Molecular phylogeny and biogeography of the Cuban genus Girardinus Poey, 1854 and relationships within the tribe Girardinini (Actinopterygii, Poeciliidae)

Ignacio Doadrio; Silvia Perea; Lourdes Alcaraz; Natividad Hernandez

Despite ongoing efforts to protect species and ecosystems in Cuba, habitat degradation, overuse and introduction of alien species have posed serious challenges to native freshwater fish species. In spite of the accumulated knowledge on the systematics of this freshwater ichthyofauna, recent results suggested that we are far from having a complete picture of the Cuban freshwater fish diversity. It is estimated that 40% of freshwater Cuban fish are endemic; however, this number may be even higher. Partial sequences (652 bp) of the mitochondrial gene COI (cytochrome c oxidase subunit I) were used to barcode 126 individuals, representing 27 taxonomically recognized species in 17 genera and 10 families. Analysis was based on Kimura 2-parameter genetic distances, and for four genera a character-based analysis (population aggregation analysis) was also used. The mean conspecific, congeneric and confamiliar genetic distances were 0.6%, 9.1% and 20.2% respectively. Molecular species identifi...

The genus Rivulus is currently comprised of two species, R. cylindraceus and R. insulaepinorum, which are endemic to Cuba. However, the taxonomic status of the latter species remains dubious because of the poor quality of the original description. In addition, a recent barcoding survey suggests that the two species may be conspecific. The aim of this study was to test the hypothesis that the two species represent a single evolutionary clade. To delimit the species and their evolutionary history, we used a combination of molecular phylogenetic analyses, with both mitochondrial and nuclear sequences, tests of phylogeographic hypotheses, combined with morphological measurements and information on known dispersal barriers and species distribution. None of the data sets support R. insulaepinorum and R. cylindraceus as separate taxa. However, a new species, restricted to the northwestern part of the main island, was identified by phylogenetic analyses, body colour pattern and geographical distribution. The evolutionary distance between the two lineages (cytb, d=15%; CAM-4, d=2.5%) indicates a long period of divergence. Phylogeographic analyses shed light on the dispersal history of R. cylindraceus, which probably originated on the Isla de la Juventud. They also suggest that each lineage had contrasting histories; Rivulus sp. is restricted to a relatively small geographic area whereas R. cylindraceus has dispersed considerably and more than once from its centre of origin, probably facilitated by sea level fluctuations. These results strengthen previous findings, i.e. that the diversity of Cuban freshwater fishes is far from well-known and deserves more in-depth studies, and that vicariance and dispersal events have resulted in a complex biogeographical landscape which has had a significant impact on the freshwater fishes of the Caribbean islands.

