium. Bot. Jahrb. 31: 314–404.
Ronquist, F., and Huelsenbeck, J.P. 2003. MrBAYES 3: Bayesian
phylogenetic inference under mixed models. Bioinformatics,
19: 1572–1574. doi:10.1093/bioinformatics/btg180. PMID:
Schischkin, B.K., and Sergievskaja, L.P. 1932. New species of the
genus Lotus L. from Asia. Animadv. Syst. Herb. Univ. Tomsk.
7–8: 5–6. [In Russian.]
Schweinfurth, G. 1896. Sammlung arabisch-aethiopischer Pflanzen
Ergebnisse von Reisen in den Jahren 1881, 88, 89, 91, 92, und
94. Bull. Herb. Boiss. 4: 9–266.
Sokoloff, D.D. 1997. Comparative study of fruit anatomy in the
genus Anthyllis (Papilionaceae, Loteae). Bot. Zhurn. 82: 58–74.
Sokoloff, D.D. 1999. Ottleya, a new genus of Papilionaceae-Loteae
from North America. Feddes Repert. 110: 89–97.
Sokoloff, D.D. 2000. New combinations in Acmispon (Legumi-
nosae, Loteae). Ann. Bot. Fenn. 37: 125–131.
Sokoloff, D.D. 2001. New records of Lotus (Leguminosae: Papilio-
noideae: Loteae) from Africa and southwest Asia. Kew Bull. 56:
Sokoloff, D.D. 2003a. On system and phylogeny of the tribe Lo-
teae DC. (Leguminosae). Bull. Moscow Soc. Naturalists. Biol.
Ser. 108: 35–48. [In Russian.]
Sokoloff, D.D. 2003b. Morphology and classification of the tribe Lo-
teae DC. of the family Leguminosae. Dr. Sci. thesis, Higher Plants
Department, Moscow State University, Moscow. [In Russian.]
Sokoloff, D.D. 2006. Cladistic analysis of the tribe Loteae (Legu-
minosae) based on morphological characters. In Plant taxonomy:
advances and relevance. Edited by A.K. Pandey, J. Wen, and
J.V.V. Dogra. CBS Publishers & Distributors, New Delhi.
Sokoloff, D.D., and Lock, J.M. 2005. Tribe Loteae. In Legumes of
the world. Edited by G. Lewis, B. Schrire, B. Mackinder,
M. Lock. Royal Botanic Gardens, Kew, UK. pp. 455–465.
Stenglein, S.A., Colares, M.N., Arambarri, A.M., Novoa, M.C.,
Vizcaino, C.E., and Katinas, L. 2004. Leaf epidermal microchar-
acters of the Old World species of Lotus (Leguminosae: Loteae)
and their systematic significance. Aust. J. Bot. 51: 459–469.
Swofford, D.L. 2000. PAUP*. Phylogenetic analysis using parsi-
mony (*and other methods). Version 4. Sinauer Associates, Sun-
Talavera, S., and Salgueiro, F.J. 1999. Sobre el tratamiento de la
familia Leguminosae en Flora Iberica. Lagascalia, 21: 155–222.
Taubert, P. 1894. Leguminosae. In Die natu
lien. Teil 3, Abt. 3. Edited by A. Engler and K. Prantl.
Engelmann, Leipzig, Germany. pp. 70–385.
Tikhomirov, V.N., and Sokoloff, D.D. 1997. Taxonomic position of
Vermifrux abyssinica (A. Rich.) Gillett and taxonomy of the
tribe Loteae s.l. (Papilionaceae). Feddes Repert. 108: 335–344.
Ujhelyi, J. 1960. Etudes taxonomiques sur la groupe du Lotus cor-
niculatus L. sensu lato. Ann. Hist.-Nat. Mus. Natl. Hung. Bot.
´s, B. 2000. Lotus L., Tetragonolobus Scop. In Flora Iberica.
Vol. 7. Edited by S. Talavera, C. Aedo, S. Castroviejo, A. Her-
rero, C. Romero Zarco, F.J. Salgueiro, and M. Velayos. Real
Jardin Botanico, Madrid. pp. 776–812, 823–829.
White, T.J., Bruns, T., Lee, S., and Taylor, J. 1990. Amplification
and direct sequencing of fungal ribosomal RNA genes for phy-
logenetics. In PCR protocols: a guide to methods and applica-
tions. Edited by M.A. Innis, D.H. Gelfand, J.J. Sninsky, and
T.J. White. Academic Press, Inc., New York. pp. 315–322.
Wojciechowski, M.F., Sanderson, M.J., Steele, K.P., and Liston, A.
2000. Molecular phylogeny of the ‘Temperate Herbaceous
Tribes’ of papilionoid legumes: A supertree approach. In Ad-
vances in legume systematics. Part 10. Higher level systematics.
Edited by B.B. Klitgaard and A. Bruneau. Royal Botanic Gar-
dens, Kew, UK. pp. 277–298.
Zharkikh, A.A., Rzhetsky, A.Yu., Morozov, P.S., Sitnikova, T.L.,
and Krushkal, J.S. 1990. A package of microcomputer programs
for sequence analysis and construction of phylogenetics. Gene,
Zohary, M. 1987. Flora Palaestina. Part 2. Jerusalem.
Appendix A. Morphological characters used
in the analysis
The full morphological data matrix is available as supple-
1. Habit. 0, tree or shrub; 1, perennial herb or suffrutescent;
2, annual or biennial herb. (Nonadditive).
