Fill and Sign the Az Joint Tax Application Form

Molecular Ecology (2011) 20, 342–356 doi: 10.1111/j.1365-294X.2010.04949.x Replicated hybrid zones of Xiphophorus swordtails along an elevational gradient Z . W . C U L U M B E R , * † H . S . F I S H E R , * M . T O B L E R , ‡§ M . M A T E O S , § P . H . B A R B E R , – M . D . S O R E N S O N * * and G . G . R O S E N T H A L * † *Texas A&M Department of Biology, 3258 TAMU, College Station, TX 77843-3258, USA, †Centro de Investigaciónes Cientı́ficas de las Huastecas ‘Aguazarca’, Calnali, Hidalgo, Mexico, ‡Department of Zoology, Oklahoma State University, Stillwater, OK 74078, USA, §Department of Wildlife and Fisheries Sciences, Texas A&M University, College Station, TX 77843, USA, –Department of Ecology and Evolutionary Biology, University of California, Los Angeles, CA 90095, USA, **Department of Biology, Boston University, Boston, MA 02215, USA Abstract Natural hybrid zones provide opportunities to study a range of evolutionary phenomena from speciation to the genetic basis of fitness-related traits. We show that widespread hybridization has occurred between two neo-tropical stream fishes with partial reproductive isolation. Phylogenetic analyses of mitochondrial sequence data showed that the swordtail fish Xiphophorus birchmanni is monophyletic and that X. malinche is part of an independent monophyletic clade with other species. Using informative single nucleotide polymorphisms in one mitochondrial and three nuclear intron loci, we genotyped 776 specimens collected from twenty-three sites along seven separate stream reaches. Hybrid zones occurred in replicated fashion in all stream reaches along a gradient from high to low elevation. Genotyping revealed substantial variation in parental and hybrid frequencies among localities. Tests of FIS and linkage disequilibrium (LD) revealed generally low FIS and LD except in five populations where both parental species and hybrids were found suggesting incomplete reproductive isolation. In these locations, heterozygote deficiency and LD were high, which suggests either selection against early generation hybrids or assortative mating. These data lay the foundation to study the adaptive basis of the replicated hybrid zone structure and for future integration of behaviour and genetics to determine the processes that lead to the population genetic patterns observed in these hybrid zones. Keywords: elevational gradient, introgression, linkage disequilibrium, local adaptation, Poeciliidae, population genetics Received 10 June 2010; revision received 21 October 2010; accepted 2 November 2010 Introduction Natural hybridization is an important evolutionary mechanism in the diversification of both plants and animals (Arnold 1997; Dowling & Secor 1997; Rieseberg et al. 2003). The mixing of divergent genomes from different parent taxa can generate new genetic combinations leading to novel, transgressive phenotypes upon which selection can act (Rieseberg et al. 1999; Bell & Travis 2005). As a result of the variation generated by hybridizaCorrespondence: Zachary W. Culumber, Fax: +1 979 845 2891; E-mail: zculumber@tamu.edu tion, hybrid populations are subject to a broader range of possible evolutionary trajectories (Guillaume & Whitlock 2007; Kalisz & Kramer 2008). The relative fitness of this broad variety of new phenotypes ultimately determines the stability and fate of hybrid zones (Barton & Hewitt 1985; Barton 2001; Burke & Arnold 2001). Much research on hybridization focuses on effects of intrinsic post-zygotic reproductive isolation, where incompatibilities between genes from both parents affect hybrid fitness reducing survivability or fertility (for review see Burke & Arnold 2001). In such cases with reduced hybrid fitness, hybrid zones can be stably maintained by continual immigration of parental forms  2010 Blackwell Publishing Ltd H Y B R I D Z O N E S A L O N G A N E L E V A T I O N A L G R A D I E N T 343 to the centre of the zone (Barton & Hewitt 1985). More recent work, however, suggests that extrinsic postzygotic isolation, in which hybrid fitness is determined by an interaction of hybrid genotype and the environment, may be more common than previously thought (Schluter 2009; Schluter & Conte 2009; Johannesson et al. 2010). In those cases, a hybrid zone can be maintained with negligible immigration of parental forms, and hybrids are expected to be at least as fit as the parentals in intermediate ecotonal habitats (‘bounded hybrid superiority’ model; Moore 1977). Since intrinsic isolating mechanisms can be relatively easily assessed in the laboratory, there is a wealth of data on genetic incompatibilities, while environmentally dependent hybrid fitness remains poorly understood (Wolf et al. 2010). Environmental effects of hybridization can be studied where different species adapt to local environmental conditions (Kawecki & Ebert 2004), and closely related species occupy proximate but environmentally distinct habitats (Fuller et al. 2007; Tobler et al. 2008). Environ- mentally intermediate zones, where locally adapted species may come in contact, may facilitate hybridization if there is a breakdown in reproductive isolation. Accordingly, natural hybrid zones commonly occur along gradients of climatic and ecological variables (Yanchukov et al. 2006; Nikula et al. 2008; Ruegg 2008). If selection along environmental gradients is driving hybrid zone formation and maintaining hybrid zone structure, as in the model of ‘bounded hybrid superiority’, hybrid zones with replicated structure should occur in independent tributaries across the landscape wherever the selective gradient occurs. Swordtail fish of the genus Xiphophorus allow us to test this hypothesis in a natural setting. Xiphophorus birchmanni and X. malinche form natural hybrids in the Sierra Madre Oriental of eastern Mexico (Rosenthal et al. 2003). X. birchmanni and X. malinche are members of the monophyletic, northern or Rı́o Pánuco clade of swordtails (Fig. 1). Kallman & Kazianis (2006) suggested that the northern swordtails diversified as a Fig. 1 Strict consensus tree based on analyses using three maximum-likelihood (ML) models implemented in PAUP. Bootstrap support values from ML analyses (GRL-Garli, PAUP and RxMRAxML) and Bayesian posterior probabilities (BYS-MRBAYES) from separate analyses are shown for the nodes of interest. Values not shown were below 50. Sequences from the present study are highlighted in grey. XM2, the only mitochondrial sequence observed in X. cortezi, is shared with X. malinche.  2010 Blackwell Publishing Ltd 344 Z . W . C U L U M B E R E T A L . result of the formation of the Sierra Madre Oriental of eastern Mexico. Uplifting and subsequent folding of the landscape produced isolated species endemic to small geographic areas, like X. malinche, X. continens, X. montezumae, X. multilineatus and X. nigrensis (Rauchenberger et al. 1990; Kallman & Kazianis 2006). Phylogenetic hypotheses have differed somewhat in the placement of X. birchmanni and X. malinche. Most phylogenetic analyses including behavioural, morphological and randomly amplified polymorphic DNA (RAPD) data have placed them as sister species (Rauchenberger et al. 1990; Borowsky et al. 1995; Marcus & McCune 1999; Morris et al. 2001), while Meyer et al.’s (1994) phylogenetic hypothesis based on sequence data from three loci placed them as distant relatives within the clade. Xiphophorus birchmanni and X. malinche differ in a number of morphological traits (Rauchenberger et al. 1990; Rosenthal et al. 2003), most notably the lack of the sexually-dimorphic sword on the caudal fin in X. birchmanni, which has been secondarily lost most likely due to a reversal of female