Long-distance dispersals and ecological transitions underlie the biogeographic expansion of the pantropical magnoliid genus Xylopia (Annonaceae)

Long-distance dispersals and ecological transitions underlie the biogeographic expansion of the pantropical magnoliid genus Xylopia (Annonaceae) David M. Johnson1, Francis J. Nge2, Gregory Stull3, Nancy A. Murray1, Keegan Floyd1, Serafin Streiff4, Carlos Rodrigues-Vaz4,5 , Vincent R. C. Soulé4, Thomas L. P. Couvreur4 1 Department of Biological Sciences, Ohio Wesleyan University, Delaware, OH 43015, USA 2 National Herbarium of New South Wales, Botanic Gardens of Sydney, Locked Bag 6002, Mount Annan, NSW 2567, Australia 3 Department of Botany, National Museum of Natural History, Smithsonian Institution, Washington, DC 20560, USA 4 DIADE, Univ Montpellier, CIRAD, IRD, Montpellier, France 5 Institut de Systématique, Evolution, Biodiversité (ISYEB), Muséum National d’Histoire Naturelle-CNRS-SU-EPHE-UA, Paris, 75005, France Corresponding author: David M. Johnson ([email protected]) Editor Roy Erkens Received 23 May 2025♦ Accepted 8 September 2025♦ Published 2 December 2025 Frontiers of Biogeography 18, 2025, e159992|DOI 10.21425/fob.18.159992 Copyright David M. Johnson et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. RESEARCH ARTICLE FRONTIERS OF BIOGEOGRAPHY The scientific journal of The International Biogeography Society Abstract Pantropical taxa with broad and changing distributions provide useful models for assessing drivers of tropical tree biodiversity. Originating at the Eocene-Oligocene boundary, Xylopia is a vertebrate-dispersed woody plant genus with ca. 190 species evenly distributed across the Tropics. How did biogeographic and ecological transitions in this genus shape its broad present-day distribution? We analysed these transitions using ancestral area, climatic and spatial phylogenetic reconstructions, based on an extensive nuclear phylogeny and a curated dataset of occurrence records. The ancestral area was reconstructed as palaeotropical. The genus then underwent two dispersals from Africa, one to the Asia-Pacific area and one to the Neotropics. While niche conservatism in continental rain forests continued, the genus repeatedly transitioned to subhumid, inundated and ultramafic environments. Transitions from rain forests to subhumid environments increased in the Afrotropics as many rain forest groups underwent extinction. Association with inundated habitats, frequent in the early evolution of the genus, became sporadic. Ultramafic transitions occurred in five clades. Xylopia is present on 51 tropical islands; single-island endemics make up ca. 90% of insular species. Repeated dispersals took place between Africa and Madagascar, the Sunda and Sahul plates in the Asia-Pacific and from Central America to the Caribbean. Island distributions indicate overdispersion to remote islands, as well as limited radiations and stepping-stone dispersals. Novel environments, including islands, acted largely as sinks, together encompassing about half the species in the genus. A suite of traits promoting long-distance dispersal by a variety of non-resident birds, combined with the capacity for habitat transitions, were fundamental drivers of pantropical expansion and diversification. These drivers operated repeatedly in all regions, while idiosyncratic historical factors determined the timing and routes of dispersals. Highlights • Age and distribution of basal grade lineages suggest that the ancestral Xylopia lineage occupied Eocene Boreotropical forests, moving southwards and diversifying independently in the Afrotropic and Asia-Pacific areas of the Palaeotropics. • Biogeographic stochastic mapping estimated high regional in situ speciation (89.8%) amongst total biogeographic events for Xylopia. • The plants provide high value seed rewards taken by birds that transit long distances and between habitats, promoting long-distance dispersal. • Xylopia exhibits overdispersion even to isolated islands, but there is no significant correlation between species richness and isolation index. • In response to Miocene aridification and shifts to monsoon climates in the Afrotropics, Xylopia dispersed across multiple lineages, including transitions to subhumid habitats. • Ultramafic species occur on islands throughout the range of Xylopia; radiations took place in Xylopia sect. Xylopia on Cuba and in Xylopia sect. Stenoxylopia on New Guinea and New Caledonia. Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 2 Keywords Aril, biogeographic and ecological transitions, inundated, islands, sarcotesta, seasonally arid/subhumid, spatial phylogenetics, ultramafic Introduction Tropical rain forests have a disjunct distribution across the globe, yet share many plant families and genera (Thorne 1972; Slik et al. 2018). Understanding drivers of present-day distribution in these clades remains a key challenge in biogeography (e.g. Armstrong et al. (2014)). Recent studies (e.g. Thomas et al. (2015)) indicate that their dispersal largely post-dates the vicariant breakup of Gondwana during the Cretaceous, which was previously thought to explain intercontinental patterns (Raven and Axelrod 1974). Instead, distributions of these clades are now attributed to the contraction of the northern Palaeocene-Eocene Boreotropical flora towards the equator (Davis et al. 2002, 2004; Givnish and Renner 2004), long-distance dispersal (Givnish and Renner 2004; Song et al. 2022) or migration across land bridges through climatically favourable regions (Tiffney 1985; Rögl 1997; Pennington and Dick 2004; Renner et al. 2010; Masters et al. 2022). However, given that historical intercontinental migrations often result in exposure of the migrants to novel environments, the biological capacity to respond to ecological change must also be considered (e.g. Hughes et al. (2013); Zizka et al. (2020); Ringelberg et al. (2023)). As the largest and only pantropical genus in the woody plant family Annonaceae (Nge et al. 2024), Xylopia (Fig. 1) provides a model to assess the interactions of these biogeographic and ecological drivers. Revisionary studies have shown that, along with continental rain forest species typical in the family (Erkens et al. 2023), Xylopia species occur in subhumid and inundated environments and are common on islands (Johnson and Murray 2015, 2018, 2020). Eighteen island species occur on ultramafic sites (Johnson et al. 2013; Johnson and Murray 2023; pers. ob. DMJ, NAM for Cuba and Philippines), extreme environments marked by soils with low levels of nutrients, high levels of toxic metals and high alkalinity (Galey et al. 2017). Xylopia species include emergent trees to shrubs (Fig. 2) and rarely shrubby lianas (3 spp.). Even on drier sites, a minimum assemblage of woody vegetation appears to be necessary for establishment in all species, for example, the open canopy of dry dipterocarp forests, Cerrado and miombo woodlands or patches of woody vegetation in shrublands, such as restinga (Johnson and Murray 2018; Gomes et al. 2021; Johnson and Murray 2022; Matias et al. 2024). As in most Annonaceae, outcrossing in the genus is promoted by protogyny and regulation of pollination chamber access by petal movements during anthesis (Fig. 3), along with other adaptations (Ratnayake et al. 2007; Saunders 2020). Perhaps contributing to high fruit set, a range of insect pollinator species has been documented in Xylopia and can vary between sites in a single species (Pombo et al. 2017; Pinheiro Saravy et al. 2021; Saravy et al. 2021); other Annonaceae may have specialised pollinators (e.g. Murray (1993); Su et al. (2005)). Limited self-pollination has also been documented in the genus (Ratnayake et al. 2007) and may contribute to fruit set in some species. Fruits are dehiscent aggregates of finger-like monocarps with contrasting endocarp and seed colours (Fig. 4). Unlike most Annonaceae, where the pericarp provides a reward (Onstein et al. 2019), in Xylopia, the seed provides unique types and combinations of arils and sarcotestas (Stull et al. 2017), nutrient-dense rewards that include lipids and protein (Sourd and Gautier-Hion 1986; Lamperti et al. 2014; Pizo et al. 2021). The asynchronously ripening multi-seeded monocarps promote movement away from a source tree while allowing removal of several seeds in a single disperser visit (Figs 2, 4). Field and herbarium observations of Xylopia indicate that fruiting is often abundant (pers. ob. DMJ, NAM). A range of vertebrate dispersers, as well as secondary dispersal by ants, have been documented for Xylopia 40°S 20°S 0 ° 20°N 100°W100°W100°W 50 °W 0 ° 50 °E 100°E 150°E Longitude Latitude 5 10 15 SR Figure 1. Global distribution of Xylopia species richness per 1° × 1° grid cell, estimated using the curated occurrence dataset and the R programme ‘phyloraster’ (Alves‐Ferreira et al. 2024). Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 3 Figure 2. Xylopia adaptations. A Stilt roots in freshwater swamp forest, Singapore (Xylopia fusca); B Deciduous shrub, with flowers, in sandy dry forest, Madagascar (X. sericolampra); C Dehiscent fruit displaying red endocarp and seeds with green sarcotesta, Gabon (X. letestui). Photographs by D. M. Johnson (A), “theophile18” (B), N. Helme (C). B and C used under the terms of the Creative Commons licence for iNaturalist (https://www.iNaturalist.org). Photographs have been cropped for emphasis. Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 4 Figure 3. Xylopia flowers. A Single flower per axil (X. bocatorena); B Multiple flowers per axil (X. brasiliensis); C Cauliflory (Xylopia sp.); D–F Petal orientation and colour variations in X. perrieri (D), X. aromatica (E) and X. collina (F); G, H Pollination chambers in X. lamarckii (G) and X. longipetala (H). Photographs by E. Henríquez (A), D. Bernardes (B), R. Hoyer (C), G. E. Schatz (D), O. M. Montiel (E), D. M. Johnson (F), C. Kaiser-Bunbury (G), A. Boupoya (H). A, D, E and H used under the terms of the Creative Commons licence for Tropicos, the botanical information system of the Missouri Botanical Garden (https://www.tropicos.org), B and C used under the terms of the Creative Commons licence for iNaturalist (https://www.iNaturalist.org), G, used with permission of the photographer. Photographs have been cropped for emphasis. Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 5 Figure 4. Fruit and seed traits of Xylopia. A–C Seed presentation in X. gracilipes (A), X. pierrei (B) and X. takeuchii (C); all with sarcotestas; D Brush aril (X. staudtii); E Fimbriate aril (X. quintasii); F Bilobed aril (X. aethiopica); G Bilobed aril and sarcotesta (X. pulcherrima); H Sarcotesta (X. pynaertii). Photographs by M. Lötter (A), N. A. Murray (B), W. Takeuchi (C), J. B. Mba (D), T. L. P. Couvreur (E and H), David Harris (F), H. Galliffet (G). A, B, C, E, F and H used with permission of the photographers, D used under the terms of the Creative Commons licence for Tropicos, the botanical information system of the Missouri Botanical Garden (https://www.tropicos.org), G used under the terms of the Creative Commons licence for iNaturalist (https://www.iNaturalist.org). Photographs have been cropped for emphasis. Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 6 (e.g. Poulsen et al. (2001); Koné et al. (2008); Christianini and Oliveira (2010); Glenn and Bensen (2013); Bello et al. (2017)). Most significant for long-distance dispersals (Nogales et al. 2012; Jordano 2016; Viana et al. 2016) are medium to large-bodied species that fly long distances or travel between habitats, including hornbills in Africa and Asia (Brosset and Erard 1986; Whitney et al. 1998; Kitamura et al. 2011) and toucans in the Americas (Ragusa-Netto 2013), as well as migratory birds, for example thrushes, tyrant-flycatchers and tanagers (Bello et al. 2017; Pizo et al. 2021; Falcón et al. 2024). Fruit-eating pigeons (Brown and Hopkins 2002; Johnson and Murray 2020; Costa et al. 2024) are known to colonise islands worldwide, but are most diverse and ecologically important in the Asia-Pacific area, where they travel in flocks and range far into the Pacific (Corlett 2017; Oliver et al. 2023; Ibanez et al. 2025). The various combinations of seed traits in Xylopia, including size and reward types, may have attracted and influenced dispersers in advantageous ways (Stull et al. 2017). Although this dispersal syndrome occurs in other tropical tree families, such as Myristicaceae (Kays et al. 2011; Kitamura and Poonswad 2013) and Meliaceae (Howe and De Steven 1979), the only other Annonaceae to have evolved dehiscent fruits with a comparable seed reward are the Neotropical genera Cardiopetalum and Cymbopetalum (Murray 1993; Johnson and Murray 1995; Lopes et al. 2024). Xylopia, however, has a wider variety of both pollinators (cf. Murray (1993) and da Silva Elias et al. (2012) vs. Ratnayake et al. (2007) and Saravy et al. (2021)) and disperser rewards (Stull et al. (2017) vs. Lopes et al. (2024)) than these two genera. For a large, broadly distributed taxon with many narrowly-distributed species, Xylopia has undergone surprisingly few dispersals between continents (Stull et al. 2017). How then have biogeographic and ecological transitions contributed to its distribution? We hypothesise that idiosyncratic historical factors interacting with the biological capacity to disperse long distances and repeatedly transition to new environments drove the expansion and present-day distribution of the genus. As a framework, we used the expanded time-calibrated phylogeny provided in our companion paper, Nge et al. (2025b), summarised in the Materials and Methods, which also analysed diversification rates in the genus. Here, we focus on using the phylogeny to assess biogeographic patterns shown in previous studies by Thomas et al. (2015) and Stull et al. (2017), as well as ecological transitions indicated in recent revisionary papers (e.g. Johnson and Murray (2018, 2023)). We assembled occurrence records for each species in the phylogeny. We conducted ancestral area reconstruction modelling to infer evolutionary timing and locations of biogeographic transitions. We used similar modelling as well as quantitative climatic reconstructions to identify the phylogenetic transitions to novel subhumid, inundated and ultramafic environments. Finally, we used spatial phylogenetic methods to assess patterns of phylogenetic diversity and endemism and discuss the implications of these for conservation concerns. Materials and methods Time-calibrated phylogeny Our biogeographic and ecological analyses required a global time-calibrated species-level phylogeny of Xylopia (Fig. 5). To this end, we used a new, extensively sampled dated phylogeny from Nge et al. (2025b). Briefly, this phylogeny includes 168/191 (88%) Xylopia species: 48/56 species from the Neotropics, 73/78 from the Afrotropics and 47/57 from the Asia-Pacific area (Suppl. material 3; Suppl. material 4 for comments on taxonomic coverage). An Annonaceae-specific baiting kit, with probes for 469 nuclear loci, was used for target enrichment (Couvreur et al. 2019; Soulé et al. 2023, 2024). Both concatenated (CON, using RAxML v.8.2.9, Stamatakis (2014); Abadi