The Evolutionary History of Antirrhinum Suggests That Ancestral Phenotype Combinations Survived Repeated Hybridizations

Home , Antirrhinum, Misopates orontium

The Plant Journal (2011) 66, 1032–1043 doi: 10.1111/j.1365-313X.2011.04563.x The evolutionary history of Antirrhinum suggests that ancestral phenotype combinations survived repeated hybridizations

Yvette Wilson and Andrew Hudson*,† Institute of Molecular Plant Sciences, University of Edinburgh, King’s Buildings, Mayﬁeld Road, Edinburgh EH9 3JH, UK

Received 29 November 2010; revised 25 February 2011; accepted 1 March 2011; published online 6 April 2011. *For correspondence (fax +44 131 650 5392; e-mail [email protected]). †Present address: Division of Plant Science, University of Dundee at SCRI, Invergowrie, Dundee, DD2 5DA, UK.

SUMMARY

The model species Antirrhinum majus (the garden snapdragon) has over 20 close wild relatives that are morphologically diverse and adapted to different Mediterranean environments. Hybrids between Antirrhinum species have been used successfully to identify genes underlying their phenotypic differences, and to infer how selection acts on them. To better understand the genetic basis for this diversity, we have examined the evolutionary relationships between Antirrhinum species and how these relate to geography and patterns of phenotypic variation in the genus as a whole. Large population samples and both plastid and multilocus nuclear genotypes resolved the relationships between many species and provided some support for the traditional taxonomic division of the genus into morphological subsections. Morphometric analysis of plants grown in controlled conditions supported the phenotypic distinction of the two largest subsections, and the involvement of multiple underlying genes. Incongruence between nuclear and plastid genotypes further suggested that several species have arisen after hybridization between subsections, and that some species continue to hybridize. However, all potential hybrids appear to have retained a phenotype similar to one of their ancestors, suggesting that ancestral combinations of characters are maintained by selection at many different loci.

Keywords: Antirrhinum, snapdragon, phylogeny, morphological evolution.

INTRODUCTION Much of our understanding of the genes involved in mor- genus Antirrhinum – in which between 17 and 27 distinct phological evolution and speciation has come from taxa that species and subspecies have been recognized in different are sufficiently different to be regarded as separate species, taxonomic accounts (Rothmaler, 1956; Webb, 1971; Sutton, but that retain the ability to form fertile hybrids. Both natural 1988). Although morphologically diverse and adapted to and artificial hybrids have been used to detect loci underly- different, often extreme, environments, all Antirrhinum ing differences between the parental species. In some cases, species can form fertile hybrids with each other and with the genes and mutations have themselves been identified. A. majus when artificially cross-pollinated. Such hybrids This has been particularly successful when the research have identified genes underlying differences in morphology infrastructure developed in a closely related model species and flower colour between their parents (Hackbarth et al., is available, for example in Drosophila (McGregor et al., 1942; Langlade et al., 2005; Schwinn et al., 2006; Feng et al., 2007; Jeong et al., 2008). 2009). Natural Antirrhinum hybrids have also identified The garden snapdragon, Antirrhinum majus (Plantagina- genes involved in flower colour variation, and have sug- ceae) has been used as a model to study inheritance, and the gested how selection acts on them (Whibley et al., 2006). genetic control of development and flower colour (Schwarz- The genus Antirrhinum can therefore provide a model for Sommer et al., 2003). Its close relatives are native to the understanding the genetic basis for patterns of phenotypic western Mediterranean region, mostly the Iberian peninsu- diversity and adaptation around the species level, which lar, and comprise a monophyletic group – the traditional may be typical of many recently evolved Mediterranean

1032 ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd Constrained evolution in the genus Antirrhinum 1033 taxa (Thompson, 2005). However, it has been difficult to relate the genetic differences between pairs of parental species to variation in the genus as a whole because the relationships between Antirrhinum species have not been resolved. For instance, it is currently not possible to infer the ancestral state of a character and whether similar phenotypes might have evolved multiple times within the genus. One obstacle to resolving evolutionary relationships within Antirrhinum is reflected in the unclear taxonomy of its species, many of which have not been reported to have a unique, fixed character (a synapomorphy; Webb, 1971). Therefore, species might not represent discrete genetic entities because they have been delimited artificially. Figure 1. The three morphological subsections of Antirrhinum. Nevertheless, support for the genetic distinction of some A representative of each subsection is shown in situ and in cultivation. Subsection Antirrhinum is represented by A. pseudomajus, subsection recognized species has been provided by allozymes and Streptosepalum is represented by A. braun-blanquetii and subsection Kick- DNA sequence variation (Mateu-Andres and Segarra- xiella is represented by A. pulverulentum. Scale bars: 150 mm for cultivated Moragues, 2000, 2003; Jimenez et al., 2005a; Mateu-Andres plants, and the ruler shown with plants in the field is 35 mm wide. and de Paco, 2005). Relationships above the species level are also unclear. The genus has been divided into three morphological subsections – Antirrhinum, Streptosepalum major subsections, and the involvement of many underlying and Kickxiella – but no subsection has been defined by a genes. However, all putative hybrid species appeared to synapomorphy, and all have been suggested to overlap in resemble one of their parents, suggesting that ancestral phenotype (Rothmaler, 1956; Webb, 1971). The subsections suites of phenotypes have survived hybridization as a result also correlate with ecology: most members of subsection of selection at multiple loci. Kickxiella are small prostrate alpines or xerophytes that grow on rock faces, whereas subsections Antirrhinum and RESULTS Streptosepalum comprise larger, more upright plants that Antirrhinum taxonomy reflects discontinuous phenotypic are able to grow in competition (Figure 1). This raises the variation possibility that each subsection represents an ecotype, and that its species have evolved similar characters indepen- Antirrhinum populations were sampled from across the dently as adaptations to a particular environment, rather geographic range of each recognized species and subspe- than sharing characters through common descent. cies, so that the depth of sampling in each taxon broadly Attempts to resolve a species phylogeny for Antirrhinum corresponded to its abundance (Figure 2; Tables 1 and S1). from DNA sequences have been unsuccessful. Relatively The only recognized species that remained un-sampled was little sequence variation has been found in the genus, Antirrhinum martenii, which could not be found at its ori- consistent with its recent origin, and the variation is not ginal collection sites in the Moroccan Rif. For convenience, distributed consistently between taxa, so that different we treated all taxa at the rank of species, including those that genes support different relationships between species are often regarded as subspecies of A. majus. (Jimenez et al., 2005b; Vargas et al., 2009). Sparse sampling To assess phenotypic variation within Antirrhinum we of the taxa used for DNA sequence analysis might also have grew plants from a representative subset of 98 populations contributed to a lack of phylogenetic resolution, given that together in a glasshouse, and recorded phenotypes for an young species are likely to share many unfixed alleles that average of 5.8 plants from each population (Table 2). We might not be represented in small taxon samples. chose phenotypes that differed between populations, with- Here, we examine evolutionary relationships within the out regard to their differences between species or sub- genus Antirrhinum by comparing populations sampled sections, to avoid any bias towards characters that might from across the geographic range of each species. Plastid automatically support traditional taxonomic divisions. Char- and multilocus nuclear genotypes resolve the relationships acters were then selected for further analysis on two genetic between many species, and further suggest that the tradi- criteria. The first attempted to reduce the use of characters tional morphological subsections largely correspond to that were strongly influenced by environment. It assumed separate evolutionary lineages. They also suggest that that members of the same species are genetically most hybridization has occurred repeatedly between the two similar to each other. Therefore, a comparison of the major ancestral lineages where they overlap in range. variation within species and between species gives an Morphometric analysis of plants grown in common garden estimate of the extent to which a phenotype is genetically conditions supported the phenotypic distinction of the two determined (approximating its broad-sense heritability).

