By merging diverse sequences into a supermatix data set, we obtained a well-resolved phylogeny with most branches strongly supported by BT values greater than 95% and PP values of 1.0. This indicates that increasing the number of taxa and the number of molecular markers improved the resultant phylogeny, and it further supports the notion that a supermatrix can be used to obtain a well-resolved and strongly supported phylogeny in cases where some data are missing [37–39]. The phylogenetic relationships of Egeria, Elodea, Ottelia, Blyxa, Apalanthe and Lagarosiphon have remained unchanged in all the earlier molecular phylogenetic studies [1, 2, 8, 9, 40, 41]. Our analyses resolved the generic relationships that are largely similar to those reported in those studies. However, incongruences existed for the other genera, which we briefly address here below.
The genus Hydrilla comprises only one species, H. verticillata. Based on rbcL, matK, trnK intron, ITS and morphological data, Les et al. (2006)  suggested that Hydrilla was most closely related to Najas, despite their being quite divergent at the phenotypic level. Our phylogenetic analyses suggested that Hydrilla is most closely related to the subclade comprising Nechamandra, Vallisneria and Maidenia (Figure 1). This position is consistent with all previous phylogenetic studies (except Les et al., 2006) based on molecular and morphological data [2, 8, 9, 40, 41]. A close relationship between the subclade (Hydrilla (Nechamandra (Vallisneria + Maidenia)) and Najas was strongly supported (BS = 96, PP = 1.0, Figure 1). This is in agreement with earlier results from rbcL+ matK+ trnK intron analyses with ML approach , rbcL , and rbcL+ matK . However, the results did not support the close affinity between the subclade and seagrasses which had been inferred from rbcL  and rbcL+ matK+ trnK intron + ITS analysis .
Stratiotes was resolved as the first diverging lineage within Hydrocharitaceae (BS = 100, PP = 1.0; Figure 1). This is in agreement with analyses based on rbcL , mtt2 and nad5  and the fossil records of Stratiotes which include the most abundant and the oldest fossils of' Hydrocharitaceae [19, 42]. However, the phylogenetic position of Stratiotes seems to be mainly derived from the mitochondrial sequences (cob, atp1) which are prone to flaws in plant phylogenetic analysis . Therefore, further studies are required to confirm the position of the genus obtained in this study.
Results of divergence time estimates are in agreement with the fossil records of Hydrocharitaceae. The 95% HPD of Najas was 11.9-34.3 Ma, consistent with the oldest fossil of this genus in the Oligocene  (Figure 1). The stem node age of Hydrocharis-Limnobium was dated around 54.7 Ma. However, the crown node age of this subclade was dated around 15.9 Ma, younger than the oldest fossil of Hydrocharis from the Upper Eocene [17, 44, 45]. This could be interpreted as an indication that Limnobium had split from the relatively ancient Hydrocharis in the Miocene (Figure 1), and the great morphological similarity between the two genera is probably due to the short evolutionary history of Limnobium. Although the present study has yielded improved divergence time estimates, it is possible that the estimates of the time of origin for some genera such as Ottelia, Vallisneria, Najas and Blyxa may have been affected by under-representation in sampling.
The age of Hydrocharitaceae estimated in this study (mean: 65.5 Ma, 95% HPD: 54.6-79.6 Ma) is in agreement with that based on rbcL analysis and external fossil calibration points (crown node age = 75 Ma) . However, the stem node age of seagrasses estimated in this study (15.9-41.3 Ma) (Figure 1), is more recent than the 119 ± 11 Ma suggested from analysis using the substitution rates of rbcL and matK . Similarly, our estimates of the stem node age of Ottelia (8.1-33.3 Ma) is more recent than the Cretaceous origin suggested by He et al. (1991) . The split between Zosteraceae and Potamogetonaceae has been dated at 47 Ma by rbcL and fossil calibration . The time is also more recent than the 100 Ma inferred from analyses using the substitution rates of rbcL and matK . These discrepancies indicate that for estimating divergence times in aquatic plants, incorporating fossil calibration point would be more reliable.
The Oriental origin of Hydrocharitaceae inferred from our analysis is supported by the known existence of regions with humid and warm conditions in southeastern Asia during the late Cretaceous and Palaeocene [46, 47] and the fact that the genetic diversity centre of this family is in tropical Asia . The ancestor of clade A was inferred to have originated and diversified in the Orient, while that of clade B dispersed from the Orient to the Southern Hemisphere during the Late Cretaceous and Paleocene (Figure 2c, d, arrow 2). Different environments and oceanic barriers among the major continents (vicariance mechanism) during the Tertiary probably contributed to the diversification of this family resulting in taxa such as the African endemic Lagarosiphon.
Most fossils of Hydrocharitaceae and its close relatives Butomaceae and Alismataceae have been found in Europe (Butomaceae in the Neogene of south Aral region, Miocene of northwest and east Caucasus [18, 48]; Alismataceae in the Tertiary of Europe, a few in North America). The fossil records seem to be inconsistent with the Oriental origin of this family. However, the absence of reports of fossils from Asia most likely reflects a bias in paleobotany, rather than an indication of the origin and past distribution of Hydrocharitaceae. A similar situation exists in Rhinolophus (Rhinolophidae), for which, although the genus is thought to have originated in Asia, fossils have only been reported from Europe and Africa but not from Asia [49, 50].