sexual preferences (Wong & Rosenthal 2006). The distributions of X. birchmanni and X. malinche also differ. Xiphophorus malinche is typically found in headwaters and highland streams, while X. birchmanni are found at lower elevations (Rauchenberger et al. 1990; Rosenthal et al. 2003; Gutiérrez-Rodrı́guez et al. 2008). Xiphophorus birchmanni has a wider distribution overall but borders the distribution of X. malinche in our study area, and the two were previously documented in sympatry without mention of hybrids (Rauchenberger et al. 1990). Natural hybrids between these species have been abundant in populations at least since the late 1990s (Rosenthal et al. 2003). Morphological and electrophoretic studies of specimens collected in 1997 revealed extensive hybridization at intermediate elevations of the Rı́o Calnali with an upstream-to-downstream gradient from malinche-type to birchmanni-type morphological and isozyme traits (Rosenthal et al. 2003). The breakdown in reproductive isolation in areas of parapatry may be facilitated by recent increases in organic pollution that resulted in impaired ability to distinguish between species-typical olfactory cues (Fisher et al. 2006). In the present study we investigated the X. birchmanni–X. malinche hybrid zones with three primary objectives: (i) To test whether the parental populations at either end of the hybrid zone are indeed phylogenetically distinct X. birchmanni and X. malinche. (ii) To develop informative SNP markers from DNA sequence data. (iii) To test for an upstream to downstream gradient in genetic markers in multiple stream reaches. We use phylogenetic analysis of mitochondrial sequence data to demonstrate that X. birchmanni and X. malinche form two distinct lineages found at opposite ends of an elevational gradient. We further show that hybrids are present in at least seven distinct stream reaches separated by mountain ridges and that each hybrid zone is characterized by elevational gradients in allele frequencies. By analysing patterns of co-segregation in single nucleotide polymorphism (SNP) markers designed from mitochondrial and unlinked nuclear DNA loci, we show substantial variation in population structure both within and among stream reaches. Methods Sampling We collected whole specimens or fin clips of X. malinche, X. birchmanni and their hybrids from 39 sites in the states of Hidalgo, San Luis Potosi and Veracruz, Mexico between 2003 and 2007 (Fig. 2, Table 1). Though collecting occurred between 2003 and 2007, for each population, samples from only one point in time were used for SNP genotyping. At a few sites, samples were collected in both 2003 and 2007, with the first sample used for DNA sequencing and the second for SNP genotyping. As previous mark–recapture experiments found no recaptures after 3 years (GGR and HSF, unpublished data), it is unlikely we re-sampled any of the same individuals. Another northern swordtail, X. cortezi, was collected from the Rı́o Axtla due to past uncertainties over its relationship with X. malinche (see sources above on placement of X. birchmanni and X. malinche). For outgroup comparison, the more distantly related variable platyfish, Xiphophorus variatus, was collected from the Rı́o Garces and the Rı́o Venado. Tricaine methylsulfonate (MS-222) was used to anesthetize fish for photographing or euthanize fish prior to preservation. For genotyping, we either removed a small piece of the upper portion of the caudal fin, or preserved the whole fish. Tissues were stored in 70–95% ethanol. DNA extraction, mtDNA and intron amplification and sequencing Whole genomic DNA was extracted from fin clips with a DNeasy tissue kit (Qiagen Inc.) according to the manufacturer’s protocol. The mitochondrial control region d-loop (CR) was amplified with polymerase chain reaction (PCR) using primers CR-A and CR-E (Lee et al. 1995). Reaction conditions for 15 lL PCRs were as follows: 10· PCR Buffer, 2.5 mM MgCl2, 0.5 lM each F and R primers, 1.5 lL dNTPs, 0.625 U Taq polymerase, 1 lL genomic DNA (gDNA) at 20 ng ⁄ lL and ddH2O to 15 lL. A hot-start PCR was used for mtDNA. Briefly, samples containing all of the above components except Taq polymerase were placed in a thermocycler. A touchdown thermocycler protocol was initiated but the  2010 Blackwell Publishing Ltd H Y B R I D Z O N E S A L O N G A N E L E V A T I O N A L G R A D I E N T 345 Fig. 2 Body shape variation in male Xiphophorus from the Rio Calnali. Depicted are mean residual principal component scores (corrected for allometric effects) and standard errors of measurement for parental species, backcrosses, and F2s. Note that backcrosses cluster with parental species. The thinplate spline transformation grids show shape variation along each principal component axis. program was paused when it reached 80 C at which time Taq polymerase was added to each sample. The program was resumed and consisted of the following: 94 C for 7 min, 12 cycles of 94 C for 15 s, 66–60 C for 15 s decreasing by 0.5 C each cycle, 72 C for 1 min followed by 28 cycles at 60 C with a final extension step at 72 C for 7 min. We amplified three nuclear introns that map to distinct genetic linkage groups in crosses of X. maculatus and X. hellerii (Walter et al. 2004). DNA ligase 1 (LIG1; linkage group VI), DNA polymerase beta (POLB; linkage group XII), tumour protein 53 (TP53; linkage group XIV). The same PCR conditions and touchdown protocol as above were used for introns except a hot-start was not used. Following all PCRs of introns and mtDNA, 2 lL of each sample was scored on a 2% agarose gel to confirm amplification success and the remaining volume was then sent for sequencing in both forward and reverse directions at either High Throughput Sequencing Solutions (University of Washington) or the Nevada Genomics Center (University of Nevada – Reno). All sequences were aligned in Sequencher 4.2 or higher (Gene Codes Corp.) with further manual adjustments made by eye. Phylogenetic analyses To reconstruct phylogenetic relationships among taxa, we used all mtDNA haplotypes of X. birchmanni, X. malinche, X. cortezi and X. variatus sequenced in this study, as well as other homologous Xiphophorus sequences downloaded from GenBank (accession nos: AF404290, DQ235814–DQ235835, DQ445669–DQ445681, and EF533642–EF533649). The monophyly of the northern  2010 Blackwell Publishing Ltd swordtails, including X. birchmanni and X. malinche, is well supported (Meyer et al. 1994; Borowsky et al. 1995; Marcus & McCune 1999; Morris et al. 2001), so we used all other Xiphophorus, the platyfish and the southern clade swordtails, collectively as an outgroup. Sequences downloaded from GenBank were trimmed such that only the portion of sequence that overlapped with our control region d-loop sequences was used. This resulted in sequences that ranged from 417 to 446 bp in length and 55 total sequences for phylogenetic reconstructions, as there were multiple haplotypes for some species (alignment submitted to TreeBase). Phylogenetic analyses were conducted using maximum-likelihood (ML) and Bayesian inference. To determine the most appropriate model of DNA substitution we used jModeltest V0.1.1 (Posada 2008), to evaluate 88 substitution models, under the Akaike Information Criterion (AIC), corrected AIC(c), and Bayesian Information Criterion (BIC). Models that accounted for 95% of the weight for all three criteria (AIC, AICc and BIC) included either a proportion of invariable sites (I), a Gamma distribution for rate variation among sites (G), or both, and several relative rates in the substitution matrix (TIM2 = 4; TrN = 3; TIM3 = 4, GTR = 6). We used these models or the closest more complex model available in ML analyses, including