Molecular Phylogenetics and Evolution 50 (2009) 16–30 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev Molecular phylogeny and biogeography of the Cuban genus Girardinus Poey, 1854 and relationships within the tribe Girardinini (Actinopterygii, Poeciliidae) Ignacio Doadrio a,*, Silvia Perea a, Lourdes Alcaraz a, Natividad Hernandez b a b Department of Biodiversity and Evolutionary Biology, Museo Nacional de Ciencias Naturales CSIC, José Gutiérrez Abascal 2, 28006 Madrid, Spain Instituto de Medicina Tropical Pedro Kouri, Apdo 601, Marianao 13, Ciudad de La Habana, Cuba a r t i c l e i n f o Article history: Received 26 March 2008 Revised 16 September 2008 Accepted 19 September 2008 Available online 30 September 2008 Keywords: Girardinini Molecular phylogeny Biogeography Taxonomy Caribbean islands a b s t r a c t Phylogenetic relationships among members of the freshwater fish tribe Girardinini were inferred to test existing colonization and diversification hypotheses for this group in the Caribbean. The genetic material examined was mitochondrial (cytochrome b, 1140 bp) and nuclear (RAG-1 and b-actin, 2450 bp) DNA from 161 specimens representing 44 ingroup and three outgroup taxa. Our mtDNA and combined data matrix (mtDNA + nuclear DNA) results rendered a well-supported phylogeny for the tribe Girardinini and suggest the need to review the group’s current taxonomy. From the data presented here, it may be inferred that the Girardinini diverged from other poeciliid fishes approximately 62 Mya ago in the Palaeocene period. This estimate, however, conflicts with the hypothesis that today’s vertebrate fauna is the result of the more recent colonization of the Antillean islands during the Early Oligocene (35–33 Mya ago). The isolation of western, central and eastern Cuba during the Miocene and that of the Juventud Island and Guanahacabibes Peninsula during the Pliocene, are the main geologic events that could have promoted speciation in this group. ! 2008 Elsevier Inc. All rights reserved. 1. Introduction The tribe Girardinini Hubbs, 1924 was proposed to comprise 6 genera and 15 species and subspecies of endemic freshwater fishes from the main island of Cuba and Juventud island (former Pinos) (Rivas, 1958; Barus et al., 1991). However, according to the taxonomy of Rosen and Bailey (1963), the tribe is only considered to have three genera (Girardinus Poey, 1854; Quintana Hubbs, 1934 and Carlhubbsia Withley, 1951) and a number of species (from ten to twelve) that depends on the describing author (Rosen and Bailey, 1963; Parenti and Rauchenberger, 1989; Ghedotii, 2000; Lucinda and Reis, 2005). The monotypic genus Quintana is represented by the species Quintana atrizona Hubbs, 1934 inhabiting both Juventud Island and western Cuba Island. The genus Carlhubbsia has two recognized species: C. kidderi (Hubbs, 1936) from Mexico and Guatemala, and C. stuarti Rosen and Bailey, 1959 from Guatemala and Belize. In turn, the genus Girardinus has eight recognized species (Lucinda, 2003): G. metallicus Poey, 1854; G. uninotatus Poey, 1860; G. creolus Garman, 1895; G. denticulatus Garman, 1895; G. cubensis (Eigenmann, 1903); G. falcatus (Eigenmann, 1903); G. microdactylus Rivas, 1944 and G. rivasi Barus and Wohlgemuth, 1994. * Corresponding author. Fax: +34 915645078. E-mail address: mcnd147@mncn.csic.es (I. Doadrio). 1055-7903/$ - see front matter ! 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2008.09.014 The genus Girardinus is among the endemic freshwater fishes that inhabit both the islands of Cuba and Juventud. This genus shows its highest diversity in western Cuba and is absent from central Cuba south of the Sancti-Spiritus district and from some eastern areas of Cuba Island, such as short rivers south of the Sierra Maestra and west of Sagua-Baracoa. Compared to other freshwater fishes of the Antillean islands, Girardinus is a good model for deciphering the evolutionary history of Cuban hydrographical systems because of its wider range of distribution throughout Cuba, its higher diversity, its presence on two islands of known geological age and its ecology, being more restricted to freshwater systems. Studies on the salt tolerance of representatives of Girardinini have failed to recover fishes from salt, brackish, or fresh water with a Cl! content higher than 1000 parts per million (Rivas, 1958, 1986). We can therefore assume that dispersion in Girardinini occurs only via connected freshwater systems (Rivas, 1958). Models of the palaeogeography of Cuba based on biological data can be useful for geological reconstructions, since Cuba occupies one of the most geologically complex areas of the planet. Cuba’s main and Juventud islands boast over 600 fluvial systems mostly crossing karst landscapes. The country’s karst topography determines that some rivers are fragmented by waterfalls or disappear for kilometers underground and may also be the cause of local speciation events in freshwater fishes such as Girardinus with a limited capacity for dispersal. By identifying the barriers that caused the fragmentation and dispersion of freshwater fishes and the timing 17 I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 of these events in a geological scenario, insight can be gained into the evolution of organisms and areas. Effectively, studies on the biogeography of freshwater fishes in Central America and the Caribbean Islands have contributed significantly to palaeobiogeography models (Rosen 1975, 1978, 1985; Nelson and Rosen, 1981). The genus Girardinus has also been examined as a means of biologically controlling disease-transmitting mosquitoes (Rodríguez et al., 2004; Menéndez Díaz et al., 2007); other studies have focussed on its complex taxonomy (Barus et al., 1981, 1986, 1991, 1998; Barus and Libosvarsky, 1986; Barus and Wohlgemuth, 1994). Most of the latter works have focused on basic aspects such as taxonomy, geographical distribution and ecological requirements. In contrast, there are only a few models of the phylogenetic relationships and biogeography of this genus. The first and most developed model is that proposed by Rivas (1958), in which the ancestor of today’s Girardinus colonized Cuba from the eastern Yucatan during the Upper Miocene–Pliocene across an intercontinental Yucatan-Cuba landbridge. Western Cuba is considered the center of origin of Girardinus, where it shows its highest diversity, and its speciation is thought to have occurred during the Pleistocene. The ancestral species of the genus is considered by Rivas (1958) to be similar to Girardinus creolus, which is the only species confined to upper stream reaches (Fig. 1). Rodriguez et al. (1992) corroborated the model suggested by Rivas (1958) considering Girardinus creolus to be the ancestral species of the genus Girardinus. An eastern Yucatan-Cuba connection was also supported by the phylogenetic relationships proposed by Rosen and Bailey (1963), who considered the genera Quintana from Cuba and Carlhubbsia from eastern Yucatan and Guatemala, as sister genera of Girardinus. However, recent studies on poeciliid fishes question the Yucatan-Cuba connection as well as the relationship of Carlhubbsia as sister group of the genera Girardinus and Quintana (Hrbek et al., 2007). An alternative model has been developed to explain the presence of Poeciliidae representatives in the Antilles, including the tribe Girardinini across GAARlandia (Greater Antilles + Aves Ridge) (Iturralde-Vinent and MacPhee, 1999; Iturralde-Vinent, 2006). This model of geological evolution of the Caribbean basin suggests that the Greater Antilles did not permanently emerge until the Eocene-Oligocene boundary (30–35 mya), when they were connected to northern South America by the Aves Ridge. The three biogeographical models (Rivas, 1958; Rosen and Bailey, 1963; Hrbek et al, 2007) are represented in Fig. 2. The present study was designed to test previous phylogenetic relationships of the tribe Girardinini (Rivas, 1958; Rosen and Bailey, 1963) and the three biogeographical models proposed for the Cuba and Juventud islands (Fig. 2) using the most complete molecular dataset for the tribe Girardinini available to date. Neoheterandria (= Allogambusia) A Upper Miocene/Pliocene ~11.6-3.6 Mya B Eocene/Oligocene ~30-35 Mya C Cretaceous ~65 Mya Chicxulub impact Fig. 2. Main biogeographical hypotheses proposed to explain the colonization of the Antillean islands. (A) Biogeographical scenario in agreement with Rivas (1958). (B) Biogeographical scenario in agreement with Iturralde-Vinent and MacPhee (1999) (C) Biogeographical scenario in agreement with Rosen (1975). (Pictures modified from original authors). Toxus serripenis Toxus creolus Yucatan-Cuba split Girardinus microdactylus 2. Materials and methods Girardinus metallicus 2.1. Sampling Dactylophallus denticulatus Allodontium cubense Glaridichthys falcatus Glaridichthys uninotatus Pliocene Pleistocene Fig. 1. Taxonomy and implied phylogenetic hypothesis for the tribe Girardinini according to Rivas (1958). DNA samples were analyzed in 111 individuals representing all described species of the genus Girardinus (except G. cubensis). Populations were collected in the wild from 48 localities of the Cuba and Juventud islands including the type locality when possible (Table 1); (Fig. 3). DNA was extracted from the dorsal muscle, which was preserved in liquid nitrogen or 70% ethanol. Voucher specimens of these species were deposited in the collections of the Tropical Medicine Institute Pedro Kouri, La Habana, Cuba and the Museo Nacional de Ciencias Naturales, Madrid, Spain. The 18 Table 1 Samples of the tribe Girardinini, collection sites, number of populations sampled in the phylogenetic tree and GenBank accession numbers Locality/Source Designation in phylogenetic tree GenBank Acc. No. (CB/RAG1/b-actin) New taxonomical denomination Number in map Alfaro cultratus Alfaro cultratus Alfaro cultratus Anableps anableps Belonesox belizanus Brachyrhaphis cascajalensis Brachyrhaphis hartwegi Brachyrhaphis hartwegi Brachyrhaphis parismina Brachyrhaphis rhabdophora Brachyrhaphis terrabensis Carlhubbsia kidderi Carlhubbsia kidderi Carlhubbsia stuarti Cnesterodom decemmaculatus Crenichthys baileyi Fluviphylax pygmaeus Gambusia affinis Girardinus creolus Girardinus creolus Girardinus creolus Girardinus creolus (G. riddley) Girardinus creolus Girardinus creolus Girardinus creolus Girardinus creolus Girardinus creolus (G. serripenis) Nicaragua Lake. San Carlos. Nicaragua Nicaragua Lake. San Carlos. Nicaragua El Monje River. San Miguel. Nicaragua Hrbek et al., 2007 Hrbek et al., 2007 Costa Rica Aquarium stock Hrbek et al., 2007 Costa Rica Hrbek et al., 2007 Hrbek et al., 2007 Mamantel River. San Antonio. México Mamantel River. San Antonio. México Hrbek et al., 2007 Hrbek et al., 2007 Doadrio and Domínguez, 2004 Hrbek et al., 2007 Hrbek et al., 2007 La Legua River. Entronque de Herradura. Cuba La Legua River. Entronque de Herradura. Cuba La Legua River. Entronque de Herradura. cuba San Francisco River. San Cristóbal. Cuba San Cristóbal River. La Hoya. Cuba San Cristóbal River. La Hoya. Cuba San Cristóbal River. La Hoya. Cuba San Cristóbal River. La Hoya. Cuba Taco-Taco River. Upstream cave entrance. Rangel. Cuba Taco-Taco River. Upstream cave entrance. Rangel. Cuba Taco-Taco River. Upstream cave entrance. Rangel. Cuba Taco-Taco River. Downstream cave. Rangel. Cuba Taco-Taco River. Downstream cave. Rangel. Cuba Taco-Taco River. Upstream cave entrance. Rangel. Cuba Guaso River. Argeo Martínez. Cuba Alfaro cultratus 1 Alfaro cultratus 2 Alfaro cultratus 3 Anableps anableps Belonesox belizanus Brachyrhaphis cascajalensis Brachyrhaphis hartwegi 1 Brachyrhaphis hartwegi 2 Brachyrhaphis parismina Brachyrhaphis rhabdophora Brachyrhaphis terrabensis Carlhubbsia kidderi 1 Carlhubbsia kidderi 2 Carlhubbsia stuarti Cnesterodom decemmaculatus Crenichthys baileyi Fluviphylax pygmaeus Gambusia affinis TCRE1 TCRE2 TCRE3 TCRE4 TCRE5 TCRE6 TCRE7 TCRE8 TCRE9 FJ178772*/!