et al. (2019)) and multi-species coalescent (MSC, using ASTRAL v.5.6.3, Zhang et al. (2018)) phylogenetic reconstructions were performed. Divergence times for lineages within Xylopia were estimated (BEAST v.2.6.6, Bouckaert et al. (2014)). Xylopia was recovered as monophyletic in both CON and MSC phylogenies (Nge et al. 2025b). Phylogenetic trees were well resolved, with a strongly supported backbone and topological conflicts restricted to shallow regions. All sections sensu Stull et al. (2017) were recovered as monophyletic; Xylopia sect. Verdcourtia (Johnson and Murray 2018), however, was nested in X. sect. Stenoxylopia (Fig. 5, as ST-Verdcourtia clade). For the CON-tree, which was used for all subsequent analyses, only four nodes were poorly supported (Bootstrap Percentage (BS) < 75%): the sister relationship of X. danguyella and X. ghesquiereana (BS = 39%) in STcapuronii, as well as the position of X. anomala within that clade (BS = 58%), the X. dibaccata–X. takeuchii subclade in ST-peekelii (BS = 56%) and the position of X. flexuosa within the XY-aethiopica clade (BS = 48%) (Nge et al. 2025b). An Early Oligocene age was recovered for the crown node of the genus (31.8 Ma, 95% confidence interval (CI): 22.6– 41.02 Ma), based on the BEAST analysis (Fig. 5, Suppl. material 2: fig. S1). The crown node ages of all five sections (Xylopia sect. Rugosperma, 20.79 Ma; X. sect. Neoxylopia, 10.54 Ma, X. sect. Ancistropetala, 8.46 Ma; X. sect. Stenoxylopia, 17.24 Ma; X. sect. Xylopia, 14.53 Ma) were dated to the Miocene, with most of the extant diversity arising in the Middle to Late Miocene after the Mid-Miocene Climatic Optimum (Steinthorsdottir et al. 2020) (Fig. 5). Species occurrence data We required data on spatial distribution of each species for climatic reconstructions and spatial phylogenetic analyses. We assembled a dataset of occurrence records for all species sampled in the phylogeny (Suppl. material 5). Most records were based on taxonomic treatments of the genus by Johnson et al. (2013) and Johnson and Murray (2015, 2018, 2020, 2023). Neotropical records were based primarily on specimens determined and annotated in the last five years, as well as determinations of digitised specimens Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 7 Figure 5. Ancestral area reconstruction with BioGeoBEARS, using the concatenated RAxML tree from Nge et al. (2025b). Colours on pie charts correspond to colours on map. Dashed vertical lines represent the individual time bins in the time-stratified BioGeoBEARS analysis simulating gradual tectonic collision of S with O and C with N. “TC-” numbers are uniform sample numbers designated in the larger Annonaceae project of which this is a part. Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 8 of L, MO, NY, P, S, U, US and WAG (acronyms from Thiers et al., continuously updated) by DMJ and NAM. Friis and Weber (2024) provided records from Ethiopia, South Sudan and Sudan. Occurrences from Australia were verified in the Australian Virtual Herbarium database (avh.chah.org. au). Several records for X. collina, X. tomentosa, X. carinata and X. sericolampra were based on diagnostic photographs (https://www.inaturalist.org/observations; Suppl. material 5). Records spanned the geographic and ecological breadth of each species, although sampling biases are inevitable given unequal sampling effort (Sosef et al. 2017; Daru et al. 2018). While it is common practice to obtain occurrence records from sources such as GBIF or iDigBio, the survey of Xylopia data available on GBIF revealed extensive misidentification of specimens at the species, genus and even family level. We therefore avoided using data directly from these sources. Collections that lacked geographic coordinate data were georeferenced using gazetteers, atlases and online sources. We mapped the occurrence data to check for errant coordinates; duplicate occurrences were removed prior to analysis. Several specimens of Xylopia aethiopica on which the label data suggested deliberate cultivation as an ornamental or spice plant were excluded from the dataset. There are no known introduced or invasive species. To assess endemism, we adopted the IUCN threshold for Vulnerable status, i.e. an Extent of Occurrence (EOO) smaller than 20,000 km2, the criterion for narrow-range endemic trees used by Tokarz and Condit (2021). For islands, we calculated Pearson correlation coefficients for total species richness and endemic species richness as functions of island size and island isolation index in R (v.4.3.1; R Core Team (2023)). To calculate confidence intervals, we applied a Fisher z-transformation using the DescTools package in R (v.0.99.60; Signorell (2025)). Values for island size and isolation index (sum of the square root of distance to the nearest continent, island and island group, in km), with islands below 250 km2 excluded, were taken from Pyron and Burbrink (2014, table S4) (Suppl. material 1: tables S1, S2). Biogeographic reconstructions To reconstruct the biogeographic history of Xylopia, we used BioGeoBEARS v.1.1.2 (Matzke 2013) in R v.3.5.1 (R Core Team 2016). BioGeoBEARS includes a variety of biogeographic models that account for dispersal and vicariance processes: dispersal-extinction-cladogenesis (DEC; Ree and Smith (2008)), dispersal-vicariance (DIVA; Ronquist (1997)) and models excluding vicariance (BAYAREA; Landis et al. (2013)). The software compares these models in a ML framework and assesses model fit via Akaike Information Criterion scores (AIC; Akaike (1974)). The software also includes the option of incorporating additional model parameters, such as jump dispersal (j), dispersal as a function of geographic distance (x) and time-stratified parameters to more accurately model the probability of dispersal between regions at different time periods, for example, due to changes in proximity from tectonic shifts. For geographic delimitation across the distribution, we defined seven discrete operational regions largely corresponding to continents (Fig. 5): S, South + Southeast Asia; O, Australia + New Guinea (West Oceania); P, Pacific (New Caledonia + Fiji); A, Africa; M, Madagascar; C, Central America; N, South America (southern Neotropics). Halmahera was included in the “O” region, following Weber’s Line, as Xylopia species there showed strong floristic affinities with the New Guinea Region (Johnson and Murray 2023). The Chocó Region (as delimited by Pérez-Escobar et al. (2019), while geologically part of Central America, was regarded as part of South America for this study. Inclusion of the Chocó with Central America would have added Xylopia columbiana to Central America and removed it from South America in the dataset. Species presence and absence in the seven regions are recorded in Suppl. material 6. We ran BioGeoBEARS with default settings of the parameters across six different models (DEC, DIVA and BAYAREA), with the addition of jump dispersal “j” for each (see Matzke (2022) for justification of this approach). We implemented the models with additional parameters that incorporate tectonic movement through time via different dispersal matrices across specified time intervals (Couvreur et al. 2011a; Thomas et al. 2015; Nge et al. 2021; Nge et al. 2022; Nge et al. 2023; Lopes et al. 2024). For Xylopia, two geological events were particularly relevant, the Miocene collision of the Sunda and Sahul plates (ca. 25 Ma; Crayn et al. (2015)) and the formation of the Panamanian Land Bridge (ca. 10–3 Ma; Bacon et al. (2015); O’Dea et al. (2016)). We incorporated these events into our