Plant Syst. Evol. 231: 109±132 (2002)

Karyological evolution and molecular phylogeny in Macaronesian dendroid spurges (Euphorbia subsect. Pachycladae) J. Molero1, T. Garnatje2, A. Rovira1, N. Garcia-Jacas2, and A. Susanna2 1 2

Department of Botany, Faculty of Pharmacy, University of Barcelona, Barcelona, Spain Botanical Institute of Barcelona (CSIC-Ajuntament de Barcelona), Barcelona, Spain

Received March 3, 2001 Accepted October 28, 2001

Abstract. Euphorbia subsect. Pachycladae is a taxon of primarily Macaronesian distribution, de®ned by morphological and biogeographical criteria. On the basis of morphological data, it is a heterogeneous group within which at least three complexes of species can be distinguished. To ascertain whether it is a natural group and discover its phylogenetic relations, we performed a cladistic analysis of the sequences of ribosomal nuclear DNA and a karyological study. The results of the two studies are concordant and show that the sub-section is polyphyletic and includes three di€erent groups. The ®rst monophyletic group is made up of the Macaronesian endemics E. atropurpurea complex and E. lamarckii complex, which form a polytomy with E. dendroides as the basal species. The lauroid species E. longifolia and E. stygiana represent the second monophyletic group, which derive from Mediterranean forms of E. sect. Helioscopia Dumort. Both species are paleopolyploid (2n ˆ 44) with highly symmetrical karyotypes. Finally, E. balsamifera, with a Canarian, African and Arabian distribution, remains isolated in a basal position. Its karyotype, with 2n ˆ 20 chromosomes, di€ers from the Macaronesian model and displays analogies with African cactiform spurges. On the basis of the results, some hypotheses are formulated about speciation processes in the three groups.

Key words: Biogeography, Euphorbiaceae, Euphorbia, ITS phylogeny, Karyology, Macaronesia, Speciation.

Introduction In the Macaronesian region, 44 species of Euphorbia have been catalogued (Hansen and Sunding 1993). The level of Macaronesian endemics for this genus is over 30%, slightly below the estimates of endemicity for the Canary Islands, which is approximately 40% (Francisco-Ortega et al. 2000), but above the percentages attributed to the remaining Macaronesian archipelagos (Hobohm 2000). The largest concentration of endemic taxa is found in the Canary Islands (9 species, with a nucleus of primary speciation of the genus in the islands of Tenerife and La Gomera), followed by Madeira (2 species), the Azores (2 species) and one species for the Cape Verde archipelago. Boissier (1862) placed all the species of Euphorbia endemic to the Macaronesian region, which are sub-succulent dendroid shrubs and small trees, in the taxon of unspeci®ed category [§] Pachycladae, later combined with

110

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

sub-section rank [E. sect. Tithymalus Boiss. subsect. Pachycladae (Boiss. in A. DC.) Pax in Engler & Prantl, 1891]. At the sectional level, the prioritary name is E. subgen. Esula Pers. sect. Balsamis Webb & Berthel. In this sub-section Pachycladae, Boissier (1862) de®ned two groups. In the Macaronesian group ``Canarienses et Mediterraneae'' he put nine endemic taxa which are of great interest in these islands on account of their importance in the landscape and their rarity in habitats associated with laurel forest (SjoÈgren 1972, Montelongo et al. 1984, Rivas-Martõ nez et al. 1993). He also included the more widely distributed species E. balsamifera Ait. and the Mediterranean E. dendroides L., but excluded the Canarian endemics E. aphylla Brouss. ex Willd. and E. lamarckii Sweet ( ˆ E. obtusifolia Poir.; E. broussonetii Willd. ex Link) which he placed in E. sect. Tirucalli Boiss. on account of their very ¯eshy branches and their small deciduous bracts. In a second group, ``Polinesiae et Sundaicae-species anomalae'', he included four species from very distant geographical areas, which are currently part of the complex E. plumerioides Teijsm. ex Hassk. (Forster 1994). The criterion used by Boissier (1862) to group all these taxa together was essentially biogeographical. According to our observations, and on the basis of new morphological evidence recollected by our team in the past years, the sub-section associated with the Macaronesian and Mediterranean area is a heterogeneous group within which at least three complexes of species, separated by obvious morphological gaps. Our grouping di€ers in some aspects from the proposal by Govaerts et al. (2000). 1. The group of endemic xerophylous and mesophylous Macaronesian and Mediterranean species. This complex, of Macaronesian and Mediterranean distribution, is represented by two closely related groups: E. atropurpurea complex and E. lamarckii complex. It includes shrubby, succulent or sub-succulent monoecious plants, with a well-developed pleiochasial syn¯orescence (pseudo-umbellate) and carunculate seeds.

The E. atropurpurea complex is a Canarian endemic group characterized by its large, semipersistent leaves, highly developed syn¯orescences with a double pleiochasium, and large (10±20 mm), connate, persistent sub-cyathial bracts. From the ecological point of view it is a stenoic group living in mesoxerophilous and mesophilous habitats at moderate altitudes (300±1100 m) in Tenerife and La Gomera. Euphorbia bourgeauana Gay ex Boiss. ( ˆ E. lambi Svent.), an evergreen species with large connate bracts, is found on very moist slopes in GuÈimar (Tenerife) and La Gomera, in a few steep enclaves in sparse Erica-Myrica scrub and laurel forest. Euphorbia atropurpurea Brouss. ex Willd. of S Tenerife and E. bravoana Svent. of NW La Gomera are two closely related, interinsular vicariant species, which are clearly distinguishable by their reddish colour and connate bracts which are smaller than in the preceding species. They live in gulleys or at the foot of cli€s in belts of juniper scrub or Erica-Myrica scrub. The Euphorbia lamarckii (``obtusifolia'') complex (Molero and Rovira 1998) is the most numerous group and is closely related to the foregoing group, from which it is di€erentiated morphologically by the fact that it comprises sub-succulent xerophytes with very rapidly deciduous leaves, a loose pleiochasial syn¯orescence with simple or forked pleiochasial radii with small (up to 10 mm in length) sub-cyathial bracts, free to the base. This group includes E. lamarckii of the western Canary Islands, E. berthelotii Bolle ex Boiss. of La Gomera, E. regis-jubae Webb & Berthel. of the eastern Canary Islands and the Atlantic coast of Morocco, E. pedroi Molero & Rovira from the coast of Portugal (Sesimbra), E. anachoreta Svent. of the Salvage Islands, E. piscatoria Ait. of Madeira and E. tuckeyana Steud. ex Webb, which is widely distributed throughout the Cape Verde archipelago. This group has specialized in thermic and more or less dry habitats in the basal vegetation belts of the Macaronesian islands, where it occupies wide areas. However, in islands with an altitudinal gradient they can extend up to more mesophilous habitats on the

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

sunny slopes of gulleys as high as 1500 m. Euphorbia tuckeyana has the most variable ecology of all the species in the group, displaying xerophilous or mesophilous behaviour according to whether it lives in semiarid coastal zones or in sub-moist or moist mountain habitats, between 300 and 1200 m (Carter et al. 1984, Brochmann 1997). Euphorbia dendroides is very similar to the preceding species, though it is woodier with reddish, striated stems and smooth, laterally compressed seeds. It is distributed along the coastline of the countries surrounding the Mediterranean, in rocky thermophilous enclaves, no higher than 400 m (Bianco et al. 1991). E. aphylla is a xerophilous, aphyllous, very ¯eshy shrub with a crowded syn¯orescence and tiny sub-cyathial, easily deciduous bracts, which is exclusive to the Canary Islands (Gran Canaria, Tenerife and La Gomera). Because of its morphological characteristics Boissier (1862) included it in section Tirucalli. Deil (1991) relates this Canarian species to the complex E. schimperi Presl-E. uzmuk S. Carter & J. R. I. Wood of Southwestern Arabia and Socotra and E. larica Boiss. of South-East Africa. But the ease with which E. aphylla hybridizes with the complex E. atropurpurealamarckii indicates that it is closely linked to subsect. Pachycladae. 2. The hygrophilous lauroid group. This second group is made up of E. longifolia Lam. ( ˆ E. mellifera Ait.) of the western Canary Islands and Madeira, and E. stygiana H. C. Watson of the Azores. Both are hygrophilousmesohygrophilous woody shrubs or trees (from 3±10 m high) which live exclusively in gulleys or on slopes of damp or extremely damp laurel forests exposed to trade winds. They are characterized by their large, lauroid, persistent leaves, complex paniculate cymose syn¯orescence, pubescent cyathium and verrucose capsule. 3. The xerophilous Euphorbia balsamifera s. l. This species has a wide Canarian-African-Arabian distribution area. It is a stirp of ¯eshy xerophilous, dendroid shrubs, which has diversi®ed into three (geographically vica-