/–* FJ178774*/!/–* FJ178773*/!/! EF017511/ EF017408/! EF017514/ EF017411/! FJ178734*/ FJ185096*/ FJ194984* FJ178735*/!/! FJ178736*/!/! FJ178737*/!/! FJ178738*/!/! FJ178739*/!/! FJ178740*/ FJ185104*/ FJ194995* FJ178741*/!/ FJ194996* FJ178742*/!/! Alfaro cultratus Alfaro cultratus Alfaro cultratus Anableps anableps Belonesox belizanus Brachyrhaphis cascajalensis Brachyrhaphis hartwegi Brachyrhaphis hartwegi Brachyrhaphis parismina Brachyrhaphis rhabdophora Brachyrhaphis terrabensis Carlhubbsia stuarti Carlhubbsia stuarti Carlhubbsia stuarti Cnesterodom decemmaculatus Crenichthys baileyi Fluviphylax pygmaeus Gambusia affinis Toxus creolus Toxus creolus Toxus creolus Toxus creolus Toxus creolus Toxus creolus Toxus creolus Toxus creolus Toxus creolus – – – – – – – – – – – – – – – – – – 9 9 9 14 15 15 15 15 12 TCRE10 FJ178743*/!/! Toxus creolus 12 TCRE11 FJ178744*/!/! Toxus creolus 12 TCRE12 TCRE13 TCRE14 FJ178745*/!/! FJ178746*/!/! FJ178747*/!/! Toxus creolus Toxus creolus Toxus creolus 12 12 12 DRAM FJ178733*/!/ FJ194993* Dactylophallus ramsdeni 31 Aguas Azules, Las Chambas. Cuba Yao River. Nuevo Yao. Cuba Yao River. Nuevo Yao. Cuba Yao River. Nuevo Yao. Cuba Contramaestre River. Contramaestre. Cuba Laguna River. Pilon. Cuba Aguacate River. Mayajigua. Cuba Malnombre River. Macahuabo. Las Tozas. Cuba Manaquita River. Norte Viñas. Cuba Manaquita River. Norte Viñas. Cuba Canoao River. Esmeralda. Cuba Bautista River. San Juan y Martinez. Cuba Los Palacios. Cuba Cañada del Infierno. Las Terrazas. Cuba San Cristobal. Cuba Hawa River. Cienaga Zapata. Cuba Aguas Azules. Las Chambas. Cuba Guayabo Dam. La Fe. Juventud I. Cuba DDEN1 DDEN2 DDEN3 DDEN4 DDEN5 DDEN6 DDEN7 DDEN8 DDEN9 DDEN10 GFAL1 GFAL2 GFAL3 GFAL4 GFAL5 GFAL6 GFAL7 GFAL8 FJ178723*/!/! FJ178724*/ FJ185102*/ FJ194991* FJ178725*/!/! FJ178726*/!/! FJ178727 */!/! FJ178728*/!/! FJ178729*/!/! FJ178730*/!/! FJ178731*/!/! FJ178732*/ FJ185100*/ FJ194989* FJ178749*/!/! FJ178750*/ FJ185094*/ FJ194981* FJ178751*/!/! FJ178752*/!/! FJ178753*/!/! FJ178754*/!/! FJ178755*/!/! FJ178756*/!/! Dactylophallus denticulatus Dactylophallus denticulatus Dactylophallus denticulatus Dactylophallus denticulatus Dactylophallus denticulatus Dactylophallus denticulatus Dactylophallus denticulatus Dactylophallus denticulatus Dactylophallus sp Dactylophallus sp Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus 26 34 34 34 33 35 24 25 22 22 27 5 10 20 14 21 26 40 Girardinus creolus (G. serripenis) Girardinus creolus Girardinus creolus Girardinus creolus Girardinus creolus Girardinus denticulatus (G. ramsdeni) Girardinus denticulatus Girardinus denticulatus Girardinus denticulatus Girardinus denticulatus Girardinus denticulatus Girardinus denticulatus Girardinus denticulatus Girardinus denticulatus Girardinus denticulatus Girardinus denticulatus Girardinus falcatus Girardinus falcatus Girardinus falcatus Girardinus falcatus Girardinus falcatus Girardinus falcatus Girardinus falcatus (G. atherinoides) Girardinus falcatus EF017508/ EF017405/! EF017519/ EF017416/! FJ178767*/!/! FJ178769*/ FJ185090*/ FJ194976* EF017521/ EF0174198/! FJ178768*/!/! EF017522/ EF017419/! EF017520/ EF017430/! FJ178777*/!/! FJ178778*/ FJ185091*/! EF017532/ EF017417/! EF017529/ EF017427/! AF510819/ FJ185089*/ FJ194975* I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 Species Carapechibey, Juventud I. Cuba Carapechibey. Juventud I. Cuba Pino Alto Lagoon. Mella. Juventud I. Cuba Cienaga de Lanier. Cayo Potrero Juventud I. Cuba Guanahacabibes Peninsula. Cuba La Fe. Pinar del Río. Cuba El Cayuco. Cuba Hondo River. Pilotos. Cuba Surgidero de Batabano. Batabano. Cuba Cienaga de Lanier. Cayo Potrero. Juventud I. Cuba San Juan River. San Juan y Martinez. Cuba Bautista River. San Juan y Martínez. Cuba Guayabo Dam. La Fe. Juventud I. Cuba Surgidero de Batabano. Batabano. Cuba Hawa River. Cienaga de Zapata. Cuba La Pastora River. Artemisa. Cuba La Pastora River. Artemisa. Cuba Surgidero de Batabano. Batabano. Cuba Cienaga de Lanier. Cayo Potrero. Juventud I. Cuba Yaguajay. Cuba Yaguajay. Cuba Mayajigua. Cuba Yao River. Nuevo Yao. Cuba Yao River. Nuevo Yao. Cuba Guaimaro River. Guaimaro. Cuba Gascon River. Santiago de Cuba. Cuba El Sábalo. Cuba La Fe. Pinar del Río. Cuba El Cayuco. Cuba Barranco del Infierno. Las Terrazas. Cuba San Francisco River. San Cristobal. Cuba San Francisco River. San Cristobal. Cuba Taco-Taco River. Aspiro. Cuba Taco-Taco River. Aspiro. Cuba Barranco del Infierno. Las Terrazas. Cuba Barranco del Infierno. Las Terrazas. Cuba Barranco del Infierno. Las Terrazas. Cuba San Francisco River. San Cristobal. Cuba San Juan River. Las Terrazas. Cuba San Juan River. Las Terrazas. Cuba Barranco del Infierno. Las Terrazas. Cuba Barranco del Infierno. Las Terrazas. Cuba San Cristobal River. La Hoya. Cuba Cañada del Infierno. Las Terrazas. Cuba San Juan River. Las Terrazas. Cuba Mosquito River. Minas. Cuba Reparto Militar Lagoon. Minas. Cuba Canoao River. Esmeralda. Cuba Nueva Gerona River. Nueva Gerona. Juventud I. Cuba Demajagua River. Demajagua. Juventud I. Cuba Las Tunas. Juventud I. Cuba Los Colonos. Juventud I. Cuba Mazareno River. Nueva Gerona. Juventud I. Cuba Mazareno River. Nueva Gerona. Juventud I. Cuba Cienaga de Lanier. Cayo Potrero. Juventud I. Cuba Curva de la Muerte River. Juventud I. Cuba La Fe. Juventud I. Cuba La Fe. Juventud I. Cuba Granja La Libertad. Juventud I. Cuba Argelia Libre. Juventud I. Cuba GFAL9 GFAL10 GFAL11 GFAL12 GFAL13 GFAL14 GFAL15 GMET1 GMET2 GMET3 GMET4 GMET5 GMET6 GMET7 GMET8 GMET9 GMET10 GMET11 GMET12 GMET13 GMET14 GMET15 GMET16 GMET17 GMET18 GMET19 GMET20 GMET21 GMET22 GMIC1 GMIC2 GMIC3 GMIC4 GMIC5 GMIC6 GMIC7 GMIC8 GMIC9 GMIC10 GMIC11 GMIC12 GMIC13 GMIC14 GMIC15 GMIC16 GMIC17 GMIC18 GMIC19 GRIV1 GRIV2 GRIV3 GRIV4 GRIV5 GRIV6 GRIV7 GRIV8 GRIV9 GRIV10 GRIV11 GRIV12 FJ178757*/!/! FJ178758*/!/! FJ178759*/ FJ185098*/ FJ194987* FJ178760*/!/! FJ178761*/!/ FJ194979* FJ178762*/!/! FJ178763*/!/ FJ194983* FJ178653*/!/! FJ178654*/!/! FJ178655*/!/! FJ178656*/!/! FJ178657*/!/! FJ178658*/!/! FJ178659*/!/! FJ178660*/!/! FJ178661*/!/ FJ194994* FJ178662*/!/! FJ178663*/!/! FJ178664*/!/! FJ178665*/!/! FJ178666*/!/! FJ178667*/!/! FJ178668*/ FJ185103*/ FJ194992* FJ178669*/!/! FJ178670*/!/! FJ178671*/!/! FJ178672*/ FJ185095*/ FJ194982* FJ178673*/!/! FJ178674*/!/! FJ178675*/!/! FJ178676*/!/! FJ178677*/!/! FJ178678*/ FJ185097*/ FJ194985* FJ178679*/!/– FJ178680*/!/– FJ178681*/!/! FJ178682*/!/! FJ178683*/!/! FJ178684*/!/! FJ178685*/!/! FJ178686*/!/! FJ178687*/!/! FJ178688*/!/! FJ178689*/!/– FJ178690*/!/! FJ178691*/!/! FJ178692*/!/! FJ178693*/ FJ185101*/ FJ194990* FJ178694*/!/–* FJ178695*/!/! FJ178696*/!/! FJ178697*/!/! FJ178698*/!/! FJ178699*/!/! FJ178700*/!/! FJ178701*/ FJ185099*/ FJ194988* FJ178702*/!/! FJ178703*/!/ FJ194986* FJ178704*/!/! FJ178705*/!/! Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Glaridichthys falcatus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus metallicus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus microdactylus Girardinus sp Girardinus sp Girardinus sp Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi Girardinus rivasi 44 44 42 43 1 2 3 8 21 43 6 5 41 20 21 19 19 20 43 23 23 24 34 34 30 32 4 2 3 17 14 14 11 11 17 17 17 14 17 17 17 17 15 17 18 29 28 27 36 47 48 38 37 37 43 41 39 39 46 45 (continued on next page) 19 falcatus falcatus falcatus falcatus falcatus falcatus falcatus metallicus metallicus metallicus metallicus metallicus metallicus metallicus metallicus (G. pygmaeus) metallicus metallicus metallicus metallicus metallicus metallicus metallicus metallicus metallicus metallicus metallicus metallicus metallicus metallicus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus microdactylus rivasi rivasi rivasi rivasi rivasi rivasi rivasi rivasi rivasi rivasi rivasi rivasi I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus Girardinus 20 Table 1 (continued) Locality/Source Designation in phylogenetic tree GenBank Acc. No. (CB/RAG1/b-actin) New taxonomical denomination Number in map Girardinus rivasi Girardinus rivasi Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Girardinus uninotatus Heterandria bimaculata Heterophallus milleri Limia dominicensis Limia melanogaster Limia tridens Limia vittata Limia vittata Neoheterandria tridentiger Phallopytchus januarius Phallichthys amates Phallichthys pittieri Phallichthys pittieri Phallichthys tico Phallichthys tico Poecilia latipunctata Poecilia mexicana Poecilia reticulata Poeciliopsis fasciata Priapella compressa Priapella intermedia Priapella olmecae Priapichthys festae Quintana atrizona Quintana atrizona Scolichthys greenwayi Xenodexia ctenolepis Xenophalus umbratilis Xenophalus umbratilis Xiphophorus helleri Xiphophorus sp Guayabo Dam, La Fe. Juventud I. Cuba La Fe. Juventud I. Cuba Santiago River. Soroa. Cuba Santiago River. Soroa. Cuba Santiago River. Soroa. Cuba Santiago River. Soroa. Cuba Guaimaro River. Guaimaro. Cuba Hawa River. Cienaga de Zapata. Cuba Barranco del Infierno. Las Terrazas. Cuba Pinar del Río. Cuba Pinar del Río. Cuba Barranco del Infierno. Las Terrazas. Cuba Santiago River. Soroa. Cuba San Francisco River. San Cristobal. Cuba San Francisco River. San Cristobal. Cuba La Legua River. Entronque de Herradura. Cuba La Legua River. Entronque de Herradura. Cuba Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 La Boca Lagoon. Camagüey. Cuba Abra River. Juventud I. Cuba Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 Sardinas River. Colon. Nicaragua Sardinas River. Colon. Nicaragua Hrbek et al., 2007 Nicaragua Lake. Nicaragua Hrbek et al., 2007 Col River. Veracruz. Mexico Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 Guanahacabibes Peninsula. Cuba Hrbek et al., 2007 Hrbek et al., 2007 Hrbek et al., 2007 El Monje River. San Miguel. Nicaragua Hrbek et al., 2007 Tabar River. Tabara de abajo. Dominican Republic GRIV13 GRIV14 GUNI1 GUNI2 GUNI3 GUNI4 GUNI5 GUNI6 GUNI7 GUNI8 GUNI9 GUNI10 GUNI11 GUNI12 GUNI13 GUNI14 GUNI15 Heterandria bimaculata Heterophallus milleri Limia dominicensis Limia melanogaster Limia tridens Limia vittata 1 Limia vittata 2 Neoheterandria tridentiger Phallopytchus januarius Phallichthys amates Phallichthys pittieri 1 Phallichthys pittieri 2 Phallichthys tico Phallichthys tico Poecilia latipunctata Poecilia mexicana Poecilia reticulata Poeciliopsis fasciata Priapella compressa Priapella intermedia Priapella olmecae Priapichthys festae Quintana atrizona 1 Quintana atrizona 2 Scolichthys greenwayi Xenodexia ctenolepis Xenophalus umbratilis 1 Xenophalus umbratilis 2 Xiphophorus helleri Xiphophorus sp FJ178706*/!