bootstrap, using three different programs: (i) PAUP*4b10 (Swofford 2001), in which parameters values were fixed to those obtained with jModeltest (models: TIM2 + G, TIM2 + I+G, TIM2 + I); (ii) RaxML 7.0.4 (Stamatakis 2006a,b; Stamatakis et al. 2008) with two different models (GTR+G and GTR+G+I) and number of bootstrap replicates determined automatically, as implemented in the CIPRES portal (http://www.phylo.org/); and (iii) 26 27 28 29 24 25 11 12 13 14 15 16 17 18 19 20 21 22 23 4 5 6 7 8 9 10 1 2 3 Drainage Rı́o Claro Rı́o Claro Tributary of Rı́o Claro Apantla Trib. of Rı́o Claro Xuchipantla Rı́o Claro Tenexco Rı́o Claro Huitzitzilingo Arroyo Tultitlán ChiquitlaM Rı́o Huazalingo Totonicapa Rı́o Huazalingo Cocalaco Trib. of Rı́o Huazalingo San Pedro Rı́o Huazalingo Achiquihuixtla Rı́o Huazalingo ChicayotlaM Arroyo Xontla T-Dubs Arroyo Xontla Spider Arroyo Xontla Calnali-High Rı́o Calnali Calnali-Mid Rı́o Calnali Aguazarca Rı́o Calnali Calnali-Low Rı́o Calnali CulhuacánM Arroyo Pochutla Pochutla-High Arroyo Pochutla Pochutla-Trib Arroyo Pochutla Nicolasia Trib. of Arroyo Pochutla Tula Rı́o Tula Ahuamole Trib. of Arroyo Pochutla Papatlata Rı́o Atlapexco Huitznopala Rı́o Atlapexco El Arenal Rı́o Atlapexco MalilaM Rı́o Conzintla TlatzintlaM Tamala Tlatemaco No. Locality FIS 272 244 219 1364 422 869 384 186 1003 986 921 1168 1007 981 920 1272 877 916 440 352 193 122 161 1499 720 450 658 320 480 8 5 5 6 5 6 6 5 6 6 6 6 6 6 6 6 12 6 0.79 0.6 0.03 30 1 19 30 30 1 0.27 0.53 0.6 0.63 0.61 0.71 30 19 30 30 41 28 14 0.41 0.91 0.63 39 30 30 0.74 0.71 0.53 0.93 30 23 24 19 0 22 0.646† 0.039 )0.025 0.063 0.202 )0.063 - )0.414† - 0.350 0.246 )0.102 0.040 - 1* 0.077 )0.094 0.016 0.217 0.659† 0.384 0.133 0.084 )0.040 0.856* 0.032 0.315 )0.333 )0.160 1* )0.417† )0.427† 0.223 0.996* 0.523† 0.253 0.269 )0.404 )0.110 )0.033 0.167 0.478* 0.510* 0.535* 0.163 - 0.298 - 0.856* )0.381 0.386† 0.142 0.191 0.93 0 0 0.033 0.429 1 0.947 0.433 0.333 0.39 0.1 0 )0.102 )0.070 0.28 0.798* 0.484* 0.466* 0.373 0.414 0 1 0.042 0 1 1 )0.173 )0.019 0.177 )0.198 )0.092 )0.204 )0.082 0.042 )0.164 )0.081 - - 0 0 0 0 0 0 4 16 2 1 0 2 0 0 0 0 0 0 3 1 0 0 0 0 1 0 0 0 0 0 2 1 0 0 0 5 5 3 1 7 6 0 0 2 3 0 0 6 7 8 0 0 0 0 0 0 0 0 2 0 0 0 2 1 0 0 0 0 5 0 0 0 0 0 0 2 0 0 2 0 0 0 2 0 0 0 0 0 4 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 4 0 2 1 0 0 0 4 7 1 0 0 0 3 0 6 0 0 0 6 0 0 0 0 0 0 0 0 0 0 0 28 0 0 0 0 0 0 0 6 7 22 9 8 7 0 4 6 6 8 0 0 0 6 1 4 0 6 0 0 9 0 0 0 0 0 0 0 0 0 0 22 0 1 0 0 4 2 4 10 12 2 4 5 0 4 12 11 3 2 0 0 1 0 25 9 5 1 3 2 0 0 0 0 0 0 1 0 0 12 3 2 5 8 4 8 10 6 9 10 4 0 0 0 0 0 0 0 0 0 0 3 2 0 0 0 8 13 2 9 8 2 5 0 0 0 0 0 4 6 6 11 14 4 0 0 0 3 0 LIG1 ⁄ POLB LIG ⁄ TP53 POLB ⁄ TP53 mtDNA 1 2 Hybrid Index – no. of X. malinche nuclear alleles LD 0.478 )0.055 )0.034 0.234 0.463 0.495 0.322 0.057 0.604* 0.609* 0.416† 0.414† 0.216 0.861* 0.546* 0.371 0.425† 0.799* 0.428† 0.424 )0.256 )0.174 0.141 - )0.335 - )0.007 )0.211 )0.091 )0.180 0.785† 0.184 1† )0.161 - TP53 0.183 - POLB 0.102 0.016 - E (m) SEQ SNP HYB LIG1 N Table 1 Collecting localities for DNA sequencing and SNP genotyping. Superscript M and B following population names indicate pure, distal populations used for SNP discovery. Indicated for each population are: elevation in meters (E) and sample size for sequencing of mtDNA and nuclear introns (SEQ) and SNP genotyping (SNP). For populations with SNP data, the following measures are shown: the proportion of hybrid multilocus genotypes (HYB), deviations from HWE for three nuclear markers (FIS), linkage disequilibrium (LD) corrected for allele frequency [D ⁄ (piqipjqj)1 ⁄ 2], frequency of X. malinche mtDNA (mtDNA), and a hybrid index showing number of individuals in each population with 0–8 X. malinche nuclear alleles. FIS and LD values where P < 0.05 after correction for multiple comparisons are indicated by an asterisk (*) and where P < 0.05 before but not after correction (†) 346 Z . W . C U L U M B E R E T A L .  2010 Blackwell Publishing Ltd 17 4 0 0 0 0 0 0 0 0 4 0 0 1 0 0 0 0 0 0 1 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 2 1 0 14 2 0 0 11 0 0 0 0 0 0 0 0 0 0 0 cortezi variatus variatus 6 5 12 0 19 7 3 4 6 6 X. X. X. 0 0.759 0.4 0 )0.017 0.570* 1* 0.166 )0.054 0.656* 0.903* 0.184 )0.160 - - )0.054 0.711* 0.833* 0.738* 0.785* 0.816† 1* 1* 0.903* )0.120 )0.167 1 )0.153 0.37 0.24 0.1 0.59 7 3 4 6 6 5 6 5 6 30 29 10 29 6 5 4 3 2 1 TP53 POLB Rı́o Conzintla Rı́o Conzintla Rı́o Conzintla Rı́o Tianguistengo Rı́o Tianguistengo Rı́o Garces Rı́o Zontecomatlán Rı́o Zontecomatlán Arroyo Tenango Arroyo Mamey Rı́o Axtla Rı́o Venado Rı́o Garces Xochicoatlán Mixtla Comala Soyatla Tlacolula GarcesB Agua Frı́aB Benito JuárezB Tenango Mamey Xilitla Atlatipa Garces 30 31 32 33 34 35 36 37 38 39 40 41 42 E (m) SEQ SNP HYB LIG1 Drainage No. Locality FIS N Table 1 (Continued)  2010 Blackwell Publishing Ltd GARLI v.0.96beta8 (Zwickl 2006) with three different models (GTR+G, GTR+I and GTR+G+I) and 100 bootstrap replicates, also implemented in CIPRES. In addition, we conducted Bayesian analyses (models: GTR+G, GTR+I and GTR+G+I) using the Parallel version of MrBayes v 3.1.2 (Ronquist & Huelsenbeck 2003) implementing four runs, with four chains each, for 50 000 000 generations sampled every 1000 generations (all other parameters were default). The appropriate ‘burnin’ length (i.e. samples discarded prior to reaching stationarity) was determined based on small and stable average standard deviation of the split frequencies, potential scale reduction factor close to 1, and stable posterior probability values (see MrBayes manual). Single nucleotide polymorphism markers 1012 941 383 1287 469 229 241 272 274 292 1889 172 229 0 LIG1 ⁄ POLB LIG ⁄ TP53 POLB ⁄ TP53 mtDNA 7 Hybrid Index – no. of X. malinche nuclear alleles LD 8 H Y B R I D Z O N E S A L O N G A N E L E V A T I O N A L G R A D I E N T 347 Sequences from specimens collected in 2004 and earlier from pure X. birchmanni (N = 14 individuals from three populations) and pure X. malinche (N = 36 individuals from five populations) were compared by manual alignment in Sequencher. Populations and sample sizes within populations for SNP development are given in Table 1. In order to identify informative differences between the two species, we sequenced X. birchmanni and X. malinche from distal populations that matched the documented distribution and morphological traits of the two species (Rauchenberger et al. 1990; Rosenthal et al. 2003; Gutiérrez-Rodriguez et al. 2008). Furthermore, in all distal populations used to determine SNPs, all mature males matched the diagnostic morphological characteristics of only one species (GGR, unpublished data). Thus, we avoided using any populations in which hybrids existed. We used sequence alignments of all alleles for mtDNA (CR) and the three nuclear loci (LIG1, POLB and TP53) to identify SNPs separating X. birchmanni and X. malinche. These SNPs were used to design bi-allelic Custom TaqMan SNP Genotyping Assays (Applied Biosystems). Briefly, each species-typical TaqMan SNP probe has its own dye colour and fluorescence of that colour is given off when a copy of that probe is incorporated during PCR. Following PCR on a real-time PCR system, an end-plate reading of fluorescence levels allows assignment of individual genotypes as heterozygous or homozygous for one probe or the other. All SNP PCRs were carried out with 8 ng of genomic DNA in 25 lL reactions. The 25 lL reactions contained 12.5 lL TaqMan genotyping Master Mix (Applied Biosystems), 1.25 lL custom SNP assay mix (20·), 2 lL of 4 ng ⁄ lL genomic DNA and 9.25 lL ddH20. Samples were run and analysed on a 7500 Fast Real-Time PCR System (Applied Biosystems) by performing a post-run read of