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 1034 Yvette Wilson and Andrew Hudson

Figure 2. The distribution of recognized Antirrhinum species. Species ranges were estimated from collection sites of this and previous studies. The range of A. tortuosum extends further eastwards than is shown here. The names of members of subsection Antirrhinum are shown in red, members of Kickxiella are shown in blue and members of Streptosepalum are shown in orange. Each species range is coloured to represent its geographic location, so that species with similar locations are shown in colours with similar hues. For A. tortuosum and A. linkianum, subpopulations that occur in distinct geographic regions are shown with different colours.

Phenotypes for which more than 35% of the total variance flower size, is accompanied by a narrower leaf shape and occurred within species (stomatal indices in leaves, stems or more reflexed dorsal petals. PC2 and PC3 capture other petals, and leaf epidermal cell size) were excluded from aspects of shape and size variation in both leaves and further analysis. flowers, whereas PC4 mainly describes the extent to which The second criterion for screening phenotypes attempted flowers vary independently of leaves (Figure 3b). Because to minimize repeated sampling of the same genetic differ- the PCs are uncorrelated to each other, each was treated as a ences. Differences in leaf and flower size and shape (allom- separate phenotypic character. etry), for example, which have been used as taxonomic We also avoided using other phenotypes that high characters in Antirrhinum, do not vary independently of correlations suggested might be developmentally con- each other because they are affected by the same genes: i.e. strained (e.g. the lengths of the style and stamens), and the characters are developmentally constrained (Langlade chose flower colour phenotypes to represent the effects of et al., 2005; Feng et al., 2009). Instead of treating them genes known to be involved in interspecies differences separately, we therefore quantified variation in the allome- (Schwinn et al., 2006; Martin et al., 1987; see Experimental tries of leaves and flowers together using a computer shape procedures for details). model of flattened node-4 leaves, flattened dorsal petals For the 22 remaining characters, average values were and side views of intact flowers, in which other aspects of calculated for each of the 98 populations, and the range of variation are apparent (e.g. the angles at which petal lobes the mean values were adjusted to a common linear scale of are presented). The shape model captures co-variation in 0–22. These rescaled values were used in de-trended corre- the positions of points placed around the organ outlines as spondence analysis (DCA) to position each population in orthogonal principal components (PCs; Figure 3). The first phenotypic space, defined by the two major axes of four PCs accounted for 93% of the variance within the genus, co-variation in the data set (Parnell and Waldren, 1996). allowing the leaf and flower allometry of each plant to be A notable feature of the distribution of populations in described accurately with these four PC values. The types of phenotypic space was that members of subsection Kickxiella variation captured by the PCs are shown in Figure 3(b). PC1 formed a cluster distinct from subsections Antirrhinum and describes mainly size variation, showing that organ size is Streptosepalum, with Kickxiella populations mainly sharing the major difference between plants, although this is corre- higher values along DCA axis 1 (Figure 4a). Kickxiella lated with other aspects of shape variation. For instance, an therefore appears to be phenotypically distinct from increase along PC1, which involves an increase in leaf and the other two subsections. The only exception involved

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 Constrained evolution in the genus Antirrhinum 1035