/! FJ178707*/!/! FJ178708*/!/! FJ178709*/!/! FJ178710*/!/! FJ178711*/!/! FJ178712*/!/! FJ178713*/!/! FJ178714*/!/! FJ178715*/!/! FJ178716*/!/! FJ178717*/!/! FJ178718*/!/! FJ178719*/ FJ185093*/ FJ194980* FJ178720*/!/! FJ178721*/!/! FJ178712*/!/! EF017533/ EF017431/! EF017517/ EF017414/! EF017533/ EF017431/! EF017523/ EF017420/! EF017535/ EF017433/! FJ178765*/!/! FJ178766*/!/! EF017526/ EF017423/! EF017530/ EF017428/! EF017513/ EF017410/! FJ178770*/!*/ FJ194977* FJ178771*/!/! Girardinus rivasi Girardinus rivasi Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Glaridicthys uninotatus Heterandria bimaculata Heterophallus milleri Limia dominicensis Limia melanogaster Limia tridens Limia vittata Limia vittata Neoheterandria tridentiger Phallopytchus januarius Phallichthys amates Phallichthys pittieri Phallichthys pittieri Phallichthys tico Phallichthys tico Poecilia latipunctata Poecilia mexicana Poecilia reticulata Poeciliopsis fasciata Priapella compressa Priapella intermedia Priapella olmecae Priapichthys festae Quintana atrizona Quintana atrizona Scolichthys greenwayi Xenodexia ctenolepis Xenophalus umbratilis Xenophalus umbratilis Xiphophorus helleri Xiphophorus sp 40 39 16 16 16 16 30 21 17 7 7 17 16 14 14 9 9 – – – – – – – – – – – – – – – – – – – – – – – 1 – – – – – – EF017512/ EF017409/! AF412127/!/! EF017539/ EF017436/! FJ178776*/!/! EF017536/ EF017434/! EF017546/ EF017443/! EF017554/ EF017451/! EF017553/ EF017450/! EF017555/ EF017452/! EF0175543/ EF017440/! EF017556/ EF017453/! FJ178764*/ FJ185092*/ FJ194978* EF017541/ EF017438/! EF017557/–/– EF017527/ EF017424/! FJ178775*/!/! EF017548/ EF017445/! FJ194974*/ !/ ! GenBank accession numbers with an * are new sequences obtained in this study. In the first column the taxonomic identification of specimens collected in the field are indicated. Column five indicates the new taxonomical denomination proposed after this study. I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 Species 21 I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 22 . . .. . 1 20 26 27 28 29 30 21 Sancti-Spiritus 25 13 16 17 18 10 9 24 23 . 34 . .. Sagua-Baracoa 33 Sierra Maestra 5 35 . . 32 31 14 19 11 15 12 2 6 4 7 8 48 47 1 3 46 45 44 38 37 36 39 40 41 43 42 Fig. 3. Map of sampling localities and areas lacking Girardinini representatives in the Cuba and Juventud islands. Numbers of localities correspond to those indicated in Table I. following non-Girardinus species of Poeciliidae were also analyzed: Alfaro cultratus (Regan, 1908); Belonesox belizanus Kner, 1860; Brachyrhaphis cascajalensis (Meek and Hildebrand, 1913); Brachyrhaphis hartwegi Rosen and Bailey, 1963; Brachyrhaphys rhabdophora (Regan, 1908); Brachyrhaphis terrabensis (Regan, 1907); Carlhubbsia kidderi (Hubbs, 1936); Carlhubbsia stuarti Rosen and Bailey, 1959; Cnesterodon decemmaculatus (Jenyns, 1842); Gambusia affinis (Baird and Girard, 1859); Heterandria bimaculata (Heckel, 1848); Heterophallus milleri Radda, 1987; Limia dominicensis (Valenciennes, 1846); Limia melanogaster (Günther, 1866); Limia tridens (Hilgendorf, 1889); Limia vittata (Guichenot, 1853); Neoheterandria tridentiger (Garman, 1895); Pamphorichthys hollandi (Henn, 1916); Phallichthys pittieri (Miller, 1907); Phallichthys tico Bussing, 1963; Phalloceros caudimaculatus (Hensel, 1868); Phalloptychus januarius (Hensel, 1868); Poecilia latipunctata Meck, 1904; Poecilia mexicana Steindachner, 1863; Poecilia reticulata Peters, 1859; Poeciliopsis fasciata (Meck, 1904); Priapella compressa (Alvarez, 1948); Priapella intermedia Alvarez and Carranza, 1952; Priapella olmecae Meyer and Espinosa Pérez, 1990; Pseudopoecilia festae (Boulenger, 1898); Quintana atrizona Hubbs, 1934; Scolicthchys greenwayi Rosen, 1967; Tomeurus gracilis Eigenmman, 1909; Xenodexia ctenolepis Hubbs, 1950; Xenophallus umbratilis and Xhipophorus helleri (Meek, 1912). The outgroups selected for phylogenetic analysis were the species Fluviphylax pygmaeus (Myers and Carvalho, 1955) of the subfamily Fluviphylacinae considered one of the sister subfamilies of Poeciliinae (Parenti, 1981) and Anableps anableps (Linnaeus, 1758), which belongs to the family Anablepsidae, considered the sister family of Poeciliidae. We also included a more distantly related outgroup, the goodeid species Crenichthys baileyi (Gilbert, 1893). The choice of these outgroups was based on published molecular phylogenetic hypotheses for the Poeciliidae (Hrbek et al., 2007). With the exception of Girardinus cubensis (Eigenmann, 1903), we included in our molecular analysis all Girardinus species recognized as valid by Lucinda (2003): G. creolus Garman, 1895; G. denticulatus Garman, 1895; G. falcatus (Eigenmann, 1903); G. metallicus Poey, 1854; G. microdactylus Rivas, 1944; G. uninotatus Poey, 1860 and G. rivasi Barus and Wohlgemuth, 1994. Girardinus serripenis was considered a junior synonym of G. creolus, as proposed by Rodriguez et al. (1992). The original description of Girardinus cubensis was based on three female specimens from Los Palacios and Pinar del Río. This was followed by a description of the G. cubensis male based on fishes from the Taco-Taco River in Aspiro (Howell-Rivero and Rivas, 1944), and recently Barus et al. (1998) found individuals of G. cubensis in La Pastora River between Artemisa and Las Mangas. Despite our sampling efforts in Los Palacios, Pinar del Río, TacoTaco River in Aspiro and La Pastora River between Artemisa and Las Mangas, we only found the species G. metallicus, G. microdactylus and G. falcatus in these localities. All fishes were captured by electrofishing and hand nets. Besides the seven Girardinus species examined, we also analyzed specimens from the terra typica of species described as Girardinus but considered invalid (Table 1): G. riddlei (Eigenmann, 1903) synonym of G. creolus from San Cristobal; G. atherinoides Rivas, 1944 synonym of G. falcatus from Mabuya; G. serripenis, synonym of G. creolus from the Taco-Taco River in Rangel (Rodriguez et al., 1992); G. ramsdeni, synonym of G. denticulatus from Rio Guaso Guantanamo; G. pygmaeus (Rivas, 1944), synonym of G. metallicus from the Rio Negro Hatiguanico system (Rosen and Bailey, 1963); and G. torralbasi (Eigenmann, 1903) synonym of G. uninotatus from Pinar del Río (Rosen and Bailey 1963). 2.2. DNA extraction, amplification and sequencing Total cellular DNA was extracted from tissues by standard proteinase K and phenol/chloroform procedures (Sambrook et al., 1989). Two overlapping fragments of the mitochondrial cytochrome b gene (total of 1140 bp) were amplified via polymerase chain reaction (PCR) from each individual DNA sample. In all species, the primers used for cytochrome b were those mentioned in Machordom and Doadrio (2001). The amplification process was conducted as follows: 94 "C (2 min), 35 cycles at 94 "C (45 s), 48 "C (1 min), 72 "C (1.45 min). PCR mixtures were prepared in 25 ll volumes with a final concentration of 0.5 mM each primer, 0.2 mM each dNTP, 1.5 mM MgCl2, and 1 unit of Taq DNA polymerase (Invitrogen). We also sequenced two nuclear genes to compare maternal and paternal lineages of the tribe Girardinini. To this end, we amplified 22 I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 Fig. 4. Phylogenetic tree (50% majority rule consensus tree) rendered by Bayesian analysis of the mitochondrial cytochrome b dataset. Numbers above nodes indicate posterior probabilities; numbers below nodes indicate bootstrap values for the MP analysis. I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 23 Fig. 5. Phylogenetic tree (50% majority rule consensus tree) rendered by Bayesian analysis of the mitochondrial cytochrome b plus nuclear (RAG1 and b-actin) dataset. Numbers above nodes indicate posterior probabilities values; numbers below nodes indicate bootstrap values for the MP analysis. 24 I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 the nuclear genes RAG-1 (1452 bp) and b-actin (998 bp aligned including gaps, although PCR products varied in size from 929 bp in Girardinus metallicus, G. microdactylus and G. rivasi to 977 bp in the outgroup Crenichthys baileyi) in a subset of mitochondrial samples. The primers used for RAG-1 and b-actin were those described in Perdices et al. (2005) and Robalo et al. (2006), respectively. For nuclear RAG-1, the amplification process was conducted as follows: 94 "C (5 min), 35 cycles at 94 "C (1 min), 54 "C (1 min), 72 "C (1.30 min), and for nuclear b-actin as: 95 "C (5 min), 35 cycles at 95 "C (30 s.), 55 "C (40 s.), 72 "C (1.30 min). After checking the PCR products on 1.5% agarose gels, these were sequenced using the Big Dye Deoxy Terminator cyclesequencing kit (Applied Biosystems Inc.). All samples were sequenced using an Applied Biosystems 3700 DNA sequencer according to the manufacturer’s instructions. Chromatograms and alignments were visually inspected and verified. All sequence data were deposited in GenBank under the Accession numbers: cytochrome b (FJ178653-FJ-178778, FJ194974); RAG1 (FJ185089-FJ185104); b-actin (FJ194975-FJ194996). 