biogeographic analyses via decreasing dispersal distance towards the present for these regions, across time intervals of 10 million years from 35–15 Ma and 5 million years from 15 Ma towards the present (Suppl. material 1: table S3). Distances were coded on a relative scale from 1–100 (100 being the furthest apart). As the crown radiation of Xylopia (31.8 Ma, Nge et al. (2025b)) post-dates the tectonic separation of Gondwana and the Eocene Boreotropical flora, these events were not included in our models. We estimated the number and mode of biogeographic events for the geographic regions using biogeographic stochastic mapping (BSM) implemented in BioGeoBEARS (Matzke 2015). BSM simulates, across nodes and branches of our phylogeny, possible biogeographic events (dispersal, vicariance, extinction) and provides a summary of mean events across the phylogeny. We performed 50 BSM counts using the best-fitting model of our biogeographic analysis (DIVALIKE + j; Suppl. material 1: table S5). Subhumid, inundated and ultramafic reconstructions We also used BioGeoBEARS to assess the direction and frequency of ecological transitions within the phylogeny. We reconstructed ancestral states for: 1) humid (i.e. rain forest) and subhumid environments; 2) inundated and non-inundated habitats (to distinguish habitat from climate) and 3) non-ultramafic and ultramafic substrates. We defined Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 9 subhumid environments on the basis of precipitation levels below those typically found in rain forest habitats. We used mean precipitation data to score species as occurring in either humid or subhumid environments. Species with < 1800 mm mean annual precipitation plus < 300 mm mean precipitation in the driest quarter were scored as occurring in subhumid environments, following DRYFLOR (2016) and Chen et al. (2019); species with either value above those thresholds were scored as humid (Suppl. material 6). We used direct precipitation data instead of qualitative biome categories to study subhumid transitions, as this approach is more reproducible and better accommodates climatic variability found within biome categories. We scored as “inundated” species for which the habitat descriptors included the words “swamp”, “inundated”, “riparian”, “gallery forest”, “riverbank”, “marshy”, “várzea”, “igapó” or “tahuampa”, based on revisionary work (Johnson et al. 2013; Johnson and Murray 2015, 2018, 2020, 2023) and specimen labels (Suppl. material 6). Taxa occurring on ultramafic substrates were scored using criteria of Johnson et al. (2013), Johnson and Murray (2023) and data from herbarium labels. For Madagascar, three species were scored as facultatively present on ultramafic, given their occurrence in the Ambatovy ultramafic area (Melluso et al. 2005). Facultative ultramafic species were coded as occurring on both non-ultramafic and ultramafic substrates and obligate species as occurring on ultramafic substrates only (Suppl. material 6). We used two approaches to model different aspects relating to transitions into and out of these habitats. First, we modelled the formation of these habitats/biomes using BioGeoBEARS and connectivity (dispersal) across different habitats. Similar to the biogeographic analyses above, the best-fitting model was determined, based on the lowest AIC score for each category (biogeographic and habitat). While we computed the AIC scores for BAYAREA models, we excluded these from the AIC assessment for best model fit as they do not account for vicariance. Second, we conducted ancestral state reconstructions for these characters (ultramafic, inundation, precipitation) as a proxy for traits that have allowed lineages to transition to, colonize and subsequently radiate in different environments. Thus, for the humid/subhumid and non-ultramafic/ultramafic characters, we also undertook an ancestral state reconstruction using SIMMAP, a Bayesian stochastic mapping approach that allows discrete traits to change along phylogenetic branches (Huelsenbeck et al. 2003; Bollback et al. 2006). We used the phytools v.2.4.4 (Revell 2024) “make.simmap” function and undertook 100 simulations. We chose the asymmetric model of character change (all rates different; ARD, Likelihood = -97.21619) rather than the equal rates model (ER, Likelihood = -116.4487) because it had a better likelihood, based on our data. Climate occupancy To further explore the phylogenetic distribution and timing of present-day transitions to subhumid environments, we conducted ancestral state reconstructions using 19 bioclimatic variables from WorldClim v.2 (30s/1 km2 dataset; Fick and Hijmans (2017)). After extracting climatic data for each occurrence, we calculated mean values for each variable for each species. These values were used for the following analyses, all performed in R v.4.4.1 (R Core Team 2021); they were also used as the basis for scoring species as humid or subhumid for the analyses described above. We ordinated the 19 bioclimatic variables using Principal Components Analyses (PCA) and used the package ‘factoextra’ v.1.0.7 (https://cloud.r-project.org/web/packages/ factoextra/index.html) to examine variation explained by each principal component and to determine which climatic variables contributed most to PC1 and PC2. We performed separate ancestral state reconstructions of PC1, PC2 and the original climatic variables (bio1–bio19) using the dated Xylopia phylogeny (Nge et al. 2025b). We fitted the data to two different ML models for continuous data (Brownian Motion and Ornstein-Uhlenbeck) and selected the model with the lowest AICc score, using the phytools package (Revell 2024). This package was also used to plot ancestral state reconstructions. These analyses allowed us to assess the phylogenetic distribution and timing of shifts between humid and subhumid environments using quantitative climatic data. We also performed analyses to identify “jumps” in climatic evolution, focusing on climatic variables pertaining to precipitation, in particular, annual precipitation (bio12) and precipitation of the driest quarter (bio17). These analyses involved inferring ancestral states for each node and then taking the difference between each (child) node and its parent node. Branches with a value change in the 90th or 95th percentile towards drier conditions from parent to child node were deemed significant and were plotted on the phylogeny for comparison with other climatic reconstructions. Spatial phylogenetics To assess phylogenetic diversity and endemism and relate them to conservation concerns, we used the occurrence data and the dated phylogeny in spatial diversity analyses with the R package ‘phyloraster’ v.2.2.0 (Alves‐Ferreira et al. 2024), implementing the methods of Faith (1992), Williams et al. (1994), Crisp et al. (2001), Isaac et al. (2007), Rosauer et al. (2009) and Redding et al. (2014) (additional method details in Suppl. material 4). Analyses were performed at a spatial resolution of 1° × 1° grid cells (Daru et al. 2020). We generated a presence/absence community matrix at this resolution, using the coordinates of the occurrence data to determine the presence/absence of each species across the grid cells. This matrix and the dated phylogeny were used for all spatial analyses. The following indices for each grid cell were estimated: phylogenetic diversity (PD), evolutionary distinctiveness (ED), phylogenetic endemism (PE) and weighted endemism (WE), as outlined in Alves‐Ferreira et al. (2024). Note that these indices, especially phylogenetic diversity, are influenced by levels of