111

riant) subspecies: Euphorbia balsamifera subsp. adenensis (De¯ers) Bally ex Bally & Carter is found all over the southern Arabian Peninsula and Somalia. Euphorbia balsamifera subsp. sepium (N. E. Br.) Maire extends from Niger to Senegal. Euphorbia balsamifera subsp. balsamifera is found on the Atlantic coast of Morocco and Mauritania, and throughout the Canary Islands. It is undoubtedly the taxon which displays the most noteworthy di€erential characteristics: dioecy, a syn¯orescence reduced to a single large ¯ower with numerous stamens (50±100), very large globular capsules with a very thick pericarp and subspheric, smooth, ecarunculate seeds. The morphological anities between this species and other groups of Euphorbia are still unknown. Karyology has proved useful in the genus Euphorbia for recognizing phylogenetic relationships and explaining speciation processes (Simon and Vicens 1999). The study of the ITS region of ribosomal nuclear DNA has proved an ecient tool for elucidating problems related to the delimitation and origin of Marcaronesian complexes (Kim et al. 1996; Francisco-Ortega et al. 1997a, b; Susanna et al. 1999; Vargas et al. 1999). The joint use of both methods seeks to attain the following objectives: 1 To verify the monophyly of Euphorbia subsect. Pachycladae, which was de®ned by Boissier on basically biogeographical grounds and, according to morphological evidence, is not a natural group. 2 To discover the anities and relationships within the group of Macaronesian species and between them and other groups of Euphorbia. 3 To clarify as far as possible the origin and evolution of the section, with special reference to speciation and colonization. Materials and methods A. Sequencing Selection of species included in the phylogenetic study aimed to encompass the three major groups

112

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

(Canarian, lauroid and Circumafrican). Outgroup species (E. margalidiana Kuhbier & Lewej., E. helioscopia L. and E. tirucalli) were chosen among related species. Euphorbia margalidiana is an insular Mediterranean endemic species, frutescent and sub-¯eshy, restricted in area to an islet in the Balearics and closely related to the endemic Iberian species E. squamigera Loisel. (Simon and Vicens 1999); E. helioscopia is an annual which is also of Mediterranean origin and has become subcosmopolitan. Both are representatives of two of the most common biotypes of the biogeographical Mediterranean area and were chosen to assess the possible relationships between the Macaronesian and Mediterranean elements. The origin of the materials is detailed in Table 1. Total genomic DNA was extracted following the CTAB method of Doyle and Doyle (1987) as modi®ed by Soltis et al. (1991) from silica gel dried leaves collected in the ®eld, or fresh leaves of plants cultivated in the Botanical Institute of Barcelona. Double stranded DNA ampli®cation of the ITS region was performed using the polymerase chain reaction (PCR) following the protocol described in Soltis and Kuzo€ (1993) and thermostable DNA polymerase (Ecotaq, Ecogen S. R. L.). The ITS1 and ITS2 regions were ampli®ed separately. For ampli®cation of ITS1, we used the 1406F primer (Nickrent et al. 1994) as the forward primer and the reverse primer was ITS2 (White et al. 1990). The ITS2 region was ampli®ed using the ITS3 primer (White et al. 1990) as the forward primer and 307R (Nickrent et al. 1994) as the reverse primer. Puri®cation and direct sequencing of the double stranded PCR product was carried out following the procedures described in Susanna et al. (1999). Phylogenetic analysis. DNA sequences were aligned visually by sequential pairwise comparison (Swo€ord and Olsen 1990). Data matrices are available on request from the ®rst author. Parsimony analysis involved heuristic searches conducted with PAUP version 4.0b4a (Swo€ord 1999) using TBR branch swapping with character states speci®ed as unordered and unweighted. Indels were coded as ``®fth base''. All mostparsimonious trees (MPT) were saved. To locate islands of most-parsimonious trees (Maddison 1991), we performed 100 replications with random taxon addition, also with TBR branch swapping.

Bootstrap (BS) analysis with 100 replicates and TBR branch swapping (Felsenstein 1985) and decay index (DI) analysis (Bremer 1988, Donoghue et al. 1992) were performed to obtain estimates of support for the branches of the consensus tree. B. Karyology We worked with radicular meristems of specimens cultivated in the glasshouses of the Barcelona Botanical Garden, obtained from seeds or taken from woodland populations. The populations studied are listed in Table 1. Geographical distribution of sampling is illustrated in Fig. 1. Root tips were pre-treated with 0.05% aqueous colchicine for two hours at room temperature and ®xed in absolute ethanol-glacial acetic acid (3:1, v/ v). They were hydrolysed in 1 N hydrochloric acid at 60°C for 10 min and stained overnight in 2% acetic orcein at room temperature. Meristems were squashed and mounted in 9:1 45% acetic acid: glycerol and preparations were sealed with DPX. The preparations were observed with a ZeissAxioplan microscope at a magni®cation of 1000 x. The best metaphase plates were photographed on ISO 9 Kodak ®lm and drawn at a magni®cation of 4500 x using a camera lucida. The biometry data were obtained on metaphase plates with fairly contracted chromosomes (colchicine e€ect); the ®gure for the overall (relative) length of the karyotype should be interpreted in the light of these conditions. Length of chromosome arms was measured following the recommendations of Bentzer et al. (1971) and LoÈve and LoÈve (1975). We used (3)5±10 best metaphase plates to prepare the karyotypes. Numerical data were analysed statistically by a computer program written by Dr. J. Simon and running on Lotus 1-2-3 for Windows (Simon and Vicens 1999). The cluster analysis was obtained by means of the multivariate analysis program Statgraphics Plus 4.1 (Statistical Graphics Corporation, U.S.A:, 1999) using the Group Average Method and Euclidean metric distance; the variables used were: somatic chromosome number (2n), karyotype P length [length of the long arm: L% ˆ 100L/P (L+S); length of the short arm: S% ˆ 100S/ (L+S); contribution of each arm of each chromosome to the total length of the karyotype] according to Bazzichelli (1967), and A1

20 20 20

AN AT1 AT2

E. anachoreta Svent. E. atropurpurea Brouss. ex Willd.

E. bourgeauana Gay ex Boiss. 20 20 18

BU1 BU2 DE3

20

BE2

20

BAb4 20 20 20

20

BAb3

BAs1 BAs2 BE1

20

BAb2

E. balsamifera Ait. subsp. sepium (N. E. Br.) Maire E. berthelotii Bolle ex Boiss. in DC.

20

BAb1

E. balsamifera Ait. subsp. balsamifera

20 20

BAa1 BAa2

E. balsamifera Ait. subsp. adenensis (De¯ers) Bally ex Bally & Carter

20

AP

E. aphylla Brouss. ex Willd.

2n

Code

Taxon

** Spain, Canary Is., Tenerife: Teno, pr. Las Casas, 260 m, Molero & De la Fuente, 4-V-1996 (BCF49890) [AF334250, AF334265]. Portugal, Salvage Is.: Fora, culta in Bot. Gard. Canario (BCF39350). ** Spain, Canary Is., Tenerife: Santiago del Teide, 950 m, Molero, 11-VI-1990 (BCF37844) [AF334242, AF334257]. Spain, Canary Is., Tenerife: Tijoco de Arriba, 900 m, Molero, 9-III-1989 (BCF37843). Yemen, Sanna: Huth, 2000 m, Molero, 6-II-1996 (BCF411587). Yemen, Al Mukalla: El Ladwas, 1300 m, Molero, 29-II-1996 (BCF41589). Morocco: Tarfaya, Oued Ouma Fatma, 50 m, Molero, 30-IV-1992 (BCF37830). Spain, Canary Is., Gomera: Hermigua, 160 m, Molero, 7-X-1993 (BCF38672). Spain, Canary Is., Gran Canaria: Agaete, 150 m, Molero, 8-VI-1990 (BCF37822). ** Spain, Canary Is., Tenerife: Teno, pr. Las Casas, 230 m, Molero & De la Fuente, 4-V-1996 (BCF49891) [AF334249, AF334264]. Mali: Mopti, BoveÂ, 400 m, Molero, 4-I-1990 (BCF37824). Mali: Gao, Molero, 4-I-1990 (BCF37826). Spain, Canary Is., Gomera: Pico Gomero, 600 m, Molero, 11-III-1989 (BCF37950). Spain, Canary Is., Gomera: Antoncayo, 500 m, Molero, 6-X-1993 (BCF38408). Spain, Canary Is., Tenerife: Ladera de GuÈimar, 700 m, Molero & PeÂrez de Paz, 11-VI-1990 (BCF37979). Spain, Canary Is., Gomera: Chorros de Epina, 1100 m, Molero, 10-VI1990 (BCF37982). ** Spain, Girona: CadaqueÂs, 20 m, Molero & Rovira, 30-IV-1989 (BCF37983) [AF334246, AF334246].

