Species delimitation and taxonomy of rice
Correct species delimitation is a prerequisite for DNA barcoding. Although considerable efforts have been made on the taxonomy ofOryza , consensus has not been reached on the number of species in the genus and some controversies remain. So far, the phylogeny of all species is incomplete. Phylogeny based on the chloroplast genome (Fig. 1) indicates that species of the E (O. australiensis ) and F (O. brachyantha ) genome types are monospecific and relatively isolated from other species. Species pairs have been found between O. meyeriana and O. neocaledonica of theG genome type, between O. longiglumis and O. ridleyi of the HJ genome type, and between O. coarctataof the KL genome type and O. schlechteri of theHK genome type (Lu & Ge, 2003). Phylogeny based on the nuclear N78+R22 marker (Fig. 2) revealed that the L genome did not exist, while O. coarctata belonged to the HK rather than the KL genome type. Species belonging to the HJ andHK genome types share a common paternal progenitor with theH genome, a now-extinct species originating somewhere in Irian Jaya, Indonesia or Papua New Guinea.
Major identification problems exist among species of the A ,B , and C genome types. As with the H genome type, the D genome type is found only in South and Central American species, such as O. alta , O. latifolia , andO. grandiglumis , with CCDD genomes. Interestingly, theD genome type isolated from the sample BOP022669 was identified as O. latifolia and formed a clade with O. australiensisof the E genome type (Fig. 2). Phylogeny indicates that theD genome type is very likely a variant of the E genome type, if not E itself, confirming earlier results (Bao & Ge, 2004; Ge, Sang, Lu, & Hong, 1999).
There is a general correlation between molecular divergence and species delimitation (Lefébure, Douady, Gouy, & Gibert, 2006). Little chloroplast genome divergence was observed between O. alta andO. grandiglumis and their conspecific nature was suggested by (Bao & Ge, 2004) based on nuclear genes. Considering the trivial morphological difference between O. alta and O. grandiglumis , the former becomes often a synonym of the latter instead of O. latifolia Desv., as for example on “The Plant List” (http://www.theplantlist.org/tpl1.1/record/kew-426597).
Within the BC genome type, the two Asian species O. malampuzhaensis and O. minuta originated by hybridization between O. punctata as maternal parent and O. officinalisas paternal parent (Zou et al., 2015). In contrast, for the African species O. schweinfurthiana , O. eichingeri served as maternal parent and O. punctata as paternal parent. Considering insignificant morphological differences between O. malampuzhaensis and O. minuta , the former could be regarded as a synonym of the latter. Given that O. schweinfurthiana is an allotetraploid with a different maternal parent compared to O. minuta , it should be considered a distinct species instead of merging it within O. punctata .
Misidentification of plant material is very common within the Agenome type due to incorrect discrimination between species. All these species diverged within a short period by a radiation event (Wambugu, Brozynska, Furtado, Waters, & Henry, 2015; Zhang et al., 2014). Some species pairs exhibit neither obvious morphological difference nor remarkable genetic divergence. A first instance of confusion involves the African cultivated rice O. glaberrima and its wild progenitorO. barthii . No obvious genetic divergence has happened between their chloroplast genomes, which confirms similar results based on nuclear genes (Li et al., 2011; Wang et al., 2014). They often grow side by side in the field without ecological niche differentiation. Hence,O . barthii should be considered a synonym of O. glaberrima or a wild type.
A second confusing case involves the Asian cultivated rice O. sativa and its wild progenitors O. nivara and O. rufipogon . The Asian cultivated rice was divided into two subspecies, subsp. indica and subsp. japonica , in spite of naked names. Although the two subspecies are reproductively isolated, differ significantly in morphology and physiology, and were domesticated separately in the Himalayan mountain range and southern China (Londo, Chiang, Hung, Chiang, & Schaal, 2006), their taxonomic status has never been questioned. Our molecular phylogenies and almost all previous studies such as that by Wambugu et al. (2015) have confirmed that the two cultivated subspecies have the closest wild species of their own. It is very clear now that O. sativa subsp. indica is domesticated from O. nivara and that O. sativa subsp.japonica comes from O. rufipogon . Because the type ofO. sativa belongs to O. sativa subsp. japonica ,O. sativa must be retained in this cultivated subspecies with an autonomous name. Therefore, the two subspecies should be detached and renamed as O. sativa subsp. sativa (syn. O. sativasubsp. japonica ) and O. nivara subsp. indica (syn.O. sativa subsp. indica ). The names of their wild progenitors, O. nivara and O. rufipogon , have to be changed accordingly to O. sativa subsp. rufipogon (syn.O. rufipogon ) and O. nivara subsp. nivara (syn.O. nivara ). In 1970, a male sterile interspecific hybrid betweenO. nivara subsp. indica (= O. sativa subsp.indica ) and O. sativa subsp. sativa (= O. sativa subsp. japonica ) was discovered at a farm in Hainan province, China. The reproductive isolation between these subspecies was broken artificially and partially fertile F1 hybrid rice was used to produce fertile F2 hybrids as a new cultivar, which exhibited considerable hybrid vigor. Subsequent hybridization, however, created taxonomic problems regarding the correct identification of the two kinds of rice and their wild progenitors, resulting in many incorrectly labeled sequences being deposited in GenBank.
After synonymizing O. longistaminata under O. glumipatulaand including Porteresia coarctata (Roxb.) Tateoka intoOryza (= O. coarctata Roxb.), 21 species are now recognized in the Oryza genus (supporting text S1).