fluorescence after amplification 348 Z . W . C U L U M B E R E T A L . at 95 C for 10 min followed by 40 cycles of 92 C for 15 s and 60 C for 1 min. All mtDNA and nuclear intron sequences have been deposited on GenBank under accession nos HM003579–HM003607. Geometric morphometric analysis To further confirm that our SNP discovery method produced informative markers, we conducted a geometric morphometric analysis of mature males collected from a population in the Rio Calnali, Calnali-Mid, where X. malinche, X. birchmanni and hybrid multilocus SNP genotypes are all found together. For the analysis, lateral photographs were taken of all mature males for which we had SNP genotypes. We digitized landmark points on each image using the software program tpsDig (Rohlf 2004). Landmarks included the tip of the upper jaw (i); the anterior (ii) and posterior (iii) junction of the dorsal fin with the dorsal midline; the junction of the caudal fin with the dorsal (iv) and ventral midline (v); the anterior junction of the anal fin with the ventral midline (vi); the bottom of the head where the operculum breaks away from the body outline (vii); the upper edge of the operculum (viii), the upper (ix) and lower (x) insertion of the pectoral fin; and the centre of the eye (xi). We also included four semi-landmarks to estimate the dimensions of the dorsal and caudal fin (including the sword – an extension of the lower caudal fin ray) by digitizing the length of the first and last fin rays in both fins. To remove potential effects of differential fin expansion, x and y coordinates were standardized for the dorsal and caudal fin (i.e. the location of semi-landmarks only varied in the distance but not the angle from other landmarks). Based on the coordinates of the digitized landmarks, a geometric morphometric analysis was performed (e.g. see Zelditch et al. 2004 for an introduction to geometric morphometric analyses). Landmark coordinates were aligned using least-squares superimposition as implemented in the program tpsRelw (Rohlf 2007) to remove effects of translation, rotation, and scale. Based on the aligned coordinates, we calculated centroid size and partial warp scores with uniform components for each individual (weight matrix). The weight matrix was subjected to a principal component analysis based on a covariance matrix to reduce the data to true dimensionality. Dimensions with eigenvalues >1 were retained as shape variables. Unless otherwise stated, all statistical analyses were performed using SPSS 17 (SPSS, Inc. 2008). To test for phenotypic differentiation among parental species, backcrosses, and F2 and later hybrid individuals, we used multivariate analyses of covariance (MANCOVA) to analyse body shape variation (seven principal components accounting for over 77% of the total varia- tion). Putative F1s are not included as none were observed in this sample of individuals. Assumptions of multivariate normal error and homogeneity of variances and covariances were met for all analyses performed. F-values were approximated using Wilks’ lambda and effect strengths by use of partial eta squared (gp2). We tested for effects of ‘centroid size’ to control for multivariate allometry and included genotype (parental X. birchmanni, backcross X. birchmanni, F2 and later hybrid, backcross X. malinche, or parental X. malinche) as a factor. The interaction term was not significant, so only main effects were analysed (F28,300 = 1.345, P = 0.119). Shape variation along the first two principal component axes was visualized with thin-plate spline transformation grids using tpsRegr (Rohlf 2005). To test for congruence between genotype categories and phenotypes, we conducted a discriminant function analysis (DFA) to determine the percentage of specimens that could be correctly classified to the correct genotype solely based on body shape. To facilitate the DFA, we first removed the effect of allometry by using the residuals of a preparatory MANCOVA. In this MANCOVA, the partial warp scores with uniform components were used as dependent variables and centroid size as a covariate. Geographic structure of hybrid zones To characterize elevational gradients in the geographic distribution of each parental species and their hybrids, we mapped the frequency of each species and hybrids for each locality sampled. Individuals with any combination of both X. birchmanni- and X. malinche-typical SNP markers (including those inferred from original sequence data) were classified as hybrids. Estimates of hybrid frequencies could be biased upwards, if SNPs are not completely fixed within the two parental species. However, estimates could also be biased downwards because some hybrid individuals will be homozygous at all loci and erroneously classified as a parental. We then conducted MANCOVA on allele frequencies of all four markers. In the MANCOVA, we tested for drainage and elevation effects on allele frequency and tested the interaction terms (drainage*marker and elevation*marker). We further included a repeated measure to test for differences in allele frequencies among the four markers. The analysis included all 23 populations sampled for SNP genotyping (Table 1). Population genetic analyses Deviations from Hardy–Weinberg equilibrium (HWE) in a population can occur if individuals do not mate at random, if there is gene flow into the population, or if  2010 Blackwell Publishing Ltd H Y B R I D Z O N E S A L O N G A N E L E V A T I O N A L G R A D I E N T 349 there is strong selection (Hartl & Clark 2007). The inbreeding coefficient, FIS, was used to measure deviation from HWE proportions resulting from heterozygote deficiency (f, Weir & Cockerham 1984) for each nuclear SNP locus for all populations using GENEPOP V4 (Raymond & Rousset 1995). If there is sufficient gene flow among populations, connectivity is maintained and prevents genetic differentiation among populations. To test for genetic differentiation among populations, we calculated FST (h, Weir & Cockerham 1994) across the data set and assessed significance using the log-likelihood G-statistic in FSTAT (Goudet 1995; Goudet et al. 1996). The estimate of h as implemented in FSTAT was also used to calculate pairwise-FST values between neighbouring localities within stream reaches. In hybrid zones, high LD can represent recent or ongoing hybridization, but can also be a sign of population structure (Jiggins & Mallet 2000). Interspecific crosses generate perfect linkage disequilibrium (LD) in F1 offspring. If F1 offspring backcross and ⁄ or mate with each other, this LD will then erode in each generation. Selection, assortative mating, or continual immigration of parental genotypes could maintain LD in a population. We tested for LD among all pairwise comparisons of nuclear SNP markers in each population according to Hill (1974) which provides a method to estimate of disequilibrium, D, and a test statistic that has an approximate v2 distribution. Values of D were divided by allele frequency [D ⁄ (piqipjqj)1 ⁄ 2] to standardize LD estimates across populations. We also calculated genotypic disequilibrium for cases when phase is not known using GENEPOP. All P-values were adjusted for multiple comparisons within populations. Results Phylogeny and SNP markers Sequencing of 160 individuals (Table 1) for the mitochondrial control region d-loop produced 10 distinct haplotypes among X. birchmanni, X. malinche, their putative hybrids and X. cortezi. The only haplotype observed in X. cortezi (haplotype XM2; Fig. 1) was also observed in putative X. birchmanni ⁄ X. malinche hybrids and is mutationally intermediate among other X. malinche haplotypes. As such, it was considered and designated as a haplotype