Location and Voucher

Table 1. Locality and vouchers of the studied populations of Euphorbia subsect. Pachycladae. The coded abbreviations of populations are used in Fig. 1, Fig. 3, Fig. 4 and Fig. 5, and Table 3. Those populations used for the molecular analysis of the ITS region are marked with **. Gen Bank accesions numbers are given in brackets (ITS-1 and ITS-2 respectively). Herbarium acronyms: BCF, Laboratori de BotaÁnica, Facultat de FarmaÁcia. Universitat de Barcelona. AZU, HerbaÁrio da Universidade dos AcËores, Terra ChaÃ, Angra do Heroismo

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae 113

ST TU

E. tirucalli L. E. tuckeyana Steud. ex Webb

± 20

44

20

RJ3

E. stygiana H. C. Watson

20 20

20

PI2 RJ1 RJ2

20

PI1

E. piscatoria Ait.

E. regis-jubae Webb & Berthel.

20

PE

44

LO4

E. pedroi Molero & Rovira

44

LO3

±

44

LO2

20

LM3 44

20

LM2

LO1

20

Turkey: Kas, 10 m, Blanche & Molero, 18-V-1989 (BCF38140). ** Spain, Barcelona: Sant Pol de Mar, Garnatje & Molero, 23-X-1999 (BCF49889) [AF334253, AF334268]. Spain, Canary Is., Tenerife: Guõ a de Isora, 230 m, Molero, 11-III-1990 (BCF37691). ** Spain, Canary Is., Gomera: Alojera, Molero & Rovira, 6-VI-1997 (BCF44443) [AF334244, AF334259]. Spain, Canary Is., La Palma: Tigalete Hondo, PeÂrez de Paz, 5-V-1997 (BCF49888). ** Spain, Canary Is., Gomera: Parque Nacional de Garajonay, cumbre de TagaqueÂ, Los Noruegos, 1200 m, Molero & Rovira, 7-VII-1997 (BCF44452) [AF334248, AF334263]. Portugal, Madeira Is., Madeira: ChaÃo dos Louros, 740 m, S. Fontinha, Molero & Rovira, 5-V-1990 (BCF38133). Portugal, Madeira Is., Madeira: Pico Ferreiro, 980 m, S. Fontinha, Molero& Rovira, 5-V-1990 (BCF38135). Portugal, Madeira Is., Madeira: Ribeiro Frio, 840 m, S. Fontinha, Molero & Rovira, 5-V-1990 (BCF38138). ** Spain, Balearic Is., Eivissa: Ses Margalides, Gradaille, VI-1990 (BCF38629) [AF334252, AF334267]. Portugal: Sesimbra, Cabo de Ares, 6 m, Molero & Rovira, 26-VII-1990 (BCF49873). ** Portugal, Madeira Is., Madeira: Ribeira Brava, 230 m, S. Fontinha, Molero & Rovira, 5-V-1990 (BCF37984) [AF334245, AF334260]. Portugal, Madeira Is., Porto Santo: Pico Ana Ferreira, 240 m, Molero & Rovira, 7-V-1990 (BCF38009). Morocco: Sa®, 30 m, Molero & Vicens, 13-V-1994 (BCF49887). Spain, Canary Is., Lanzarote: Los Valles, 520 m, J.M. Montserrat, 12-VI-1990 (BCF37855). Spain, Canary Is., Gran Canaria: Sta. Ma de Guõ a, Molero & Rovira, 30-V-1999 (BCF39362). ** Portugal, AcËores Is.: Pico Landroal, 825 m, 5-VIII-1991, E. Dias (AZU2816) [AF334247, AF334262]. ** Unknonw origin, culta in Bot. Gard. Barcelona [AF334251, AF334266]. ** Cabo Verde, SaÃo Tiago: Assomada, culta in C.I.T.A. Tenerife (BCF49886) [AF334240, AF334255].

18 ±

DE4 LM1

Location and Voucher

2n

Code

E. margalidiana Kuhbier & Lewej

E. longifolia Lam. (=E. mellifera Ait.)

E. lamarckii Sweet

E. helioscopia L.

Taxon

Table 1 (continued) 114 J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae and A2 asymmetry indices according to Romero (1986).

Results A. Molecular analysis Size and composition of the ITS region. Numerical results of the study are summarized on Table 2. The ITS1 and ITS 2 alignment of 14 taxa consisted of 519 positions and contained 131 phylogenetically informative substitutions and 43 phylogenetically informative indels. ITS 1 mean pairwise distances (as calculated by PAUP) within ingroup varied from 0% (between Euphorbia atropurpurea, E. regis-jubae and E. bravoana, and between E. stygiana and E. longifolia) to 7.9% (between E. aphylla and E. longifolia). Pairwise distance between ingroup and outgroup varied from 1.9% (between Euphorbia margalidiana and E. longifolia) to 7.4% (between E. margalidiana and E. atropurpurea). ITS 2 mean pairwise distances (as calculated by PAUP) within ingroup varied from 0% (between Euphorbia atropurpurea and E. regis-jubae) to 7.9% (between E. stygiana and E. bravoana). Pairwise distance between ingroup and outgroup varied from 1.2% (between Euphorbia margalidiana and E. longifolia) to 8.1% (between E. margalidiana and E. tuckeyana). Phylogenetic analysis. The analysis resulted in 15 most-parsimonious trees (MPTs) 238 steps long, all in one island. The strict consensus tree of all the MPTs is shown in Fig. 2. The Canarian group forms a robust monophyletic clade (95% BS, 9 DI), but this clade does not include E. balsamifera, which is placed instead in a basal position. Euphorbia dendroides is sister to the Canarian clade, with very high support (100% BS, 21 DI). The two lauroid treelets E. stygiana and E. longifolia form a well-supported clade (100% BS, 12 DI), to which the two representatives of E. subgen. Esula sect. Helioscopia Dumort. (E. margalidiana and E. helioscopia) are successive sisters with high support (96% BS and 9 DI, 100% BS and 17 DI, respectively).

115

B. Karyological studies Chromosome number, karyotypes and idiograms. The somatic chromosome numbers of the 36 populations studied belonging to 17 taxa of Euphorbia subsect. Pachycladae are given in Table 1. The chromosomal characteristics [karyotype length, length range, chromosomal formula, presence of satellites, Bazzichelli index (Bazzichelli 1967), Romero A1 and A2 asymmetry indices (Romero 1986) and Stebbins index (Stebbins 1971)] of the 20 chosen populations which represent the 17 taxa included in the sub-section to subspecies level are given in Table 3. Figure 3 shows the idiograms of each population represented in Table 3. Figure 4 illustrates mitotic metaphases of the majority of taxa that appear in Table 3. The detailed chromosomal characteristics of the karyotypes of the 20 populations studied, which represent 15 species and 2 subspecies, are not included, but they can be obtained from the ®rst author. A cluster showing the phenetic relationships among the karyotypes of the taxa studied is presented in Fig. 5. Discussion A. Phylogenetic analysis The phylogenetic analysis of the ITS sequences con®rms that Euphorbia subsec. Pachycladae, as de®ned by Boissier (1862), is not a natural group. On the contrary, it is a polyphyletic taxon within which it is possible to di€erentiate at least three clades of divergent evolution which coincide basically with the three groups we de®ned in the introduction on both morphological, ecological and biogeographical grounds. 1. The Mediterranean-Macaronesian clade. This clade encompasses the E. atropurpurea complex, E. lamarckii complex and Euphorbia dendroides, with strong support (BS ˆ 100%, DI ˆ 21). Euphorbia dendroides, located in the basal position, is shown to be sister to this well-supported Macaronesian clade (BS ˆ 95%, DI ˆ 9). The Macaronesian group emerges as a polytomy in a derived position with respect to

116

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

Fig. 1. Geographical distribution of studied taxa and populations. The code numbers assigned to the populations correspond to those of Table 1 Table 2. Numeric results of the nrDNA analysis Data Set

ITS1

ITS2

Total aligned characters (bp) Informative substitutions (bp) Informative indels (bp) Length of the indels (bp) Length of the informative indels (bp) Ingroup divergence (%) Ingroup-outgroup divergence (%)

290 73 32 1±44 1±35 0±7.9 1.9±7.4

229 58 11 1±4 1±2 0±7.9 1.2±8.1

G/C content, Mediterranean-Macaronesian clade (%) G/C content, lauroid clade (%) G/C content, E. balsamifera (%)

63.22 55 62

the Mediterranean stock, as occurred with other Macaronesian groups such as Argyranthemum Webb ex Sch. Bip. (Francisco-Ortega et al. 1997a, b) and Cheirolophus Cass. (Susanna et al. 1999). ITS is not capable of distinguishing between the age and derivation of the di€erent taxa, and this indicates the absence of synapomorphic mutations, from which it may be inferred that the genetic divergence between the species is very low and

the speciation is recent. One noteworthy point is that E. aphylla, traditionally considered as part of section Tirucalli, is con®rmed as belonging to the Macaronesian group, a ®nding that we will discuss later in the karyological section. The solidity of the clade formed by E. dendroides and the group of Canarian species suggests again that the western Mediterranean is the point of origin of the Canarian species, as has been shown for other Macaro-

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

nesian groups such as Aeonium Webb & Berth. (Mees et al. 1996; Van Ham et al. 1998), Argyranthemum (Francisco-Ortega et al. 1997a, b), Cheirolophus (Susanna et al. 1999) and Sonchus L. (Kim et al. 1996). 2. The lauroid clade E. longifolia-E. stygiana. The tree de®nes a robust monophyletic clade (BS ˆ 100% y DI ˆ 12) for these two species, which form another major clade along with E. margalidiana and E. helioscopia of the outgroup (both belonging to subgen. Esula sect. Helioscopia). Euphorbia helioscopia, an annual weed of Mediterranean origin, now sub-cosmopolitan, is sister to the remaining species in the clade (BS ˆ 100%, DI ˆ 17); the sub-¯eshy Mediterranean Balearic shrub E. margalidiana is itself sister (BS 96%, DI ˆ 9) to the E. stygiana-longifolia clade. The relations established by the clade must be seen in relation to the remaining Macaronesian groups: any interpretation in the broader context of the genus Euphorbia warrants the utmost caution, since the number of representatives analyzed is too small to draw de®nitive

117

conclusions. What can be deduced for this laurel forest group is a Mediterranean origin, though di€erent from the Macaronesian-Mediterranean clade related to E. dendroides, since the connection between the two clades is unsupported (BS 55%, DI ˆ 2). It should be realized that di€erentiation within this woody lauroid group may be more recent than expected and that it probably originated from a perennial herbaceous plant or a shrub. This again calls into question the relationship between a woody, original habit and a herbaceous, derived habit. Several cases have now appeared which illustrate the opposite direction, from a herbaceous taxon adapted to thermic and xeric conditions to a derived woody taxon in damp and uniform climates. An example is the Sonchus genus in the Canary Islands (Kim et al. 1996); or the pachycaul treelets of the genus Centaurodendron Johow in the Juan FernaÂndez archipelago, which come from Chilean herbaceous species of the genus Plectocephalus D. Don (Wagenitz and Hellwig 1996). In our case

Fig. 2. Strict consensus tree of the 15 most parsimonious trees generated by the ITS matrix. Length: 238 steps. Consistency index excluding uninformative characters: 0.673; retention index: 0.753; homoplasy index: 0.222. BAL ˆ E. balsamifera group

BAa2 BAb1 BAb4 BAs1

balsamifera balsamifera balsamifera balsamifera balsamifera

E. E. E. E. E.

adenensis balsamifera balsamifera sepium

LO1 LO2 ST

longifolia-E. stygiana model longifolia longifolia stygiana

E. E. E. E.

model subsp. subsp. subsp. subsp.

