Multiple introductions from the Iberian peninsula are responsible for invasion of Crupina vulgaris in western North America
2002; Wiley; Volume: 154; Issue: 2 Linguagem: Inglês
10.1046/j.1469-8137.2002.00382.x
ISSN1469-8137
AutoresTeresa Garnatje, Roser Vilatersana, Cindy Talbott Roché, Núria Garcia‐Jacas, Alfonso Susanna, D. C. Thill,
Tópico(s)Plant and animal studies
ResumoNew PhytologistVolume 154, Issue 2 p. 419-428 Free Access Multiple introductions from the Iberian peninsula are responsible for invasion of Crupina vulgaris in western North America T. Garnatje, T. Garnatje Molecular Systematics, Botanic Institute of Barcelona (CSIC-ICUB), Av. Muntanyans s.n. E-08038 Barcelona, Spain;Search for more papers by this authorR. Vilatersana, R. Vilatersana Molecular Systematics, Botanic Institute of Barcelona (CSIC-ICUB), Av. Muntanyans s.n. E-08038 Barcelona, Spain;Search for more papers by this authorC. T. Roché, C. T. Roché Department of Plant, Soil and Entomological Sciences, University of Idaho, Moscow, ID 83844–2339, USASearch for more papers by this authorN. Garcia-Jacas, N. Garcia-Jacas Molecular Systematics, Botanic Institute of Barcelona (CSIC-ICUB), Av. Muntanyans s.n. E-08038 Barcelona, Spain;Search for more papers by this authorA. Susanna, Corresponding Author A. SusannaAuthor for correspondence: A. Susanna Tel: +34 93 3258050 Fax: +34 93 4269321 Email: [email protected]Search for more papers by this authorD. C. Thill, D. C. Thill Department of Plant, Soil and Entomological Sciences, University of Idaho, Moscow, ID 83844–2339, USASearch for more papers by this author T. Garnatje, T. Garnatje Molecular Systematics, Botanic Institute of Barcelona (CSIC-ICUB), Av. Muntanyans s.n. E-08038 Barcelona, Spain;Search for more papers by this authorR. Vilatersana, R. Vilatersana Molecular Systematics, Botanic Institute of Barcelona (CSIC-ICUB), Av. Muntanyans s.n. E-08038 Barcelona, Spain;Search for more papers by this authorC. T. Roché, C. T. Roché Department of Plant, Soil and Entomological Sciences, University of Idaho, Moscow, ID 83844–2339, USASearch for more papers by this authorN. Garcia-Jacas, N. Garcia-Jacas Molecular Systematics, Botanic Institute of Barcelona (CSIC-ICUB), Av. Muntanyans s.n. E-08038 Barcelona, Spain;Search for more papers by this authorA. Susanna, Corresponding Author A. SusannaAuthor for correspondence: A. Susanna Tel: +34 93 3258050 Fax: +34 93 4269321 Email: [email protected]Search for more papers by this authorD. C. Thill, D. C. Thill Department of Plant, Soil and Entomological Sciences, University of Idaho, Moscow, ID 83844–2339, USASearch for more papers by this author First published: 30 April 2002 https://doi.org/10.1046/j.1469-8137.2002.00382.xCitations: 13AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Summary • Crupina vulgaris is a relatively recent invader to grasslands and other open habitats in western North America. Like related Centaurea species, it was introduced from the Mediterranean region, where it does not exhibit ruderal behavior. Determining the number and sources of invasion founders allows fuller interpretation of colonization dynamics and recognition of potential intercontinental carriers, both critical factors for curbing the spread of invasive species. • We chose the molecular technique of random amplified polymorphic DNA (RAPD) to identify the number and sources of invasion founders from the eastern hemisphere, by comparing indigenous and invasive populations. • Our results indicated that the five North American populations derived from three or more successful invasion events whose founders originated in the Iberian peninsula. • Also inferred by the similarity clustering among eastern hemisphere populations is a more ancient origin of the genus to the east of the Mediterranean, a concept supported by the scarcity of suitable nonanthropic habitat in Spain. Its epizoochoric association with migratory movements of domestic herds suggests probable routes of migration first to southern Europe, then later to North America. Introduction Crupina vulgaris (Asteraceae: Cynareae) is a diminutive winter annual forb from the Mediterranean region, whose original name, Centaurea crupina