Table 1 Antirrhinum species names and abbreviations populations of the same species tended to cluster together, although many overlapped. The overlap between species Numbers used in in a space involving 22 phenotypes does not rule out the Species Abbreviation morphmetricsa possibility that individual species might be defined by fewer phenotypes or by characters that were not consid- Subsection Antirrhinum ered here. A. australe Rothm. au 8/48 A. barrelieri Boreau ba 3/30 The DCA allows co-variation between characters to be A. boissieri Rothm. bo 1/12 represented in the same phenotype space as the populations A. cirrhigerum Filcahob ci 4/33 (Figure 4b), illustrating their effects on the separation of A. graniticum Rothm. gr 7/33 populations. The spread of characters along axis 1 sug- A. grosii Font Quer go gested that the phenotypic distinctiveness of subsection A. latifolium Miller la 5/10 A. linkianum Boiss. & Reuterb li 6/25 Kickxiella, which has higher axis-1 values, does not depend A. litigiosum Paub lt 7/29 solely on one type of character, for example size or flower A. majus L.b ma colour. This was further supported by the finding that A. pseudomajus ps 3/15 subsection Kickxiella differed significantly from the other b Ferna´ ndez-Casas two subsections for all phenotypes, except PC2, PC4 and Ve A. siculum Millar si 7/16 A. striatum Ferna´ ndez-Casasb st 6/17 (P £ 0.05 in Student’s t-tests). Therefore, subsection Kickxi- A. tortuosum Vent.b to 9/99 ella can be defined by a combination of phenotypes that are Subsection Kickxiella likely to reflect variation in a number of different genes. A. charidemi Lange ch 1/4 In contrast, all subsections overlapped along DCA axis 2, A. hispanicum Chav. hi 3/16 which also reflects variation in a number of different A. hispanicum Mh 3/16 (Moroccan accessions) characters, although members of the same species tended A. lopesianum Rothm. lo 1/3 to cluster with each other along this axis. A. microphyllum Rothm. mi 1/3 A. molle Lange mo 6/47 Plastid haplotypes support distinct evolutionary lineages A. mollisimum Rothm. ms 3/30 One explanation for the partial overlap of Antirrhinum spe- A. pertegasii Rothm. pe 1/6 A. pulverulentum Lazaro pu 2/24 cies in phenotypic space is that the genus behaves as a A. rupestre Rothm. ru 5/20 single interbreeding population, in which the genetic A. sempervirens Lapeyr. se 2/11 differences between individuals reflect their geographic A. subbaeticum Guemes su 1/3 separation (isolation by distance). In this case, the similar A. valentinum Font Quer va 1/3 phenotypes of Kickxiella species should either reflect their Subsection Streptosepalum A. braun-blanquetii Rothm. bb 1/9 closer genetic relatedness and geographic proximity to each A. meonanthum me 1/6 other, or may be independent of both relatedness and Hoffmans. & Link geography if similar phenotypes have evolved independently. We took this scenario of a single population as the a The number of populations of each species used in morphometric null hypothesis against which to test alternatives, including analysis is shown, followed by the total number of individuals sampled. the possibility that members of subsection Kickxiella share bThese taxa have also been considered subspecies of A. majus, e.g. similar phenotypes because they descend from a separate A. majus ssp. striatum or A. majus ssp. majus. evolutionary lineage. We first examined the relatedness of individuals from their plastid haplotypes. Because previous studies had populations from Morocco that had previously been detected relatively little sequence variation in Antirrhinum included in Antirrhinum hispanicum (subsection Kickxiella; plastids (Jimenez et al., 2005b), we compared 19 non-coding Rothmaler, 1956). These mapped in the same phenotypic loci across 10 Antirrhinum species, and identified the three space as subsections Antirrhinum and Streptosepalum, most variable (trnD-trnT, trn-S-trnR and trnS-trnFM; Shaw whereas Spanish populations of A. hispanicum, from which et al., 2005, 2007). Their sequences were then obtained from the species was originally described, fell within the Kickxi- 90 populations that represented the geographic range of ella cluster. Because Moroccan populations could not be each species, and were found to contain 54 polymorphisms classified as A. hispanicum on grounds of morphology, arranged into 34 different haplotypes. The relationships they were subsequently treated as a separate taxon within between haplotypes were inferred by parsimony. To allow subsection Antirrhinum. for the resolution of a purely branching haplotype network Although DCA supported the phenotypic distinctiveness (Figure 5), 12 single-nucleotide polymorphisms that were of Kickxiella, subsections Streptosepalum and Antirrhinum present in rare haplotypes and not fixed in any species were were not separated from each other. At the species level, assumed to be homoplasious to mutations supporting

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 1036 Yvette Wilson and Andrew Hudson

Table 2 Phenotypes used in de-trended correspondence analysis (DCA)

Character H2a Description Scaleb

PC1 81 First PC from the allometry model of leaves and flowers C PC2 83 Second PC from the allometry model of leaves and flowers C PC3 75 Third PC from the allometry model of leaves and flowers C PC4 73 Fourth PC from the allometry model of leaves and flowers C Anth 91 Intensity of red anthocyanin pigmentation in the corolla D, 5 Branch 67 Ratio of the length of the longest axillary branch to Ht (below) C CarpL 88 Carpel length (from ovary base to stigma tip) C El 73 Presence of anthocyanin outside the corolla face and dorsal petal sinus D, 2 FlNode 67 Number of the node at which flowers were first produced D, 29 Ht 73 Length of the stem from cotyledons to first flower C InfNode 84 Maximum number of nodes bearing open flowers at any one time D IntHairs 88 Density of stem trichomes between nodes 1 and 2 C LfHair 90 Density of adaxial epidermal trichomes in node-4 leaves C PeCell 65 Density of adaxial epidermal cells in dorsal petals C PedL 75 Pedicel length C StemW 60 Diameter of the stem midway between nodes 1 and 2 C StRatio 73 Ratio of ventral to dorsal stamen lengths C Sulf 87 Presence of widespread yellow aurone pigment in the corolla D, 2 Ve 68 Presence of darker anthocyanin pigmentation over corolla veins D, 2 YelFace 78 Extent of yellow aurone pigmentation in the corolla face D, 4 YelHair 75 Extent of yellow trichomes within the ventral corolla tube D, 5 YelTube 68 Extent of yellow pigmentation within the ventral corolla tube D, 4 aPercentage of the total variance in each character that is contributed by differences between species, which approximates to the heritability (H 2)of the character in the broad-sense. bCharacters were either measured on a continuous scale (C) or scored in discrete categories (D), in which case the number of character states is given (e.g. D, 2 refers to a binary character). Units are not given for continuous characters because they were subject to linear rescaling, which made them dimensionless. deeper branches. Sequences from Misopates and Chae- (subsection Antirrhinum) has only clade-IV Kickxiella norrhinum, each of which has been suggested to be sister haplotypes, and A. sempervirens, A. lopesianum, to Antirrhinum (Ghebrehiwet et al., 2000; Oyama and A. pertegasii and A. molle in subsection Kickxiella appear Baum, 2004; Vargas et al., 2004), both rooted the Antirrhinum fixed for clade-II haplotypes usually found in subsections haplotype network in the same position. Antirrhinum and Streptosepalum. Secondly, A. tortuosum, The network contained four main clades (Figure 5). A. barrelieri and A. australe from subsection Antirrhinum, Clade I was restricted to Antirrhinum siculum. Clade II and A. pulverulentum from subsection Kickxiella, carry occurred mainly in subsections Antirrhinum and Strepto- both clade-II and clade-IV haplotypes. sepalum, and Clades III and IV occurred mainly in The depth of taxon sampling was increased by genotyp- subsection Kickxiella. This distribution of haplotypes ing another 273 individuals from an additional 156 popula- cannot be explained solely by geographic separation, tions (Table S1). Clade-IV haplotypes were identified by the and therefore is inconsistent with the null hypothesis that presence of an MseI site in trnD–trnT, and clade-II haplo- the genus behaves as a single, unstructured population. types were identified by a TfiI site in trnS–trnR. Individuals For instance, members of subsection Antirrhinum carrying lacking both these sites were assumed to carry clade-III a clade-II haplotype occur throughout the geographic haplotypes, unless they were accessions of A. siculum range of the genus, and overlap with Kickxiella species (clade I). The additional samples confirmed the distribution carrying clade-IV haplotypes. A more plausible explana- of haplotypes identified by DNA sequencing, and also tion is that the distribution of plastid haplotypes reflects revealed that populations of A. litigiosum (subsection the evolutionary relationships between species: i.e. clade- Antirrhinum) from the Valencia region carried clade-IV III and -IV haplotypes were present in the Kickxiella (Kickxiella) haplotypes, whereas those from central Spain ancestor; clade-II haplotypes were present in the separate carried clade-II haplotypes only. lineage leading to subsections Antirrhinum and Strepto- One explanation for the cases of incongruence between sepalum; and clade-I haplotypes were present in the plastid haplotype and morphology is that ancestral poly- lineage represented by A. siculum only. However, several morphisms have persisted in the lineages, leading to taxa have haplotypes that are incongruent with their different subsections (lineage sorting). However, this can- phenotypes, in one of two respects. First, A. latifolium not easily account for the presence of highly diverged