DE3

atropurpurea-E. lamarckii model anachoreta AN aphylla AP atropurpurea AT1 berthelotii BE2 bourgeauana BU1 bravoana BR1 lamarckii LM3 pedroi PE piscatoria PI1 regis-jubae RJ1 regis-jubae RJ3 tuckeyana TU

20 20 20 20

44 44 44

18

20 20 20 20 20 20 20 20 20 20 20 20

Code 2n

E. dendroides model E. dendroides

E. E. E. E. E. E. E. E. E. E. E. E. E.

Taxon

90,64 80,64 85,48 71,30

93,38 78,88 82,10

30,22

57,92 56,04 59,46 59.44 54,50 50,48 51,16 50,68 63,18 45,30 47,36 62,88

6.01±3.49 5.64±3.05 5,84±3,20 4.73±2.67

3.79±0.91 2.86±0.87 3.03±0.96

2.22±1.29

3.58±2.37 3.35±2.15 3.51±2.44 3.65±2.35 3,28±2,33 3,15±2.02 3.00±2.09 3,24±1,87 3.58±2.70 2.54±1.72 2.68±1.99 3.59±2.62

Karyotype Length range length (lm) (lm) st sm + 1 st sm st st sm + 2 st sm + 2 st sm + 2 st st st sm + 7 st

3 7 9 1 4 7 7 7 2 2 2

1M 3m 3m 3m

+ + + +

3 2 2 2

m + 6 st sm + 5 st sm + 5 st sm + 5 st

1 M + 8m + 12 sm + 1 st 1 M +10 m + 11 sm 14 m + 8 sm

9m

7 sm + 2m+ 1m+ 9 sm + 6 sm + 1m+ 10 sm 1m+ 1m+ 8 sm + 8 sm + 1m+

Chromosomal formula

26,64 31,96 29,12 29,41 25,80 30,01 28,67 29,00 27,67 28,55 29,65 25,74

S%

0.63 0,52 0,58 0.57 0,65 0,56 0.59 0.58 0.61 0.60 0.57 0.64

A1

0.10 0,12 0,10 0.12 0,09 0,13 0.11 0.15 0.08 0.10 0.10 0.08

A2

3A 3A 3A 3A 3A 3A 3A 3A 3A 3A 3A 3A

Steb.

5S 67,08 68,68 10S 66,06 2S 67,28

32,92 31,32 33,94 32,72

0.49 0,53 0,48 0,50

0.19 0,19 0,18 0,16

3A 3A 3A 3A

63,62 36,38 0.39 0.33 2A 61,25 36,55 0,36 0,28 2A 60,25 39,75 0,36 0,27 2A

2S 59,63 40,37 0.32 0.16 2A

3S 73,36 68,04 2S 70,88 3S 70,59 74,54 7S 69,99 71,33 5S 70,93 72,33 6S 71,45 6S 70,35 74,26

Sat L%

Table 3. Karyological data of the populations studied. The code used to determine studied populations in Table 1. Abbreviations are as follows: 2n, somatic chromosome number; chromosomal formulae are according to Levan et al. (1964); Sat, satellite (e.g. 8 S, indicates to the satellite observed in the short arm of the chromosome pair numbers; L% and S%, indices that express the contribution of each arm of each chromosome to the length of the karyotype, according to Bazzichelli (1967); A1 and A2 , Romero (1986) asymmetry indices; Steb., Stebbins symmetry type

118 J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

119

Fig. 3. Haploid idiograms of the 20 studied populations of Euphorbia subsect. Pachycladae. For each taxon, population code is indicated between parentheses. The code is the same used in Tables 1 and 2 and in Fig. 3. Scale bar =3 lm.

molecular data also appear to support Carlquist's hypothesis (1970, 1974) that the predominance of woody endemics on islands arises from increased woodiness as a response to the uniformity of insular climates.

3. Euphorbia balsamifera appears in a basal position between the outgroup species, with no relation to the remaining species in the sub-section. Nor does it show anities with the outgroup of chosen Mediterranean

120

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

Fig. 3 (continued)

Euphorbia, nor with E. tirucalli, which occupies an isolated position in the tree. B. Karyological data Chromosome numbers. Of the 36 populations studied in Table 1, the numbers obtained for

E. balsamifera subsp. balsamifera, E. balsamifera subsp. adenensis, E. pedroi, E. longifolia and E. stygiana constitute a new scienti®c ®nding. The remaining chromosome numbers coincide, broadly speaking, with bibliographical references in indices of plant chromosome counts (Bolkhovskikh et al. 1969; Moore 1974,

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

121

Fig. 4. Mitotic metaphases. a E. anachoreta (AN); b E. aphylla (AP); c E. balsamifera subsp. balsamifera (BAb4); d E. balsamifera subsp. adenensis (BAa2); e E. balsamifera subsp. sepium (BAs1); f E. berthelotii (BE2); g E. bourgeauana (BU2); h E. bravoana (BR); i E. lamarckii (LM3); j E. longifolia (LO3); k E. regis-jubae (RJ3); l E. stygiana (ST); m E. tuckeyana (TU). Scale bar: 3 lm

122

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

Fig. 5. Cluster analysis showing the relationships between the karyotypes of the studied populations

1977; Missouri Botanical Garden Cumulative Database W3 TROPICOS [http//mobot.mobot.org/egi-bin/search_pick]) and in studies dealing more speci®cally with the Canary Islands (ArdeÂvol et al. 1993) and Cape Verde Islands (Brochmann et al. 1997). Some of the data proposed in the bibliography, however, do not match ours and deserve comment. Thus we have detected identi®cation errors with E. lamarckii (western Canary Islands) being mistaken for E. regis-jubae (eastern Canary Islands) and vice versa. These are easy to correct when the place of origin is know, since all these species are allopatric. In this respect, the diploid number (2n ˆ 20), which is here indicated for the ®rst time for E. lamarckii, coincides with the gametic number (n ˆ 10) obtained by Linder and Lambert (1965) using Tenerife populations identi®ed as E. regis-jubae. The diploid number indicated herein for E. bourgeauana (2n ˆ 20) in Tenerife (GuÈimar) coincides with that o€ered by Bramwell et al. (1972) for Buenavista, though in fact the taxon studied by these authors was E. atropurpurea forma lutea (Santos 1988). With respect to E. piscatoria (2n ˆ 20), the diploid number we obtained does not coincide with the tetraploid number (2n ˆ 40) o€ered by Aldridge and

Ortega (1976). For E. longifolia ( ˆ E. mellifera), the number obtained (2n ˆ 44) does not match the one reported by Aldridge and Ortega (1976) and Dalgaard (1991), who indicate 2n ˆ 40 chromosomes for this species. For E. balsamifera s. l., there is a chromosome count of 2n ˆ 20 from Senegal (MieÁge 1960) which coincides with our data for E. balsamifera subsp. sepium in Mali. The gametic number n ˆ 21 obtained by Brunel and Laplace (1977) on materials from Togo determined as E. balsamifera requires con®rmation, since the determination could be erroneous. For E. dendroides (2n ˆ 18), our results coincide with many other previous counts. Basic chromosome number and ploidy levels. Euphorbia subsect. Pachycladae reveals three di€erent basic numbers (x ˆ 9, 10 and 11) and two ploidy levels (2x and 4x). Euphorbia atropurpurea complex and E. lamarckii complex have the same diploid level as E. aphylla, with a basic number of x ˆ 10. In these groups the chromosome number has remained constant during chromosome evolution, without giving rise to processes involving polyploidization or dysploidy, except in the Mediterranean species E. dendroides, a diploid with 2n ˆ 18 chromosomes. Euphorbia bals-