2.3. Phylogenetic analysis Nucleotide saturation was assessed by plotting the absolute number of transitions and transversions against patristic distance values. Analyses were performed independently on complete gene datasets (cytochrome b, RAG-1, or b-actin) and on the total number of base pairs sequenced (3590 bp) (‘‘Total evidence”, Kluge, 1989). Nucleotide composition for all genes was determined using the v2 homogeneity test of base frequencies implemented in PAUP* 4.0b10 (Swofford, 2002). To establish divergences between lineages, we calculated uncorrected p distances and represented them as percentages. Modeltest version 3.07 (Posada and Crandall, 1998) was used to infer the best-fit model of evolution based on the Bayesian Information Criterion (BIC) (Schwarz, 1978). Model selection was performed for each individual gene and for the combined dataset, in which each gene partition was allowed to follow its own model. The aligned data were analyzed independently by Bayesian inference (BI) and maximum parsimony (MP) methods. BI analysis was performed using MrBayes 3.1.2 (Ronquist and Huelsenbeck, 2003), simulating two simultaneous Markov chain analyses (MCMC) for 3,000,000 generations each to estimate the posterior probability distribution. Topologies were sampled every 100 generations and a majority-rule consensus tree was estimated after eliminating the first 104 generations in each analysis. MP analysis was performed with the PAUP* 4.0b10 package (Swofford, 2002) through heuristic searches with 10 random stepwise additions and TBR branch swapping. Maximum parsimony results were based on a 6:1 Ti/Tv weighting, according to the empirically determined ti/tv ratio (ratio=5.85). MP trees obtained using different weighting schemes (10:1, 8:1 or equal weights) were similar and congruent (trees not shown). Confidences for this analysis were estimated by bootstrapping (1000 repetitions) (Felsenstein, 1985). The incongruence length difference test (Farris et al., 1995) as implemented in PAUP* 4.0b10 was used to determine if different gene partitions had significantly different signals, for their use in the combined dataset analyses. Since no significant differences were found (partition-homogeneity test, p > 0.05), all analyses were performed on the combined data set. To determine whether a particular tree topology corresponded to a significantly better or worse interpretation of the data than an alternative tree, we used the Shimodaira–Hasegawa test (Shimodaira and Hasegawa, 1999), as implemented in PAUP* 4.0b10. Our phylogenetic analyses were conducted in two successive steps. First, we analyzed the mitochondrial cytochrome b datasets, which represented most of the species and populations of the tribe Girardinini. From this analysis, we inferred phylogenetic relationships within the tribe Girardinini. Second, we selected a subset of the mitochondrial dataset and combined it with the corresponding taxonomic sampling of the nuclear genes RAG1 and b-actin. From this, we inferred relationships between Girardinini and other American poeciliids. We also tried to resolve phylogenetic relationships among the lineages of Girardinini unresolved by cytochrome b analysis. 2.4. Molecular clock and divergence times A semi-parametric penalized likelihood (PL) analysis was conducted using the r8s package (Sanderson, 2002) to estimate the divergence times of taxa. The PL method maximizes the log likelihood of a model with different rates on each branch, but includes a roughness penalty that costs the model more as the magnitude of changes in rates increases. The relative contribution of two components (log likelihood and roughness penalty) is determined by increases in a smoothing parameter. The TN (truncated Newton) algorithm was implemented, and smoothing parameters were chosen by cross-validated assessment (Sanderson, 2002) ranging from 100 to 108 in exponential increments of 0.5. The optimal smoothing value was 1.0. The branches that constituted a polytomy in the tree were collapsed. Divergence times were estimated using the calibration point of separation between Limia vittata from Cuba Island and L. dominicensis and L. tridens from Hispaniola. This calibration point was based on the sister relationship proposed for representatives of the group Limia domicensis from Hispaniola and Limia vittata from Cuba by Hamilton (2001). The separation time assumed between the islands of Hispaniola and Cuba (wind passage) was 14!17 Mya, as the most recent estimate (Iturralde-Vinent, 2006). Confidence intervals for divergence dates are based on the curvature of the likelihood surface (Cutler, 2000) as implemented in r8s. 3. Results 3.1. Phylogenetic analysis The mitochondrial (cytochrome b) and two nuclear (RAG-1 and b-actin) genes sequenced rendered 3590 characters (1140 mitochondrial and 2450 nuclear). Base frequencies were homogeneous across all sites (cytochrome b: v2 = 122.22, df = 384, p = 1.0; RAG-1: v2 = 16.68, df = 72, p = 1.0, b-actin: v2 = 4.30, df = 72, p = 1.0). For cytochrome b, 665 sites were variable and 538 were parsimony informative (47.2%). According to codon position, the most informative was the third (335 parsimony informative characters), followed by the first (94 characters). Plots of the absolute number of sites of each codon position against patristic distances indicated an absence of nucleotide saturation at any position (plots not shown). In the joint mitochondrial and nuclear analysis, 1105 sites were variable, and 392 were parsimony informative. MP cytochrome b analysis generated one most parsimony tree with the following parameters: length = 5239, CI = 0.209, HI = 0.791, RI = 0.708. MP analysis of all genes (mitochondrial plus nuclear) produced three trees (length = 4144, CI = 0.371, HI = 0.629, RI = 0.584). The best-fit models selected following the Bayesian Information Criterion (BIC) were the TrN + G + I model for cytochrome b, and the K80 + G model for both RAG-1 and b-actin. Rate matrices for cytochrome b were R(a) = 1.00, R(b) = 19.78, R(c) = 1.00, R(d) = 1.00, R(e) = 8.53, R(f) = 1.00. Base frequencies were: cytochrome b, A = 0.28, C = 0.35, G = 0.11, T = 0.26. RAG1 and b-actin had equal base frequencies. Among site variation was approached by the gamma distribution shape parameter a (cytochrome b: 25 I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 Crenichthysbaileyi Fuviphylaxpygmaeus Anablepsanableps Xenodexiactenolepis Poeciliopsisfasciata Poeciliareticulata Pamphorichthys hollandi Poeciliamexicana Poecilialatipunctata Limiam elanogaster Limiat ridens Limia dominicensis Limia vittata1 Limia vittata 2 Phalloptychus januarius Cnesterodomdecemmaculatus Xiphophorushelleri Heterandriabimaculata Belonesoxbelizanus Gambusia affinis Heterophallus milleri Scolichthys greewayi Neoheterandria tridentiger Carlhubbsiastuarti Carlhubbsiakidderi1 Carlhubbsiakidderi 2 GMIC1 GMIC2 GMIC3 GMIC4 GMIC5 GMIC6 GMIC7 GMIC8 GMIC9 GMIC10 GMIC11 GMIC12 GMIC13 GMIC14 GMIC15 GMIC16 GRIV9 GRIV10 GRIV11 GRIV12 GRIV13 GRIV14 GRIV1 GRIV2 GRIV3 GRIV4 GRIV5 GRIV6 GRIV7 GRIV8 GMIC17 GMIC18 GMIC19 GMET20 GMET21 GMET22 GMET19 GMET18 GMET17 GMET16 GMET15 GMET14 GMET13 GMET12 GMET11 GMET10 GMET9 GMET8 GMET7 GMET6 GMET5 GMET4 GMET3 GMET2 GMET1 TCRE4 TCRE5 TCRE6 TCRE7 TCRE8 TCRE1 TCRE2 TCRE3 TCRE9 TCRE10 TCRE11 TCRE12 TCRE13 TCRE14 TCRE15 GUNI1 GUNI11 GUNI12 GUNI13 GUNI2 GUNI3 GUNI4 GUNI5 GUNI14 GUNI15 GUNI6 GUNI7 GUNI8 GUNI9 GUNI10 GFAL1 GFAL8 GFAL9 GFAL10 GFAL11 GFAL12 GFAL2 GFAL3 GFAL4 GFAL5 GFAL6 GFAL7 GFAL13 GFAL14 GFAL15 DRAM1 DDEN9 DDEN10 DDEN1 DDEN2 DDEN3 DDEN4 DDEN5 DDEN6 DDEN7 DDEN8 Quintanaatrizona 1 Quintanaatrizona 2 Priapellacompressa Priapellaolmaceae Priapellaintermedia Priapichthysfestae Alfarocultratus 1 Alfarocultratus 2 Alfarocultratus 3 Brachyrhaphis cascajalensis Brachyrhaphis parismina Phallichthyspittieri 1 Phallichthyspittieri 2 Phallichthyspittieri 3 Phallichthys tico 1 Phallichthys tico 2 Xenphalusumbratilis 1 Xenphalusumbratilis 2 Xenphalusumbratilis 3 Brachyrhaphi s terrabensis Brachyrhaphis rhabdophora Brachyrhaphis hartwegii 1 Brachyrhaphis hartwegii 2 Girardinus microdactylus Girardinus rivasi Girardinus sp. Girardinus-Toxus split Girardinus metallicus Glaridichthys split Toxus creolus Dactylophallus split Glaridichthys uninotatus Quintana split Glaridicthys falcatus Girardinini split Dactylophallus ramsdeni Dactylophallus sp Dactylophallus denticulatus Mya Geological time Cretaceous Paleocene Eocene Oligocene Miocene Pleistocene Pliocene Fig. 6. Divergence times of the major cladogenetic events occurring in the evolutionary history of the tribe Girardinini. The scale bar below tree shows the time scale indicated by a relaxed molecular clock. 26 I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 a = 0.75; RAG-1: a = 0.30 and b-actin: a = 0.16). For cytochrome b, the proportion of invariable sites was I = 0.34. These parameters were used in the subsequent phylogenetic analyses. Considering the parameters estimated and the empirical frequencies of the nucleotides, MP and BI of the mitochondrial and mitochondrial/nuclear datasets rendered congruent topologies that in no case supported the monophyly of the tribe Girardinini (Rosen and Bailey, 1963; Parenti and Rauchenberger, 1989; Ghedotii, 2000; Lucinda and Reis, 2005) yet did support major lineages and population divergences (Figs. 4 and 5). In the mitochondrial and mitochondrial/nuclear analyses, the genus Quintana appeared as sister group of Girardinus whereas Carlhubbsia did not appear in the monophyletic clade Girardinus-Quintana. Relationships between the clade Girardinus-Quintana and the remaining poeciliids examined were not fully resolved. The combined data analysis grouped with low support (81% posterior probability) the clade Girardinus-Quintana with another two clades. The first clade was comprised of species of the genera Brachyrhaphis, Xenophallus and Phallichthys, all of them from lower Central America. The second clade clustered the species of the genera Poecilia, Limia, Cnesterodom and Pamphorichthys showing a wide distribution range across America and Greater Antilles. Relationships among the three clades were not resolved. Within the genus Girardinus, four strongly supported evolutionary independent lineages were identified: Group I included the populations of G. metallicus, G. microdactylus and G. rivasi; Group II was composed of populations of the single species G. creolus; Group III included the populations of G. uninotatus and G. falcatus; and finally, Group IV contained the populations of G. denticulatus. When we analyzed the combined data matrix, intergroup relationships were not fully resolved and we could only appreciate the sister-taxon relationship of Group IV with respect to the remaining three groups of Girardinus (Fig. 5). In contrast, our mitochondrial gene analysis resolved the relationships among these three other clades but with low support (85% posterior probability) (Fig. 4) and assigned Group III a sister position with respect to Groups I and II. For cytochrome b, genetic uncorrected ‘‘p” distances among the four lineages were x = 9.6 ± 1.2 (7.1–12.9). The minimum distance was found between Group I and Group II (x = 8.29 ± 0.52 (7.1– 9.6)) and the maximum between Group III and Group IV (x = 11.7 ± 0.52 (10–12.9)). Monophyletic Group I corresponded to the genus Girardinus as defined by Rivas (1958). Within this group, we identified two different clades, one corresponding to the populations of G. metallicus and the other to the populations of G. rivasi from Juventud Island plus G. microdactylus from Cuba Island. Within the clade formed by the G. metallicus populations, the western populations close to the Guanahacabibes Peninsula formed a lineage well separated from the others. Uncorrected ‘‘p” distances according to cytochrome b of this western population were x = 2.6 ± 0.1 (2.4–2.8). When we considered all G. metallicus populations except the one from the Guanahacabibes Peninsula, uncorrected ‘‘p” distances for populations of a wide distribution range in Cuba Island were x = 0.3 ± 0.05 (0.26–0.35). Uncorrected ‘‘p” distances for cytochrome b found between individuals from the terra typica of G. pygmaeus and the populations of G. metallicus, x = 0.4 ± 0.3 (0.09–1.1), fell within the intra-population divergence range shown by G. metallicus. Within the clade formed by the populations