species richness (Alves‐Ferreira et al. 2024). However, species richness in Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 16 Zhou et al. 2012; Baker and Couvreur 2013; Armstrong et al. 2014; Couvreur 2015, table 1; Li et al. 2024). Thus, by ca. 11 Ma, the genus was established in the Afrotropics, the Asia-Pacific area and the Neotropics (Fig. 5). The primary disjunction in X. sect. Stenoxylopia between the Asian-Pacific area and the Afrotropics is congruent with expansion of tropical habitats and migratory corridors during the Mid-Miocene Climatic Optimum (Foster et al. 2012; Zhou et al. 2012; Steinthorsdottir et al. 2020; Couvreur et al. 2021; Morley 2024). Xylopia sect. Xylopia, with a crown age of 14.5 Ma (95% CI 10.07–19.99 Ma) and an ancestral area reconstructed as the Afrotropics, showed a single dispersal to the Neotropics, resolving the conflicting results of Thomas et al. (2015), one dispersal, and Stull et al. (2017), two dispersals. At the time of dispersal (11.6 Ma), cool temperatures (Denk et al. 2010) and the absence of key land bridges (Tiffney 1985; Rögl 1997) at higher latitudes would have made Boreotropical dispersal highly unlikely. Thus, overwater dispersal is implicated in the migration of X. sect. Xylopia to the Neotropics. Long-distance dispersal and colonisation of islands Long-distance dispersal is necessary to account for the broad global distribution of Xylopia, particularly overwater dispersals to isolated islands and from Africa to South America. Despite lacking adaptations typically associated with island dispersing taxa (Nogales et al. 2012, 2024; Weigelt et al. 2015; Roeble et al. 2024), the genus reached 51 tropical islands with isolation indices up to ca. 90 (Suppl. material 1: table S2). Multiple lineages colonising the same island, i.e. overdispersion, occurred repeatedly. For example, five lineages dispersed to Madagascar (Fig. 5) and, of nine islands with three Xylopia species, all but one show overdispersion (Suppl. material 1: tables S1, S2). As is typical with plant endemism and founder effects on islands more generally (e.g. Matzke (2014); Schrader et al. (2024)), colonisation of islands resulted in high in situ speciation in Xylopia and single-island endemics make up ca. 90% of insular species. Thus, islands are major biogeographic sinks in Xylopia, although stepping-stone dispersals took place in at least three lineages (Fig. 5). Increased dispersal pressure via widespread species (Viana et al. 2016; Wu et al. 2023) may have operated in Xylopia. The broad present-day distributions of X. aethiopica and X. aromatica, across continents and multiple islands, represent this possibility. In summary, evidence indicates that long-distance overwater dispersals of Xylopia were surprisingly common for a vertebrate-dispersed rain forest group. In Xylopia, possible vectors for overwater dispersal are volant vertebrates and ocean currents. Volant vertebrates typically direct dispersal to suitable habitats, i.e. climate matching (Gillespie et al. 2012). Dispersal by bats is unknown in Xylopia and, given the strongly conserved array of diurnal visual traits attractive to birds, is unlikely to be significant (Fig. 4). As detailed in the Introduction, a variety of birds are known to disperse Xylopia seeds. Two impediments to longer overwater dispersals by birds are the predominant north-south orientation of bird migration pathways across the globe, which would carry seeds away from the Tropics, and the short seed retention times recorded in experimental studies. Latitudinal bird movement in the Tropics, however, can occur when birds move along east-west island chains as has been documented in the southern Pacific (Kingston et al. 2003; Gillespie et al. 2012), as well as when birds are carried by tropical storms. For example, 13 vagrant African/Eurasian bird species have been documented in French Guiana and Brazil in the last 20 years (Lenrumé et al. 2024). While captive bird studies have indicated seed retention times of only a few hours, contra-indicating longer distance dispersals (e.g. Wotton et al. (2008); Kays et al. (2012); Bracho-Estévanez et al. (2024)), field observations and modelling studies have indicated dispersal up to 400 km (e.g. McConkey et al. (2004); Viana et al. (2016); Kleyheeg et al. (2019)) and there are indications that the 40°S 20°S 0 ° 20°N 100°W 100°W 100°W 50 °W 0 ° 50 °E 100°E 150°E Longitude Latitude PD 50 75 100 125 150 Figure 10. Phylogenetic diversity in Xylopia. Values are plotted by 1° × 1° grid cell across the geographic distribution, calculated using the curated occurrence dataset and the R programme ‘phyloraster’ (Alves‐Ferreira et al. 2024). Grid cells with PD values less than 50 are shown in light grey. The vertical scale represents the sum of the branch lengths for all species in a grid cell (with individual branch lengths from the phylogeny representing the estimated number of substitutions per site). Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 17 upper limit may be higher: one of the vagrant species noted above, fitted with a satellite transmitter, flew non-stop from the Netherlands for 5,600 km before its satellite signal was lost (van der Winden et al. 2010), a distance well above that from Africa to South America. Many Myristicaceae have arillate seed rewards, which can be compared to those of Xylopia. However, in contrast to the broad distribution of Xylopia, genera within the tropical Myristicaceae family have limited distributions, which have been attributed to poor long-distance dispersal capacity (Kitamura and Poonswad 2013; Frost et al. 2021). Most Myristicaceae have larger seeds: Kitamura and Poonswad (2013) indicate mean seed lengths > 2 cm (and up to 7 cm) for Asian-Pacific genera of Myristicaceae, except Horsfieldia (mean length < 2 cm). Xylopia seeds vary in length from 0.5–2.2 cm, but there are only a few species, with small geographic distributions, that have seeds at the upper end of this range (e.g. Johnson et al. (2013); Johnson and Murray (2015, 2018, 2020, 2022, 2023)). We suggest that the range of seed sizes and reward types in Xylopia increased capacity for both dispersal and transitions to new environments through the wider array of potential dispersers. Ocean current transport was not inferred for Xylopia, but dispersal by rafting has been supported by a number of studies in other taxa (e.g. Dick et al. (2003); Dick and Pennington (2019); Scarpetta et al. (2025)). In general, rafting follows predictable routes, is slow (Australia to New Zealand takes 2–3 years under good conditions, Gillespie et al. (2012)) and requires tolerance of sea water. And, although there are exceptions, most ocean current dispersals end in the littoral zone (Gillespie et al. 2012). A well-documented example of rafting in plants, which included fossil evidence, is the Africa/South America disjunction of the rain forest tree Symphonia globulifera (Dick et al. 2003), a species with the capacity for vegetative reproduction. Xylopia lacks vegetative reproduction; other impediments to rafting in the genus include dehiscent fruits with limited seed dormancy (Garwood 1995; Christianini and Oliveira 2010; Kanmegne et al. 2017), and absence from the littoral zone. In addition, the single dispersal of a Xylopia ancestor from Africa to the Neotropics occurred during middle to late Miocene, when east-west equatorial currents that exist today were absent in the Atlantic (Butzin et al. 2012) and the Congo River flowed through an arid region and possibly had a lower discharge volume than at present (Senut et al. 