shared with X. malinche (incomplete lineage sorting). There was one haplotype observed for X. variatus. The optimal trees from all ML searches were compared to one another and to the consensus tree from two Bayesian analyses. The consensus tree from ML analyses in PAUP implementing the three best-fit models of sequence evolution is reported here along with bootstrap support values from all ML analyses and posterior  2010 Blackwell Publishing Ltd probabilities from MrBayes shown for nodes of interest (Fig. 1). Here we are primarily concerned with the relationships among X. birchmanni, X. malinche and X. cortezi, thus support values for relevant nodes are given. All runs gave strong support for monophyly of the five X. birchmanni haplotypes sequenced in this study along with all other previously sequenced X. birchmanni haplotypes. All runs also placed X. malinche together with X. cortezi, X. multilineatus and X. nigrensis. There were five mitochondrial control region d-loop haplotypes for X. birchmanni and four for X. malinche. X. cortezi and X. malinche shared one haplotype, accounting for the total of five haplotypes that clustered with other X. malinche on the phylogenetic tree. The majority of populations (19 of 27) contained only one mtDNA control region haplotype. Both the mtDNA control region d-loop and DNA ligase 1 exhibited an accumulation of substitutions between the species, consistent with historical divergence and more recent secondary contact and introgression (Table 2). Both the polymerase beta and tumour protein 53 gene showed fewer interspecific differences but contained informative SNPs nonetheless. The complete consensus sequences between X. birchmanni and X. malinche are shown in Table 2 with interspecific SNPs in brackets and reporter sequences for TaqMan SNP assays underlined. Geometric morphometric analysis Among 99 males, body shape was significantly influenced by centroid size, indicating allometric effects (F7,87 = 12.461, P < 0.001, gp2 = 0.501), and genotype (F28,315.1 = 4.045, P < 0.001, gp2 = 0.242). Visualizing the first two axes of shape variation indicates that backcrosses cluster with the respective parental species, with F2s intermediate between the two parental species (Fig. 2). This is also indicated by the discriminant function analysis that assigned over 77% of the specimens (compared to the expected 17% under a null hypothesis of no pattern) to the correct genotype solely based on geometric morphometric data (Table 3); misclassifications were relatively common between parentals and backcrosses. Over 30% of F2 or later hybrids were misclassified, indicating overlap in morphospace occupation with other groups. Replicated geographic structure There were elevational clines in the distribution of each species and hybrids, with X. malinche at high elevations, X. birchmanni at low elevations and hybrids at intermediate elevations. These gradients were replicated across stream reaches as demonstrated in the distribution of multilocus genotypes (Figs. 3 and 4). MANCOVA on SNP 350 Z . W . C U L U M B E R E T A L . mtDNA Control Region d-loop TTTCCACCTCTAACTCCCAAAGCTAGGGTTCTAATTTAAACTATTCTTTGA CCGGACTCTGCCCCTCCT[T ⁄ A]AGTACATGTATGTATTATCCCCATTAATA GATTTTAACCATTTAAAGT[G ⁄ A]ATGTAATTCTACATTAATGAAAAATCAAA A[G ⁄ :]TTATA[A ⁄ G]GAACTTAAATACATTA[T ⁄ C]ATCATCAAATAAATATGAA GGTAGACATAAACCA[C ⁄ T]T[A ⁄ G]AA[C ⁄ T]TT[C ⁄ T]AAA[C ⁄ T]T[C ⁄ T]CATTA AA[T ⁄ C]ATGTTA[T ⁄ A]AAAAATGACGATATTGAATTG[T ⁄ C]CCTATCA[T ⁄ C]A ACTCTCATCAGTCTAGATATACCAGGACTCACCAC CTCTGCAAGT[C ⁄ A]AGAGTCAAATG[T ⁄ C]AGTAAGAGACCACCATCAGTTG ATTTCTTAATGTACACGTTTATTGATGGTCA[A ⁄ C]GGACAAAAATCGTGGG GGT[A ⁄ C]GCACACAGTGAACTATTCCTGGCATCTGGTTCCTACTTCAGG DNA Ligase 1 GCCCTCGCAGGAGTCTGCACAAACAAACACATTAATGCACCAAAACACGA GTCTGGATGTGCAGAAAGGAAATTAAGAGATTATCTGTTGGATATACGGA GTGTCTTTGTGTTGGTGGGTCTCAAAGGGAAACAAACGGGTTTTGGGAAA CAAAGAGGCCGTGGACAAGGACGGGGACATAACTCATATGCACAAGGAAA ACACATCCTTTATACGTCAGAATCACACACACACACAACACAAAATAAATCC AT[T ⁄ C]ATACAACCCACAAGTTCTGGCGCTGCTTTGGATACACTCACACAGA CAGTTTACACGTTAAGGCAGCTGAGAAAGACACAAATAAAATGAAACTTTTA CTCAAAGCACCCACACACG[C ⁄ A]AGAGATTCAGGGGAGCCAACAAACAGGA A[G ⁄ A]ATGACACAAACTAGAAGATGACACAAAC[: ⁄ T]AG[G ⁄ A]A[: ⁄ G]ATGACA CAAAGGAAGGAAAACGAGAAGAAAAAGTTTGCCTCTGTTTCACAGACCTCG GACG DNA Polymerase Beta CTTTCCTATCCACA[G ⁄ T]TACAGAATAAAAACATTATCGAATGGCTAAAAGAA GTATTTATAAAGAAAGTACTTTAAAG[C ⁄ T]TTTAGTAATATGATATGAACAAGC AAGCTAGTTTACAATACCTTGATCTCATGTGCCATACATTATACAATGAATAT ATGATTTTGTTATTGTAAAAGGATTTTAAAACGTCATTGTCAGAGTATACTCTA AATATCTGACCGGAAGTGAGTGAAAAATGAAGCAGAAATAAAACCTTTTGAG CAGATAAATTAAGTCTATCCAAAATAACCACTGAATATGTGGATTCTACTCGT ATCTGCAAATAAAAGCAAAGGTTTTTAGGAACTCACCTGTATGCATTGTACTT AT Tumour Protein 53 AAACCTGGAAAAAAGTGGGACTAAGCAGAC[G ⁄ A]AAGAAAAGAAGTATGCTT TTAAATTGTTTTATATGTATGTATAGTAGACTAAAGTTTTATTCTATCTGTCACT TATCATGAACATTCTTCATTTTAGAGAGTGCTCCTGCTCCAGATACCTCCACC GCAAAAAAGTCCAAGTCTGCCTCTAGTGGAGAGGATGAGGACAAGGAGATTT ACACTCTCTCTGTAAGGCCTGTTTCTGCAACTGGATGACATCACAATATTAGG ATGAATGAAGTAACTTTTTTACTTTCTAACTTTGTTGCAGATCCGGGGCCGTA ATCGTTATCTGTGGTTCAAGAGC Table 2 Sequence consensus between X. birchmanni and X. malinche for the four genes used to design SNP markers. Interspecific base substitutions are denoted in brackets [X. birchmanni base ⁄ X. malinche base] and the underlined portion is the reporter sequence used in multilocus genotyping with TaqMan SNP chemistry. Brackets containing a colon represent and insertion ⁄ deletion event rather than base substitution Table 3 Classification results of the discriminant function analysis Predicted group membership Original Count % X. birchmanni X. birchmanni backcross F2 X. malinche backcross X. malinche X. birchmanni X. birchmanni backcross F2 X. malinche backcross X. malinche X. birchmanni X. birchmanni backcross F2 X. malinche backcross X. malinche 17 4 1 0 2 81.0 17.4 7.7 0.0 8.7 3 17 2 0 1 14.3 73.9 15.4 0.0 4.3 1 1 9 0 0 4.8 4.3 69.2 0.0 0.0 0 0 1 19 5 0.0 0.0 7.7 100.0 21.7 0 1 0 0 15 0.0 4.3 0.0 0.0 65.2  2010 Blackwell Publishing Ltd H Y B R I D Z O N E S A L O N G A N E L E V A T I O N A L G R A D I E N T 351 Fig. 3 Sampling localities, frequencies, and geographic distribution of parental species and hybrids. All localities used for initial DNA sequencing and SNP genotyping are shown with X. malinche (red), X. birchmanni (yellow) and hybrids (green) based on multilocus genotypes of four SNP markers. Two tributaries (C and D) highlighted by a box at the centre of the figure are enlarged in the inset image. (a) (d) Calnali X. malinche allele frequency 0.6 Claro 1.0 0.3 0.5 0.0 1200 0.0 700 500 300 Elevation (m) 100 1100 1000 (e) 900 Pochutla 1.0 (b) 1.0 Huazalingo 0.5 0.5 0.0 0.0 1600 1200 1000 800 600 400 (f) 1300 1000 700 (c) 400 100 1.0 Conzintla Xontla 1.0 0.5 0.9 0.0 1500 0.8 1000 950 CR LIG1 POLB TP53 900 1250 1000 750 500 250 (g) 1.0 Tianguistengo 0.5 0.0 1200 1000 800  2010 Blackwell Publishing Ltd 600 400 200 Fig. 4 Allele frequency clines along each of the seven tributaries as a function of elevation. 