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

amifera also has a basic number of x ˆ 10 and a diploid level, though the karyotype structure indicates it has no direct relationship with the two preceding groups (Fig. 3). Euphorbia longifolia complex shows a tetraploid level (2n ˆ 44) with a basic number of x ˆ 11 and is unrelated to the remaining Macaronesian groups. This basic number is extremely rare for this genus (Perry 1942, 1943). Brochmann et al. (1997) suggest that x ˆ 10 in groups of Macaronesian endemics could have arisen from the tetraploidization of the basic number x ˆ 6 (which they consider the basic number of the genus) followed by descending dysploidy, along the lines of the model described by Borgen (1997) for Lobularia canariensis. This hypothesis is not easy to prove: moreover there is no agreement about the primary basic number in the genus Euphorbia. Perry (1942, 1943) considers it to be x ˆ 8; Bolkhovskikh et al. (1969) propose x ˆ 6; and Hans (1973) suggests x ˆ 7. But x ˆ 10 is the basic number of various sections of the genus and is becoming increasingly frequent as more chromosome counts on Euphorbia become available. Some authors (Webster 1967, Mehra and Choda 1978, Wiebecke 1989) consider it the oldest for the genus, as it is usually associated with the ``cactiform'' and ``coralloid'' forms which are considered archaic (Croizat 1958) and are distributed in very old areas of relict ¯ora in South Africa, East Africa and Madagascar. The basic number x ˆ 10 is the most widespread among the endemics of the African continent and Macaronesia. In section Tirucalli Boiss., which is morphologically close, diploids with a base of x ˆ 10 are also frequent. But the origin of this basic number may be di€erent according to the taxonomic group and geographical origin. For instance x ˆ 10 is widespread in section Esula, of Eurasian distribution, some species of which may be closely related to subsect. Pachycladae (Kuzmanov 1964). Phenogram, karyotypes and idiograms. The phenetic dendrogram based on karyological results (Fig. 5) shows considerable resemblance with the consensus tree of ITS sequenc-

123

es (Fig. 2) and yields very similar results. The phenogram presents four main branches, differentiated by their greater Euclidean distance, in which the karyotypes are grouped together in four di€erent structural models. 1. Euphorbia atropurpurea-E. lamarckii model. A ®rst model comprises the same species as the monophyletic clade of the ITS tree de®ned by E. atropurpurea complex and E. lamarckii complex, with the exception of E. dendroides, which is removed very far from the rest. In the phenogram, the taxa belonging to this model form a vaguely heterogeneous cluster in which the populations studied from the two complexes are mixed at random, so that neither karyology nor ITS sequences analysis distinguish between the two morphological groups. These are schizo-endemics (sensu Favarger and Siljak-Yakovlev 1986) and the di€erent taxa present very similar karyotypes (Fig. 4). Figure 3 enables us to de®ne a common model for the species belonging to this group: moderately small chromosomes, among which submetacentric chromosomes (70±100%) predominate, with a few metacentric (0±20%) and few subtelocentric chromosomes (10±30%, except in E. tuckeyana, where they are predominant). Table 3 reveals that the chromosomal formulas vary somewhat between the di€erent taxa because many pairs lie on the boundaries of the categories, though they correspond to the same model, as can be seen from the idiograms (Fig. 3). In 40% of the populations studied the short arm of di€erent pairs presents a satellite which is not visible in all the plates studied. According to Romero's indices, intrachromosomal asymmetry (A1) is moderately high with few oscillations (0.57±0.64) between the di€erent taxa; on the other hand, interchromosomal asymmetry (A2) is very low (0.08±0.22). All the taxa fall within Stebbins's category 3A (1971). In the tree (Fig. 5), the Canarian endemic Euphorbia aphylla, which has usually been included in section Tirucalli on account of its morphological characteristics, falls within the cluster de®ned for this model. The chromo-

124

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

some characteristics of this species, which basically coincide with the data contributed by Wiebecke (1989), do not di€er substantially from the other Macaronesian representatives of the group. In contrast Wiebecke (1989) highlights the karyological di€erences between this species and other representatives of section Tirucalli (E. tirucalli L. and E. stenoclada Baill.) from East Africa and Madagascar, notably in terms of their di€erent nuclear structure, di€erent meiotic behaviour and the absence in the latter species of subtelocentric chromosomes; karyotype symmetry is greater in this East African group, with metacentric and submetacentric chromosomes clearly predominant. These results tally with the ITS phylogeny, in which E. tirucalli, unlike E. aphylla, is not related to the Macaronesian clade or the Mediterranean species of the outgroup. Intersectional natural hybrids of E. aphylla ´ E. lamarckii (E ´ jubaephylla Svent.) have been described from Tenerife: these are totally viable and present di€erentiated phenology. In Gran Canaria, E. aphylla crosses with E. regisjubae, also producing stable hybrids (E. ´ marreroi Molero & Rovira ined.). Euphorbia ´ petterssonii Svent. is the hybrid of E. atropurpurea ´ E. aphylla, described in Tenerife (Teno). The presence of these hybrids shows the weakness of the reproductive barriers and the genetic anities between this species and representatives of subsect. Pachycladae. Though it does not live in the most arid habitats of the Canary Islands, its ¯eshy, aphyllous habit might be the result of a process of adaptative convergence towards the conditions of the habitat: xeric, thermic slopes and cli€s close to the sea, exposed to strong, moist, salt-laden winds. 2. Euphorbia dendroides model. This species poses some intriguing karyological problems. This Mediterranean species is closely related to the E. lamarckii complex both morphologically and in terms of ITS analysis. On the other hand, in terms of chromosome characteristics it occupies an isolated position in the phenogram (Fig. 5). It is found to be closer to the E. longifolia complex on account

of the symmetry of its karyotype, which matches Stebbins's model 2B (1971), though its origin is very di€erent. The karyotype (Fig. 3) is made up of small chromosomes (1.29±2.22 lm) whose overall length is substantially lower than in related Canary Islands species. The much more symmetrical complement consists of 9 metacentric pairs. Intrachromosomal asymmetry (A1) is much lower than in Macaronesian representatives of the group. Our results coincide basically with those of Wiebecke (1989), except in the number of satellite-bearing pairs observed. The latter author also found other structural di€erences between the interphase nuclei of E. dendroides and E. cf. regis-jubae-E. piscatoria (number and shape of the chromocentres). There has been a widespread assumption that evolutionary progress in angiosperms is associated with an increase in karyotype asymmetry and a decrease in the number of chromosomes (Stebbins 1971, El-Lakany and Dugle 1972). Nowadays, however, these concepts are being questioned and it is becoming apparent that they do not have the same value in the di€erent taxonomic groups (Stace 2000). Euphorbia dendroides features a more symmetrical karyotype than the taxa of Macaronesian stock (archaic character) and a basic number (x ˆ 9) which presumably derived from the older basic number, common to the Macaronesian group, of x ˆ 10. One explanation for these karyological di€erences is that E. dendroides might be related, not only to the Canarian group, but to other North African or Eurasian stocks which we have not studied, more precisely with certain representatives of section Esula, which have the same basic number and ploidy level and present a degree of morphological anity. But the more pronounced morphological anity and ITS analysis both relate it to the E. lamarckii complex, of which it turns out to be sister (Fig. 2). It is possible to postulate a process of descending dysploidy (x ˆ 10 ® x ˆ 9) from a hypothetical late Tertiary African ancestor which it shared with the Macaronesian group. In the Macaronesian group this dysploidy did not occur,

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

owing perhaps to the greater stability of the island habitat. 3. Euphorbia longifolia-E. stygiana model. In the phenogram (Fig. 5) the species of the lauroid group form an isolated cluster with the greatest Euclidean distance of all the groups, except E. dendroides, with which they share the same type of 2B asymmetry. From the cytogeographical point of view, it is a karyologically isolated tetraploid complex with a base of x ˆ 11, which can be catalogued as paleopolyploid using the endemicity criterion developed by Favarger and Contandriopoulos (1961), or as schizo-endemic in terms of the more precise nomenclature adopted by Favarger and Siljak-Yakovlev (1986). The three karyotypes studied correspond to the following outline: moderately small metacentric and submetacentric chromosomes (length range 0.87±3.79 lm), de®ning a fairly symmetrical karyotype (category 2A, Stebbins 1971). Comparative analysis of the two populations studied reveals certain diagnostic di€erences. The Canarian population of E. longifolia displays the least symmetrical karyotype (which is also somewhat longer), possessing a ®rst st pair, 8 m pairs and one pair of M chromosomes in terminal position. The Madeiran population shows a somewhat more symmetrical karyotype, with a ®rst sm pair and an increase in metacentric chromosomes to 10 pairs. Euphorbia stygiana presents the most symmetrical karyotype of the three, with a greater number of metacentric chromosomes (up to 14 pairs) and a ®rst metacentric pair. The ®rst pair of each karyotype is the most obvious di€erential trait for distinguishing between the three populations. If we accept the criterion that the most symmetrical karyotypes are the oldest in karyological evolution, a criterion which appears to be supported in the genus Euphorbia (Simon and Vicens 1999), our results demonstrate that E. longifolia in the Canary Islands has a more highly evolved karyotype than in Madeira and that both are more evolved than E. stygiana in the Azores. 4. Euphorbia balsamifera model. Despite its 2n ˆ 20 chromosomes, the structural di€er-