L., indicates its close relationship with the genus Centaurea (Bremer, 1994; Garcia-Jacas et al., 2001). Upon discovery of a naturalized population of Crupina vulgaris in Idaho in 1968 (Stickney, 1972), scientists were alarmed by its invasive potential, because Centaurea comprises some of the most aggressive invaders of western North American arid and semiarid wildlands (e.g. C. diffusa, C. solstitialis, C. stoebe, C. virgata). As a new discovery, Crupina vulgaris was listed as a Federal noxious weed, and was subsequently found in three other states: California, Washington and Oregon. Current estimates of the infested area exceed 26 000 ha. On overgrazed rangelands, it forms dense, nearly pure stands (Miller & Thill, 1983), a concern because of its relatively low palatability to cattle, the primary resource use (Zamora et al., 1989). Because Crupina vulgaris is a tap-rooted annual that primarily infests steep south-facing slopes, the potential for soil erosion is aggravated when it replaces fibrous-rooted grass species, which have more persistent residue. Also, dense populations of this exotic invader reduce biodiversity in native ecosystems (Prather et al., 1991). History of C. vulgaris as invader in North America and Australia Contrary to previous reports that the first North American record of C. vulgaris was in Idaho in 1968 (Stickney, 1972), the earliest collections of this species date from Massachusetts (Sorrie & Somers, 1999). Specimens at the Harvard University Herbaria, collected by C. E. Perkins in 1877 and 1879 from Boston, and South Boston Flats, Suffolk County, Massachusetts, USA, indicate that C. vulgaris was among the numerous species introduced in ship’s ballast (NEBC, W. Kittridge, pers. comm.). Crupina vulgaris was not collected again and it does not appear in current floras, indicating that it failed to establish in the northeastern United States. Thus, the discovery in Idaho was the first successful introduction. Over the next two decades three additional populations were reported: in 1975 at Santa Rosa, Sonoma County, California (Miller & Thill, 1983), in 1984 at Lake Chelan, Chelan County, Washington (Alverson & Arnett, 1986), and in 1987 at Dry Creek, Umatilla County, Oregon (Couderc-LeVaillant & Roché, 1993). The Sonoma County population was declared eradicated by 1982 (Miller & Thill, 1983), but it was rediscovered 7 y later about 1 km from the original infestation (Davis & Sherman, 1991). In 1990 another new population was reported in Modoc County, California (Calif. Department Food & Agric. records). These five infestations, detected between 1968 and 1990 in four western States, remain widely disjunct from each other (Fig. 1), unconnected by any known satellite or transitional populations. Figure 1Open in figure viewerPowerPoint Location of Crupina vulgaris colonies in western North America. A report of the introduction and eradication of C. vulgaris in British Columbia during the 1980s (Zamora et al., 1989) appears to have been in error (R. Cranston, pers. comm.). In South Australia, where C. vulgaris was discovered in 1936 around a reservoir in Hope Valley near Adelaide, scientists believe it was introduced in packing material for the heavy construction equipment, imported from Europe, which was used to build the reservoir (Ising, 1937). Although it has persisted in this location for 65 y, it has not dispersed (Jessop & Toelken, 1986), and is not considered a weed in Australia (Lazarides et al., 1997). Multiple introductions suspected Although the means of intercontinental migration was not known, multiple introductions have been suspected in North America, due to morphological and phenological differences between populations (Couderc-LeVaillant & Roché, 1993; Roché, 1996). Morphological differences include cypsela size and pappus length: populations from Modoc and Chelan counties have uniformly smaller cypselas (19 mg) with shorter pappus (4–5 mm) and the remaining three populations (Idaho, Oregon, and Sonoma County, California) are characterized by a larger cypsela (36 mg) with a longer pappus (7–9 mm). The same differentiation among populations was seen in rosette characteristics when populations were