The central and southern Andes have a high diversity of plant species (Pérez-Escobar et al., 2022). The history of this diverse flora is linked to the orogeny of the Andes and the development of the South American Arid Diagonal (Simpson, 1983; Antonelli et al., 2009; Luebert and Weigend, 2014), a strip of arid climate that extends along the Andes from Peru to Argentinian Patagonia (Villagrán and Hinojosa, 1997; Abraham et al., 2020). Aridification occurred gradually during the Oligocene and Miocene as a consequence of the strong rain shadow produced by the uplift of the Andes (Dunai et al., 2005). The establishment of a phase of global climate cooling since the middle Miocene (Zachos et al., 2001) and the reinforcement of the cold Humboldt Current in the Pliocene intensified aridification (Hinojosa and Villagrán, 1997; Hartley et al., 2005; Garreaud et al., 2010).

The South American Arid Diagonal forms a biogeographic corridor that has allowed the expansion and diversification of plant genera of different biogeographic origins (Simpson, 1983; Antonelli et al., 2009; Luebert and Weigend, 2014; Glade-Vargas et al., 2021). In the high Andes, plant lineages that have migrated from high latitudes tracking cold-humid climates coexist with lineages that have shifted repeatedly between warmer and drier habitats at low elevations and colder and wetter habitats at high elevations (Antonelli, 2015). Transitions between habitats may be accompanied by shifts in the strategy of plants that allow surviving periods of extreme cold or drought (Ogburn and Edwards, 2015; Pérez et al., 2020); or alternatively, might be favored by the preexistence of “enabler” traits that allow species to cope with both conditions (Harvey and Pagel, 1991; Edwards and Donoghue, 2013; Zanne et al., 2014).

Plants have evolved different strategies that permitted them to survive unfavorable seasons (Grime, 1977; Stearns, 2000). The annual life strategy is part of a fast development strategy that allows plants to complete the life cycle within the favorable season of a single year (Evans et al., 2005; Kooyers, 2015; Friedman and Rubin, 2015; Pérez et al., 2020). Annuals are favored in hot and dry seasonal climates (Friedman, 2020; Boyko et al., 2023), but tend to be excluded from high elevation habitats, where short growing seasons and repeated frosts inhibit seedling establishment and reduce the probability of completing a life cycle in one season (Bliss, 1971; Korner, 2003; Givnish, 2015). Perennial plants also have adaptations that allow survival of unfavorable seasons. Perennial herbs can shed aboveground tissues during unfavorable seasons, leaving belowground buds in rhizomes (woody underground stems) or specialized reserve organs (such as bulbs, tubers or corms) that enable resprouting (Raunkiaer, 1934; Bliss, 1971) and allow plants to survive in cold and dry conditions (Howard et al., 2019). Woody plants can shed leaves and protect buds with scale-like modified leaves or become smaller, leaving buds close to the ground (Bliss, 1971; Korner, 2003; Zanne et al., 2014). According to these strategies, plants can be classified in those that persist as seeds or underground storage organs and those that have perennial buds at, near or above the soil surface (Raunkiaer, 1934; Taylor et al., 2023).

Accelerated rates of diversification can be triggered by shifts in traits related with plant life history strategies, including life span (Givnish, 2010; Friedman, 2020), woodiness (Nürk et al., 2019), plant size (Boucher et al., 2017) and reproductive strategy (de Vos et al., 2014). Annual and herbaceous clades often have more species than their sister woody clades (Givnish, 2010), indicating that the shorter generation time of herbs can accelerate speciation (Smith and Donoghue, 2008). Likewise, small plants have higher diversification rates than do larger plants (Boucher et al., 2017). Shifts from annual to perennial life history strategy have been proposed as key innovations that trigger accelerated species diversification in high elevation habitats (Drummond et al., 2012; Hughes and Atchison, 2015; Nevado et al., 2016; but see Givnish, 2015). Evolution of secondary woodiness from herbaceous ancestors also appears to act as a key innovation driving plant radiations on island systems (Nürk et al., 2019).

The Andean uplift and aridification drove the rapid diversification of several plant lineages, including some of the most species-rich genera of Fabaceae (Kenicer et al., 2005; Hughes and Eastwood, 2006; Scherson et al., 2008). In southern South America, the richest genus of Fabaceae is Adesmia DC. (Dalbergieae, Faboideae), which comprises around 230 species of annuals, perennial herbs, small cushions and shrubs ( Figure 1 ) grouped in 43 series and two subgenera: Acanthadesmia (that includes all spiny shrubs) and Adesmia (that includes all unarmed forms) (Burkart, 1967). Species grow mainly along the Arid Diagonal from northern Peru to Patagonia, with some species extending to Southern Brazil. Adesmia has two centers of diversification, one in Chile and the other in Argentina (Burkart, 1967). Both areas have been proposed as centers of origin, with the Andes acting as a vicariant barrier (Mihoc, 2012) or as a biogeographic corridor from south to north (Burkart, 1967). A phylogenetic study has been conducted for Adesmia sect. psoraleoides, a monophyletic series endemic to southern Brazil (Iganci et al., 2013). This study, which included some species from other regions, found a north to south pattern of diversification, suggesting an expansion from arid regions of Bolivia and Chile to Patagonia, followed by colonization of southern Brazil.

