Cryptic species
Phenotypic and genetic differences, as well as diversification patterns, were used to assess classifications as distinct species. First, the leaf phenotypes of species P (C. parvifolia ) and G (C. glabra ), as well as ecotype SD (C. subpubescens ), exhibited distinct and distinguishable distributions in the PCA of leaf morphology (Fig. 5). However, the ecotypes of C. subpubescens , excluding ecotype SD, displayed PCA plots with less clarity in terms of separation.
Callicarpa parvifolia , endemic to the Chichijima Islands grows in very dry and sunny rocky areas and is notably distinguishable from otherCallicarpa species and ecotypes owing to its small, rounded, remarkably hairy, and thick leaves (Fig. S11). Similarly, C. glabra , also endemic to the Chichijima Islands grows in the dry scrub understory and is characterized by its small, slender, hairless, and rather thick leaves. Phylogenetically, these two species form a sister clade, each maintaining monophyly (Fig. 3b). Accordingly, the current taxonomy of the two species appears reasonable.
Ecotype SD, found only in the Hahajima Islands, forms a canopy in dry scrubs, with leaves that are small, round, hairy, and thick, similar toC. parvifolia (Fig. S11). However, it can be distinguished fromC. parvifolia by significantly lower leaf hair density and thickness. Phylogenetically, ecotype SD and C. parvifolia were located different clades (Fig. 3b). A parallel evolutionary process of adaptation to dry environments likely resulted in a morphology closely resembling that of C. parvifolia . Additionally, compared with ecotype S from the Chichijima Islands, treated as C. subpubescens , ecotype SD differs both in its leaf morphology and flowering phenology, with prolonged flowering from summer to winter (Table 1), warranting recognition as a new species.
Although ecotype ST in the Hahajima Islands could not be clearly distinguished from other ecotypes of C. subpubescens except ecotype SD in leaf morphology (Fig. 5), it can be distinguished from other ecotypes using flowering phenology (the flowering season of ecotype ST is autumn, whereas those of ecotypes S, SG, and SH are summer; Table 1). Phylogenetically, we consider ecotype ST to be a cryptic species, as it diverged from the ancestral species at an age more than twofold greater than that of other genetic groups based on population demography analysis (Fig. 4).
Both ecotype S in the Chichijima Islands and ecotype SG in the Hahajima Islands flower in summer (Table 1). Ecotype S is identical to the registered type specimen of C. subpubescens (specimen no. K000674714), exhibiting fine, soft hairs on its leaves, whereas ecotype SG has almost no hairs on its leaves. However, the number of hairs did not differ compared with ecotype SG (Fig. S11), and the plot distributions of ecotypes S and SG overlapped in PCA (Fig. 5). Despite belonging to different genetic groups, no obvious phenotypic differences were observed. Several examples exist in the Bonin Islands of the same species differentiated genetically due to gene flow restriction caused by the different island groups they inhabit (Setsuko et al., 2017; Setsuko et al., 2022; Setsuko et al., 2020; Sugai et al., 2013). Therefore, considering ecotype SG as homologous to ecotype S is reasonable.
Ecotype SH, thought to be derived from a hybrid of ecotypes ST and SG, is found only on Hahajima Island (Figs. 1 and 2a). Ecotypes SH and ST possessed similar leaves, with ecotype SH occupying part of the broader distribution of ecotype ST in the PCA plot (Fig. 5). However, the plot distributions of ecotypes SH and SG in PCA did not overlap. As the flowering season of ecotypes SG and SH is summer and that of ecotype ST is autumn (Table 1), ecotypes SH and ST are distinguished by their flowering periods, whereas ecotypes SH and SG are roughly distinguished by their leaf morphologies. In naturally distributed individuals, habitat information can aid taxonomic classification of ecotypes SG and SH, as ecotype SH inhabits high-elevation cloud forests and forms the forest canopy layer, whereas ecotype SG inhabits the understory of mesic forests, except in the high-elevation areas of the Hahajima Island (Setsuko et al., 2023). Based on these findings, we propose thatC. subpubescens can be divided into three species in addition to one hybrid-derived taxon, rather than one species.
Concerning ecotypes in the Mukojima Islands, the flowering of STm and Sm was investigated only once in July, with these ecotypes found to be in the early and late stages of flowering, respectively. July marked the beginning of flowering for ecotype ST in the Hahajima Islands, and the end of flowering for ecotype S in the Chichijima Islands (Table 1). Combining flowering information with the phylogenetic tree results from this study, Sm was considered the same ecotype as S, whereas STm was considered the same ecotype as ST. Determining whether ecotypes STm and Sm in the Mukojima Islands can be considered identical to ecotypes ST and S, respectively, necessitates a more thorough investigation into the flowering periods of ecotypes STm and Sm. Furthermore, although only a minimal number of hybrid individuals (two) were found in the Mukojima Islands (Figs. 1 and 2), the extent to which they form hybrid zones, akin to ecotype SH on Hahajima Island (Setsuko et al., 2023), remains unclear. However, given the coexistence of lineages from different origins on the same island, there is a possibility that this codistribution contributes to hybrid speciation (Kagawa & Takimoto, 2018). Further research should include more comprehensive investigations into hybridization in the Mukojima Islands.