grown in a uniform garden study. In this common garden, the shiny, dark green rosette leaves of Chelan plants grew close to the soil surface at all times, in contrast to the lighter green rosette leaves of plants from the Oregon and Idaho populations, which grew upward at an angle away from the soil, except during cold winter conditions (Roché, 1996). In the same study, plants from the Oregon and Idaho populations flowered up to 10 d earlier than the population from Chelan. The Chelan population also required longer vernalization at cold temperatures and was more sensitive to damage from high temperatures than the Oregon and Idaho populations (Roché, 1996). A pattern of isolated foci is consistent with relatively recent introductions and the species’ propagule dispersal system, which, relative to common ruderal species, is highly inefficient (Roché & Thill, 2001). As a self-compatible annual, still relatively early in the invasion process, C. vulgaris was a good candidate for use of a molecular marker technique to investigate the invasion genetics among North American populations and compare them with native populations. In the eastern hemisphere, C. vulgaris is distributed around the Mediterranean basin, with its western limits in the Iberian peninsula and Morocco, and extending across southern Europe, Crimea, northern Greece and Turkey, northern Iran and Iraq, to the Caucasus region and Uzbekistan, Turkmenistan and northeastern Afghanistan (Kupicha, 1975; Dittrich et al., 1980; Couderc-LeVaillant, 1984; Czerepanov, 1995). Objectives and RAPD analysis Isozymes have been used successfully to investigate genetic variation among and within populations of a variety of invading species, for example Bromus tectorum (Novak et al., 1993; Novak & Mack, 1993), Capsella bursa-pastoris (Neuffer & Hurka, 1999), and Lythrum salicaria (Strefeler et al., 1996). However, a limitation of isozyme analysis is that variation may be too low for detection of genetic diversity (Nissen et al., 1995), and preliminary studies showed that this was the case for C. vulgaris (T. Garnatje, unpublished data). Randomly amplified polymorphic DNA (RAPD) analysis has effectively identified genotypes and relationships among populations of a variety of organisms (Hadrys et al., 1992). RAPDs often represent the appropriate scale of genetic analysis for studies of colonizing plants, tracking migration pathways, identifying the number of introductions and linking them to ancestral populations (Stiller & Denton, 1995; Neuffer, 1996). Although RAPD markers are dominant, making them a less powerful tool than codominant isozyme markers, this limitation is offset by the greater number of loci that can be compared. Chloroplast DNA restriction fragment length polymorphism (RFLP) has been used to investigate more detailed within-population variability and maternal inheritance (Rowe et al., 1997). We selected RAPDs as the suitable scale of genetic detail to determine the number of introductions and source locations for an outcrossing, but self-compatible annual species. The objectives of this study were first to examine the appropriateness of RAPD-based DNA fingerprinting to characterize the genetic structure of the invasive C. vulgaris populations, second to determine the number of introductions responsible for the North American invasion, and third to identify the most probable source locations in the eastern hemisphere of the invasion founders. This information would be useful for recognizing potential intercontinental carriers and understanding colonization dynamics, as well as a base for various weed management strategies such as biological control, competitive vegetation and prevention tactics. Materials and Methods Plant material We sampled natural populations in Eurasia (Armenia, France, Greece, Iran, Italy, Spain, Switzerland and Uzbekistan) and North America, collecting a single cypsela per plant. Some populations were not used in the analysis because an insufficient number of seeds germinated. A few others were eliminated because clear, reproducible band patterns could not be obtained from their DNA. Collection sites of the 48 populations included in the study are shown in