Upstream colonisation from the Indo-Australian Archipelago
While there is strong evidence that upstream colonisation from islands to continents does occur, it has until recently been considered an anomalous trajectory (Bellemain & Ricklefs, 2008). The islands of the IAA appear to be the source area for at least one global radiation of birds (Jønsson et al. , 2010). Recent macroevolutionary analyses of Columbiformes have also emphasised the importance of islands in their early evolutionary history, but did not focus on spatial or temporal patterns of dispersal. Conversely, other data indicate that the biota of continental Asia serves as a strong filter on upstream colonisation by either blocking successful colonisation completely (Oliver et al. , 2018b; White et al. , 2021) or by strongly filtering taxa by ecology (Letsch et al. , 2020).
Limitations of our phylogeny (70% sampling) and poor resolution at basal nodes) and historical reconstruction methods mean that we are able to give an accurate summary of trends, but not a detailed reconstructions of all events. Nonetheless, in support of Lapiedra (et al. 2021) we found that in the extant radiation of Columbiformes upstream colonisations from the IAA dominated downstream colonisations through much of the Oligo-Miocene. The importance of islands in the IAA is particularly striking given their comparatively small areal extent when compared to nearby continents (Fig. 2). The regional geography and ecological context of the earliest inferred upstream shifts into the Old World are relatively unclear, with the alternative hypotheses being one or two shifts from islands into the broader “Old World” in a weakly supported clade spanning Treron -Turtur . The Oligo-Miocene timing and direction of these colonisations is potentially comparable with upstream colonisation of the Oscine passerines (Jønsson et al. , 2010; Moyle et al. , 2016), although the outcomes in terms of diversification differ starkly (less than 40 species versus over 5000). A more strongly resolved phylogenetic tree, and ideally fossils, would be needed to shed more light on the trajectory of potential early upstream shifts in these lineages.
The geographical, temporal and ecological context of more recent inferred upstream colonisations in Columbiformes is clearer. The specialist fruit-eating arboreal pigeons (fruit doves) of the Ptilinopini have colonised nearby continents on at least ten separate occasions (and probably more if missing taxa such as Ptilinopus alligator and intraspecific populations are added to the phylogeny). Ecological niche shifts to arboreality in the ancestor of the Ptilinopini underpinned extensive radiation in the IAA and Pacific (~100 species), and potentially also predisposed them to repeatedly colonise nearby continents (Lapiedra et al. , 2021). In contrast, speciation within continental regions appears to have been very limited. Indeed, around half the inferred upstream shifts involve species level taxa that occur across islands and continents. Furthermore most Ptilinopini in continental areas (and especially Asia) remain closely associated with coastal habitats and islands (e.g. Ducula aenea , Ptilinopus jambu, P. melanospilus ), or other habitats with often less biodiverse communities such as montane areas (e.g.Ptilinopus porphyreus , Ducula badia ) (Baptista et al. , 1997). This conforms with a hypothesis that upstream colonisation is challenging (White et al. , 2021), and even when frequent, ecological filtering may limit new arrivals to comparatively species poor environments on ‘mainlands’ (Mayr & Diamond, 2001).
Upstream colonisation by terrestrial-feeding Raphinini and Phabines has been even more limited, again suggesting ecological filtering at continental margins. Raphinini are entirely restricted to islands, despite their evident history of dispersing across vast distances into the Indian and Pacific oceans. The one striking outlier is provided by the Phabines, which are inferred to have colonised Australia from islands in the early Miocene and subsequently radiated across the continent. The relative success of this upstream colonisation could have been mediated by major climatic changes (aridification) and the concomitant opening up of niches and expansion of key food sources for terrestrial taxa, especially grasses (Toon et al. , 2015). Like the Raphinini, the Phabines have been largely unable to colonise continental Asia. The only exception is in the genus Geopelia , which originated in Australia, but has likely colonised the expanding savannahs of the Sundaland only recently during the Plio-Pleistocene. Thus, our data support the hypothesis that islands have been a net source area for continents (Lapiedra et al. , 2021), while simultaneously indicating that continents (and especially the lowland rainforests of Asia) present a formidable ecological filter to upstream colonisation (White et al. , 2021).