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 Constrained evolution in the genus Antirrhinum 1037

(a) Nuclear genotypes support species delimitations and evolutionary lineages

Because maternally inherited plastid genomes provide only limited information about the relationships between individuals, particularly for hybrids, we also examined variation in nuclear genes. Previous studies had found relatively (b) little nuclear sequence variation and inconsistent distribution between Antirrhinum species (Gu¨ bitz et al., 2003; Vargas et al., 2009). We therefore sampled multiple nuclear loci as amplified fragment length polymorphisms (AFLPs) from the set of 293 accessions that had been used for plastid haplotype analysis. Removal of potentially homoplasious markers left 381 informative AFLPs. The AFLPs were used to infer the population structure within Antirrhinum using a Bayesian model, STRUCTURE, which assigns individuals to a specified number of populations by minimizing genetic disequilibria within each population (Falush et al., 2003). This method is not dependent on prior taxonomic assumptions, and is able to represent hybrids as genetic admixtures of other populations. When STRUCTURE was used to assign each accession to Figure 3. Models describing organ shape and size variation in the genus one of two populations (K = 2 in Figure 7), one population Antirrhinum. comprised subsection Streptosepalum and most of sub- (a) Points were placed around node-4 leaves, flattened dorsal petals and section Kickxiella, and the second consisted of subsection images of intact corollas. Green points were placed manually and red points were spaced automatically between them. Variation in the positions of all the Antirrhinum and the Kickxiella species from south-east Spain points within the genus were described by principal components (PCs). The (A. hispanicum, A. mollissimum, A. rupestre and A. chari- effects of variation of Æ2SDs in the first four PCs are shown relative to demi). This division therefore supported the genetic distinc- the mean organ outlines in black. Var shows the percentage of the variation in the genus that is captured by each PC. tion of Streptosepalum and northern Kickxiella species from subsection Antirrhinum and southern Kickxiella. A. latifolium and A. molle, which showed incongruence of plastid haplotypes and morphology consistent with hybrid origins, haplotypes within the same species. For example, the were suggested to have admixed nuclear genomes, as was clade-II and clade-IV haplotypes found in A. tortuosum A. siculum. (subsection Antirrhinum) differ by up to 21 mutations, and A maximum of 14 different populations could be identi- those in Spanish A. hispanicum (subsection Kickxiella) differ fied by STRUCTURE (Figures 7 and S3). Only two of these – by up to 17 mutations, although only 24 mutations distin- A. graniticum and A. latifolium – consisted almost entirely of guish the most dissimilar haplotypes within the genus as unique haplotypes, suggesting a high level of coalescence a whole (Figure 5). Such cases of incongruence are more within these species. The two Streptosepalums remained consistent with transfer of haplotypes by hybridization within the same population as A. grosii and A. lopesianum between lineages. Further evidence for hybridization is from subsection Kickxiella. The central Spanish Kickxiella provided by the location of populations with incongruent species, which are all local endemics and do not overlap haplotypes. They occur where Kickxiella species overlap in range, were assigned to another population, which with members of the other subsections: for example, with we subsequently refer to as core Kickxiella. Many of the subsection Antirrhinum in the Sierra Nevada of south-east remaining species fell into geographic groups with similar Spain (Figure 6). In contrast, haplotypes are congruent with genetic compositions. In south-east Spain, for example, morphology where subsections grow in isolation (e.g. A. mollissimum was designated an admixture of genomes subsection Antirrhinum in Morocco, western Portugal and from neighbouring A. charidemi and A. rupestre, and the northern Spain, and Kickxiella in central Spain). The only A. rupestre genome was also found admixed in A. hispan- notable exception is that of A. latifolium, which carries a icum and A. barrelieri nearby. clade-IV haplotype, is restricted to the Alps and does not Cultivars of A. majus were assigned to the same popula- currently co-occur with a Kickxiella species, although this tion as A. pseudomajus and A. striatum from around the does not rule out the possibility of hybridization with a eastern Pyrenees, supporting the domestication of A. majus previously sympatric Kickxiella. from this population.

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 1038 Yvette Wilson and Andrew Hudson

Figure 4. Distribution of Antirrhinum species in phenotype space. The two major axes of co-variation in 22 different phenotypic characters were identified by de-trended correspondence analysis. The mean values for 98 populations were plotted in the space define by these axes (a).Members of subsection Antirrhinum are shown as triangles, members of subsection Streptosepalum are shown as squares and members of subsection Kickxiella are shown as circles. Populations are coloured as in Figure 2, to reflect their geographic origins. The label mh denotes Moroccan accessions previously assigned to A. hispanicum. The contribution of each character to the phenotypic space is shown in (b). Characters and their abbreviations are detailed in Table 2, and accession numbers for populations are presented in Figure S2.