Table 1, with voucher specimens deposited at the Botanic Institute of Barcelona, Spain (BC). Seeds were germinated and DNA was extracted from a fully developed cotyledon of five plants from each population. A duplicate DNA extraction from one plant per population served as a control. Table 1. Plant collections of Crupina vulgaris used as sources of DNA. The depository for voucher specimens is the Botanic Institute of Barcelona, Spain (BC) Country and population Province or state Collector and number France Briançonnet Alpes Maritimes Roché & Susanna 2020 Briançon Hautes Alpes Korfhage & Roché 110 Gap Hautes Alpes Korfhage & Roché 103 Puy St. Richard Hautes Alpes Korfhage & Roché 119 Les Matelles Hérault Sobhian s.n. St. Jean de Cuculles Hérault Sobhian s.n. Draguignan Var Roché & Susanna 2021 St. Maximim la Ste. Baume Var Roché & Susanna 2025 Apt Vaucluse Roché & Susanna 2028 Saignon Vaucluse Sobhian s.n. Greece Kalpáki (Ioánina) Epiros Roché & Susanna 1979 Kónitsa (Ioánina) Epiros Roché & Susanna 1978 Elefteropouli (Kavala) Macedonia Roché & Susanna 1999 Siátista (Kozáni) Macedonia Roché & Susanna 1973 Skepasto (Seres) Macedonia Roché & Susanna 2007 Skopos (Flórina) Macedonia B. Roché s.n. Thermi (Thessaloníki) Macedonia Roché & Susanna 1953 Panagia (Tríkala) Thessalía Roché & Susanna 1983 Didimotiho (Evros) Thraki Roché & Susanna 1996 Italy Aosta Aosta Korfhage & Roché 116 Susa Torino Korfhage & Roché 118 Spain Maeztu-Corres Álava Vilatersana 102 Covarrubias Burgos Roché & Susanna 1932 Rebolledo de la Torre Burgos Garcia-Jacas & Susanna 2079 Santibáñez del Val Burgos Roché & Susanna 1933 Huete (Cañaveras) Cuenca Susanna 1599 Solán de Cabras Cuenca Susanna 1813 Linares de la Sierra (Aracena) Huelva Garnatje 56 & Vilatersana Candasnos Huesca Roché & Susanna 1949 Cazorla, Fuente del Oso Jaén Garnatje 77 & Vilatersana Vegacervera León Garcia-Jacas & Susanna 2074 Torrelaguna Madrid Garcia-Jacas & Susanna 1455 Reinoso de Cerrato Palencia Roché & Susanna 1928 El Burgo de Osma Soria Roché & Susanna 1936 Pont del Diable Tarragona Garcia-Jacas, Martín, Susanna 1472 & Vallès Huerta de Valdecarábanos Toledo Susanna 1884 Pollos Valladolid Roché & Susanna 1924 Cervera de la Cañada Zaragoza Roché & Susanna 1940 Frasno-Saviñán Zaragoza Roché & Susanna 1943 Malanquía Zaragoza Roché & Susanna 1938 Salvatierra de Esca Zaragoza Vilatersana 87 Villarroya de la Sierra Zaragoza Roché & Susanna 1939 United States Kelly Springs, Modoc Co. California Raymond s.n. Santa Rosa, Sonoma Co. California Lane & Pitcairn 2027 Lawyer Cyn., Clearwater Co. Idaho Roché 2028 Dry Creek, Umatilla Co. Oregon Roché Lake Chelan, Chelan Co. Washington Roché Uzbekistan 30 Km E of Bukhara Bukhara Khassanov s.n. We had two choices when planning the RAPDs amplifications. We could carry out an in-depth analysis with 10–15 primers, if we sacrificed breadth in sampling. Alternatively, we could sample extensively, including more populations, and use fewer primers. After a preliminary survey with 20 primers, we chose the second option. Crupina vulgaris is self-compatible, and we found very low levels of intrapopulational variation, as compared with interpopulational differences. Thus we were able to obtain useful inferences with fewer primers than normally used for outcrossing plants. DNA extraction and RAPD amplifications Total genomic DNA was isolated from fresh leaf material following the protocol of Doyle & Doyle (1987), as modified by Cullings (1992). Twenty primers (A4, B8, B20, D10, E5, F6, G1-G3, G19, J1-J3, and J4-J10, Operon Technologies, Alameda, CA, USA) were screened to obtain five primers (B8, E5, G2, G19, and J10) that produced reproducible, polymorphic bands. Primers were tested with different DNA concentrations to determine the concentration for optimal RAPD patterns while minimizing artefactual bands. DNA amplification was carried out in 10 µl reaction volumes containing 1.0 µl of template DNA; 5.6 µl sterile distilled water; 1 µl of 10 mM MgCl2; 1.0 µl 10× loading buffer (Promega Corp, Madison, WI, USA); 0.8 µl 1.25 mM dNTP (Promega) and 0.4 µl of 5 pmol primer, and 1 unit (0.2 µl) of Taq polymerase (Promega). Reaction volumes were placed in 0.2 ml microfuge tubes, covered