In this study, we established phylogenetic relationships using four nuclear DNA markers of 80 Adesmia species that occur in Chile and Argentina from latitude 18° to 52°S and from sea level to 4500 m elevation ( Figure 2 ). We reconstructed ancestral states for area, climatic niche and growth form, to understand how the Andean uplift and the onset of arid and semiarid conditions promoted diversification of Adesmia. We also included the presence/absence of spines to explore whether the two subgenera (Acanthadesmia and Adesmia) are monophyletic. In addition, we examined whether shifts in traits related to plant life history strategy (life span, woodiness and plant height) evolved in relation to climate. Finally, we determined whether the high species diversity of Chilean Adesmia is due to uniform speciation, or to events of accelerated speciation triggered by shifts in growth form.

Our findings suggest that the most likely growth form of the ancestor of Chilean Adesmia was a perennial herb, which is also the predominant growth form in its sister genera Chaetocalyx, Zornia, and Amicia. However, incorporating phylogenetic uncertainty makes this conclusion less certain. In 82% of the 1,000 sampled trees, the most probable ancestor of Adesmia was identified as a perennial herb. On average, the likelihood that the ancestor was a perennial herb (53%) was twice as high as the probability that it was an annual (26%) or a shrub (21%). Our results also suggest that the ancestor of Chilean Adesmia likely originated in the high Andes of northern and central Chile. The low and middle elevations of central Chile were colonized in the late Miocene. These events correlate with the last pulses of Andean uplift in northern and central Chile (Ramos and Giglione, 2008; Giambiagi et al., 2016; Garzione et al., 2017) and the intensification of aridification produced by the establishment of a phase of global climate cooling (Hinojosa and Villagrán, 1997; Hartley et al., 2005; Zachos et al., 2001).

Patagonia was colonized by perennial herbs in the Pliocene (Clade D), and later by spiny shrubs. Shifts between the high Andes and the Patagonian steppe are common (Heibl and Renner, 2012), but diversification patterns are generally consistent with south-to-north migration (Nicolas and Plunkett, 2012; Simpson et al., 2009). We found that the climatic niche associated with temperature is highly conserved in the groups that reached Patagonia (clade D and subclade E1), supporting the hypothesis that Andes acted as a corridor facilitating the dispersion of lineages that track cold conditions (Luebert and Weigend, 2014).

Atacama Desert was colonized repeatedly since the middle Pliocene, first by lineages that acquired an annual habit, and later by several lineages of perennial herbs and spiny shrubs. Multiples plant genera colonized the coastal Atacama Desert in the last 10 Mya, including Heliotropium sect. Cochranea (Luebert and Wen, 2008), Malesherbia (Gengler-Nowak, 2002), Nolana (Tu et al., 2008), Cristaria (Bohnert et al., 2022) and Argylia (Glade-Vargas et al., 2021). Different scenarios have been proposed to explain these diversification patterns. Glade-Vargas et al. (2021) suggested that intensification of aridification during the late Miocene and Pliocene, together with the elevation of coastal range during Pleistocene, generated a biogeographic corridor along the Pacific Coast. According to this hypothesis, we observed a close relationship between coastal Atacama and Mediterranean species from subclades E2 and E3. However, we also identified a close relationship between species from different elevations, indicating that east-west migrations also occurred. Activation of east-west oriented canyons during the brief pluvial phases that interrupted hyperarid conditions in the Pliocene and Pleistocene of northern Chile (Jordan et al., 2014) might have facilitated species migration. Recently Bohnert et al. (2022) showed that Cristaria lineages colonized different habitats in the Atacama Desert during these pluvial phases.

Adesmia showed multiple and bidirectional transitions between growth forms that are non-significantly associated with climate. Interestingly, annuals, perennial herbs, and shrubs can coexist under the same climatic conditions. This pattern has also been described by Hjertaas et al. (2023), who suggest that similar environments can select for different life history strategies, or alternatively, that the evolution of life strategies can be constrained by historical and developmental factors. Cousins-Westerberg et al. (2023) also failed to detect a relationship between cold tolerance and growth form in Salicornieae. These results also parallel findings in Aeonium (dos Santos et al., 2022), but in this case, the authors demonstrated that growth forms modulate the response of leaf size, height, and other plant traits to climate rather than being adaptations per se. Our results contrast to other comparative studies indicating that climate strongly influences the evolution of annuality (Friedman, 2020; Boyko et al., 2023), plant height (Bliss, 1971; Korner, 2003; Zanne et al., 2014), and woodiness (Howard et al., 2019; Klimes et al., 2022).