2009; Couvreur et al. 2021). In short, although rafting cannot be ruled out, barriers to establishment appear to be greater for rafting than for bird dispersal. Successful dispersals to oceanic, as opposed to continental, islands are rare in vertebrate-dispersed woody plants. In addition to the evolutionary limitations of the finite lifespan of oceanic islands (Gillespie et al. 2012; Swenson et al. 2019), barriers to successful colonisation include lack of dispersal capacity and absence of communities supporting establishment (Matthews and Triantis 2021). On the isolated oceanic Mascarene Islands, where Xylopia species are the only native Annonaceae, the genus overcame both dispersal and establishment barriers. Their distribution intersects with that of frugivorous blue pigeons, Alectroenas spp. (Heinen et al. 2023; cf. also Oliver et al. (2023)). The birds occur throughout the western Indian Ocean, including Madagascar, where they are known to take Xylopia seeds (Johnson and Murray 2020). The pigeon species native to Mauritius is now extinct, but was part of a superspecies that includes the birds on Madagascar (Heinen et al. 2023). Radiation of Xylopia on Mauritius, in which two of the three species are so divergent in floral morphology that they were initially classified in a different genus (Fig. 3G; Safford (1913); Johnson and Murray (2020)), suggests adaptation to the existing island community through selection for pollinator specialisation, which often occurs on oceanic islands (Wang et al. 2025). Reduced genetic connectivity to source populations, driving endemism and in situ speciation, varies with both island age and isolation (Gillespie et al. 2012). In Xylopia, dispersals to the oceanic Mascarene and Gulf of Guinea islands, comparable in size, but with disparate ages and isolation factors, illustrate this point. In the Mascarene Islands, dated ca. 7.8 Ma (McDougall and Chamalaun 1969) with isolation indices of 73–86 (Suppl. material 1: table S2), a single radiation of three endemic species occurred. In contrast, in the Gulf of Guinea islands, dated ca. 0.1 Ma with isolation indices of 17–40, four independent colonisations took place. Only one population on the most isolated island has differentiated from mainland populations, perhaps representing a distinct species (Strauß et al. 2024). In accord with biogeographic theory (reviewed in Whittaker et al. (2017); Suppl. material 2: fig. S2, Suppl. material 1: table S1), total and endemic species richness on islands in Xylopia were positively correlated with island size. Theory also predicts a negative correlation of richness with island isolation due to lower colonisation and higher extinction rates, but we found no correlation with isolation index (Suppl. material 2: fig. S2, Suppl. material 1: table S2). We attribute this to repeated overdispersion and the capacity of the genus to transition to new habitats, in particular to ultramafic sites on isolated islands. Overdispersion, along with the repeated endemism and radiation of Xylopia on islands, suggests the operation of taxon cycles (Keppel et al. 2023), meriting further study beyond the scope of this paper. Overall, the dispersal of Xylopia has been rare between continents, but occurred repeatedly within the major tropical areas, congruent with distributions of documented bird dispersers. Subhumid, inundated and ultramafic adaptation Xylopia transitioned to subhumid, inundated and ultramafic environments repeatedly throughout its evolutionary history and across its distribution. An assemblage of woody vegetation appears necessary for establishment, which likely encompasses environmental and community components required by many woody animal-dispersed Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 18 plants. Expansion into subhumid environments was associated with adaptations including reduced stature (0.4–4 m), sprawling habit, deciduous, leathery, revolute or tomentose leaves and indurated bud scales (Johnson et al. 2013; Johnson and Murray 2018, 2020). Small leaf size, found even in the early-diverging lineages of the genus, may also have been an advantageous preadaptation to lower precipitation environments, but this idea has not been tested in Xylopia. Two African species re-sprout after fire (Johnson and Murray 2018), suggesting a geoxylic habit (Gomes et al. 2021). Detarioid woodlands, for example, miombo, have existed since the Eocene (Jacobs et al. 2010; Neumann et al. 2017) and may provide a subhumid forest assemblage for Xylopia establishment (Gomes et al. 2021); 15 species of Xylopia are associated with 18 genera of these ectomycorrhizal legumes, mostly in Africa (Johnson and Murray 2018; Suppl. material 1: table S10). Occupancy of inundated habitats is common only in the basal grade of Xylopia (Fig. 6). These habitats were more extensive in the late Eocene/early Oligocene Palaeotropics, but declined as the climate continued to cool and dry (Couvreur et al. 2021; Morley 2024). Species in early diverging Xylopia sect. Rugosperma and X. sect. Neoxylopia grow in swamp forest and have stilt roots, a wetland-associated adaptation rare within Annonaceae (Fig. 2A; Corner (1978); Johnson and Murray (2015, 2018, 2023); Randi et al. (2022)). Limited diversification or elevated extinction occurred within these clades. Adaptations to higher light levels, disturbance and seasonally drier conditions common in swamp forest and floodplain plants may have allowed Xylopia species to transition to new habitats, as in X. staudtii (Savill and Fox 1967) and X. ferruginea (pers. ob. DMJ, NAM) and as has been documented in widespread temperate tree species (e.g. Acer rubrum, Abrams (1998); Juglans nigra and others; Hanberry (2022)). Isolated rivers and swamp forests in low precipitation areas have also allowed inundated species to persist in subhumid environments, as in some African Melastomataceae (Argyrella and Anaheterotis; Veranso‐Libalah et al. (2018)). In Xylopia, X. rubescens (Friis and Weber 2024) and X. katangensis occur only in swamp forest, but X. longipetala and X. elliotii track rivers into otherwise dry regions. The existence of water corridors has been suggested as a means of geographic spread and speciation from Amazônia to the Brazilian Atlantic coastal forest across dry Cerrado (Fabaceae; Nicholls et al. (2025)), but we have no evidence for such speciation in Xylopia. Plants growing on ultramafic substrates tolerate drought stress, metals such as nickel, chromium and cobalt and high levels of magnesium (Brady et al. 2005; van der Ent et al. 2014; Galey et al. 2017). In Xylopia, all ultramafic species occur on islands, but adaptations differ depending on local conditions. On dry ultramafic sites in New Caledonia and Cuba, species manifest parallel evolution of traits that characterise extreme xeromorphy (e.g. Iturralde (2001); Brady et al. (2005); Garnica-Díaz et al. (2022)). However, species of Xylopia on high rainfall ultramafic sites, such as in New Guinea, also show adaptations to edaphic stress. Takeuchi (2003) documented that these forests, while superficially resembling those of nearby non-ultramafic vegetation, included smaller-leaved species, few trees with buttresses and fewer vining, cauliflorous and compound-leaved plants. Takeuchi (2003) also noted the smaller crowns of canopy trees on ultramafic sites, a feature observed by collectors of X. pachysericea and X. musella (Johnson and Murray 2023). Within the New Caledonian ultramafic clade, soil specialisation has been documented: X. pancheri occurs primarily on weathered, coarse-grained peridotite and