The remarkable congruence of New and Old World savanna origins.

Abstract

‘Despite its high biodiversity and importance to humanity, the scientific study of savanna and – perhaps more alarmingly – its conservation have been neglected relative to its better-known cousin, the tropical rain forest.’ The method used by Maurin et al. is to estimate the timing of the appearance of fire-adapted lineages, specifically the geoxylic suffrutex growth form – strange plants with massive underground woody axes or lignotubers that are hidden from surface fires, and with aerial shoots that are short in height and duration (Figs 1, 2). Frank White, one of the most influential 20th Century workers on African vegetation and plant taxonomy, called such plants, very appropriately, ‘underground trees’ (White, 1976). White's (1976: 59) observation that, ‘suffrutices not only are closely related to large trees and have presumably evolved from large trees or lianes, but occur in genera which except for their suffruticose members consist exclusively of large woody plants' was highly prescient in foreseeing the results found by Maurin et al. using modern phylogenetics, nearly 40 yr later. White was inferring that these suffrutices were trees that had gone underground in an evolutionary sense, which is confirmed by Maurin et al. Previously, savanna woody plant evolution has only been considered in South America (Simon et al., 2009; Simon & Pennington, 2012), where phylogenetic studies revealed recent evolution of the geoxylic suffrutex growth form and of other woody species with adaptations to fires, such as thick corky bark, from lineages in surrounding biomes (Simon et al., 2009). These recent origins in the past 4 million yr (Myr) are consistent with the fossil record of C4 grasses that suggests a savanna expansion driven by the invasion of flammable grasses (Edwards et al., 2010). Maurin et al. use a comprehensive dataset, sampling 1400 woody species including more than a quarter of the estimated 200 geoxylic suffrutices in the Zambesian region of southern Africa. They demonstrate multiple independent origins of the geoxylic growth form, mostly starting in the Pliocene (5–2.5 Myr ago), with the majority of divergences occurring within the past 2 Myr. This abrupt, rapid and phylogenetically scattered evolution of plants with fire adaptations is consistent with a savanna biome origin and expansion in African that is startlingly congruent in time with that in the Americas, adding important new evidence supporting the late Miocene/Pliocene origin of the savanna biome worldwide (Beerling & Osborne, 2006; Edwards et al., 2010). Such transcontinental synchronicity of the frequent evolution of the geoxyle habit in the Pliocene strongly suggests a common global explanation. In this regard, a complex set of shared climate–fire–vegetation feedback mechanisms seems especially relevant. Several studies have suggested a key role for fire, whereby increased climatic seasonality contributed to enhanced and intensified fire activity, which triggered the global expansion of C4 grasslands and the establishment of the world's savannas (Beerling & Osborne, 2006; Edwards et al., 2010; Scheiter et al., 2012; Hoetzel et al., 2013), as well as widespread evolution of the geoxyle and other fire-adapted growth forms that are one of the hallmarks of the tropical savanna biome today. The Maurin et al. study supports conclusions that the evolution of adaptations to fire in woody plants may be a relatively simple process in developmental (Simon et al., 2009) and genetic (Simon & Pennington, 2012) terms. This ease of adaptation to fire is what underlies the numerous evolutionary transitions into the savanna biome documented in Africa (Maurin et al.) and South America (Simon et al., 2009), and it may reflect that fire adaptations such as thick, corky bark and the ability to root sprout may only require changes in gene regulation rather than structural mutation (Simon & Pennington, 2012). What seems to be especially striking here is the ease with which plants can reallocate their biomass underground, something that chimes with the apparent evolutionary lability of plant growth forms more generally, such as shifts between annual and perennial habit, or the evolution of climbing habit. In the case of the underground trees, Maurin et al. suggest that they are an example of heterochrony – a change in timing of development – as the trees flower when they have a dwarf stature; however, this does not explain the key change of the massive transfer of woody biomass underground and there appear to be rather few data to show conclusively that geoxyles flower at a younger age than their arborescent congeneric counterparts. Savannas appear to be an example where adaptive shifts from other biomes have played a key role in the generation of high tropical species diversity. The savanna biome boundary appears to have been permeable over evolutionary timescales to the ingress of woody lineages, or their in situ adaptation across a shifting biome boundary, because fire does not pose a significant adaptive barrier (Simon & Pennington, 2012). In both the Neotropical cerrado and African savanna, there is no need to invoke dispersal of pre-adapted lineages from other geographically distant fire-prone environments, but rather in situ adaptation of locally available lineages to fire. In the Neotropics, this contrasts with phylogenetic patterns seen in dry tropical forests – closed canopy forests that grow in similar seasonal climates to savannas, but on richer, often calcareous, rocky soils which may retain less water (Oliveira-Filho et al., 2013). The available, though limited, phylogenetic evidence suggests that these dry forests are characterised by lineages that are often entirely confined to this biome across large geographical disjunctions (Schrire et al., 2005), suggesting less frequent shifts to the dry forest biome – a pattern often characterised as phylogenetic niche (Donoghue, 2008) or biome (Crisp et al., 2009) conservatism. Perhaps counterintuitively, it seems that fire may have been less of an adaptive barrier to woody plants in the tropics than switching between major soil types. Future research might focus on the nature of the morphological, physiological (Edwards & Donoghue, 2013; Donoghue & Edwards, 2014) and genetic (Simon & Pennington, 2012) transitions required to make an evolutionary switch from one major biome and to survive in another (Edwards & Donoghue, 2013; Donoghue & Edwards, 2014). While the congruent recency of origin and phylogenetic lability of geoxyle evolution in the New and Old Worlds revealed by this study are striking, much remains to be explained, such as why so many species of plants have accumulated in the main area of South American savanna, the cerrado of central Brazil and Bolivia, which occupies a much smaller total area than the tropical savannas of Africa. The cerrado region has as many angiosperm species (c. 11 000) as the Brazilian Amazon (Forzza et al., 2010), and the Sudanian and Zambesian savannas have a similar combined total (White, 1983). The species richness of the cerrado appears to be a result of many examples of in situ radiation of up to 50 or more species per clade (Simon et al., 2009), which display various fire adaptations, including the geoxyle growth form. We suspect that such in situ radiations are less frequent in African savannas, and future work clearly needs to prioritise densely sampled, species-level phylogenies across numerous clades if we are to fully understand the evolution of savanna plants (Hughes et al., 2013; Donoghue & Edwards, 2014). A strength of the Maurin et al. study is its taxonomic breadth, covering the entire woody flora of African savannas, but its weakness is a lack of dense species sampling of related species in other biomes. This lack of species sampling precludes identifying in many cases the exact phylogenetic point, and timing, of shifts between biomes. In comparison, the study of Simon et al. (2009) of South American savannas included dense species sampling, but focused on only four clades, three of which were legumes, so its generality was limited. Combining both approaches – phylogenetic breadth and dense sampling of species – for clades including savanna species in the Americas, Africa, Asia and Australia, may hold the key to further understanding the origin and expansion of the global tropical savanna biome. We thank Michael Donoghue and Erika Edwards for sharing unpublished manuscripts, Peter Linder for information about African savannas, and Rosemary Wise for permission to reproduce the illustration in Fig. 1.