with mineral oil (Sigma-Aldrich, St. Louis, MO, USA) and placed in a thermal cycler (PTC 100, MJ Research Inc., San Francisco, CA, USA). Amplification was carried out with the following thermal profile: (i) initial denaturation at 95°C for 1 min (ii) melting at 95°C for 30 s, annealing at 37°C for 1 min and extension at 72°C for 2 min for 44 cycles (iii) final extension at 72°C for 5 min, after which the samples were maintained at 4°C until removal. Absence of contamination in each run was verified by including a negative control in which DNA was omitted. Reproducibility was confirmed by duplicating randomly selected reactions for DNA samples of each population for each primer. Amplification products were resolved electrophoretically in 2% agarose gels in 1 × TBE buffer run at 10 V cm−1 for about 5 h in 0.5× TBE, and visualized by staining with ethidium bromide. PGEM DNA marker (Promega) was used as a molecular size marker. The gels were photographed under UV light with a Kodak digital camera (DC220) (Eastman-Kodak, Rochester, NY, USA). Data analysis Analysis of the RAPD products (loci) included only reliably scored and repeated bands, and a representative of the dominant pattern was selected from each population to use for comparison among populations. Each band was assigned a molecular weight using Lane Manager 2.1a (T.D.I.S.A., Madrid, Spain), which performed a semilogarithmic regression with the DNA molecular weight standard on each gel. Bands were then manually scored for each population, 1 for presence and 0 for absence, to generate a binary matrix. This binary matrix of polymorphic bands from all five primers was analyzed using the program NTSYS-pc (Rohlf, 1997), employing SIMQUAL (similarity for qualitative data) and the similarity coefficient, Dice. This coefficient was used in order to generate an unweighted pair-group (UPGMA) dendrogram, through the routine SAHN within NTSYS. The coefficient Dice is equivalent to that of Nei & Li (1979), which was recommended by Lamboy (1994) for analyzing RAPD data. However, dendrograms generated using the Jaccard index were identical except for the length of the branches. To evaluate the fit of the cluster analysis, a Mantel test was carried out as described by Rohlf (1997). To analyze the possible relationship between genetic and geographic distances in the sampled populations, we generated a matrix of geographic distances by calculating the distance in kilometers between a subset of 33 populations, using the geographic co-ordinates (Bartish et al., 2000). American populations were excluded from this analysis. This matrix was compared with the genetic distance matrix used for the cluster analysis. A Mantel test was carried out with MXCOMP in NTSYS-pc, with 9000 permutations to compute the significance of the correlation. To further assess the correlation between geographic and genetic distances among Iberian and French populations, we carried out a second Mantel test, but this time with only 16 populations from Spain and France. Results Clustering of the genetic distances The five primers employed generated a total of 75 RAPD bands, ranging in size between 200 bp and 1.9 kb. The number of bands scored per primer varied between 11 (J-10) and 21 (E-5). The total number of polymorphic bands was 70, representing 93.3% of all bands. Details of the number of bands and percent polymorphism for each primer are shown in Table 2. Table 2. Base sequences of the primers used for DNA amplification and the number of polymorphic bands produced by each primer Primer Sequence, 5′ to 3′ No. total bands Polymorphic bands Polymorphic bands (%) B-08 GTCCACACGG 12 10 83.3 E-05 TCAGGGAGGT 21 21 100 G-02 GGCACTCAGG 13 11 84.6 G-19 GTCAGGGCAA 18 18 100 J-10 AAGCCCGAGG 11 10 90.9 The relationships among populations is presented as a dendrogram generated by UPGMA clustering (Fig. 2). The Mantel test result was r = 0.74556, which indicates a poor fit for the cluster (Rohlf, 1997). The similarity coefficient between the 48 populations varied between 0.53 and 0.94. The clustering is not clean geographically, as populations from different countries did not separate into distinct clusters. Nevertheless, despite