The shrub habit evolved at least three times, with two reversions to the herbaceous form. Across angiosperms, multiple transitions between herbaceous and woody habits have occurred (Klimes et al., 2022), encompassing varying degrees of woodiness, from extensive wood production (in shrubs or trees) to reduced secondary growth limited to the basis of stems (typically seen in perennial herbs) or to fascicular areas (Rowe and Paul-Victor, 2012; Trueba et al., 2015). Woodiness has been observed in island (Dulin and Kirchoff, 2010; Zizka et al., 2022) and continental lineages (Kidner et al., 2016), where it evolved in response to multiple factors such as drought resistance (Frankiewicz et al., 2021; Klimes et al., 2022), extended plant longevity resulting from a stable climate or reduced herbivory (Zizka et al., 2022), and adaptation to toxic soils (Kidner et al., 2016). Conversely, the herbaceous habit is linked to tolerance to frost and shade (Klimes et al., 2022).

The annual habit originated five times independently from a perennial herbaceous ancestor. Three transitions matched events of desert colonization, in agreement with other comparative studies that show than annual form is favored in warm and arid environments (Evans et al., 2005; Friedman and Rubin, 2015; Pérez et al., 2020). However, it also has been suggested that annual habit might evolve under other seasonally stressful conditions, such as flooding and erosion (Hjertaas et al., 2023) or frosty seasons (Kadereit et al., 2007) as an escape strategy. Accordingly, the annual habit in Adesmia also evolved twice in the High Andes, where short growing seasons and repeated frosts are thought to inhibit seedling establishment (Bliss, 1971; Korner, 2003; Givnish, 2015). Warm and moist soils under the snowpack is also expected to difficult the persistence of seed banks in high elevations (Hjertaas et al., 2023). The evolution of the annual habit in A. capitellata is accompanied by the acquisition of autonomous selfing and reduced allocation to floral structures (Arroyo and Uslar, 1993), suggesting that the annual strategy might have evolved as part of a rapid development strategy that copes with the impoverished and variable conditions for animal pollination of high Andes (Arroyo et al., 2006).

We also detected at least one reversion from the annual form to the perennial form. Transitions from perennial to annual habitat have occurred thousands of times in flowering plants, but reversions are less frequent (Friedman, 2020; Boyko et al., 2023; Hjertaas et al., 2023). Derived perennial habits have been described, for example, in Orobanchaceae and Montiaceae (Tank and Olmstead, 2008; Ogburn and Edwards, 2015).

BAMM analyses indicated a constant rate of diversification, suggesting that the high species diversity of Chilean Adesmia is due to uniform speciation rather than to accelerated events of speciation. No evidence of radiations was also found in Oxalis (Heibl and Renner, 2012), which colonized Andean and desert habitats several times. Accelerated speciation events have been detected in lineages that acquired a perennial habit, which acted as a key innovation in high elevation habitats (Drummond et al., 2012; Hughes and Atchison, 2015; Nevado et al., 2016). In contrast, our data indicates that shifts in growth form in Adesmia did not lead to an increase in speciation. Furthermore, perennial herbs and spiny shrubs shifted repeatedly between low and high elevations habitats, suggesting that these are successful adaptations to survive both drought and unfavorable cold seasons. However, it is important to highlight that our study sampled only the 64% of Chilean Adesmia species. According to Cusimano and Renner (2010), sampling less than 80% of the species can introduce biases toward downturns. These biases are particularly pronounced when deeper nodes are oversampled. Nevertheless, we believe this issue does not significantly affect our study because we made a deliberate effort to include species representing the full range of morphological and taxonomic diversity within Chilean Adesmia.

It is also important to note that our study only sampled the 34% of the species in the genus. Little is known about phylogenetical relationships of the remaining species of Adesmia, and accordingly, expanding the sample size could potentially alter the outcomes of this study, particularly the ancestral reconstruction of area and growth form. A phylogenetic study has been conducted for Brazilian species of sect. psoraleoides (Iganci et al., 2013). This study, which included some species from other regions, found a north to south pattern of diversification, suggesting an expansion from arid regions to Patagonia, followed by colonization of southern Brazil. These results indicate that Chilean species might be paraphyletic.

Our phylogenetic analysis based on five nuclear regions of Chilean Adesmia does not match Burkart (1967) classification of Adesmia into two subgenera: Acanthadesmia, which groups spiny shrubs, and Adesmia, which groups unarmed forms. Ancestral reconstruction for spine presence/absence suggest that spiny structures were acquired by the ancestor of Clade E. These structures would have lost at least once in each subclade (E1, E2 and E3) is not monophyletic. Discrepancy between morphological and molecular data was also detected by Iganci et al. 2013. Most series described for Adesmia by Burkart (1967) were also not recovered in the molecular phylogeny of Chilean Adesmia, indicating that the systematics of the genus should be revisited.

Overall, our results showed that growth forms are evolutionarily labile in Adesmia. However, contrary to expectations, transitions were not significantly associated with climate, suggesting that annual, perennial herbaceous, and woody habits are alternative and successful strategies to survive unfavorable seasons in both the Atacama Desert and the high Andes. The high species diversity in these regions resulted from repeated events of colonization facilitated by adaptations that allowed plants to cope with both drought and cold.