of G. rivasi and G. microdactylus, we found that some of these populations of G. microdactylus from eastern Cuba Island were more closely related to G. rivasi from Juventud Island than to G. microdactylus from western Cuba Island. Uncorrected ‘‘p” distances based on mitochondrial cytochrome b were as follows: x = 1.4 ± 0.2 (1.1–1.8) between populations of G. microdactylus from eastern Cuba and G. rivasi; x = 2.6 ± 0.1 (2.4–3) between the two populations of G. microdactylus, and x = 2.8 ± 0.2 (2.6–3.2) between populations of G. rivasi and G. microdactylus from western Cuba. Group II corresponded to the genus Toxus as proposed by Rivas (1958). This group includes the single species Girardinus creolus. We found three different clades formed by G. creolus populations from Taco-Taco River, La Legua River in western Cuba and the San Francisco-San Cristobal rivers. San Cristobal is the terra typica of G. riddlei (Eigenmann, 1903) and is considered by Rosen and Bailey (1963) a synonym of G. creolus. Relationships among the three clades were not resolved. The Shimodaira–Hasegawa test for cytochrome b revealed no significant differences (p > 0.05) among the three possible hypotheses, but indicated a better relationship for the San Cristobal population as sister group to the La Legua populations (–ln L = 2622.33403) than the San Cristobal population as sister to the Taco-Taco populations (–ln L = 2622.99649) or TacoTaco as sister to the La Legua populations (–ln L = 2626.78896). Within the three clades, mitochondrial genetic uncorrected ‘‘p” distances were X = 0.1 ± 0.00 (0.00–0.06). The minimum distance was detected between populations of G. creolus from La Legua and the San Francisco–San Cristobal rivers x = 1.4 ± 0.2 (1.1–1.9) and the maximum distance between the La Legua and Taco-Taco populations x = 2 ± 0.2 (1.7–2.4). Populations of the Upper Taco-Taco in Rangel, terra typica of G. serripenis, did not differ from populations downstream of the Taco-Taco falls. Genetic distances within the Taco-Taco populations were x = 0.05 ± 0.08 (0.00–0.18). Monophyletic Group III corresponded to Glaridichthys as was defined by Rivas (1958). We distinguished two different clades, one formed by G. uninotatus and the other clade comprised of the populations of G. falcatus. This G. falcatus clade showed a geographical population-structure, with four different lineages. These four lineages corresponded to populations of G. falcatus from Juventud Island, Guanahacabibes Peninsula, central Cuba Island and western populations from Cuba Island. The most differentiated lineage was that formed by the populations of the Guanahacabibes Peninsula. For cytochrome b, uncorrected ‘‘p” distances for these Guanahacabibes populations with respect to the other three lineages were x = 3.4 ± 0.2 (3.0–3.7). The other three populations exhibited less divergence. Minimum distances x = 1.0 ± 0.2 (0.6–1.4) were those among populations of G. falcatus from western Cuba. Genetic uncorrected ‘‘p” distances between the populations of G. falcatus from central Cuba versus the western Cuban populations were x = 1.6 ± 0.19 (1.3–1.7) and between central Cuba and Juventud Island were x = 2 ± 0.28 (1.7–2.4). No significant genetic differences were found between individuals of the terra typica of G. atherinoides (Mabuya-Las Chambas) and individuals of G. falcatus from the central and western area of Cuba Island (x = 0.6 ± 0.24 (0.3– 0.9)). Specimens of G. uninotatus lacking the typical black spot from several localities of Pinar del Rio province, described as G. torralbasi (Eigenmann, 1903), and typical individuals of G. uninotatus showed no genetic differences according to their cytochrome b sequences. Monophyletic Group IV corresponded to Dactylophallus as described by Rivas (1958). This group showed three different clades. The first clade was composed of the individuals from the Guaso River, terra typica of G. ramsdeni. The second clade contained specimens from central Cuba Island in Viñas district and the third clade was comprised of the remaining populations from central and eastern Cuba. For cytochrome b, minimum uncorrected ‘‘p” distances were observed between the populations from Viñas and those from central and eastern Cuba (x = 1.9 ± 0.1 (1.7–2.1). Maximum distances were those between the Guaso River populations (terra typica of G. ramsdeni) and those of central and eastern Cuba (x = 2.4 ± 0.1 (2.3–2.5). Distances between Guaso and Viñas were x = 2.2 ± 0.06 (2.2–2.3). The relationships among the three clades were unresolved by both the mitochondrial and combined mitochondrial/nuclear datasets. The Shimodaira–Hasegawa test I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 revealed no significant differences among the three different hypotheses (p > 0.05) but indicated a better relationship for the Guaso population as sister group to the central and eastern populations of G. denticulatus (–ln L = 2692.61030) than the Guaso populations as sister group to Viñas (–ln L = 2693.27613) or than the Viñas population as sister group to the central and eastern populations (–ln L = 2696.47910). 3.2. Molecular clock and divergence times A likelihood ratio test rejected the hypothesis of evolutionary rate constancy across all taxa analyzed (–lnL with clock = !11638.74, –lnL without clock = !11481.81, df = 127, p < 0.000). In the absence of a constant rate of evolution, we used the semi-parametric penalized likelihood method (Sanderson, 2002) to estimate the divergence times of taxa. Splitting of the Girardinini tribe from other poeciliids occurred approximately 62 mya (CI 95%, 44–77 mya), in the Paleocene (Fig. 6). Within the tribe Girardinini, splitting of the genus Quintana was dated in the Eocene (approximately 50 mya, CI 95% 32–67 mya). The separation of the different genera formerly considered Girardinus was ascribed to the Oligocene–Eocene periods. The first genus to split was Dactylophallus (29 mya, CI 95% 18–39 mya), followed by Glaridicththys (21 mya, CI 95% 12–27 mya) and Girardinus from Toxus (18 mya, CI 95%, 10–25 mya). 4. Discussion 4.1. Phylogeny of Girardinini and taxonomical implications Our data resulted in a well-resolved phylogeny for the tribe Girardinini, although relationships between Girardinini and other poeciliids could not be resolved using our mitochondrial and nuclear gene data. Our analyses did not support the monophyly of the tribe Girardinini as proposed by most authors (Rosen and Bailey, 1963; Parenti and Rauchenberger, 1989; Ghedotii, 2000; Lucinda and Reis, 2005). Also contrary to the findings of previous morphological studies (Rosen and Bailey, 1963; Ghedotii, 2000; Lucinda and Reis, 2005), while the genus Quintana emerged as the sister group of Girardinus, the genus Carlhubbsia seems more related to other poeciliids than to the genera Girardinus and Quintana. These data are in agreement with a previous molecular phylogeny of the poeciliids in which Carlhubbsia was assigned to a clade comprised of the genera Xiphophorus, Heterandria, Belonesox, Gambusia, Scolichthys and Priapella (Hrbek et al., 2007). Our phylogenetic analysis recovered this clade based on mitochondrial data but our nuclear gene data could not resolve the relationships between the genera Priapella and Scolichthys and the clade comprised of Carlhubbsia, Xiphophorus, Belonesox, Heterandria and Gambusia. The results of this study indicate that the tribe Girardinini should only contain species of Girardinus and Quintana, contrasting with the findings of previous morphological approaches based on the morphology of the gonopodium and on osteological data (Rosen and Bailey, 1963; Ghedotii, 2000; Lucinda and Reis, 2005). Suggestions that Girardinus is closely related to Allogambusia (= Neoheterandria) and Phalloceros (Rivas, 1958) were rejected by our analyses. The sister group of Girardinini was unresolved in our analyses because of low posterior probability support. Previous molecular phylogenies of poeciliids have established that Girardinus is a basal group to a high number of species from North and Central America (Hrbek et al., 2007). However, some of the clades recovered by our phylogeny suggest a high level of homoplasy at the morphological level has probably confounded poeciliid classification. This is obvious for traditional characters such as the gonopodium, which prob- 27 ably evolved multiple times in response to similar ecological pressures or maybe the result of sexual selection (Hankison and Morris, 2002; Langerhans et al., 2005; Meyer et al. 2006). In effect, as for the Girardinini, recent morphological and molecular studies are now beginning to reject the monophyly of these poeciliid tribes (Ghedotii, 2000; Lucinda and Reis, 2005; Hrbek et al., 2007). Within the tribe Girardinini, the phylogenetic relationships defined here are in disagreement with those proposed by Rivas (1958) on the basis of morphological characters. Notwithstanding, our monophyletic clades corresponded to the four genera: Girardinus Poey, 1854; Dactylophallus Howell-Rivero and Rivas, 1944; Glaridichthys Garman, 1896; Toxus Eigenmann, 1907 recognized as Girardinini by Rivas (1958) along with the genus Quintana Hubbs, 1934. These four genera were monophyletic and their genetic divergence was similarly high to that found in other Cyprinidontiform genera (8–11% (x = 10%)) such as those of the Goodeidae family (Doadrio and Domínguez, 2004). Hence, the genera recognized by Rivas (1958) and corroborated by our molecular phylogeny, are of evolutionary significance and strongly supported by morphological and genetic characters. The tribe Girardinini should thus comprise the four genera recognized by Rivas (1958): Dactylophallus, Girardinus, Glaridichthys and Toxus plus the genus Quintana. The genus Carlhubbsia should be excluded from the tribe Girardinini on the basis of our molecular phylogeny and that reported by Hrbek et al. (2007). The species Toxus creolus has been considered the ancestor of all other Girardinini representatives (Rivas, 1958; Rodriguez et al., 1992). Our phylogeny based on sister groups rather than ancestor–descendant histories revealed that the genus Dactylophallus was the one that first split from the remaining three genera (Girardinus, Glaridichthys and Toxus). Girardinus and Glaridichthys are sister taxa. This idea that Dactylophallus was the first to break away is consistent with the separation of the islands of Cuba and Juventud into two endemic regions on the basis of their freshwater fishes (Rauchenberger, 1988). Dactyllophalus is the only Girardinini genus that inhabits easternmost Cuba and has a population in central Cuba, while Girardinus, Glarydichthys and Toxus inhabit central and western Cuba Island and Juventud Island yet not the easternmost part of Cuba. Our phylogenetic hypothesis has further major taxonomic implications. The described species Girardinus pygmaeus, Toxus riddlei, T. serripenis and Glaridichtys atherinoides showed no significant genetic differences with respect to Girardinus metallicus, Toxus creolus and Glaridichthys falcatus respectively. These results are in agreement with a previous morphological study that failed to consider the former four species as valid (Rosen and Bailey, 1963). In contrast, Dactyllophalus ramsdeni varied sufficiently (genetic differences of 2.2% and 2.4% versus the populations of G. denticulatus) to be assigned its own identity and should be considered a different species to G. denticulatus and not its subspecies. To solve the problem of paraphyly in the genera Girardinus and Dactyllophallus, we proposed the existence of taxa not formally described. Unexpectedly, we found populations of Girardinus microdactylus in central Cuba outside its distribution range in western Cuba. These populations showed less divergence with respect to G. rivasi from Juventud Island than to coespecific populations from the western area of Cuba. Divergence between G. rivasi and G. microdactylus was sufficiently high but the latter is not considered a recent introduction. To avoid the paraphyly problem, we provisionally considered G. microdactylus from central Cuba as a different species. Similarly, within Dactylophallus denticulatus, we recognized the species D. ramsdeni from the Guaso basin, the same locality of the endemic cichlid Nandopsis ramsdeni (Fowler, 1938), and a non-formally described species from central Cuba (Viñas district). 