Molecular Phylogenetics and Evolution 51 (2009) 269–280 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev Allopolyploidy in Fragariinae (Rosaceae): Comparing four DNA sequence regions, with comments on classification Magnus Lundberg a,*, Mats Töpel b, Bente Eriksen b, Johan A.A. Nylander a, Torsten Eriksson a,c a Department of Botany, Stockholm University, SE-10691, Stockholm, Sweden b Department of Environmental Sciences, Gothenburg University, Box 461, SE-40530, Göteborg, Sweden c Bergius Foundation, Royal Swedish Academy of Sciences, SE-10405, Stockholm, Sweden article info abstract Article history: Potential events of allopolyploidy may be indicated by incongruences between separate phylogenies Received 23 June 2008 based on plastid and nuclear gene sequences. We sequenced two plastid regions and two nuclear ribo- Revised 25 February 2009 somal regions for 34 ingroup taxa in Fragariinae (Rosaceae), and six outgroup taxa. We found five well Accepted 26 February 2009 supported incongruences that might indicate allopolyploidy events. The incongruences involved Aphanes Available online 5 March 2009 arvensis, Potentilla miyabei, Potentilla cuneata, Fragaria vesca/moschata, and the Drymocallis clade. We eval- uated the strength of conflict and conclude that allopolyploidy may be hypothesised in the four first Keywords: cases. Phylogenies were estimated using Bayesian inference and analyses were evaluated using conver- Allopolyploidy gence diagnostics. Taxonomic implications are discussed for genera such as Alchemilla, Sibbaldianthe, Cha- Fragariinae Incongruence maerhodos, Drymocallis and Fragaria, and for the monospecific Sibbaldiopsis and Potaninia that are nested Molecular phylogeny inside other genera. Two orphan Potentilla species, P. miyabei and P. cuneata are placed in Fragariinae. Bayesian convergence diagnostics However, due to unresolved topological incongruences they are not reclassified in any genus. Ó 2009 Elsevier Inc. All rights reserved. 1. Introduction species were thought to be polyploid, ranged between 30% and 52% (Müntzing, 1936; Darlington, 1937; Stebbins, 1950; Grant, The importance and detection of hybridisation and its role in 1963, 1981). More recent studies based on guard cell sizes suggest species formation have been, and continue to be, major foci in evo- that more than 70% of all extant angiosperms are of polyploid ori- lutionary plant research, notably because hybridisation followed gin (Masterson, 1994), and polyploidy has been suggested to be the by polyploidisation is suggested to be an important mode of speci- cause of at least 2–4% of all recent speciation events (Otto and ation among vascular plants. Following hybridisation, alternations Whitton, 2000), thus suggested to be an important mode of speci- at genomic and gene levels occur in the offspring with potential ation (e.g. Mallet, 2007). Such estimates, in particular the older advantages of the polyploid species compared to its diploid pro- ones, may be rather inaccurate. As an example of the problems genitors. This may be caused by changes of the newly formed hy- in making such estimates, the model plant Arabidopsis thaliana is brids by reproductive isolation, or by changes in biochemical, defined as a functional diploid with a relatively small genome physiological and developmental flexibility; changes that are ex- (The Arabidopsis Genome Initiative, 2000). However, results from pressed, for example in plant size, flowering time and reproductive whole genome sequencing indicates that there have been two or output, but also in new combinations of characters (Levin, 1983; more rounds of genome duplication events in the evolution of that Schranz and Osborn, 2000; Song et al., 1995). Speciation through species (Vision et al., 2000; Blanc et al., 2003; Bowers et al., 2003; polyploidy was discovered in the beginning of the last century by Simillion et al., 2002). Recent genetic and genomic studies suggest Winkler (1916) and Winge (1917). To distinguish between the that most or perhaps all angiosperms have undergone one or sev- two main types of polyploidy commonly recognised, Kihara and eral rounds of polyploidisation followed by extensive diploidisa- Ono (1926) introduced the terms allopolyploidy (duplication of tion (Wolfe, 2001; Eckhardt, 2001), the evolutionary process chromosomes in interspecific hybrids) and autopolyploidy (dupli- where the genomic content of a polyploid species degenerates into cation of chromosomes within a species). a diploid state again. Many estimates of the frequency of polyploidy among angio- Before molecular methods were developed, polyploid specia- sperms have been presented. Early estimates, based on chromo- tion was detected by chromosome counts