2. Phyllotaxis on vegetative shoots. 0, spiral; 1, distichous.
3. Pulvinus at the leaf base. 0, present; 1, absent.
4. Leaf base width of lower foliage leaves. 0, leaf base does
not (or does not completely) encircle the node; 1, leaf
base completely encircles the node.
5. Stipule morphology. 0, stipules entirely membranous or
with memberanous part; 1, membranous part of stipules
6. Stipels. 0, present; 1, absent.
7. Petioles of foliage leaves. 0, always present and distinct;
1, variable (present and distinct or absent); 2, always ab-
sent or extremely short (up to 1 mm). (Nonadditive).
8. Rachis of upper foliage leaves. 0, elongated; 1, shor-
tened. (If leaflet number is more than three, then 0
means leaves pinnate and 1 means leaves palmate.)
9. Leaflet number in foliage leaves. 0, variable; 1, five; 2,
10. Shape of basal leaflets of a leaf. 0, maximum width
Degtjareva et al. 829
#2006 NRC Canada
near the middle or at the apex of the leaflet; 1, maxi-
mum width at the leaflet base.
11. Terminal leaflet shape. 0, obovate to oblanceolate or
elliptical (length to width ratio less than or equal to 3);
1, linear or narrowly-oblanceolate (length to width ratio
more than 3).
12. Basal leaflet fusion. 0, basal leaflets of a leaf free; 1,
basal leaflets fused to the rachis.
13. Peduncle length. 0, elongated (the peduncle, i.e., umbel
stalk, is much longer than its width); 1, shortened (the
peduncle is almost as long as wide).
14. Sterile bract (pseudobract – see Degtjareva et al. 2003)
presence. 0, absent; 1, present.
15. Sterile bract position. 0, typically at the base of the
partial inflorescence; 1, typically separated from the
partial inflorescence by an elongated internode.
16. Sterile bract morphology. 0, foliage leaf; 1, scale-like
17. Partial inflorescence type. 0, raceme; 1, head or umbel.
18. Flower number per partial inflorescence. 0, one; 1, two
or three; 2, four to seven; 3, eight and more. (Additive).
19. Floral bud position. 0, bent backwards; 1, not bent
20. Bract fusion. 0, always free; 1, (often) fused to each
21. Bracteoles. 0, always present; 1, always or often absent.
22. Flower size. 0, not exceeding 7 mm; 1, 7–10 mm; 2,
10–15 mm; 3, 15–25 mm; 4, more than 25 mm. (Addi-
tive). Univariate analysis was made to analyse this char-
acter. It helped to determine character states. More than
2000 individual measurements of flower length was
made (Kramina and Tikhomirov 1991; Kramina 1992,
1999 and T.E. Kramina, unpublished data).
23. Calyx symmetry (terminology after Endress 1994). 0,
polysymmetric (with five symmetry planes), 1, mono-
symmetric (with single symmetry plane).
24. Lower calyx teeth length. 0, shorter than the tube (plus
hypanthium) or as long as the tube; 1 longer than the
tube (plus hypanthium).
25. Yellow color of petals. 0, petals never yellow; 1, petals
at least sometimes yellow.
26. Red, pink or dark color of wings and standard. 0, wings
and standard never red, pink or dark (sometimes except
veins); 1, wings and standard at least sometimes or par-
tially red, pink or dark.
27. Petal claws. 0, not or slightly exceeding calyx tube; 1,
considerably exceeding calyx tube.
28. Standard indumentum. 0, absent; 1, present.
29. Keel shape. 0, rostrate (as in Figs. 171–185 in Valde
2000); 1, obtuse (as in Figs. 187–188 in Diaz Lifante
30. Keel tip shape. 0, straight; 1, incurved.
31. Stylodium surface. 0, smooth; 1, papillose.
32. Stamen filaments. 0, distally not dilated; 1, distally di-
33. Hairs along the ventral slit on ovary and fruit. 0, absent;
34. Hairs at lateral ovary and fruit surface. 0, absent; 1,
35. Ventral tooth on the stylodium. 0, absent; 1, present.
36. Dorsal tooth or outgrowth on the stylodium. 0, absent;
37. Ovule orientation pattern (terminology after Tikhomirov
and Sokoloff 1997). 0, micropylae superae; 1, micropy-
lae alternantes; 2, micropylae inferae. (Nonadditive).
38. Fruit length. 0, more than 2 times as long as the calyx;
1, shorter than or up to 1–2 times as log as calyx.
39. Fruit shape. 0, straight or almost straight; 1, incurved
toward ventral side; 2, incurved toward dorsal side.
40. Dorsal fruit dehiscence. 0, present; 1, absent.
41. Ventral fruit dehiscence. 0, present; 1, absent.
42. Transversal fruit breaking. 0, absent (i.e., fruits not lo-
mentaceous); 1, present (i.e., fruits lomentaceous).
43. Structure of pericarp parchment layer (see Sokoloff
1997 and Tikhomirov and Sokoloff 1997, for details
and illustrations). 0, fibres form single stratum in each
fruit valve or parchment layer absent; 1, fibres form
two strata in each fruit valve.
44. Seed shape. 0, rounded or slightly elongated; 1, consid-
erably elongated or linear.
45. Seed surface. 0, with large conspicuous papillae; 1,
without large papillae.
46. Basic chromosome number. 0, x= 10; 1, x=8;2,x=
7; 3, x=6.(Additive).
830 Can. J. Bot. Vol. 84, 2006
#2006 NRC Canada