Charles Darwin’s “abominable mystery”, as reported in the correspondence with Joseph Hooker, Gaston de Saporta and Oswald Heer between 1875 and 1881 (Darwin and Seward 1903), has ended up representing all the unresolved problems regarding the origin and early evolution of angiosperms. Actually, Darwin was not directly interested in the angiosperm fossils, identification of the ancestor of flowering plants, character homology or the evolutionary history of flowering plants, all subjects that since that time have worried generations of botanists (Friedman 2009). Instead, Darwin was plagued by the sudden origin and fast radiation of angiosperms all around the world in the mid-Cretaceous, an event that was in contrast with his belief of a gradual, slow evolution. To give an explanation to this problem, Darwin speculated about an early, pre-Cretaceous birth and differentiation of flowering plants in a 'lost continent', which was submerged after angiosperm radiation all around the world: “plants of this great division must have been largely developed in some isolated area, whence owing to geographical changes, they at last succeeded in escaping, and spread quickly over the world”. (letter to O. Heer, 8 March 1875; see Darwin and Seward 1903: 204) or, again, “I have sometimes fancied that development might have slowly gone for an immense period in some isolated continent or large island, perhaps near the South Pole.” (letter to J.D. Hooker, 12 August 1881; see Darwin and Seward 1903: 26).
Even if more fossil and molecular data are now available, as yet, it is not possible to determine the site of origin of the angiosperms (Briggs and Walters 2016). In the past, contrasting views have been developed in this regard, placing the site of origin in the Arctic region (Seward 1927), in alpine biomes of northern latitudes (Stebbins 1974), in SE-Asia (Takhtajan 1969), in E-Asia (Sun et al. 1998; Sun et al. 2001), at high boreal latitudes (Hochuli and Feist-Burkhardt 2004), or in unrelated scattered areas from multiple seed plant lineages (see the polyphyletic-polychronic-polytopic hypothesis by Wu et al. 2002). Overall, there is no consensus on the location(s) of angiosperm origin.
However, if we consider the geographic distribution of the 26 early branching angiosperm families (i.e. ‘Archaeangiospermae’ sensu Stuessy 2010): ANA grade, Magnoliids and Chloranthaceae (see also APG IV 2016), all of them show a distribution including, at least for tiny areas, Gondwanan territories. Overall, 21 early branching families (81%) are found in SW Pacific Gondwanan areas, i.e. New Guinea, E Australia, Fiji, New Zealand, or New Caledonia, with Amborellaceae, Austrobaileyaceae, and Degeneriaceae endemic to these areas (Fig. 1). Moreover, the worldwide maximum concentration of archaeangiosperm families (about 60%) falls in small portions of the NE Australian coast and Papua Nuova Guinea (Fig. 1).
Although current ranges of these families may significantly differ from their past distributions, taken together these data suggest a possible SE Gondwanan origin of the angiosperms. A more general Gondwanan origin was previously hypothesised also by Cronquist (1988), Retallack and Dilcher (1981) and MacDonald (2003). The presence of early divergent lineages (Amborellaceae, Atherospermataceae, Chloranthaceae, and Winteraceae) in New Caledonia, an island re-emerged from Zealandia only ca. 37 million years ago (Ma) after a period of inundation (Cluzuel and Chiron 1998), is not in contrast with this view. There is increasing evidence that the present New Caledonian biota is the product of late Cenozoic immigration events by means of long-distance dispersal from nearby territories (e.g., E Australia, New Zealand, New Guinea). Also Amborella trichopoda, the only extant representative of Amborellaceae, became extinct elsewhere after New Caledonia colonization (Pillon 2012; Grandcolas et al. 2014; Nattier et al. 2017).
Luyendyk (1995) first proposed the name Zealandia for a collection of continental fragments between the Australian and Pacific plates (Smith and Sandwell 1997; Stagpoole 2002). In the last two decades, the progressive accumulation of bathymetric, geological, and geophysical data, has led Mortimer et al. (2017) to define Zealandia not as a collection of partly submerged continental fragments, but a coherent 4.9 M km2 continent (Fig. 1), that was formerly part of Gondwana. After its separation from Gondwana (85–130 Ma from Antarctica, 60–85 Ma from Australia; Lewis et al. 2006), Zealandia was 94 % submerged during the late Cretaceous (60–85 Ma; Goldberg et al. 2008).
There is a striking coincidence between the geological evidence of this continent that disappeared in the late Cretaceous and the distribution, around or inside the limits of Zealandia, of many early branching angiosperm families (Fig. 1). The age estimates of the divergence of crown group angiosperms using molecular clock data vary considerably, between 140 and 240 Ma or earlier (Martin et al. 1989a,b; Soltis and Soltis 2004; Moore et al. 2007; Bell et al. 2010; Silvestro et al. 2014; Foster et al. 2017). The reconstruction of the genome of the most recent common angiosperm ancestor suggests an age of 214 Ma for its appearance (Murat et al. 2017). By contrast, reliable angiosperm fossil records older than 130–140 Ma have been not found so far (Doyle 2012; Foster et al. 2017, Herendeen et al. 2017). Even though the lack of fossil records before the early Cretaceous suggested to several scholars that angiosperms must have undergone extremely fast evolution early in their history, Cascales-Miñana et al. (2016) argued that for 100 million years of their evolutionary history angiosperms remained relatively rare, maybe growing in habitats poorly represented in the fossil records. Zealandia was quite a large portion of Gondwana, mostly submerged during the late-Cretaceous (60–85 Ma; Goldberg et al. 2008). This area could have been the cradle of the ancestor and of the early evolution of the main angiosperm lineages during Triassic-Jurassic periods.
Putting the pieces together, our hypothesis is that the flowering plant ancestor could have appeared in the ‘lost continent’ Zealandia some 200 million years ago or earlier, perhaps from Paleozoic-Mesozoic seed ferns (Taylor and Taylor 2009) after a whole genome duplication event (Amborella Genome Project 2013) and/or rapid genome downsizing (Simonin and Roddy 2018), slowly differentiating the main angiosperm lineages (Archaeangiosperms, Monocots and Eudicots) for millions of years. Then, these already established lineages could have spread outside Zealandia, starting from the early Cretaceous, in a rapid worldwide radiation, before its submersion in the late Cretaceous.
In a recent work (Sauquet et al. 2017), the ancestral angiosperm flower was reconstructed. According to these authors, this ancestral flower was bisexual, showing undifferentiated, whorled tepals and stamens with three elements per whorl. This combination of features is not observed in any extant angiosperm, including the earliest-divergent lineages (e.g., ANA grade). If the ancestral angiosperm flower model and our hypothesis are right, we expect that palaeobotanical investigations in the submerged Zealandia region may allow to find (micro)-fossil records showing these putatively ancestral features. Possibly, Darwin’s intriguing speculation on the appearance and early evolution of angiosperms in a 'lost continent' was correct.