Does the origin fit with dispersal?
The modern continents viz. South America, Africa, Eurasia, Australia and North America have been separated by oceans since at least ca. 90 Ma [46, 51, 52], earlier than the origin of Hydrocharitaceae. Therefore, dispersal must have played a dominant role in the transoceanic distribution of this family. This contradicts the view that the transoceanic distribution of Ottelia mainly resulted from vicariance . The role of dispersal in transoceanic distribution has been supported by evidences from the studies of geological events and land plant families. Ocean currents are a viable means of dispersal of plants , and a tropical westward-flowing ocean current had spanned the world from the Cretaceous to Paleocene [54, 55]. Island chains existed in the Tethys from Cretaceous to Eocene, which served as a stepping-stone in biotic dispersal between S.E. Asia, Africa and southern Europe [32, 56, 57]. The Malay Archipelago probably facilitated biotic dispersal between S.E. Asia and Australia during the Miocene . The North Atlantic Land Bridge (NALB) aided plant migration between North America and Europe during the late Cretaceous and early Tertiary [59–61]. The Bering Land Bridge (BLB) was open from at least the early Paleocene until its closure ca. 7.4-4.8 Ma . Several recent studies of angiosperms based on molecular and fossil data have supported dispersal as the dominant factor responsible for transoceanic distribution, e.g., in Cucurbitaceae , Sapindaceae , Chrysophylloideae (Sapotaceae) , Burseraceae  and Malphigiaceae . It is probable that Hydrocharitaceae have dispersed to all continents of the world via island chains, land bridges and ocean currents.
Biogeographic studies have suggested that the sub-cosmopolitan distribution of the aquatic plant family Alismataceae has mainly resulted from dispersals (the work will be reported in a separate paper). It is probable that dispersal is the dominant factor, accounting for transoceanic distribution of aquatic angiosperms. However, more studies on aquatic angiosperms are required to investigate this idea further.
Historical biogeography of some genera of Hydrocharitaceae
The ancestor of Stratiotes was suggested to have dispersed from Orient into Europe during the late Cretaceous and Palaeocene (Figure 2b & 2c, 2d, arrow 1), which coincided with the existence of the Tethys seaway (TESW) . Alternatively, the ancestor may have migrated from Orient to Europe across Eurasia. Abundant fossils (15 fossil species) of this genus in Europe  suggested that the genus had diversified widely in this region adapting to wet swamps in the Late Cretaceous .
The genus Hydrilla is native to Eurasia and Australia , and introduced to Americas  and parts of Africa . The centre of differentiation of the genus was thought by Cook and Luond (1982) to lie in tropical Asia . This idea got support from genetic diversity analysis which revealed that the highest diversity is located in China and with lower albeit similar genetic types occurring in Africa, India and USA . Hydrilla might have arisen in the Orient dispersing to Europe and Australia (Figure 2, e, arrow 1 & 5).
The MRCA of the seagrasses within Hydrocharitaceae were suggested to have lived in Oriental area during the Oligocene and Miocene. The result is in agreement with the view that seagrasses possibly originated in the S.E. Asia [72, 73]. The result is supported by the environments of S.E. Asia which was characterized by abundant islands, spacious shallow-seas, warm temperature and plenty of isolated seas . However, the result denied the Cretaceous origin of the group which has been suggested in previous studies [13, 75, 76]. The seagrasses were suggested to have been dispersed from Oriental to other regions (Figure 2c), probably by ocean currents [73, 77]. For example, the warm northward Kuroshio Current carried seagrasses from the equatorial region to the Nansei Islands . Seagrasses are capable of surviving during the LDD between major ocean systems .
Vallisneria has a world-wide distribution, with the highest number of species in Australia [79, 80]. Les et al. (2008)  resolved the phylogeny of this genus, but they conceded that the geographical origin is difficult to pinpoint. In this study by DIVA analysis, Oriental and Australasian areas were suggested as the co-existed ancestral areas of Vallisneria. However, Oriental area is more likely the centre of origin considering the following facts: the closest relative of Vallisneria namely Nechamandra is confined to Asia ; the ancestral species in Vallisneria namely Vallisneria spinulosa, V. spiralis and V. denseserrulata are confined to the Old World [67, 80].
Evolution of morphological characters
Ancestral state reconstruction of reproductive system in Hydrocharitaceae provides empirical evidence that evolution of dioecy in plants has been a bidirectional, viz. from dioecy to hermaphroditism, and from hermaphroditism to dioecy (Figure 3a). This view is supported by Delph (2009)  and Canovas et al. (2011) , but rejects the view that hermaphroditism is the ancestral state in Hydrocharitaceae .
The evolution of leaf habit and leaf shape in Hydrocharitaceae provides several cases of evolutionary adaptation to diverse habitats. The evolution from aerial-submerged leaf to submerged leaf is probably due to change in habitat from shallow to deep waters . The reverse evolution from submerged leaf to aerial-submerged leaf in Ottelia is probably an adaptation to change in habitat from deep to shallow water or some other disadvantageous habitat(s). Taxa with broad-circular leaves (e.g., Ottelia and Hydrocharis) usually occur in still water, while those with ribbon like leaves such as Enhalus and Thalassia occur in coastal waters with strong waves .