and crossing experi- some numbers in related species or arbitrary levels above which ments (Grant, 1981). Drawbacks of this method include that it assumes recent polyploidy events and that parental species are ex- * Corresponding author. Fax: +46 (0)8 165525. tant. More recently, several analyses using molecular data have E-mail address: [email protected] (M. Lundberg). been developed to detect and reconstruct hybrid speciation, e.g. 1055-7903/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2009.02.020 270 M. Lundberg et al. / Molecular Phylogenetics and Evolution 51 (2009) 269–280 splits decomposition (Bandelt and Dress, 1992; Huson, 1998) and flicting topologies and with apparent high clade credibilities linkage disequilibrium (e.g. Rieseberg et al., 2003). A third, and in (Huelsenbeck et al., 2002; Nylander et al., 2008). It is possible to our view the currently most powerful way to reconstruct allopoly- investigate the performance of the MCMC runs and to reduce the ploid speciation is to use phylogenies based on nuclear single or risk of analysis artifacts by using convergence diagnostics low copy DNA markers (e.g. Sang and Zhang, 1999; Smedmark (Nylander et al., 2008). et al., 2003; Mason-Gamer, 2001; Popp et al., 2005). In the ideal Fragariinae is a subclade of Rosoideae in Rosaceae (Morgan case, one copy of the gene would be present in the phylogeny for et al., 1994; Eriksson et al., 2003; Potter et al., 2002, 2007). The each ploidy level. Such a complex but bifurcating tree can then clade Fragariinae was discovered by Eriksson et al. (1998) based be turned into a more realistic estimation of the phylogeny: a net- on nuclear ribosomal sequence data and it was further supported work picturing both reticulate and bifurcating branches. This by combining nuclear ribosomal data with plastid sequences method has several advantages: (1) The entire phylogeny of the (Eriksson et al., 2003). Fragariinae is the sister group of Potentilla group is analysed and allopolyploidy can be detected even if ances- in strict sense. Different segregates of genera in Fragariinae have tors are not extant. (2) Parentage is traceable because allopolyp- previously often been classified together with Potentilla, but genera loids have one gene copy from each parent, at least initially. (3) such as Alchemilla and Potaninia have in most cases been placed It may be possible to trace several rounds of allopolyploidy events. elsewhere in Rosaceae (Focke, 1894; Hutchinson, 1964; (4) It is possible to distinguish allopolyploidy (where gene copies Schulze-Menz, 1964; Kalkman, 1988, 2004). Eriksson et al. (1998, from single organisms are distantly related) from autopolyploidy 2003) redefined Potentilla and noted that the rather few species (where gene copies are sisters). There is now a software package that had been classified as Potentilla by earlier authors but were available that aims to turn a bifurcating tree with gene copies into found to belong to Fragariinae had already also been classified in a reticulate tree (‘‘Padre”; Huber et al., 2006). other genera. However, a few orphan Potentilla species still remain However, a difficulty of this method is that it involves exten- in Fragariinae without having been reclassified. sive cloning and sequencing of low copy genes, and in order to Fragariinae contain 10 genera and a few orphan Potentilla spe- be able to obtain sequences from all copies it is often required cies: Alchemilla in the wide sense (at least 350 spp.) including to work with fresh plant material. For this reason, it is practical Aphanes and Lachemilla (Gehrke et al., 2008), Comarum (2 spp.), to first screen a group for potential allopolyploidy events and Dasiphora (ca. 4 spp.), Potaninia (1 sp.), Sibbaldia (2–6 spp.), Sibbal- then focus on specific parts of the phylogeny, using low copy dianthe (2 spp.), Sibbaldiopsis (1 sp.), Chamaerhodos (5–8 spp.), Fra- gene sequences. Screening involves sequencing a large set of spe- garia (ca. 20 spp.) and Drymocallis (ca. 30 spp.), most of which cies for easily amplified DNA regions. Phylogenies are then esti- occur in the northern hemisphere with their highest diversity in mated for each data set and topologies are compared, temperate regions. Previously used taxon names are presented in contrasting plastid sequence data that are generally maternally Table 1. There are no known distinctive morphological synapomor- inherited with nuclear data that are potentially paternal because phies of Fragariinae but it is strongly supported by DNA sequence of biparental inheritance. This comparative step allows us to de- data. The Fragariinae share a number of non-molecular characters tect and evaluate potential conflicts between the separate data