dispersal of individual populations from different countries throughout the dendrogram, several general tendencies are apparent (Fig. 2). The Uzbekistan population does not cluster, reflecting the large geographic and genetic distance. The majority of the Greek populations, instead, form a cluster. The rest of the clusters are an intermingling of populations from northern Spain (four) and the Alps (two Italian and three Alp and Maritime Alp populations from France), with two northern Greek populations from western Macedonia (Siátista, Skepasto). The upper branch of the dendrogram comprises 27 populations, including all five American populations. The remainder are populations from northern and central Spain and southern France, except for one Greek population and two southern Spanish populations. Figure 2Open in figure viewerPowerPoint Map of Spain, showing locations of populations. The five American populations (Fig. 2) segregate into three branches of the dendrogram. Modoc population (California) clustered at 0.90 similarity with the Chelan (Washington) population. These two populations are found associated at 0.87 similarity with two populations from the northern part of the Iberian peninsula, Reinoso de Cerrato and El Burgo de Osma (Fig. 3). In turn these four populations cluster with five more populations from northern Spain and one from Kavala (eastern Macedonia), Greece (Fig. 2). The population from Oregon clustered at 90% similarity with a Spanish population from Huerta de Valdecarábanos. Similarity between the Sonoma (California) and Idaho populations was 0.88 and these were joined at 0.82 similarity to a larger cluster, which included the Oregon population and six Spanish populations: Solán de Cabras, Rebolledo de la Torre, Salvatierra de Esca, Linares de la Sierra, Huerta de Valdecarábanos, and Torrelaguna (2, 3). Figure 3Open in figure viewerPowerPoint Dendrogram constructed from UPGMA showing the relationships among populations of Crupina vulgaris collected from North America (USA) and France (FR), Greece (GR), Italy (IT), Spain (SP), and Uzbekistan (UZ). Site names are listed in Table 1. Correlation between geographic and genetic distances We found a significant correlation between geographic and genetic distances with our first analysis (normalized Mantel statistic Z r = 0.47851, p[random Z > = observed Z] = 0.0002), including populations from Greece, France, Italy, Spain and Uzbekistan. However, when comparing geographic and genetic distances among Iberian and French populations only, the correlation was significantly lower (normalized Mantel statistic Z r = 0.14127, p[random Z > = observed Z] = 0. 1792). Discussion Colonization pattern in North America Clustering of the North American populations by the molecular marker data yields a pattern lacking geographical proximity: that is Idaho with Sonoma County, California; Chelan, Washington, with Modoc County, California; and Oregon separate (Fig. 2). This may indicate one exotic population founding another, but does not preclude separate introductions by the same vectors from similar origins. Unraveling the invasion dynamics would be simpler if one could assume that discovery closely followed initial introduction. But the inconspicuous, delicate stature of Crupina vulgaris makes it almost invisible when populations are sparse, so that a considerable amount of time (i.e. decades) could easily elapse between introduction and detection. All six original discoveries of C. vulgaris in the western United States were made by professional botanists or weed scientists, and none were reported by casual observers or landowners. In spite of this limitation, a probable invasion order may be inferred from corroborating evidence, such as size and shape of populations and land use. Because the Idaho and Sonoma County discoveries (1968 and 1975) significantly predate the others (1984, 1987, 1990), they likely represent the first introduction(s). This is consistent with the size and distribution pattern of the Idaho population, where more than 20 000 ha are infested, with roadside stringers and satellite populations up to 60 km distant from the perimeter of the core infestation. Which of the