28 I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 Further, our phylogenetic analysis suggests the existence of several well-differentiated populations that could be considered as different species after a morphological study. Girardinus metallicus and Glaridichthys falcatus from Guanahacabibes Peninsula showed a high degree of differentiation similar to that of valid species such as Girardinus microdactylus and G. rivasi. The populations of Toxus creolus showed high divergence, probably due to isolation of the species in highland streams. These observations prompt the need for a taxonomic revision of all these populations. A provisional taxonomy based on the results of this genetic study is provided in Table 1. 4.2. Divergence times and biogeography The evolution of the Antillean islands has puzzled biogeographers for over two centuries (Wallace, 1881). Research on the origins of Antillean fauna has rendered three main models with implications for the origin and evolution of poeciliids. The classic vicariant model was essentially proposed by Rosen (1975, 1985). This model hypothesized that poeciliids inhabited the proto-Antillean arc during the Cretaceous (144–65 Mya), when interchange between North America and South America was possible, and became isolated when the arc was displaced eastwards. The dispersalist hypothesis as proposed by Myers (1966), essentially involved recent colonizations of Central America during the Neogene by freshwater fish fauna. This author indicated that Central America has been the Poeciliidae family’s center of origin since the Neogene. The third model combines dispersal and vicariant hypotheses to suggest that the current Antillean fauna came from South America, reaching the islands through GAArlandia (Greater Antilles + Aves Ridge) in the Eocene–Oligocene (33–35 Mya) (Iturralde-Vinent and MacPhee, 1999; Iturralde-Vinent, 2006). For poeciliid fishes, this last model has recently been proposed as the most plausible based on molecular clock approaches (Hrbek et al., 2007). The GAArlandia model, nevertheless, poses some inconsistencies for freshwater fish fauna. An example is the presence of populations of the genus Ophisternon in western Cuba and Juventud Island since its sister group occurs in the Yucatan Peninsula and is absent from other Central and South American regions (Perdices et al., 2005). According to the GAArlandia model, there was no connection between the Yucatan and western Cuba and Juventud Island during the Cenozoic (Iturralde-Vinent and MacPhee, 1999; Iturralde-Vinent, 2006). For mammals, the presence of the insectivore Solenodon in Cuba and Hispaniola with a divergence with respect to other insectivore mammals dated during the Cretaceous (Roca et al., 2004) created conflict between the GAArlandia model and molecular clock datings. In other vertebrates, the dispersalist model recovered the inconsistencies between molecular clocks and biogeographical models (Matthew et al, 2007). However, for fishes such as Girardinus with a low dispersal capacity restricted to contact between different freshwater systems, the dispersion by flotsam proposed by dispersalist models (Matthew et al., 2007) is highly unlikely. If inconsistencies between molecular clock and palaeogeographical data exist, explanations must be looked for in the concepts of these doctrines rather than trying to fit models to the wide dispersion range of freshwater fishes. The estimated divergence time of the endemic genera of Cuba Girardinus and Quintana from the other poeciliids was approximately 62 Mya (CI 95%, 44–77 Mya), in the Upper Cretaceous-Middle Eocene. This is in agreement with datings provided by other vertebrates such as Solenodon (Roca et al., 2004) and Cricosaura (Hedges et al., 1991, 1992). Several arguments have been proposed to reject the ancient colonization during the Upper CretaceousMiddle Eocene, of current fauna of the West Antillean islands. One such theory is that the extinction of sister taxa could affect the molecular clock calibration (Roca et al., 2004). A further explanation for this estimated divergence could be the use of dispersal by flotsam models (Hedges 1996). Based on our data, we cannot rule out the extinction of a probable sister group of endemic Cuban taxa on the main or other Antillean islands. The use of the geological event of separation between Cuba and Hispaniola for the genus Limia could also be a possible source of error. The model of Hrbek et al. (2007) for poeciliid fishes is more congruent with the GAArlandia theory (Iturralde-Vinent and MacPhee, 1999) in terms of colonization of the Caribbean islands. The divergence time for the main radiation of poeciliid fishes has been estimated at 44 Mya (CI 95% 40.78–48.97 Mya) (Hrbek et al, 2007), which is a little earlier than the date of 35–33 Mya assigned to the formation of GAArlandia (Iturralde-Vinent, 2006). The molecular clock calibration for poeciliids (Reznick et al., 2002; Hrbek et al., 2007) uses the formation of the Trans-Mexican Volcanic Belt to calculate the maximum age for the separation of several species of Poeciliopsis. However, the Trans-Mexican Volcanic Belt formation process was a long and intermittent one involving several pulses. For instance pulses dated at 10, 23 and 30 Mya ago have been reported (Ferrari et al., 1999). The use of this type of barrier to date the separation of organisms has the limitation that it is difficult to establish a fixed date for a given area. We believe that the separation between the Caribbean islands is more reliable to determine calibration points. The use of the genus Limia as a calibration point, which shares its habitat with Girardinus and is similar in body size to this species, avoids the problems of calculating substitution rates based on organisms with different metabolic rates (Estabrook et al., 2007). The lack of congruence between the separation date of the Girardinini inhabiting Cuba from sister groups on the mainland could be attributable to the extinction of some sister groups (see Iturralde-Vinent and MacPhee, 1999). Otherwise, we would have to refute the GAArlandia hypothesis and freshwater fishes could have inhabited western Cuba since 60 Mya (Cretaceous period). Fossil data are essential for verifying palaeogeographical data or molecular clock calibrations. However, freshwater fish fossils from the Cenozoic of Cuba have not been examined. Within the tribe Girardinini, the separation between Girardinus and the other Cuba and Juventud island endemism Quintana occurred approximately 50 Mya ago (CI 95%, 32–67 Mya). Girardinus is widely distributed throughout Cuba and Juventud Island. Quintana exhibits a more reduced distribution area than Girardinus, restricted to western Cuba and Juventud Island. This distribution cannot be explained by current palaeogeographical models if we accept that speciation is explained by allopatry. Radiation of the main lineages of the tribe Girardinini took place in the late Oligocene-Early Miocene, Dactylophallus being the first genus to split (29 Mya, CI 95% 18–39 Mya). This radiation is consistent with the end of the GAArlandia stage and the isolation of different portions of land on the islands of Cuba and Juventud (Iturralde-Vinent and MacPhee, 1999). Populations of clades such as G. metallicus and G. falcatus from the Guanahacabibes Peninsula diverged from other Cuban populations in the Pliocene (approximately 5–3 Mya). This date suggests that Guanahacabibes became isolated from the rest of Cuba Island during the Pliocene. However, there is insufficient palaeogeographical information for this period to explain the divergence times of these populations. The palaeogeographical data on Juventud Island suggest its ancient isolation from Cuba Island at least since the Oligocene, and then a more recent isolation during the Plio-Pleistocene. It is not possible to infer the ancient isolation of Juventud Island from the current distribution of freshwater fishes. However, the presence of Quintana, the sister group of former Girardinus, in Juventud Island could be explained by this ancient event of isolation followed I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 by dispersion to the few localities inhabited by the species in Cuba Island. This theory, nevertheless, awaits the support of population genetics studies. The possibility of a more recent isolation of Juventud Island could be examined by determining the presence of the endemic species G. rivasi or the well-differentiated populations of G. falcatus. The data for these freshwater fishes indicate the isolation of Juventud Island during the late Pliocene (2–1.3 Mya). This date must be considered with caution since the calibration point used to calculate the molecular clock is older. A 1–2 Mya time frame has been described as critical for modifying molecular rates (Ho and Larson, 2006) and could overestimate divergence times (Waters et al., 2007). The palaeogeographical data on Girardinus seem to be adequate for testing the different biogeographical models. Divergence times for these vertebrates seem to support Rosen’s vicariant ideas to explain the high divergence between the tribe Girardinini and its relatives from the mainland. However, the high extinction rate assumed for the Chixchulub impact (Izzet et al. 1991; Swisher et al, 1992), could confound the data since the sister group of Girardinini could be extinct. The absence of Girardinini from the other Antillean islands, especially Hispaniola, and its high diversity in western Cuba does not explain its