1 INTRODUCTION
Pollination mutualism is one of the major interaction systems between plants and animals, and through this interaction, flower visitors contribute to the reproduction of plants in different ways (Galen, 1996; Scobell & Scott, 2002; Herrera et al., 2006; Inoue et al., 2007; Gómez et al., 2009; Dohzono & Suzuki, 2010; Nattero et al., 2011). Adaptation to locally different pollinator assemblages within the distribution range of a plant species leads to local morphological specialization, which may cause trait diversification and speciation in the plants (Grant & Grant, 1965; Stebbins, 1970; Galen & Newport, 1987). Geographic variation in floral traits such as flower size and shape (Gómez et al., 2006; Nagano et al., 2014), corolla tube size (Hodges, 1997; Fenster et al., 2004), odor (Pellmyr, 1986; Majetic et al., 2009), and color (Campbell et al., 1997; Newman et al., 2012) are considered to have evolved as a local adaptation to regional pollinators. In particular, morphological matching between floral spur length and pollinator proboscis length is well known, with the textbook example being Darwin’s hawkmoth and orchid (Darwin, 1877; Nilsson, 1988). In fact, geographic correlations between floral size and pollinator size have been reported in a variety of plant taxa (Alexandersson & Johnson, 2002; Herrera et al., 2006; Anderson & Johnson, 2008; Johnson & Anderson, 2010; Boberg et al., 2014; Nagano et al., 2014; Kuriya et al., 2015).
Local adaptation of floral traits to pollinators may have occurred across multiple regions, but there is little evidence as to whether variation in floral traits has occurred independently among regional populations (but see Anderson et al., 2014; Toji et al., 2021). As in textbook examples of ecological speciation (Nosil, 2012), one useful approach to understand the interaction between trait diversification and speciation in angiosperms is to combine a field analysis of local plant evolutionary adaptations with a population genetic analysis that examines genetic relationships among populations. In particular, local adaptation of floral traits to pollinators may lead to speciation via the establishment of prezygotic reproductive isolation (Grant-Stebbins model; Grant & Grant; Stebbins, 1970; Johnson & Anderson, 2010; Anderson et al., 2014), because one possible result of specialization of a trait to a particular pollinator is a lack of pollinator sharing among plant populations (Herrera et al., 2006; Anderson & Johnson, 2008; Newman et al., 2015). About 25% of plant diversification events may be associated with pollinator shifts (Van der Niet & Johnson, 2012); thus, combined analyses of local adaptation of floral traits and population genetics can shed light on the mechanisms of plant diversity (Thompson, 2005; Thompson et al., 2013). In this study, we conducted both trait and genetic analyses to determine whether the differentiation of floral traits (flower size) among plant populations in different mountain regions was the result of secondary contact between two differentiated lineages with long and short flowers, or whether flower size evolved recently in each population as an adaptation to the local pollinator size.
In genus Aquilegia (Ranunculaceae), adaptive radiation to different pollinators (bumblebees, hummingbirds, and hawkmoths) has occurred. Mainly, flower color, spur length, flower orientation, and pistil length have evolved to adapt to each pollinator (Fulton & Hodges, 1999; Hodges et al., 2004). Moreover, molecular phylogenetic evidence also indicates that pollinator shifts have led to morphological diversification and speciation within this genus (Whittall & Hodges, 2007). According to Whittall & Hodges (2007), a more ancestral floral state of Aquilegia is purple, downward facing, short-spurred flowers, which are pollinated by bumblebees. From plants with this floral state, taxa with red, downward facing flowers with protruding stamens and intermediate length spurs, which are pollinated by hummingbirds, were derived. Then, taxa with white and yellow long-spurred, lateral and upward facing flowers, which are pollinated by hawkmoths, were derived from those taxa. Their results reveal an interesting patten of species-level diversification as a consequence of pollinator shifts, although evidence for flower trait diversification at the earlier stages of speciation is lacking. To observe early stages of speciation, it is useful to investigate the pattern of evolutionary morphological diversification within a single species (Sobel & Streisfeld, 2015).
In this study, we focused on evolutionary processes leading to spur length and flower color differentiation in Aquilegia buergerianavar. buergeriana . In this species, geographic variation in spur length has previously been observed in six populations in two mountain regions, but the relationship between spur length and flower visitors in these populations is not known (Hattori et al., 2014). Yellow-flowered individuals are dominant in this species, and bumblebees seem to be the main flower visitors. In some populations, red-flowered individuals occur orthotopically with yellow-flowered individuals, but the genetic relationship between red- and yellow-flowered individuals is unknown. Differences in flower color in Aquilegia can lead to genetic isolation even between neighboring or sympatric populations and is likely to be important in speciation (Schemske & Bradshaw, 1999; Hopkins & Rausher, 2012). Here, we first investigated the correspondence between variation in floral spur length and flower-visiting insect size in 16 Aquilegia populations in four mountain regions. The results showed a morphological correlation between spur length and average visitor size in each population, even within the same mountain region; spur lengths were shorter in populations visited by smaller flower visitors, and spur lengths were longer in populations visited by larger flower visitors. Next, we identified genome-wide single-nucleotide polymorphisms (SNPs) by the MIG-seq (multiplexed inter-simple sequence repeat genotyping by sequencing) method (Suyama & Matsuki, 2015) to clarify the genetic relationships among the populations. These results showed that genetic relationships tended to be clustered by mountain region and, therefore, that spur length evolved in parallel in each mountain region. Individuals with different flower colors were not differentiated in the genome-wide SNPs analysis, however. This result suggests that pollinator isolation by flower color has not occurred in these populations. Instead, the red flower color is maintained in various populations in which most individuals have yellow flowers.