other two clusters, Oregon or Chelan/Modoc, represents the second colonization event is not clear because potential dispersal vectors and suitable habitat differ so much between sites. The population at Lake Chelan, Washington, discovered in 1984, also comprises several smaller infestations, scattered along slopes above the northeast shore of the lake for about 10 km. Its related population in Modoc County, California, discovered 6 y later, is contained within a 300-ha polygon. When the Oregon population was detected in 1987, its boundaries enclosed about 200 ha, but by 2000 a satellite population had established several km away from the core infestation. Founder populations in the eastern hemisphere The positive correlation between genetic distances and geographic distances allows us to make some general inferences on the geographic origin of the American populations. The study indicates that all five populations of C. vulgaris in North America originated in Spain. That there were at least three successful colonization events suggests that there was a vector for long distance dispersal that facilitated repeated introductions. Although the weak correlation between geographic distance and genetic distance within the Iberian Peninsula prevents us from precisely pinpointing the origins of the American populations, as was the intent of the paper, it may reveal a previously unsuspected level of migration within the Iberian Peninsula. The Lawyer Canyon and Sonoma populations are linked to a broad area from the central and the north part of the Iberian Peninsula (populations from Torrelaguna, Solán de Cabras and Huerta de Valdecarábanos and from Salvatierra de Esca and Rebolledo de la Torre, respectively), with the inclusion of extreme south population from Linares de la Sierra (2, 3). The Chelan and Modoc populations derive from northern Spain, based on similarity with populations near Reinoso de Cerrato and El Burgo de Osma (Figs 2 and 3). The Oregon population most closely resembles a population from the central meseta or plateau of Spain, near Huerta de Valdecarábanos (Fig. 2). That the invasive populations are genetically similar to populations scattered over relatively wide areas of Spain and the grouping of Spanish populations that are not geographically proximate suggests an explanation that C. vulgaris has a relatively efficient distribution mechanism (considering the size of the propagule) and may also be a colonizer in the Iberian Peninsula, albeit with a much older chronology. Origin of the genus and migration in the eastern hemisphere Several factors support this explanation that C. vulgaris may be a colonizer in the Iberian Peninsula: first the location of the center of origin of Cardueae genera, second sparcity of appropriate native habitat, and third position of populations in RAPD dendrogram (i.e. lower correlation between geography and the cluster pattern as an indicator of recent and continuous migrations). The traditional view is that the center of origin for the subtribe Centaureinae to which Crupina vulgaris belongs, is east of the Mediterranean region, from whence it spread to occupy semidesert habitats during the Miocene (Small, 1919; Bremer, 1994). Crupina may have followed a pattern of westward migration in the Mediterranean region similar to that of numerous other steppe species which expanded their range to occupy anthropic (human modified) habitats that approximated their native environment (Gamarra & Montouto, 1999). Crupina vulgaris is restricted to open habitats with a Mediterranean climate and well-drained soils, including grasslands, open wood or shrublands, or remnants of these types of vegetation adjacent to cropland (e.g. vineyards, cereal grains, olives). In the Iberian Peninsula, these habitats are almost entirely anthropic. Iberian climax vegetation is largely temperate forest (evergreen or deciduous Quercus, or Fagus, Pinus, or mixed stands), with natural grasslands occurring only above treeline in the mountains and internally drained basins (Montserrat & Fillat, 1990). Natural communities in which C. vulgaris grows, Thero-Brachypodietea, are limited to nonzonal sites (e.g. shallow soils, ro
Referência(s)