125

ences between the karyotype of E. balsamifera and the complex E. atropurpurea-E. lamarckii cause it to form an independent cluster in the phenogram (Fig. 5). This corroborates other morphological and molecular evidence and makes it possible to point to a di€erent origin. The karyotype is stable in its wide distribution area (Yemen ± sub-Saharan Africa ± Morocco ± Canary Islands) and there are hardly any di€erences between the populations studied, except the position of the satellitebearing pairs. The chromosomes are moderately small (length range 2.67±6.01 lm). The complement shows marked interchromosomal asymmetry, with 5 m-sm pairs and 5 st pairs. In subsp. balsamifera there are practically no di€erences between the Tenerife and Moroccan populations (except for the presence of a satellite-bearing pair in the Moroccan population). The karyotype of subsp. sepium is very similar to subsp. balsamifera, except for the presence of a satellite in pair II. Subsp. adenensis also shows a very similar karyotype, with the exception, nuanced by oscillation between the category limit values, of pair I (M by sm) and pair V (st by sm). Pair Y also has a satellite. In terms of symmetry, all the populations ®t into Stebbins's category 3A (1971). Because of their karyological resemblance and allopatric geographical distribution the three taxa can be considered schizoendemics. This model presents certain analogies with that described by Vosa and Bassi (1991) for certain species of Euphorbia (subgenus Euphorbia) from South Africa, which are succulent, spiky and diploid (2n ˆ 20) and display high interchromosomal bipolar asymmetry with 4±5 metacentric pairs of chromosomes and 4±5 subtelocentric-telocentric pairs. This bipolar model is repeated for other diploid species of subgenus Euphorbia (2n ˆ 20) studied by Wiebecke, such as E. resinifera Berg, E. monteiroi Hook., E. ferox Marloth, E. valida N. E. Br. and E. globosa Sims. These bimodal karyotypes seem to suggest that subgenus Euphorbia has a very ancient homoploid hybridogenous origin, with ancestors that have now probably

126

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

disappeared. Euphorbia balsamifera presents other analogies with the spiky, cactiform Euphorbia, such as the structure of the thick walled laticifers, and the morphology of the bone-shaped, discoid starch grains (Vindt 1953, Rao and Prasad 1986). C. Biogeographical implications With reference to the group of endemic Euphorbia studied herein, it was con®rmed that the colonization of the Macaronesian islands and related biogeographical territories involved at least three introductions in di€erence periods, these being necessary to explain the presence of three clades in the ITS sequence analysis. 1. The Macaronesian-North African-Mediterranean group: E. atropurpurea-E. lamarckii complex. Whereas ITS analysis indicates that this group is monophyletic, it actually turns out to be a polytomy (Fig. 2); genetic divergence between the species is low and the speciation process is probably recent. These results are coherent with karyological evolution: there are no substantial di€erences between the karyotypes of the di€erent species, all of which correspond to the same model in terms of the stability of their chromosome number, the relative length of the genome and the degree of symmetry of the karyotype. This group presents to the same pattern of eco-geographical di€erentiation as other genera strongly di€erentiate in the islands, such as Aeonium, Argyranthemum, Sonchus, etc. (Francisco-Ortega et al. 2000). The latter reveal two complementary speciation models in two different Macaronesian scenarios: a) interinsular vicariance by expansive colonization in similar habitats, and b) adaptative intrainsular radiation in suciently stable islands with a marked ecological gradient. These two speciation processes, as suggested by the polytomy of the ITS analysis, must have been rapid and simultaneous. The existence of a thermophilous, lowland North African-Mediterranean ancestor, from which E. dendroides in the Mediterranean might also descend, could be postulated.

The ®rst evolutionary entity are the lowlands thermophilous species: the E. lamarckii complex (E. obtusifolia complex sensu Molero and Rovira 1998). This group is a clear example of speciation by interinsular vicariance. Few morphological di€erences separate the taxa of this group and habitat specialization is very similar for all, with minor adaptations to di€erent microclimates. The species belonging to the group have colonized mainly hot and relatively arid areas of the basal belt of the Macaronesian Islands and the Atlantic coast of North Africa. In the larger islands, however, they display greater ecological amplitude and have adapted to more mesophilous habitats. The speciation process occurred relatively recently, probably following the same model of rapid speciation inferred for Sonchus and Argyranthemum (Kim et al. 1996; Francisco-Ortega et al. 1997a, b). Current evidence indicates that the territory of the E. lamarckii complex has been split between two large groups of species whose distribution is connected with climatic and stochastic factors. A ®rst group, related to E. regis-jubae, has been di€erentiated: In the Eastern Canary Islands and on the Atlantic coast of Morocco, the Fuerteventura, Lanzarote and Gran Canaria populations are morphologically homogeneous and colonize the same lowland habitats, though on Gran Canaria the species colonizes the entire island, practically to the summit, and displays greater morphological plasticity and adaptation to the altitude gradient. It is noteworthy that adaptative radiation and speciation have not occurred in the group on Gran Canaria whereas in Tenerife they have, although the orographic complexity of the two islands is comparable. There may be various reasons for this. One is geological instability leading to more recent eruptions that may have wiped out ancient colonizations. The other is the more unstable, arid climate of Gran Canaria, which is farther away from the in¯uence of the trade winds than the western island and Madeira (Hobohm 2000). This aridity causes arboreal vegetation to be absent

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

and explains why there are no notable gaps in the ecological gradient. The result is that high genetic ¯ow between populations is maintained and this homogenizes and slows down or prevents di€erentiation, as happens in the larger islands of the Cape Verde archipelago (Brochmann et al. 1997). Regarding African populations of E. regis-jubae [described as E. regis-jubae var. pseudo-dendroides (H. Lindb.) Maire] and the Portuguese E. pedroi, our present data do not allow any inference on the speciation processes of these taxa and the direction of the colonization of the eastern Canary Islands. The second group is formed by E. lamarckii, E. piscatoria, E. anachoreta and E. berthelotii, which are as a whole a geographical vicariant of the E. regis-jubae group in the western Canary Islands. These species has colonized similar habitats in the islands of Tenerife, La Gomera, La Palma and El Hierro, probably from a single colonization in Tenerife-La Gomera. Morphological evidence suggests that E. anachoreta of the Salvage Islands and E. piscatoria of Madeira derive from the western Canary Island populations by longdistance dispersal: they share the same type of small, deciduous sub-cyathial bracts, which constitute an excellent marker. Euphorbia. piscatoria shows minimal morphological divergence from E. lamarckii (diagnostic traits are con®ned basically to capsule size and seed shape) and its derivation is clear. Where E. anachoreta is concerned, the founder e€ect has been followed by a rapid speciation process by insular isolation with a pronounced genetic drift, a process very similar to that which occurred with E. margalidiana in the western Mediterranean (Balearic Islands). Euphorbia berthelotii is an endemic that occupies most of La Gomera, but it is allopatric with respect to E. lamarckii, which has settled on the northwestern strip of the island (opposite the south coast of Tenerife) probably as a result of later colonization. Of all the species in this group, E. tuckeyana displays the widest morphological di€erentiation. Analysis of the ITS region fails to distinguish it from its Canarian and

127

Madeiran brothers and its karyotype matches the same symmetry model. This is another example of a genus which has showed strong divergence in the Canary Islands but is represented by only one species in Cape Verde (Brochmann et al. 1997). The model strongly resembles that of Sonchus daltonii Webb, for which Kim et al. (1996) suggest colonization by long-distance dispersal from Canarian ancestors. Nonetheless, the islands are far away from one another (approximately 1600 km) and the dispersal mechanisms of Euphorbia are not comparable to those of Sonchus. Another hypothesis proposed by Brochmann et al. (1997) is that at some period in the Quaternarian a possible ancestor of MediterraneanNorth African origin could have emigrated southward, ¯eeing from climatic change, to the western coasts opposite Cape Verde. Subsequently it could have colonized the islands and disappeared from the continent. This hypothesis cannot be ruled out until the possible phylogenetic connection between the Macaronesian complex and the Cape Verde complex related to E. mauritanica L. are checked by fresh analyses. The E. atropurpurea complex is a product of two processes, adaptative radiation and interinsular vicariance,in the radiation centre of the genus that is essentially con®ned to the islands of Tenerife and La Gomera. The volcanic stability of these islands, especially La Gomera, since the Pliocene (Hobohm 2000), made it possible for these more specialized products of ecological adaptation to be preserved. Euphorbia atropurpurea occupies a broad strip of the southern and eastern slope of Tenerife. Its interinsular vicariant in La Gomera, E. bravoana, has a much more limited distribution on the NW slope of the island. Both live in mesoxerophilous or mesophilous habitats. Euphorbia bourgeauana lives in more mesophilous and even meso-hygrophilous habitats (Santos 1988), but has not di€erentiated between Tenerife and La Gomera, probably because it arose from rather more recent evolution. These taxa must have evolved from the same lowland ancestor that gave rise to