352 Z . W . C U L U M B E R E T A L . marker frequency revealed significant effects of both elevation (F = 7.655, d.f. = 6, P = 0.001) and drainage (F = 36.852, d.f. = 1, P < 0.001). There was a significant interaction between drainage and marker (F = 2.278, d.f. = 18, P = 0.013) but not between elevation and marker (F = 2.199, d.f. = 3, P = 0.101). There was no difference in patterns of allele frequency change among the four markers (Wilk’s k = 0.635, F =2.487, d.f. = 3, P = 0.106). Figure 4 shows allele frequency clines for all SNP markers in seven stream reaches. SNP population genetics The frequency of hybrids varied widely among localities, ranging from 0 to 0.93. The frequency of hybrid multilocus genotypes for all populations (i.e. localities) is given in Table 1 and depicted in Fig. 3. Hybrids were found in 22 of the 23 localities sampled for SNP genotyping. In these 22 localities population structure falls along a continuum. Two populations were composed entirely of hybrids. In 14 localities, one parent species coexists with hybrids. In the remaining six populations, both parent species and hybrids are found together. It is in these populations where tests for deviations in the distribution of SNP markers from Hardy– Weinberg expectations produced significant results (Table 1). Linkage disequilibrium (D) was highly significant in several populations (Table 1) and estimates of genotypic disequilibrium from GENEPOP produced the same results as calculating Hill’s (1974) metric D (not shown). In these populations, significant heterozygote deficiency and linkage disequilibrium, potentially indicate some degree of reproductive isolation between X. malinche, X. birchmanni and hybrids. We observed highly significant genetic differentiation among populations (h = 0.407; P = 0.001). We further tested for differentiation among populations within stream reaches along the upstreamto-downstream gradients. FST in pairwise comparisons of adjacent populations varied from 0 to 0.6335 (Table 4). Values reported as negative are effectively zero, because FST can only be greater than or equal to zero. Discussion Extensive hybridization was detected in seven separate stream reaches with two main patterns emerging throughout the hybrid zone. First, in each stream reach, there was a distinct elevational gradient between X. malinche upstream and X. birchmanni downstream. Second, population structure varied widely among sites. At some localities, there were low levels of introgressive hybridization and others were composed completely of Table 4 Pairwise FST values for each population and the next closest sampling location within stream reaches. In many cases low FST is likely correlated to the distance between sampling sites and local geography (i.e. barriers to dispersal) as sampling locations are not spaced evenly along stream reaches Drainage ⁄ comparison Rı́o Calnali Calnali-High Calnali-Mid Aguazarca Rı́o Claro Tlatzintla Tlatemaco Apantla Xuchipantla Arroyo Pochutla Nicolasia Tula Rı́o Conzintla Malila Xochicoatlán Mixtla Xochicoatlan Comala Papatlatla Rı́o Huazalingo Totonicapa Cocalaco Arroyo Xontla T-Dubs * FST Calnali-Mid Aguazarca Calnali-low 0.0541† )0.0211 0.0936† Tlatemaco Apantla Xuchipantla Tenexco 0.3086* 0.1298* 0.6335* )0.0129 Tula Papatlatla 0.1227* 0.1723* Xochicoatlán Mixtla Comala Comala Papatlatla Huitznopala 0.1064† 0.1641 0.2064† 0.562* 0.0447† 0.0106 Cocalaco San Pedro 0.1261* 0.0095 Spider 0.0435 P < 0.05. Significant before, but not after, correction for multiple tests. † hybrid individuals. By contrast, 6 of the 23 sampled populations were highly structured into three groups: hybrids and each of the two parent species. These population genetic patterns likely indicate sharp geographic differences in the dynamics of hybridization and selection. Phylogeny and SNP markers Phylogenetic analysis of mitochondrial control region d-loop sequences showed that X. birchmanni and X. malinche are in distinct and well-supported clades. X. malinche haplotypes were more closely related to those in X. cortezi, X. multilineatus and X. nigrensis, but due to limited data and ⁄ or incomplete lineage sorting, relationships among these species were not well-resolved. Our results were consistent with previous mtDNA analyses, in which these four species form a wellsupported clade that is distinct from a more distantly related X. birchmanni (Meyer et al. 1994; Marcus & McCune 1999). Earlier analyses based on nuclear  2010 Blackwell Publishing Ltd H Y B R I D Z O N E S A L O N G A N E L E V A T I O N A L G R A D I E N T 353 allozymes and RAPDs but on only one or a few individuals or populations per taxon (Borowsky et al. 1995; Morris et al. 2001) suggest a closer relationship between X. birchmanni and X. malinche, perhaps reflecting the effects of hybridization between these species. Our results, however, indicated fixed differences between pure populations of these species at several nuclear loci. Sequencing and alignment of these nuclear loci and mtDNA allowed us to identify informative SNPs that separate the two species. In addition to our conservative approach selecting only distal populations matching historical species distributions, and only those populations where 100% of the observed males matched diagnostic morphological traits of only one species, geometric morphometric analyses supported the informative nature of the SNP markers. In a large hybrid population containing X. birchmanni parentals, X. malinche parentals and hybrids, parental genotypes had parental (pure species) morphology as did their respective backcrosses while hybrid (F2 and later individuals) had intermediate, hybrid morphology. Replicated geographic structure One of the strengths of this system is the replication in multiple stream reaches. Hybrid zones between these species occur in at least seven separate streams along similar ecological gradients. Gene flow among hybrid zones and among X. malinche populations in stream headwaters is impeded by high mountain ridges, long upstream to downstream distances and geographic barriers in some places. Thus, each hybrid zone likely represents an independent outcome of secondary contact due to a shift towards higher elevation in X. birchmanni, perhaps associated with Pleistocene climate change or more recent anthropogenic effects, pushing X. malinche into multiple high elevation refugia. Alternatively, as Kallman & Kazianis (2006) suggest, an ancestral form may have been uplifted and populations isolated by folding of mountain ridges resulting in the current distribution of X. malinche. Though commonly touted as natural laboratories, replicating studies of hybridization phenomena is difficult. Nolte et al. (2009) demonstrated the utility of such natural replication, testing for loci contributing to reproductive isolation between two species of sculpin (Cottus), in two independent hybrid zones. Even more recently, a study of two independent hybrid zones of cyprinid fish demonstrated the heterogeneity of outcomes of hybridization within a species pair (Aboim et al. 2010). Taking advantage of these and other replicated hybrid zones – like the birchmanni–malinche hybrid zones – should continue to provide even greater power in studying evolutionary processes and patterns.  2010 Blackwell Publishing Ltd In the birchmanni–malinche system, each hybrid zone is characterized by an elevational gradient in the frequency of parental and hybrid individuals and of individual, species-diagnostic SNP allele frequencies. In each stream reach, hybrid zones had clear geographic structure, with X. malinche alleles being replaced along an upstream-to-downstream gradient by X. birchmanni alleles and with hybrids prevalent at intermediate elevations. Analysis of SNP marker frequencies showed significant effects of drainage and elevation. Mid-elevation hybrid populations contained a preponderance of backcross, F2 or later generation hybrid individuals, whereas F1s were rare. Out of 760 individuals genotyped for one mtDNA and three nuclear loci, 13 had genotypes consistent with first-generation hybrids. However, only a few of those occurred in populations where both parental forms are present, suggesting that the majority of ‘F1’ fish were erroneously classified backcross fish. The factors that maintain hybrid zone structure in this system are not yet known as the data here do not address fitness differences or allow for explicit tests of among hybrid zone models. However, the