Figure 5. Relationships between plastid haplotypes. Sampled haplotypes are shown in boxes containing the species names and accession numbers of the plants carrying them. The most parsimonious relationships between haplotypes are shown by lines connecting boxes with cross lines that represent the number of mutations by which haplotypes differ. Diagonal cross lines show mutations that were assumed to be homoplasious with earlier mutations. Branches in which mutations introduced an MseIorTﬁI restriction site are also shown. The tree is rooted with sequences from Misopates orontium, which differs from the in-group by 19 mutations.

The ability to progressively subdivide the genus into onomy, and suggested that AFLP genotypes might resolve populations with boundaries that corresponded to classical relationships between species, even though extensive species delimitations provided support for traditional tax- admixture was inferred for all but A. graniticum and

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 Constrained evolution in the genus Antirrhinum 1039

Figure 6. Geographic distribution of plastid haplotypes. Members of subsection Antirrhinum are shown as circles, members of subsection Kickxiella are shown as triangles and members of subsection Streptosepalum are shown as squares. Filled shapes represent haplotypes that are congruent with morphology; unﬁlled shapes show incongruent haplotypes.

A. latifolium. Therefore, genetic distances were calculated Antirrhinum, supporting the phenotypic evidence for their between all pairwise combinations of individuals, and were membership of this subsection. The third major clade then used in neighbour-joining analysis with Misopates consisted of A. meonanthum and A. braun-blanquetii orontium as the out-group. When all individuals were (subsection Streptosepalum) from north-west Iberia, included in the analysis, members of the same species together with two of their neighbouring Kickxiella species, usually clustered with each other, as they did in the STRUC- A. grosii and A. lopesianum. TURE analysis, but the relationships between species were DISCUSSION poorly resolved (Figure S4). One explanation for poor resolution is that some of the taxa have hybrid origins, and so By comparing the genotypes of Antirrhinum populations at carry alleles from multiple lineages. These shared alleles multiple nuclear loci we have identified three well-supported could reduce resolution both within and between the clades of species. The two largest clades do not correspond parental lineages. We therefore excluded potential hybrids to individuals from the same geographic locations, although – i.e. plants with incongruent nuclear and plastid genotypes, they contain geographic subgroups within them, and so or apparently admixed nuclear genomes – if they reduced cannot be explained by a genetic structure that relates solely the overall support. The remaining accessions were to geographic isolation. The clades are therefore likely to resolved into three major clades (Figure 8). One comprised represent different evolutionary lineages. the core Kickxiella, within which A. pulverulentum formed One clade consists of species from subsection Kickxiella, a supported group with its central Spanish neighbour which was originally defined by the shared phenotypes of A. microphyllum, and A. pertegasii grouped with the geo- its members. Morphometric analysis under common garden graphically distant A. subbaeticum. The second major clade conditions confirmed the phenotypic distinctiveness of consisted of species from subsection Antirrhinum, with this subsection, with common characters that include small A. siculum at its base and A. graniticum sister to the stature, small, ovate and hairy leaves, and small pale flowers remaining taxa, which included two geographic groupings: with reflexed petals. The ancestral Kickxiella is therefore A. pseudomajus, A. striatum and A. litigiosum from around likely to have shared these characters, and, like its descen- the eastern Pyrenees, and A. linkianum and A. cirrhigerum dants, to have lived on rock faces. The second major clade from the west coast of Iberia. All accessions of Moroccan comprised mainly members of subsection Antirrhinum. ‘A. hispanicum’ clustered together within subsection These are dark-pink- or yellow-flowered species that form

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 1040 Yvette Wilson and Andrew Hudson

Figure 7. Assignment of individuals to populations. Plants were assigned to different ancestral populations, each shown in a different colour, by their nuclear genotypes. Representative assignments are shown for models in which the maximum number of populations (K) was set to either two or 17. Black bars denote members of subsection Antirrhinum, dark-grey bars denote members of subsection Streptosepalum and light-grey bars denote members of subsection Kickxiella.

Figure 8. A neighbour-joining tree of Antirrhi- num species. The tree was made from pairwise nuclear genetic distances between inferred non-hybrid accessions. Support values are shown for nodes that were recovered in more than 50% of 1000 bootstrap replicates.

large, upright plants, with larger, more elongated leaves, classical division of Antirrhinum into three morphological and are able to grow in competition. The third major clade subsections therefore appears partly natural. contains both of the species from subsection Streptosep- The overlap of the three subsections along DCA axis 2 alum and two species with Kickxiella phenotypes. The implies that all subsections vary to a similar extent for the