128

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

E. lamarckii, though the more marked morphological di€erentiation is indicative of a more active speciation process, arising from adaptation to very di€erent habitats. Though available molecular data do not permit us to establish derivations and karyological data provide no additional information, the morphological characters, which are segregated with respect to the E. lamarckii group, may be seen as adaptative and derived. The gradual concrescence of the sub-cyathial bracts, along with their persistence and increased size, and the growing complexity of the syn¯orescence (ranging from a simple pleiochasium in E. lamarckii complex to an always double pleiochasium in E. atropurpurea complex), are highly illustrative derived characters. 2. Lauroid group E. longifolia-E. stygiana. The well-supported monophyly of this group of hygrophytes indicates recent divergence between these two species, which come from a taxonomically isolated stock. The morphological di€erence between them is slight, the di€erences being basically con®ned to the leaves, which are more lauroid in E. stygiana as a speci®c adaptation to the moister, fresher habitat of the laurel forest of the Azores. In principle the lauroid habit is considered ancient, though a wide variety of ¯oristic elements currently live together in laurel forest, some of which derive from the ¯ora of the Lower Tertiary while others are much more recent. One known feature of this group is its Mediterranean origin, which is di€erent from the other group of Macaronesian xerophyte and mesophyte Euphorbia related to section Helioscopia of subgenus Esula. Regarding the karyological evidence, if we accept that the most symmetrical karyotypes are the most primitive and can be associated with an earlier colonizing process, our results point to a southward colonization from the Azores to the western Canary Islands, via Madeira. However no migratory ¯ow from the Azores to the Canary Islands has been reported in the bibliography for laurel forest species, as far as we are aware, let alone for stocks of Mediterranean origin. Existing data on the

direction of colonization between the Macaronesian archipelagos for laurel forest plants, such as the results obtained by CueÂnoud et al. (2000) about Ilex, are not very conclusive. Speciation in this group has taken the form of divergence by interinsular vicariance followed by colonizing radiation in the Western Canary Islands and the Azores archipelago. Colonization between islands has been veri®ed by long-distance dispersal (Cain et al. 2000). Accidental ornithochory can be assumed, since the seeds of Euphorbia with their smooth, non mucilaginous episperm and the presence of a caruncle are only adapted to short-distance dispersal by explosive dehiscence of the capsule (primary dispersal), followed by myrmecochory (secondary dispersal), according to Baiges et al. (1991). The precise mechanism for these laurel forest species is unknown. 3. Euphorbia balsamifera belongs to the same Canarian-African-Arabian biogeographical element as other genera of similar distribution, such as Aeonium, Caralluma R. Br., Ceropegia L., Dracaena Vand. or Euphorbia sect. Aphyllis Webb & Berthel. (Bramwell 1976, 1985). The absence of molecular data makes dicult to establish the way of colonization of this species. However, observation of its present distribution and karyotype, which is very stable throughout its area, makes it possible to speculate about its origin and speciation model. The three subspecies have a very similar karyotype except for the di€erent distribution of the satellites, which usually have an adaptative meaning (Fig. 3). However, subsp. adenensis presents a somewhat more symmetrical, primitive karyotype, which may suggest that the species originated in the high plateaux of the Yemen. The probable speciation model was by geographical vicariance from an ancestor which radiated in a warm period (the late Tertiary?) and which in the cold periods of the early Quaternary experienced a fragmentation of its area, seeking refuge in the most distant points (the Canary Islands and the high plateaux of the Yemen), and undergoing genetic drift and slight di€erentiation. The

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

expansion of the desert in the late Quaternary probably made possible the recovery of the African area and speeded up di€erentiation in western Africa. Morphological and karyological anities relate subsp. sepium more closely to subsp. balsamifera. D. Taxonomical implications The characters of its syn¯orescence and cyathium caused Boissier (1862) to place § Pachycladae in subgenus Esula, near § Carunculares [ ˆ sect. Carunculares (Boiss.) Tutin] and § Galarrhoei [ ˆ sect. Helioscopia Dumort subsect. Galarrhoei (Boiss. in DC.) Pax in Engler]. After demonstrating the polyphyly of this taxon, it seems appropriate to summarize the anities of each group and suggest a suitable taxonomical restructuration, in accordance with the results of this study. The taxonomical position of E. balsamifera is still uncertain. Its morphological and genetic particularities, which are not closely linked to any of the groups studied, make it advisable to maintain an independent status for it at section rank ± as formerly proposed by Webb and Berthelot (1844±1850) ± within subgenus Esula, until its relations with other African and Arabian Euphorbia belonging to subgenera Tirucalli (Boiss.) S. Carter and Lyciopsis can be investigated. The complexes E. atropurpurea and E. lamarckii display great uniformity in their genetic and karyological characters, possess common morphological traits and occupy a common Macaronesian area; they can be grouped together in an independent taxon at section level within subgenus Esula. Euphorbia dendroides and probably E. aphylla should be included in the new section by reason of their obvious morphological anities and those established by ITS sequence analysis in this study. Even so, it would be advisable ®rst to elucidate the phylogenetic relationship of this group with the E. mauritanica complex of southern Africa and the E. plumerioides complex of Oceania and Malaysia, about which little is yet known.

129

Finally the lauroid group, E. longifolia E. stygiana, constitutes a clearly Macaronesian group, with noteworthy morphological particularities, karyologically isolated and genetically di€erentiated from the other Macaronesian groups, but with morphological and evolutive anities with Mediterranean representatives of section Helioscopia. It seems appropriate to group these two species together in an independent taxon at sectional level within subgenus Esula with which they display the greatest anities. AÂngel FernaÂndez-LoÂpez, Eduardo Dias, Susana Fontinha, AÂguedo Marrero, Jose Ma Montserrat, Pedro L. PeÂrez de Paz and Arnoldo Santos provided us with logistic assistance in the ®eld work and supplied various samples for molecular and karyological studies. Our grateful thanks to Miquel Veny for his scrupulously careful maintenance of the collections of Euphorbia in the glasshouses of the Botanical Institute of Barcelona. Authors thank Dr. N. Zimmermann for his useful suggestions for improving the manuscript. This study was funded by projects DIGCYT- PB87/10008 and PB97/1134 of the Spanish Ministry of Education and Science and by grant 1999SGR 00332 for Consolidated Research Groups from the Catalan autonomous government (Generalitat de Catalunya).

References Aldridge A. E., Ortega J. (1976) Estudios en la ¯ora de macaronesia: algunos nuÂmeros de cromosomas II. Bot. Macar. 2: 13±20. ArdeÂvol J. F., Borgen L., PeÂrez de Paz P. L. (1993) Checklist of chromosome numbers counted in Canarian vascular plants. Sommerfeltia 18: 1±59. Baiges J. C., Espadaler X., Blanche C. (1991) Seed dispersal in W Mediterranean Euphorbia species. Bot. Chron. 10: 697±705. Bazzichelli G. (1967) Studi del ciclo del Leucanthemum atratum (Jacq. ) DC. sensu amplo. Ann. Bot. Roma 29: 385±891. Bentzer B., Bothmer R. V., Engstrand L., Gustafsson M., Snogerup S. (1971) Some sources of error in the determination of arm ratios of chromosomes. Bot. Notiser 124: 65±78. Bianco P., Medagli P., Bedalov M., D'Emerico S., Mastropasqua L. (1991) ConsideÂrations sur Euphorbia dendroides L. espeÁce steÂno-meÂditer-

130

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

raneÂenne macaroneÂsienne. Bot. Chron. 10: 689± 696. Boissier E. (1862) Euphorbia L. In: De Candolle A. P. (ed.) Prodromus systematis naturalis regni vegetabilis, vol. 15(2). Treutel et WuÈrz, Paris, pp. 1±188. Bolkhovskikh Z., Grif V., Matvejeva T., Zakharyeva O. (1969) Chromosome numbers of ¯owering plants. Leningrad. Borgen L. (1997) Genetic di€erentiation in endemic Lobularia (Brassicaceae) in the Canary Islands. Nord. J. Bot. 16: 487±503. Bramwell D. (1976) The endemic ¯ora of the Canary Islands; distribution, relationships and phytogeography. In: Kunkel G. (ed.) Biogeography and ecology in the Canary Islands. The Hage, pp. 207±240. Bramwell D. (1985) ContribucioÂn a la biogeogra®a de las islas Canarias. Bot. Macar. 14: 3±34. Bramwell D., Humphries C. J., Murray B. G., Owens S. J. (1972) Chromosome studies in the ¯ora of Macaronesia. Bot. Notiser 125: 139± 152. Bremer K. (1988) The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795±803. Brochmann C., Rustan H., Lobin W., Kilian N. (1997) The endemic vascular plants of the Cape Verde Islands, W Africa. Sommerfeltia 24: 1± 356. Brunel J. F., Laplace A. (1977) IOPB chromosome number reports LVIII. Taxon 26: 565±577. Cain M. L., Milligan B. G., Strand A. E. (2000) Long-distance seed dispersal in plant populations. Amer. J. Bot. 87: 1217±1227. Carlquist S. (1970) Hawaii: a natural history. The Natural History Press, New York. Carlquist S. (1974) Island biology. Columbia University Press, New York. Carter S., Lobin W., Radcli€e-Smith A. (1984) A collection of notes on Euphorbiaceae from the Cape Verde Islands. Senckenbergiana Biol. 64: 429±451. Croizat L. (1958) An essay on the biogeographic thinking of J. C. Willis. Arc. Bot. Biogeogr. Ital. 34: 90±116. CueÂnoud P., del Pero M. A., Loizeau P. A., Spichiger R., Andrews S., Manen J. F. (2000) Molecular Phylogeny and Biogeography of the genus Ilex L. (Aquifoliaceae). Ann. Bot. 85: 111± 122.