consistency of overall structure across all stream reaches along the same elevation gradient suggests that selection by the environment could play a role. The hybrid zones in different stream reaches all exhibited consistent upstream-to-downstream structure, but there were differences among the zones. In the Rı́o Claro, X. malinche were found at much lower elevation than observed in other stream reaches. This may reflect differences in historical distributions of species or differences in environmental conditions among stream reaches. Most of the hybrid zones were bounded on each end by pure populations of parental species, but no pure X. malinche population is observed in the Rı́o Calnali and no pure populations of X. birchmanni occur before the 60-m waterfall at Chahuaco. Hybrid populations were composed of hybrids and one parental (or only hybrids) in some hybrid zones, while hybrids and both parentals occur in sympatry in populations in other hybrid zones. This heterogeneity could be a consequence of any number of biotic or abiotic variables such as differences in migration, historical frequency of parentals, connectivity of populations and variation in natural or sexual selection. SNP population genetics Heterozygote deficiency and LD were generally low in most sampled populations. Low FIS and LD suggest that hybrids are both viable and fertile and have backcrossed extensively with the parental species; thus, there is little evidence of reproductive isolation in most locations. This is supported by a low frequency of F1s 354 Z . W . C U L U M B E R E T A L . and high frequencies of F2 (or later) and backcross individuals in almost all populations. By contrast, when heterozygote deficiency and LD were observed, they co-occurred in the same populations. Both parent species coexist with hybrids in those populations that deviate from HWE and have LD, with hybrids accounting for 10–61% of the population with the exception of Calnali-high where only X. birchmanni and hybrids were observed (Table 1). Genetic differentiation across all populations was significant as expected, and FST between populations within stream reaches was large and significant in some but not all cases. However, pairwise comparisons should be interpreted with caution because distance between populations is not constant and may account for the large variation in pairwise FST. It should be noted, however, that even over distances as short as 200–300 m (22 m elevation) there was significant differentiation in some cases even in the absence of apparent physical barriers to migration. As with overall geographic patterns, the extent to which migration and selection from the biotic and abiotic environment play in shaping population-level patterns is unknown. Deviations from HWE and LD could be maintained by migration-selection balance such as in a tension zone model (Gay et al. 2008; Reugg 2008) or if hybrids are relatively fit compared to parentals, with some degree of assortative mating in some populations, as in hybrid superiority (Good et al. 2000). Studies of mate choice have shown that X. birchmanni has strong preferences for both visual and olfactory signals of their own males over X. malinche males (Wong & Rosenthal 2006; Fisher et al. 2006). This may play a role in mating patterns in structured populations with high FIS and LD. Additionally, hybrids may have an advantage due to sexual selection. Certain hybrid males are likely preferred by females because they have recombinant phenotypes not possible in pure species males (Fisher et al. 2009). Future work will be able to build on our current understanding of the hybrid zones and the evolutionary processes within them. Persistence of substantial frequencies of parental fish in populations with no source of pure parentals and a tendency for assortative mating in X. birchmanni suggest that mating preferences could play a role in population structure. However, measurements of gene flow and fitness-related traits are necessary to begin to weigh the relative contributions of environmental selection to overall geographic structure and the role behaviour, migration and selection in both stream and population-level patterns in the hybrid zones. The mechanisms by which population structure arises and potentially leads to speciation in hybrid populations are not well understood. The breadth of pre-zygotic and post-zygotic factors contributing to these processes is rarely tested (but see Mendelson et al. 2007). However, it is the combination and interaction of these various factors that determine the fate of hybrids, population structure and, consequently, whether or not speciation occurs (Bolnick 2009), making hybrid zones useful study systems. The malinche–birchmanni hybrid zones and molecular markers described here provide an opportunity to test the effects of selection at different points along the pre- to post-zygotic continuum. Together, behavioural and molecular approaches in this and other systems can better clarify the processes of hybridization and hybrid speciation. Acknowledgements We would like to thank the federal government of Mexico for collection permits. We thank Charles Criscione for comments on the manuscript. We also thank the Texas A&M University Brazos HPC cluster for computation for phylogenetic analyses . Research was supported by National Science Foundation grant IOS-0923825 to GGR. ZWC was supported by National Science Foundation IGERT-0654377 in Applied Biodiversity Science at TAMU. References Aboim MA, Mavarez J, Bernatchez L, Coelho MM (2010) Introgressive hybridization between two Iberian endemic cyprinid fish: a comparison between two independent hybrid zones. Journal of Evolutionary Biology, 23, 817–828. Arnold ML (1997) Natural Hybridization and Evolution. Oxford University Press, New York. Barton NH (2001) The role of hybridization in evolution. Molecular Ecology, 10, 551–568. Barton NH, Hewitt GM (1985) Analysis of hybrid zones. Annual Review of Ecology and Systematics, 16, 113–148. Bell MA, Travis MP (2005) Hybridization, transgressive segregation, genetic covariation, and adaptive radiation. Trends in Ecology & Evolution, 20, 358–361. Bolnick DI (2009) Hybridization and speciation in Centrarchids. In: Centrarchid Fishes: Diversity, Biology and Conservation (ed Cooke SJ and Philipp DP). pp. 39–69, Oxford University Press, Oxford. Borowsky RL, McClelland M, Cheng R, Welsh J (1995) Arbitrarily primed DNA-Fingerprinting for phylogenetic reconstruction in vertebrates - The Xiphophorus model. Molecular Biology and Evolution, 12, 1022–1032. Burke JM, Arnold ML (2001) Genetics and the fitness of hybrids. Annual Review of Genetics, 35, 31–52. Dowling TE, Secor CL (1997) The role of hybridization and introgression in the diversification of animals. Annual Review of Ecology and Systematics, 28, 593–619. Fisher HS, Wong BBM, Rosenthal GG (2006) Alteration of the chemical environment disrupts communication in a freshwater fish. Proceedings of the Royal Society B-Biological Sciences, 273, 1187–1193. Fisher HS, Mascuch S, Rosenthal GG (2009) Multivariate male traits misalign with multivariate female preferences in the swordtail fish, Xiphophorus birchmanni. Animal Behaviour, 78, 265–269.  