X. dibaccata on friable serpentinites (Johnson et al. 2013). These examples illustrate the flexible responses of Xylopia to both climatic and edaphic variables on ultramafic substrates at ecological (i.e. plasticity) and evolutionary scales (i.e. adaptation). The genetics of ultramafic adaptation are still not understood. Quantitative trait loci (Brady et al. 2005) and dispensable genes (presence/absence variants that can be enriched for functions, such as responses to toxic metals or abiotic stress, Hu et al. (2022)), are possible mechanisms. The specialised adaptations associated with plant growth on ultramafic substrates appear to have largely constrained Xylopia lineages within these sites, as has been found in the Alseuosmineae clade of Asterales and genera of Celastrales, Oxalidales and Malpighiales (Pillon et al. 2019; Gotty et al. 2022). Perhaps associated with its capacity to enter subhumid habitats or simply because of the outsized presence of the genus on islands, Xylopia has the highest proportion (ca. 10%) of ultramafic-occurring species in Annonaceae (cf. van Heusden (1996); Johnson and Murray (1999); Su and Saunders (2006); Turner (2014); van der Ent et al. (2014); Bayas et al. (2018); Pillon et al. (2019)). African ecological transitions Most Annonaceae, including Xylopia, occur in tropical rain forests (Erkens et al. 2023). In Africa, however, nearly half of Xylopia species occupy subhumid environments, arising as aridification began following the Mid-Miocene Climatic Optimum (Couvreur et al. 2021). Most humid forest species in Africa are nested within the larger Xylopia sect. Stenoxylopia lineage (Fig. 6) and likely represent reversions to higher rainfall habitats following the onset of the Indian Ocean monsoon regime at ca. 8 Ma (Nge et al. 2025b). Subhumid ancestry may, however, have provided species, such as X. hypolampra (ST-acutiflora), with the ecological breadth to occupy areas of Central Africa where low irradiance levels during the dry season allow persistence of evergreen forest on lower precipitation sites than in the Asian or American Tropics (Philippon et al. 2019). Shifts to subhumid environments also occurred in Afrotropical species of the Annonaceae genus Artabotrys (Chen et al. 2019) and in East African species of Isolona and Monodora (Couvreur et al. 2011b). In the tribe Monodoreae, however, significant extinction (range contraction) in response to aridification has been reported, based on biogeographic reconstructions (Dagallier et al. 2024); only 4/90 species Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 19 in the tribe are reported as dry-adapted. African plant extinction driven by increasing Tertiary aridification has also been inferred in other families, for example, Arecaceae and Fabaceae (Parkia) (Blach-Overgaard et al. 2013; Couvreur 2015; Couvreur et al. 2021; Oliveira et al. 2021). African PD levels are elevated for Xylopia, especially across the humid Guineo-Congolian Region, where the presence of the two oldest African lineages, as well as several more recent narrow-range endemics overlap (Fig. 10, Suppl. material 2: figs S7, S8). Low PD in Africa has been reported for other groups (Dagallier et al. 2020; Qian et al. 2023; Tietje et al. 2023), but this result may be due to the broader taxonomic and geographic scales of these studies, which make direct comparisons with Xylopia difficult. In subhumid eastern and southern areas of the continent, narrow-range endemism in Xylopia predominates (Johnson and Murray (2018, fig. 6)). The ecological breadth of Xylopia in Africa raised species diversity there to the level of Asia-Pacific and the Neotropics, in contrast to the typical “odd man out” status of African plant diversity (Couvreur 2015; Raven et al. 2020; Silva de Miranda et al. 2022). Madagascar and Mascarene Islands dispersals Our study confirmed five independent Xylopia dispersals between Africa and Madagascar (Stull et al. 2017; Johnson and Murray 2020). This represents the largest number of unique dispersal events amongst Annonaceae genera on Madagascar (Couvreur et al. 2008; Zhou et al. 2012; Chen et al. 2019; Dagallier et al. 2024; Ravomanana et al. 2025), but comparable to dispersals in other groups, such as the Ochnaceae tribe Ochneae (Schneider et al. 2022). Our results show no evidence in Xylopia of dispersal from Asia to Madagascar (Fig. 5), as has been documented in genera such as Nepenthes (Murphy et al. 2020) and Canarium (Federman et al. 2015). Madagascar has been isolated since the Jurassic Period (ca. 160 Ma) and how the island’s flora was assembled has long been debated (Simpson 1940; Ali and Hedges 2022; Antonelli et al. 2022). Relevant here is the recent proposal of sequential land bridges existing between Africa and Madagascar during the Tertiary, which is supported by palynological and macrofossil data (Masters et al. 2021; Génin et al. 2022; Aslanian et al. 2023). A suggestion by Génin et al. (2022, table 1) that Xylopia arrived from Africa in the Oligocene does not accord with our results, but a Late Miocene land bridge at 12–5 Ma (Génin et al. 2022) intersects with the crown node ages of all five Malagasy Xylopia lineages, dated as 6.5–5.4 Ma, resulting from dispersal events dated 15–5 Ma (Fig. 5). This time period also coincides with onset of the Indian monsoon augmentation of precipitation in the region (Nge et al. 2025b). The development of this monsoon climate may have provided open niches facilitating successful dispersals. Species of ST-acutiflora and ST-odoratissima both dispersed from Africa to Madagascar; their Malagasy species are deeply nested within African clades. In the XY-aethiopica clade, our results gave African origin a higher probability than a Malagasy origin (Fig. 5). Rain forests in the Eastern Arc mountains of Tanzania, at their fullest extent ca. 30 Ma (Dagallier et al. 2024), exist today as rain forest islands, harbouring the single extant African species of this clade. The timing of dispersal suggests that these forests provided the source for the Malagasy species, as in the Annonaceae genus Isolona (Dagallier et al. 2024). For both ST-Verdcourtia and ST-capuronii, however, ancestral area reconstruction showed an ambiguous stem node (Nge et al. 2025b). Madagascar has high PD (Fig. 10, Suppl. material 2: figs S7, S8), as has been documented in many taxa (e.g. Pyron and Burbrink (2014); Weigelt et al. (2015)). In Xylopia, we attribute this in large part to the diversity of the source biota represented in the multiple Africa–Madagascar transitions. With 30 mostly narrowly endemic species, the island has served as a major sink for Xylopia (Suppl. material 1: table S6). Madagascar was, however, the source area for dispersal of Xylopia to the Mascarene Islands, a unique dispersal in the Annonaceae. Asian-Pacific island dispersals and rain forest niche conservatism The Asian-Pacific clades of Xylopia all arose after the collision of the Sunda and Sahul plates in the Miocene, which reduced a major biogeographic barrier to plant dispersal (Su and Saunders 2009; Sniderman and Jordan 2011; Crayn et al. 2015; Joyce et al. 2021; Morley 2024; Olivar et al. 2024). After this event, lowland rain forest plants largely migrated from west to east (van Balgooy 1976; Crayn et al. 2015; Joyce et al. 2021; Kuhnhäuser et al. 2025). We inferred three west to east dispersals in the Xylopia sect. Stenoxylopia clade. Atypically, Australia+New Guinea functioned as a secondary source area for most eastward movement of this clade into the Pacific (Fig. 5). This is reflected by the greater strength of Australia+New