Dalgaard V. (1991) Chromosome studies in ¯owering plants from Macaronesia II. Willdenowia 20: 139±152. Deil U. (1991) Rock communities in tropical Arabia. In: Engel T., Frey W., KuÈkchner H., Cramer J. (eds.) Flora et Vegetatio Mundi IX. Berlin & Stuttgart, pp. 175±187. Donoghue M. J., Olmstead R. G., Smith J. F., Palmer J. D. (1992) Phylogenetic relationships of Dipsacales based on rbcL sequences. Ann. Missouri Bot. Gard. 79: 333±345. Doyle J. J., Doyle J. L. (1987) A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. 19: 11±15. El-Lakanny M. H., Dugle R. (1972) DNA content in relation to phylogeny of selected boreal forest plants. Evolution 26: 427±434. Favarger C., Contandriopoulos J. (1961) Essai sur l'endeÂmisme. Ver. Schweiz Bot. Ges. 71: 383± 408. Favarger C., Siljak-Yakovlev S. (1986) AÁ propos de la classi®cation des taxons endeÂmiques baseÂe sur la cytotaxonomie et la cytogeÂneÂtique. In: Colloque International de Botanique PyreÂneÂenne, La Cabanasse (PyreÂneÂes±Orientales). SocieÂte Botanique de France, Groupement Scienti®que Isard, pp. 287±303. Felsenstein J. (1985) Con®dence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783±791. Forster P. I. (1994) Revision of Euphorbia plumerioides Teisjsm. ex Hassk. (Euphorbiaceae) and allies. Austrobaileya 4(2): 245±264. Francisco-Ortega J., Santos-Guerra A., Hines A., Jansen R. K. (1997a) Molecular evidence for a Mediterranean origin of the Macaronesian endemic genus Argyranthemum (Asteraceae). Amer. J. Bot. 84: 1595±1613. Francisco-Ortega J., Crawford D. J., SantosGuerra A., Jansen R. K. (1997b) Origin and evolution of Argyranthemum (Asteraceae: Anthemideae) in Macaronesia. In: Givnidh T., Sytsma K. (eds.) Molecular evolution and adaptative radiation. Cambridge University Press, Cambridge, pp. 407±431. Francisco-Ortega J., Santos-Guerra A., Kim S.Ch., Crawford D. J. (2000) Plant genetic diversity in the Canary Islands: a conservation perspective. Amer. J. Bot. 87(7): 909±919. Govaerts R., Frodin D. G., Radcli€e-Smith A. (2000) World checklist and bibliography of

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae Euphorbiaceae (with Pandaceae). Royal Botanic Gardens, Kew. Hans A. S. (1973) Chromosomal conspectus of the Euphorbiaceae. Taxon 22: 591±635. Hansen A., Sunding P. (1993) Flora of Macaronesia. Checklist of vascular plants: 4. Revised edition. Sommerfeltia 17: 1±297. Hobohm C. (2000) Plant species diversity and endemism on islands and archipelagos with special reference to the Macaronesian Islands. Flora 195: 9±224. Kim S.-Ch., Crawford D. J., Francisco-Ortega J., Santos-Guerra A. (1996) A common origin for woody Sonchus and ®ve related genera in the Macaronesian islands: molecular evidence for extensive radiation. Proc. Natl. Acad. Sci. USA 93: 7743±7748. Kuzmanov B. (1964) On the origin of Euphorbia subg. Esula in Europa (Euphorbiaceae). Blumea 12(2): 369±379. Levan A., Fredga K., Sandberg A. (1964) Womenclature for centromeric position on chromosomes. Hereditas 52: 201±220. Linder R., Lambert A. M. (1965) EÂtude caryologique d'endeÂmiques canariennes. Bull. Soc. Bot. France 112: 234±238. LoÈve A., LoÈve D. (1975) Plant chromosomes. Cramer, Vaduz. Maddison D. R. (1991) The discovery and importance of multiple islands of most parsimonious trees. Syst. Zool. 40: 315±328. Mehra P. N., Choda S. P. (1978) Cyto±taxonomical studies in the genus Euphorbia L. Cytologia 43: 217-235. Mes T. H. M., van Brederode J., Hart H. 'T. (1996) Origin of the woody Macaronesian Sempervivoideae and the Phylogenetic Position of the East African of Aeonium. Bot. Acta 109: 47±491. MieÁge J. (1960) Nombres chromosomiques de plantes d'Afrique Occidentale. Rev. Cytol. Biol. VeÂg. 21: 373±384. Molero J., Rovira A. (1998) A note on the taxonomy of the Macaronesian Euphorbia obtusifolia complex (Euphorbiaceae). Taxon 47: 321± 332. Montelongo V., Rodrigo J. D., Bramwell D. (1984) Sobre la vegetacioÂn de Gran Canaria. Bot. Macar. 12±13: 7±50. Moore R. J. (1974) Index to plant chromosome numbers for 1972. Regnum Veg. 68.

131

Moore R. J. (1977) Index to plant chromosome numbers for 1973/74. Regnum Veg. 96. Nickrent D. L., Schuette K. P., Starr E. M. (1994) A molecular phylogeny of Arceuthobium (Viscaceae) based on nuclear ribosomal DNA internal transcribed spacer sequences. Amer. J. Bot. 81: 1149±1160. Pax F. (1891) Euphorbia L. In: Engler A., Prantl K. (eds.) Die natuÈrlichen P¯anzenfamilien, vol. 19. Leipzig, pp. 208±222. Perry B. A. (1942) Chromosome number relationships in the genus Euphorbia. Chron. Bot. 7: 413±414. Perry B. A. (1943) Chromosome numbers and phylogenetic relationships in the Euphorbiaceae. Amer. J. Bot. 30: 527±543. Rao K. S., Prasad M. N. V. (1986) Typology of latex starch grains of certain Euphorbiaceae and their possible signi®cance in systematics. Plant Syst. Evol. 160: 189±193. Rivas-Martõ nez S., Widpret W., Del Arco M., Rodrõ guez O., PeÂrez de Paz P. L,. Garcõ a-Gallo A., Acebes J. R., Dõ az T. E, FernaÂndez GonzaÂlez F. (1993) Las comunidades vegetales de la isla de Tenerife (Islas Canarias). Itinera Geobot. 7: 69± 374. Romero C. (1986) A new method for estimating karyotype asymmetry. Taxon 35: 526±530. Santos A. (1988) Notas sobre Euphorbia bourgaeana Gay ex Boiss in DC. y especies a®nes. Bot. Macar. 16: 29±36. Simon J., Vicens J. (1999) Estudis biosistemaÁtics en Euphorbia L. a la mediterraÁnia occidental. Institut d'Estudis Catalans. Arx. Secc. Ci. 122. 704 pp. Barcelona. SjoÈgren E. (1972) Vascular plant communities of Madeira. Bol. Mus. Munic. Funchal 26: 45± 125. Soltis D. E., Soltis P. S., Collier T. G., Edgerton M. L. (1991) The Heuchera group (Saxifragaceae): evidence for chloroplast transfer and paraphyly. Amer. J. Bot. 78: 1091±1112. Soltis D. E., Kuzo€ R. K. (1993) ITS sequence homogeneity within and among populations of Lomatium grayi and L. laevigatum (Umbelliferae). Molec. Phylog. Evol. 2: 166±170. Stace C. A. 2000. Cytology and cytogenetics as fundamental taxonomic ressources for the 20th and 21st centuries. Taxon 49: 451±477. Stebbins G. L. (1971) Chromosomal evolution in higher plants. Edward Arnold, London.

132

J. Molero et al.: Molecular phylogeny and karyology of Euphorbia subsect. Pachycladae

Susanna A., Garnatje T., Garcia-Jacas N. (1999) Molecular phylogeny of Cheirolophus (Asteraceae: Cardueae ± Centaureinae) based on ITS sequences of nuclear ribosomal DNA. Plant Syst. Evol. 214: 147±160. Swo€ord D. L. (1999) PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sinauer, Sunderland. Swo€ord D. L., Olsen G. J. (1990) Phylogeny reconstruction. In: Hillis D., Moritz C. (eds.) Molecular systematics. Sinauer Associates, Sunderland, pp. 411±501. Van Ham R. C. H. J., Hart H. 'T. (1998) Phylogenetic relationships in the Crassulaceae inferred from chloroplast DNA restriction±site variation. Amer. J. Bot. 85: 123±134. Vargas P., McAllister C., Morton C., Jury S. L., Wilkinson M. J. (1999) Polyploid speciation in Hedera (Araliaceae): phylogenetic and biogeogrphic insights based on chromosome counts and ITS sequence. Plant Syst. Evol. 219: 165±179. Vindt J. (1953) Monographie des EuphorbiaceÂes du Maroc. Anatomie. Trav. Inst. Sci. CheÂri®en 19. Vosa G., Bassi P. (1991) Cromosome studies in the Southern African ¯ora 95±102. The basic karyotype of eight species of succulent Euphorbia L. Caryologia 44: 27±33. Wagenitz G., Hellwig F. H. (1996). Evolution of characters and phylogeny of the Centaureinae. In: Hind D. J. N., Beentje H. G. (eds.) Compos-

itae: Systematics. Proceedings of the International Compositae Conference, Kew, 1994. Royal Botanical Gardens, Kew, pp. 491±510. Webb P. B., Berthelot S. (1844±1850) Histoire naturelle des õÃ les Canaries, 3 (2, 3). Paris. Webster G. L. (1967) The genera of Euphorbiaceae in the Southeastern United States. L. Arnold Arbor. 48: 303±430. White T. J., Bruns T., Lee S., Taylor J. (1990) Ampli®cation and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: Innis M., Gelfand D., Sninsky J., White T. (eds.) PCR protocols: a guide to methods and applications. Academic Press, San Diego, pp. 315± 322. Wiebecke W. (1989) BeitraÈge zur Karyosystematik von Euphorbia und verwandten Gattungen. PhD Dissertation, University of Wien.

Addresses of the authors: JuliaÁ Molero and Anna Rovira, Departament de BotaÁnica, Facultat de FarmaÁcia, Universitat de Barcelona, Avda. Joan XXIII s. n., E-08028 Barcelona, Spain. (E-mail: [email protected]). NuÂria Garcia-Jacas, Teresa Garnatje and Alfonso Susanna, Institut BotaÁnic de Barcelona (C.S.I.C.-Ajuntament de Barcelona), Av. Muntanyans, s. n. E-08038 Barcelona, Spain.