passage from South America through GAArlandia such that the origin of the tribe remains unknown. Acknowledgments We are grateful to R. Fimia Duarte, M. Díaz Pérez, A. Chamizo, V. Rivalta, A. de Sostoa, P. Garzon and O. Dominguez for help in field collection. D. McAllister and K. de Jong sent us valuables specimens. A. Burton reviewed the English of the manuscript. We also thank L. Rojas Rivero for her support and M. Iturralde-Vinent for helpful suggestions while preparing the manuscript. This study was financed by project A/9998/07. References Barus, V., Libosvarsky, J., 1986. Some new data on Dactylophallus denticulatus (Poeciliidae) from Cuba. Folia Zool. 35 (4), 357–362. Barus, V., Libosvarsky, J., Guerra, F., 1981. Notes on Toxus creolus (Poeciliidae) from Cuba. Folia Zool. 30 (4), 363–370. Barus, V., Libosvarsky, J., Martínez, A.P., 1986. Supplements to the description of Girardinus microdactylus (Pisces: Poeciliidae). Folia Zool. 36, 173–194. Barus, V., Libosvarsky, J., Martínez, A.P., 1991. On the Toxus creolus vs. T. serripenis (Poeciliidae) and remarks to the tribe Girardinini from Cuba. Folia Zool. 40 (3), 261–272. Barus, V., Peñaz, M., Prokes, M., 1998. Some new data on Girardinus cubensis (Poeciliidae) from Cuba. Folia Zool. 47, 287–293. Barus, V., Wohlgemuth, E., 1994. Two proposed subspecies in Girardinus microdactylus (Pisces: Poeciliidae) from Isla de Cuba and Isla de la Juventud. Folia Zool. 43 (3), 245–254. Cutler, D.J., 2000. Estimating divergence times in the presence of an overdispersed molecular clock. Mol. Biol. Evol. 17, 1647–1660. Doadrio, I., Domínguez, O., 2004. Phylogenetic relationships within the fish family Goodeidae based on cytochrome b sequences data. Mol. Phylogenet. Evol. 31, 416–430. Estabrook, G.F., Smith, G.R., Dowling, T.E., 2007. Body mass and temperature influence rates of mitochondrial DNA evolution in North American cyprinid fish. Evolution 61, 1176–1187. Farris, J.S., Källersjö, M., Kluge, A.G., Bult, C., 1995. Testing significance of incongruence. Cladistics 10, 315–319. Felsenstein, J., 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39, 783–791. Ferrari, L., López-Martínez, M., Aguirre-Díaz, G., Carrasco-Núñez, G., 1999. Spacetime patterns of Cenozoic arc volcanism in central Mexico: from the Sierra Madre Occidental to the Mexican Volcanic Belt. Geology 27, 303–306. Ghedotii, M.J., 2000. Phylogenetic analysis and taxonomy of the poecilioid fishes (Teleostei: Cyprinodontiformes). Zool. J. Linn. Soc. 130, 1–53. Hamilton, A., 2001. Phylogeny of Limia (Teleostei: Poeciliidae) Based on NADH Dehydrogenase Subunit 2 Sequences. Mol. Phylogenet. Evol. 19 (2), 277–289. Hankison, S.J., Morris, M.L., 2002. Avoiding a compromise between sexual selection and species recognition: female swordtail fish assess multiple species-specific cues. Behav. Ecol. 14, 282–287. Hedges, S.B., 1996. Historical biogeography of West Indian vertebrates. Annu. Rev. Ecol. Syst. 27, 163–196. 29 Hedges, S.B., Bezy, R.L., Maxson, L.R., 1991. Phylogenetic relationships and biogeography of Xantusiid lizards, inferred from mitochondrial DNA sequences. Mol. Biol. Evol. 8, 767–780. Hedges, S.B., Hass, C.A., Maxson, L.R., 1992. Caribbean biogeography: Molecular evidence for dispersal in West Indian terrestrial vertebrates. Proc. Natl. Acad. Sci. USA 89, 1909–1913. Ho, S.Y.W., Larson, G., 2006. Molecular clocks: when times are a-changin’. Trends Genet. 22, 79–83. Howell-Rivero, L., Rivas, L.R., 1944. Studies of Cyprinodont fishes. Two new genera of the Tribe Girardinini from Cuba. Torreia 12, 1–19. Hrbek, T., Seckinger, J., Meyer, A., 2007. A phylogenetic and biogeographic perspective on the evolution of poeciliid fishes. Mol. Phylogenet. Evol. 43, 986–998. Iturralde-Vinent, M.A., 2006. Meso-Cenozoic Caribbean Paleogeography: Implications for the Historical Biogeography of the Region. Int. Geol. Rev. 48, 791–827. Iturralde-Vinent, M.A., MacPhee, R.D.E., 1999. Paleogeography of the Caribbean region: implications for Cenozoic Biogeography. Bull. Amer. Mus. Nat. Hist. 238, 1–95. Izzet, G.A., Dalyrymple, G.B., Snee, L.W., 1991. 40Ar-39Ar age of Cretaceous-Tertiary boundary tektites from Haiti. Science 252, 1539–1542. Kluge, A.G., 1989. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Syst. Zool. 38, 7–25. Langerhans, R.B., Layman, C.A., DeWitt, T.J., 2005. Male genital size reflects a tradeoff between attracting mates and avoiding predators in two livebearing fish species. Proc. Natl. Acad. Sci. USA 102, 7618–7623. Lucinda, P.H.F., 2003. Family Poeciliidae. In: Reis, R.E., Kullander, S.O., Ferraris, C.J. (Eds.), Checklist of the freshwater fishes of South and Central America. EDIPUCRS, Porto Alegre, Brazil, pp. 555–581. Lucinda, P.H.F., Reis, R.E., 2005. Systematics of the subfamily Poeciliinae Bonaparte (Cyprinodontiformes: Poeciliidae), with an emphasis on the tribe Cnesterodontini Hubbs. Neotrop. Ichthyol. 31, 1–60. Machordom, A., Doadrio, I., 2001. Evidence of a Cenozoic Betic-Kabilian connection based on freshwater fish phylogeography (Luciobarbus, Cyprinidae). Mol. Phylogenet. Evol. 18, 252–263. Matthew, P.H., Duellman, W.E., Hedges, S.B., 2007. Major Caribbean and Central American frog faunas originated by ancient oceanic dispersal. PNAS 104 (24), 10092–10097. Menéndez Díaz, Z., Manso Valdés, E., Castex Rodríguez, M., Fuentes González, O., Hernández Contreras, N., García Ávila, I., 2007. Algunos aspectos ecológicos de la ictiofauna larvífaga presente en Cayo Santa María, Cuba. Rev. Cubana Med. Trop. (3), 59. Meyer, A., Salzburger, W., Scharti, M., 2006. Sexual selection drives hybrid origin of a new species of the swordtail (Xiphophorus clemenciae, Pisces). Mol. Ecol. 15, 721–730. Myers, G.S., 1966. Derivation of the freshwater fish fauna of Central America. Copeia 4, 766–773. Nelson, G., Rosen, R., 1981. Vicariance Biogeography: A Critique: Symposium of the Systematics Discussion Group of the American Museum of Natural History, May 2–4, 1979. Columbia University Press, New York, NY. Parenti, L.R., 1981. A phylogenetic and biogeographic analysis of cyprinodontiform fishes (Teleostei, Atherinomorpha). Bull. Am. Mus. Nat. Hist. 168, 335–557. Parenti, L.R., Rauchenberger, M., 1989. Systematic overview of the poeciliins. In: Meffe, G.K., Snelson, F.F., Jr. (Eds.), Ecology and Evolution of Livebearing Fishes (Poeciliidae). Prentice Hall Inc., Englewood Cliffs, p. 453. Perdices, A., Doadrio, I., Bermingham, E., 2005. Evolutionary history of the synbranchid eels (Teleostei: Synbranchidae) in Central America and the Caribbean islands inferred from their molecular phylogeny. Mol. Phylogenet. Evol. 37, 460–473. Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14, 817–818. Rauchenberger, M., 1988. Historical Biogeography of Poeciliid fishes in the Caribbean. Syst. Zool. 37, 356–365. Reznick, D.N., Mateos, M., Springer, M.S., 2002. Independent origins and rapid evolution of the placenta in the fish genus Poeciliopsis. Science 298, 1018–1020. Rivas, L.R., 1958. The origin, evolution, dispersal, and geographical distribution of the Cuban poeciliid fishes of the tribe Girardinini. Proc. Am. Philos. Soc. 102, 281–320. Rivas, L.R., 1986. Comments on Briggs (1984): Freshwater fishes and biogeography of Central America and the Antilles. Syst. Zool. 35, 633–639. Robalo, J.I., Sousa-Santos, C., Levy, A., Almada, V.C., 2006. Molecular insights on the taxonomic position of the paternal ancestor of the Squalius alburnoides hybridogenetic complex. Mol. Phylogenet. Evol. 39, 276–281. Roca, A.L., Bar-Gal, G.K., Eizirik, E., Helgen, K.M., María, R., Springer, M.S., O’Brian, S.J., Murphy, W.J., 2004. Mesozoic origin for West Indian insectivores. Nature 429, 649–651. Rodriguez, H.L., Du-Bouchet, A., Smith, M.L., 1992. Phylogenetic position of the Cuban Poeciliid Fish, Girardinus creolus (Cyprinodontiformes). Copeia 1992, 474–476. Rodríguez, J., González, E., Hernández, N., Capó, V., García, I., 2004. Comparación morfológica e histológica del tubo digestivo de Gambusia puncticulata y Girardinus metallicus, peces utilizados en el control biológico de mosquitos. Rev. Cubana Med.Trop. 56 (1), 73–76. Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19, 1572–1574. Rosen, D.E., 1975. A vicariance model of Caribbean biogeography. Syst. Zool. 24, 431–464. 30 I. Doadrio et al. / Molecular Phylogenetics and Evolution 50 (2009) 16–30 Rosen, D.E., 1978. Vicariant patterns and historical explanation in biogeography. Syst. Zool. 27, 159–188. Rosen, D.E., 1985. Geological hierarchies and biogeographic congruence in the Caribbean. Ann. Missouri Bot. Garden 72 (4), 636–659. Rosen, D.E., Bailey, R.M., 1963. The poeciliidae fishes (Cyprinodontiformes), their structure, zoogeography, and systematics. Bull. Am. Mus. Nat. Hist. 126, 1–176. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular cloning: A Laboratory Manual, second ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sanderson, M.J., 2002. Estimating absolute rates of molecular evolution of divergence times: a penalized likelihood approach. Mol. Biol. Evol. 19, 101–109. Schwarz, G., 1978. Estimating the dimension of a model. Ann. Stat. 6 (2), 461–464. Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116. Swisher, C.C., Grajales-Nishimura, J.M., Montanari, A., Margolis, S.V., Claeys, P., Alvarez, W., Renne, P., Cedillo-Pardo, E., Maurasse, F.J.-M.R., Curtis, G.H., Smit, J., McWilliams, M.O., 1992. Coeval 40Ar/39Ar ages of 65.0 million years ago from Chixchulub crater melt rock and Cretaceous-Tertiary boundary tektiles. Science 257, 954–958. Swofford, D. L., 2002. PAUP* v. 40b10. Phylogenetic Analysis Using Parsimony (and other methods). Ver.4. Sunderland, M. A. Sinauer Associates, Inc. Wallace, A.R., 1881. Island Life or the Phenomena and causes of Insular Fauna and Floras, Including a Revision and Attempted Solution of the Problem of Geological Climates. Harper & Brothers, New York. Vol. 8, pp. 522. Waters, J.M., Rowe, D.L., Apte, S., King, T.M., Wallis, G.P., Anderson, L., Norris, R.J., Craw, D., Burridge, C.P., 2007. Geological dates and molecular rates: rapid divergence of rivers and their biotas. Syst. Biol. 56 (2), 271–282.

RELATED PAPERS

RELATED TOPICS

Log In

Molecular phylogeny and biogeography of the Cuban genus Girardinus Poey, 1854 and relationships within the tribe Girardinini (Actinopterygii, Poeciliidae

Molecular phylogeny and biogeography of the Cuban genus Girardinus Poey, 1854 and relationships within the tribe Girardinini (Actinopterygii, Poeciliidae

Molecular phylogeny and biogeography of the Cuban genus Girardinus Poey, 1854 and relationships within the tribe Girardinini (Actinopterygii, Poeciliidae

Molecular phylogeny and biogeography of the Cuban genus Girardinus Poey, 1854 and relationships within the tribe Girardinini (Actinopterygii, Poeciliidae

Molecular phylogeny and biogeography of the Cuban genus Girardinus Poey, 1854 and relationships within the tribe Girardinini (Actinopterygii, Poeciliidae

Related Papers

RELATED PAPERS

RELATED TOPICS