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 Constrained evolution in the genus Antirrhinum 1041 different morphological and flower-colour characters that Mediterranean (Kropf et al., 2003; Dixon et al., 2007). contribute to this axis. This is consistent with ancestral Contraction to alpine refugia can also explain why all high variation that has not been fixed during divergence of mountain ranges in Iberia and southern France either have subsections, although fixation may have occurred during their own endemic core Kickxiella species or, in the case speciation because members of the same species tend to of the Alps (A. latifolium), the Pyrenees (A. molle and cluster along axis 2. A. sempervirens) and the Baetic Cordillera (the south-east Distinct plastid lineages were also identified, and each Spanish species), retain vestiges of the Kickxiella lineage in was found predominantly in subsections Antirrhinum and their plastid and nuclear genotypes. Populations from the Rif Streptosepalum,orinKickxiella. Many species carry con- Mountains of Morocco have been recognized as a species gruent nuclear and plastid genomes, for instance A. micro- within subsection Kickxiella (A. martenii; Rothmaler, 1956), phyllum, A. subbaeticum and A. valentinum have genotypes suggesting that Kickxiella might once have extended into expected of direct descendants of the core Kickxiella line- North Africa. As Kickxiella species are found on dry rock age, whereas A. siculum, A. graniticum, A. pseudomajus, faces, mostly at higher elevations, contraction in their A. striatum, A. cirrhigerum and A. linkianum have both ranges could have occurred during periods of increasing plastid and nuclear genomes representative of subsection rainfall and warming. Several major events of this kind have Antirrhinum. occurred in the Mediterranean region since the end of the However, several species appear to have nuclear and last ice age (Fletcher et al., 2010). The same environmental plastid genomes from different lineages. These include changes that reduced the range of Kickxiella could have A. sempervirens, A. pertegasii and A. pulverulentum, which allowed subsection Antirrhinum to spread through lowland have subsection Antirrhinum plastids but core Kickxiella regions, perhaps from coastal refugia, and therefore to nuclei. Such cytonuclear incongruence can be explained hybridize with Kickxiella in regions of contact. if hybridization between subsections was followed by intro- Character evolution gression involving Kickxiella, leading to the capture of an Antirrhinum plastid by a pre-dominantly Kickxiella nuclear The genetic relationships between Antirrhinum species genome. Similarly, A. tortuosum could also have captured suggest that several hybridization events occurred between its diverse incongruent plastids from multiple Kickxiella subsections Antirrhinum and Kickxiella, yet the phenotypes donors. of the two subsections remain distinct from each other. This The remaining species share their nuclear genotypes with is most apparent in two species that are sympatric in the both Antirrhinum and Kickxiella lineages, and consequently Sierra Nevada of south-east Spain: A. rupestre (Kickxiella cannot be assigned to one of the three major clades. phenotype) and A. barrelieri (Antirrhinum phenotype), However, they share plastids and nuclear genes with their which have indistinguishable AFLP and plastid genotypes, geographic neighbours, suggesting that they are the result consistent with their continuing hybridization. Therefore, of hybridizations between subsections Antirrhinum and although hybridization has the potential to create new Kickxiella without significant introgression. This is particu- adaptive combinations of phenotypes (e.g., Rieseberg et al., larly apparent in the species from south-east Spain, which 2003), evolution in Antirrhinum appears to be constrained. have similar nuclear genotypes and carry both Kickxiella Two factors suggest that the evolutionary constraints and Antirrhinum plastids. Hybridization could also have within Antirrhinum are not developmental. Firstly, the given rise to the nuclear clade consisting of the two phenotypic characters that distinguish subsections Kickxiel- Streptosepalum species together with A. grosii and la and Antirrhinum were chosen to represent variation in A. lopesianum from subsection Kickxiella, because these multiple genes with independent effects. Secondly, hybrids species are found together in north-west Iberia and carry between A. majus and A. charidemi have shown that multi- diverged plastids. Alternatively, the clade might represent a ple genes underlie the differences between their Antirrhi- third ancestral lineage in which some members have num and Kickxiella phenotypes (Langlade et al., 2005; Feng captured incongruent plastids from subsection Kickxiella. et al., 2009). For instance, 10 genes with additive effects

It is not possible to resolve the order in which the three were found to affect leaf size, so that F1 hybrids and almost major lineages diverged from each other. However, the all F2 progeny had leaf phenotypes that were intermediate distribution of the core Kickxiella species between the between the two parents (i.e. they occupied the gap in mountains of central Spain suggests that this lineage was phenotypic space between subsections Antirrhinum and once more widespread, and became fragmented on con- Kickxiella). traction to its current mountain refugia. This fragmentation A more likely explanation is therefore that Antirrhinum could have contributed to the formation of distinct species. phenotypes are constrained by selection. For some charac- A similar range contraction has been proposed to account ters this might reﬂect contrasting adaptations to life on bare for the genetic relationships in several alpine species that rock faces or in competition: for instance, small organs and are currently restricted to mountains around the western dense hairs might be advantageous in limiting water loss on