2010 Blackwell Publishing Ltd H Y B R I D Z O N E S A L O N G A N E L E V A T I O N A L G R A D I E N T 355 Fuller RC, McGhee KE, Schrader M (2007) Speciation in killifish and the role of salt tolerance. Journal of Evolutionary Biology, 20, 1962–1975. Gay L, Crochet PA, Bell DA, Lenormand T (2008) Comparing clines on molecular and phenotypic traits in hybrid zones: a window on tension zone models. Evolution, 62, 2789– 2806. Good TP, Ellis JC, Annett CA, Pierotti R (2000) Bounded hybrid superiority in an avian hybrid zone: effects of mate, diet, and habitat choice. Evolution, 54, 1774–1783. Goudet J (1995) FSTAT (Version 1.2): a computer program to calculate F-statistics. Journal of Heredity, 86, 485–486. Goudet J, Raymond M, de-Meeus T, Rousset F (1996) Testing differentiation in diploid populations. Genetics, 144, 1933– 1940. Guillaume F, Whitlock MC (2007) Effects of migration on the genetic covariance matrix. Evolution, 61, 2398–2409. Gutiérrez-Rodriguez C, Shearer AE, Morris MR, de Queiroz K (2008) Phylogeography and monophyly of the swordtail fish species Xiphophorus birchmanni (Cyprinodontiformes, Poeciliidae). Zoologica Scripta, 37, 129–139. Hartl DL, Clark AG (2007) Principles of Population Genetics, 4th edn. Sinauer Associates, Sunderland, MA. Hill WG (1974) Estimation of linkage disequilibrium in random mating populations. Heredity, 33, 229–239. Jiggins CD, Mallet A (2000) Bimodal hybrid zones and speciation. Trends in Ecology & Evolution, 15, 250–255. Johannesson K, Panova M, Kemppainen P, Andre C, RolanAlvarez E, Butlin RK (2010) Repeated evolution of reproductive isolation in a marine snail: unveiling mechanisms of speciation. Philosophical Transactions of the Royal Society B-Biological Sciences, 365, 1735–1747. Kalisz S, Kramer EM (2008) Variation and constraint in plant evolution and development. Heredity, 100, 171–177. Kallman KD, Kazianis S (2006) The genus Xiphophorus in Mexico and Central America. Zebrafish, 3, 271–285. Kawecki TJ, Ebert D (2004) Conceptual issues in local adaptation. Ecology Letters, 7, 1225–1241. Lee WJ, Conroy J, Hunting Howell W, Kocher TD (1995) Structure and evolution of teleost mitochondrial control regions. Journal of Molecular Evolution, 41, 54–66. Marcus JM, McCune AR (1999) Ontogeny and phylogeny in the northern swordtail clade of Xiphophorus. Systematic Biology, 48, 491–522. Mendelson TC, Imhoff VE, Venditti JJ (2007) The accumulation of reproductive barriers during speciation: postmating barriers in two behaviorally isolated species of darters (percidae : etheostoma). Evolution, 61, 2596–2606. Meyer A, Morrissey JM, Schartl M (1994) Recurrent origin of a sexually selected trait in Xiphophorus fishes inferred from a molecular phylogeny. Nature, 368, 539–542. Moore WS (1977) Evaluation of narrow hybrid zones in vertebrates. Quarterly Review of Biology, 52, 52263–52277. Morris MR, de Queiroz K, Morizot DC (2001) Phylogenetic relationships among populations of northern swordtails (Xiphophorus) as inferred from allozyme data. Copeia, 1, 65– 81. Nikula R, Strelkov P, Vainola R (2008) A broad transition zone between an inner Baltic hybrid swarm and a pure North Sea subspecies of Macoma balthica (Mollusca, Bivalvia). Molecular Ecology, 17, 1505–1522.  2010 Blackwell Publishing Ltd Nolte AW, Gompert Z, Buerkle CA (2009) Variable patterns of introgression in two sculpin hybrid zones suggest that genomic isolation differs among populations. Molecular Ecology, 18, 2615–2627. Posada D (2008) jModelTest: Phylogenetic model averaging. Molecular Biology and Evolution, 25, 1253–1256. Rauchenberger M, Kallman KD, Moritzot DC (1990) Monophyly and geography of the Rio Pánuco basin swordtails (genus Xiphophorus) with descriptions of four new species. American Museum Novitates, 2975, 1–41. Raymond M, Rousset F (1995) GENEPOP (Version1.2) – population-genetics software for exact tests and ecumenicism. Journal of Heredity, 86, 248–249. Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and speciation. Heredity, 83, 363–372. Rieseberg LH, Raymond O, Rosenthal DM, et al. (2003) Major ecological transitions in wild sunflowers facilitated by hybridization. Science, 301, 1211–1216. Rohlf F (2004) tpsDig (ed. http://life.bio.sunysb.edu/morph/). Rohlf F (2005) tpsRegr (ed. http://life.bio.sunysb.edu/morph/). Rohlf F (2007) tpsRelw (ed. http://life.bio.sunysb.edu/morph/). Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19, 1572–1574. Rosenthal GG, De La Rosa Reyna XF, Kazianis S, Stephens MJ, Morizot DC, Ryan MJ, Garcı́a de León FJ (2003) Dissolution of sexual signal complexes in a hybrid zone between the swordtails Xiphophorus birchmanni and Xiphophorus malinche (Poeciliidae). Copeia, 2, 299–307. Ruegg K (2008) Genetic, morphological, and ecological characterization of a hybrid zone that spans a migratory divide. Evolution, 62, 452–466. Schluter D (2009) Evidence for ecological speciation and its alternative. Science, 323, 737–741. Schluter D, Conte GL (2009) Genetics and ecological speciation. Proceedings of the National Academy of Sciences of the United States of America, 106, 9955–9962. Stamatakis A (2006a) Phylogenetic models of rate heterogeneity: a high performance computing perspective. Proceedings of 20th IEEE ⁄ ACM International Parallel and Distributed Processing Symposium (IPDPS2006), High Performance Computational Biology Workshop. Rhodos, Greece. Stamatakis A (2006b) RAxML-VI-HPC: maximum likelihoodbased phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics, 22, 2688–2690. Stamatakis A, Hoover P, Rougemont J (2008) A Rapid Bootstrap Algorithm for the RAxML Web Servers. Systematic Biology, 57, 758–771. Swofford DL (2001) PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sinauer Associates, Sunderland, MA. Tobler M, DeWitt TJ, Schlupp I, de Leon FJG, Herrmann R, Feulner PGD, Tiedemann R, Plath M (2008) Toxic hydrogen sulfide and dark caves: phenotypic and genetic divergence across two abiotic environmental gradients in Poecilia mexicana. Evolution, 62, 2643–2659. Walter RB, Rains JD, Russell JE, et al. (2004) A microsatellite genetic linkage map for Xiphophorus. Genetics, 168, 363–372. Weir BS, Cockerham CC (1984) Estimating F-Statistics for the analysis of population-structure. Evolution, 38, 1358–1370. 356 Z . W . C U L U M B E R E T A L . Wolf JBW, Lindell J, Backstrom N (2010) Speciation genetics: current status and evolving approaches Introduction. Philosophical Transactions of the Royal Society B-Biological Sciences, 365, 1717–1733. Wong BBM, Rosenthal GG (2006) Female disdain for swords in a swordtail fish. American Naturalist, 167, 136–140. Yanchukov A, Hofman S, Szymura JM, Mezhzherin SV (2006) Hybridization of Bombina bombina and B. variegata (Anura, Discoglossidae) at a sharp ecotone in western Ukraine: comparisons across transects and over time. Evolution, 60, 583–600. Zelditch M, Swiderski D, Sheets H, Fink W (2004) Geometric Morphometrics for Biologists. Elsevier Academic Press, Amsterdam. Zwickl DJ (2006) Genetic Algorithm Approaches for the Phylogenetic Analysis of Large Biological Sequence Datasets Under the Maximum Likelihood Criterion. The University of Texas at Austin, Austin, TX. Z.C. is interested in mechanisms maintaining genetic variation in natural populations, particularly pertaining to the evolutionary genetics of mate choice and sexual selection. H.F.’s research focuses on the evolution of reproductive traits that are shaped by sexual selection and are pivotal to the formation and dissolution of species boundaries. M.T. works on ecological and evolutionary effects of environmental variation on populations particularly as it pertains to adaptation and speciation. M.M. currently focuses on understanding the evolution of insect-bacteria associations and phylogeography of freshwater organisms. P.B.’s research focuses on marine connectivity and the evolution of marine biodiversity and the application of this information to the development of more effective marine management policies. M.S. has longstanding interests in the application of molecular genetic data to questions in evolutionary ecology. G.R.’s research centers around the evolution and mechanisms of mate choice.  2010 Blackwell Publishing Ltd