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 1042 Yvette Wilson and Andrew Hudson dry rock faces, but a disadvantage in competing with other translated and rotated to set their centroids at the origin, and to species (Parkhurst and Loucks, 1972; Ehleringer et al., 1976). minimize variance in the positions of points (a Procrustes alignment Hybrids with intermediate phenotypes could therefore be without scaling). Principal component analysis (PCA) was used to partition the variance between plants into orthogonal PCs. For each maladapted in both parental habitats. Because the differ- type of organ, PC1 identified size as the major source of variation ences between Antirrhinum and Kickxiella phenotypes between plants (Figure S1). To minimize re-sampling this size appear to involve multiple unlinked genes, selection would variation (which could have a common genetic basis in the different have to act co-ordinately on many loci. organs), we therefore made a single shape model for all three The proposed evolutionary history of Antirrhinum, which organs together. The leaf matrices were first rescaled so that the total variance between plants in the leaf data set was the same as for involves hybridization between emerging lineages, influ- petals and flowers together. The three point matrices (leaf, petal and enced by environmentally induced range changes, might be flower) representing each individual were combined and subject to typical of the early stages of species formation in many taxa. PCA. It suggests that adaptation can exert a major constraint on Plastid haplotype analysis diversity, even when environments change rapidly and novel genotypes are produced frequently in hybrids. To compare plastid haplotypes, 19 non-coding loci (Shaw et al., One testable prediction of our hypothesis that ancestral 2005, 2007) were amplified from 10 Antirrhinum species, and sequences of the three loci with the highest proportion of nucleotide combinations of genes are reselected following hybridiza- substitutions relative to indels were obtained from all populations. tion in Antirrhinum is that similar phenotypes will have a Further accessions were genotyped by incubating the product from similar genetic basis, even in species that share little of their trnD–trnT with MseI and from trnS–trnR with TfiI. Where lack of nuclear genomes. The genus Antirrhinum, and particularly cleavage suggested a novel haploptype for the species, PCR prod- its recent hybrid species, might therefore allow adaptive ucts were sequenced to confirm the absence of the restriction site. Amplified fragment length polymorphisms (AFLPs) were pro- genes to be identified through genotype–phenotype associ- duced from DNA digested with PstI and MseI, and were amplified ation. with four primer combinations (P11–M49, P11–M41, P12–M37 and P14–M35, in Keygene nomenclature). AFLPs that did not represent EXPERIMENTAL PROCEDURES genotypes consistently were identified by carrying out 10 AFLP Taxon sampling reactions on each of four independent DNA extractions from the same set of genetically diverse plants. Fragments that were not The populations used in this study are detailed in Table S1. Popu- detected in all replicates of the same plant were removed from lations of most species were sampled in 2006–2007. Sampling sites the larger data set. Significant negative correlations between AFLP included locations used in previous taxonomic accounts to aid band size and frequency were detected for two primer combina- identification. Seeds of these accessions are available on request. tions, suggesting size homoplasy among smaller bands. Fragments Other accessions were kindly provided by Isabel Mateu-Andre´ s, smaller than 100 nucleotides were therefore excluded from the Christophe The´ baud, Thomas Gu¨ bitz, Enrico Coen and their analysis. colleagues. Individuals were clustered into populations using STRUCTURE (Falush et al., 2003, 2007). Priors and parameters relating to Phenotype analysis correlated allele frequencies and the extent of population subdivision were found to have only minimal effects on population Plants for phenotype analysis were grown from field-collected assignments and likelihoods of models fitting the data (Jakobsson seeds or from seeds produced by intercrossing members of the and Rosenberg, 2007). At least eight simulations were carried out same population. Germination was synchronized in March 2007 by for each value of K (number of populations), without varying other imbibing seeds in 10 lM gibberellin (GA ), and plants were grown in 3 parameters. Each simulation comprised 20 000 burn-in and 120 000 natural light in a glasshouse. Phenotypes were recorded as the fifth experimental replications. Results were visualized with DISTRUCT flower opened. Phenotypes and their abbreviations are listed in (Rosenberg, 2004). Table 2. Cell and hair densities were measured from impressions Pairwise genetic distances (Jaccard) between individuals were made in cyanoacrylate glue on microscope slides by counting the calculated from AFLP data in PAST (Hammer et al., 2004), and number of cells or hairs within a 2 mm2 area. Other phenotypes neighbour-joining analyses of distance matrices in PHYLIP (Felsen- were measured directly from plants or from digital images. Flower stein, 1989). images were used to score the presence of yellow aurone pigments outside the corolla face, which reflects variation in the sulfurea gene ACKNOWLEDGEMENTS (sulf; Whibley et al., 2006), darker anthocyanin in petal cells over- lying major veins (Ve, conditioned by the Venosa locus; Schwinn We are grateful to Maureen Erasmus, Kim Coulson, Mary Coulson, et al., 2006), and restriction of anthocyanin pigmentation to the Niall Wilson, Monique Burrus and Christophe The´ baud for their corolla face and between dorsal petal lobes, which is characteristic considerable help with fieldwork. This work was supported by of mutations in the Eluta gene (El; Martin et al., 1987). Although BBSRC, through grant BB/D552089/1, and a postgraduate student- these three genes show epistatic interactions, they have indepen- ship to YW, and by a Small Project Grant from the University of dent effects that allow all combinations of genotypes to be inferred Edinburgh Fund. (Martin et al., 1987). Shape models were initially made for individual organs (node-4 SUPPORTING INFORMATION leaves, flattened dorsal petals or intact flowers) using the method of Additional Supporting Information may be found in the online Langlade et al. (2005), in which points were positioned around the version of this article: outlines of each organ. Matrices of point co-ordinates were then Figure S1. Allometric variation in Antirrhinum leaves and flowers.

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043 Constrained evolution in the genus Antirrhinum 1043

Figure S2. Distribution of numbered Antirrhinum populations in late quaternary: insights from Pritzelago alpina (L.) O. Kuntze (Brassica- phenotype space. ceae). Mol. Ecol. 12, 931–949. Figure S3. Assignment of Antirrhinum accessions to between two Langlade, N.B., Feng, X., Dransfield, T., Copsey, L., Hanna, A.I., Thebaud, C., and 14 populations. Bangham, A., Hudson, A. and Coen, E. (2005) Evolution through genetically controlled allometry space. Proc. Natl Acad. Sci. USA, 102, 10221– Figure S4. Neighbour-joining analysis of all Antirrhinum acces- 10226. sions. Martin, C., Coen, E., Robbins, T., Bartlett, J., Alemaida, J. and Carpenter, R. Table S1. Antirrhinum populations used in this study. (1987) The control of floral pigmentation in Antirrhinum. Biochem. Soc. Please note: As a service to our authors and readers, this journal Trans. 15, 14–17. provides supporting information supplied by the authors. Such Mateu-Andres, I. and de Paco, L. (2005) Allozymic differentiation of the materials are peer-reviewed and may be re-organized for online Antirrhinum majus and A. siculum species groups. Ann. Bot. 95, 465– delivery, but are not copy-edited or typeset. Technical support 473. issues arising from supporting information (other than missing Mateu-Andres, I. and Segarra-Moragues, J.G. (2000) Population subdivision files) should be addressed to the authors. and genetic diversity in two narrow endemics of Antirrhinum L. Mol. Ecol. 9, 2081–2087. REFERENCES Mateu-Andres, I. and Segarra-Moragues, J.G. (2003) Allozymic differentiation of the Antirrhinum graniticum and the Antirrhinum meonanthum species Dixon, C.J., Schonswetter, P. and Schneeweiss, G.M. (2007) Traces of ancient groups. Ann. Bot. 92, 647–655. range shifts in a mountain plant group (Androsace halleri complex, McGregor, A.P., Orgogozo, V., Delon, I., Zanet, J., Srinivasan, D.G., Payre, F. Primulaceae). Mol. Ecol. 16, 3890–3901. and Stern, D.L. (2007) Morphological evolution through multiple cis-reg- Ehleringer, J., Bjorkman, O. and Mooney, H.A. (1976) Leaf pubescence: effects ulatory mutations at a single gene. Nature, 448, 587–590. on absorptance and photosynthesis in a desert shrub. Science, 192, 376– Oyama, R. and Baum, D.A. (2004) Phylogenetic relationships of north Amer- 377. ican Antirrhinum (Veronicacae). Am. J. Bot. 91, 918–925. Falush, D., Stephens, M. and Pritchard, J.K. (2003) Inference of population Parkhurst, D.F. and Loucks, O.L. (1972) Optimal leaf size in relation to envi- structure using multilocus genotype data: linked loci and correlated allele ronment. J. Ecol. 60, 505–537. frequencies. Genetics, 164, 1567–1587. Parnell, J. and Waldren, S. (1996) Detrended correspondence analysis in the Falush, D., Stephens, M. and Pritchard, J.K. (2007) Inference of population ordination of data for phenetics and cladistics. Taxon, 45, 71–84. structure using multilocus genotype data: dominant markers and null Rieseberg, L.H., Raymond, O., Rosenthal, D.M., Lai, Z., Livingstone, K., alleles. Mol. Ecol. Notes, 7, 574–578. Nakazato, T., Durphy, J.L., Schwarzbach, A.E., Donovan, L.A. and Lexer, C. Felsenstein, J. (1989) PHYLIP – phylogeny inference package (version 3.2). (2003) Major ecological transitions in wild sunflowers facilitated by Cladistics, 5, 164–166. hybridization. Science, 301, 1211–1216. Feng, X., Wilson, Y., Bowers, J., Kennaway, R., Bangham, A., Hannah, A., Rosenberg, N.A. (2004) DISTRUCT: a program for the grphical display of Coen, E. and Hudson, A. (2009) Evolution of allometry in Antirrhinum. Plant population structure. Mol. Ecol. Notes, 4, 137–138. Cell, 21, 2999–3007. Rothmaler, W. (1956) Taxonomische monographie der Gattung Antirrhinum. Fletcher, W.J., Sanchez Goni, M.F., Peyron, O. and Dormoy, I. (2010) Abrupt Feddes Rep. 136, 1–124. climate changes of the last deglaciation detected in a western mediterra- Schwarz-Sommer, Z., Davies, B. and Hudson, A. (2003) An everlasting nean forest record. Clim. Past, 6, 246–264. pioneer: the story of Antirrhinum research. Nat. Rev. Genet. 4, 657– Ghebrehiwet, M., Bremer, B. and Thulin, M. (2000) Phylogeny of the tribe 666. Antirrhineae (Scrophulariaceae) based on morphological and ndhF Schwinn, K., Venail, J., Shang, Y., Mackay, S., Alm, V., Butelli, E., Oyama, R., sequence data. Plant Syst. Evol. 220, 223–239. Bailey, P., Davies, K. and Martin, C. (2006) A small family of MYB-regulatory Gu¨ bitz, T., Caldwell, A. and Hudson, A. (2003) Rapid molecular evolution of genes controls floral pigmentation intensity and patterning in the genus CYCLOIDEA-like genes in Antirrhinum and its relatives. Mol. Biol. Evol. 20, Antirrhinum. Plant Cell, 18, 831–851. 1537–1544. Shaw, J., Lickey, E.B., Beck, J.T. et al. (2005) The tortoise and the hare II: Hackbarth, J., Michaelis, P. and Scheller, G. (1942) Untersuchungen an dem relative utility of 21 noncoding chloroplast DNA sequences for phyloge- Antirrhinum-Wildsippen-Sortiment von E. Baur. Z. Indukt. Abst. Verer- netic analysis. Am. J. Bot. 92, 142–166. bungsl. 80, 1–102. Shaw, J., Lickey, E.B., Schilling, E.E. and Small, R.L. (2007) Comparison of Hammer, Ø., Harper, D.A.T. and Ryan, P.D. (2004) PAST: paleontological whole chloroplast genome sequences to choose noncoding regions for statistics software package for education and data analysis. Palaeontol. phylogenetic studies in angiosperms: the tortoise and the hare III. Am. J. Electronica, 4,4. Bot. 94, 275–288. Jakobsson, M. and Rosenberg, N.A. (2007) CLUMPP: a cluster matching and Sutton, D.A. (1988) A Revision of the Tribe Antirrhineae. Oxford: OUP. permutation program for dealing with label switching and multimodality in Thompson, J.D. (2005) Plant Evolution in the Mediterranean. Oxford: OUP. analysis of population structure. Bioinformatics, 23, 1801–1806. Vargas, P., Rossello, J.A., Oyama, R. and Guemes, J. (2004) Molecular evi- Jeong, S., Rebeiz, M., Andolfatto, P., Werner, T., True, J. and Carroll, S.B. dence for the naturalness of genera in the tribe Antirrhineae (Scrophular- (2008) The evolution of gene regulation underlies a morphological differ- iaceae) and three independent evolutionary lineages from the new world ence between two Drosophila sister species. Cell, 132, 783–793. and old. Plant System. Evol. 249, 151–172. Jimenez, J.F., Guemes, J., Sanchez-Gomez, P. and Rossello, J.A. (2005a) Vargas, P., Carrio, E., Guzman, B., Amat, E. and Guemes, J. (2009) A geo- Isolated populations or isolated taxa? A case study in narrowly-distributed graphic pattern of Antirrhinum (Scrophulariaceae) speciation since the snapdragons (Antirrhinum sect. Sempervirentia) using RAPD markers Pliocene based on plastid and nuclear DNA polymorphisms. J. Biogeogr. Plant Syst. Evol. 252, 139–152. 36, 1297–1312. Jimenez, J.F., Sanchez-Gomez, P., Guemes, J. and Rossello, J.A. (2005b) Webb, D.A. (1971) Taxonomic notes on Antirrhinum L. Bot. J. Linn. Soc. 64, Phylogeny of snapdragon species (Antirrhinum; Scrophulariaceae) using 271–275. non-coding cpDNA sequences. Israel J. Plant Sci. 53, 47–54. Whibley, A.C., Langlade, N.B., Andalo, C., Hanna, A.I., Bangham, A., Thebaud, Kropf, M., Kadereit, J.W. and Comes, H.P. (2003) Differential cycles of range C. and Coen, E. (2006) Evolutionary paths underlying flower color variation contraction and expansion in European high mountain plants during the in Antirrhinum. Science, 313, 963–966.

Accession numbers of sequence data: FR690148-FR160233, FR690277-FR690458 and FR690868-FR690870.

ª 2011 The Authors The Plant Journal ª 2011 Blackwell Publishing Ltd, The Plant Journal, (2011), 66, 1032–1043