Guinea (Sahul) as a source area than South+Southeast Asia (Sunda) in the BSM analysis (Suppl. material 1: table S6). There was only one dispersal in Xylopia sect. Rugosperma in the Asia-Pacific area and the direction of dispersal was ambiguous (Fig. 5). An understanding of the relationships of four unsampled Southeast Asian species of the group (Suppl. material 4) to the morphologically isolated X. aenea from New Guinea might resolve the ambiguity. Insular species, occurring in all three Asian-Pacific clades, occupy 44 islands (Suppl. material 1: table S1). Movement of Xylopia amongst islands was likely facilitated by overwater long-distance dispersal beginning in the Miocene and later, during the Pleistocene, by more limited overland migration and island-hopping when lower sea levels created temporary small islands or connected separate islands, for example in the Solomon Islands (Lavery et al. 2023; Nge et al. 2025b). The islands of Fiji and New Caledonia are amongst the most isolated islands on which Xylopia occurs (Suppl. material 1: table S2) and Fiji forms the Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 20 eastern limit of many angiosperm genera (Thorne 1972). Fiji was colonised by Xylopia twice independently, once from the Solomon Islands archipelago (ST-peekelii, subclade sister to X. brunneola) and once from New Guinea (ST-malayana, sister species to X. papuana). Notably, New Caledonia did not serve as the source for the three Xylopia species known from Fiji, in contrast to several other plant lineages (e.g. Perez-Calle et al. (2024). In Annonaceae, five genera extend into the Pacific (Tang et al. 2015; Turner and Utteridge 2017; Liu et al. 2025) and only two extend further than Xylopia: Huberantha, which shows some adaptation to coastal strand habitats (Thomas et al. 2015), and Meiogyne (van Heusden 1996; Liu et al. 2025). Most Asian-Pacific lineages of Xylopia are rain forest conservative. We attribute this to the limited extent of subhumid environments in the area (Morley 2024), rather than a limited capacity to diversify in subhumid environments. Complex diversity patterns are present however. Of the three largest islands, Borneo has twice the species richness (6 endemic/11 widespread) of New Guinea as well as of Sumatra. Sumatra lacks endemics entirely, while half the species of New Guinea are endemic (Suppl. material 1: table S1). The highest PD in the area is centered on the Malay Peninsula, where species of X. sect. Rugosperma and the two ST-malayana subclades overlap extensively (Fig. 10). These patterns likely reflect different combinations of landmass ages and movements, volcanism, and ultramafic soils in this tectonically active area (e.g. Johnson and Murray (2015); Galey et al. (2017); Johnson and Murray (2023); Morley (2024)). Neotropical founder effect and expansion Dispersals from Africa to the Neotropics have been particularly rare (see Givnish and Renner (2004) for a review; Armstrong et al. (2014); Ruhfel et al. (2016); Gamisch and Comes (2019)), attributed to unfavourable wind patterns over the Atlantic (Givnish and Renner 2004). We inferred a single dispersal of Xylopia from Africa to the Neotropics at ca. 11.6 Ma, followed by rapid diversification into three clades between 11.4 and 11.2 Ma (Nge et al. 2025b). Overall, however, its Neotropical expansion was not linked to elevated rates of diversification (Nge et al. 2025b), in contrast to rates for seed plants generally (Tietje et al. 2022). We also document low PD for Xylopia in the Neotropics (Fig. 10, Suppl. material 2: figs S7, S8), again atypical for Neotropical angiosperms (Qian et al. 2023; Tietje et al. 2023), perhaps due to the single recent dispersal of the genus. An exception to the lower PD is indicated in lowland Peru and Ecuador, an area where several Neotropical lineages intersect. Additional collecting would be needed to determine whether this is a true centre of high PD or just an artefact of low collection bias in other parts of the Amazon Basin, but this is one of the two highest tree diversity areas in South America (ter Steege et al. 2023). Dispersals from South America to Central America occurred in Xylopia between 9 and 3 Ma, congruent with the closure of the Central American Seaway (Bacon et al. 2015). Northward expansion of Xylopia to Central America was limited, however, perhaps impeded by competition with earlier arrivals, including Annonaceae species (e. g. Erkens et al. (2007); Lopes et al. (2024)). We found no dispersals from Central to South America in Xylopia, in agreement with an asymmetrical model of migration for plants and animals from 6 Ma onwards (Bacon et al. 2015). Transitions to novel environments were more limited in the Neotropics than elsewhere, but occurred in all three clades. Aridification was less extreme in the Neotropics than in Africa, but a notable jump to subhumid environments did take place in XY-peruviana (Figs 8, 9). In XY-muricata, with only eight species, the full range of Xylopia habitats was exploited (Figs 6, 7), illustrating the inherent capacity of the genus for environmental transitions. XY-aromatica, which encompasses the majority of Neotropical species of Xylopia, is largely rain forest conservative, but includes two species with exceptionally broad distributions and ecological breadth (Fig. 3E). Our results accord with Hughes et al. (2013), who found that Neotropical diversification patterns were structured ecologically more than geographically, with long-distance dispersal overriding geographic barriers. Intercontinental dispersals in Annonaceae We have suggested that Xylopia had a complex sequence of intercontinental dispersals — from the Eurasian Boreotropics to the Asia-Pacific and separately into Africa during the early Oligocene at ca. 31.8 Ma, followed by mid-Miocene dispersals from Africa to tropical America ca. 11 Ma and to the Asia-Pacific ca. 13 Ma (Fig. 5) and, later, to both Central America/Caribbean and the Pacific. How does this history compare to other widespread genera in the large magnoliid family Annonaceae? Dispersal between continents has not been common in the family and only eight of 108 genera occur in more than one of the three major tropical areas (Nge et al. 2024). Of these, Duguetia appears to have migrated via the Boreotropical route from Africa to tropical America, as evidenced by its Eocene stem age/ Miocene crown age pattern (Pirie and Doyle 2012), a route similar to that of other exclusively Neotropical Annonaceae thought to have originated in Africa, for example, Guatteria (Erkens et al. 2009) and the Bocageeae lineage (Lopes et al. 2024). Xylopia reached the Neotropics from Africa after the closing of this route, thus likely by overwater dispersal. Movement overland from Africa into the Asia-Pacific during the mid-Miocene (discussed by Thomas et al. (2015) and Chen et al. (2019)) is the suggested dispersal route for Uvaria and Artabotrys. These dispersals were congruent with the dispersal of Xylopia sect. Stenoxylopia from Africa; X. sect. Rugosperma was already present in the Asia-Pacific area. The other four Annonaceae genera present in more than one tropical area have trajectories distinct from that of Xylopia. Sphaerocoryne probably originated in the Afrotropics at ca. 7 Ma, but soon spread to Sundaland; Huberantha arose in the Asia-Pacific region at ca. 13 Ma, spreading to Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 21 eastern Africa and Madagascar at ca. 6.5 Ma (Thomas et al. 2015). Annona speciated widely in the Neotropics before one or possibly two dispersals to Africa in the Miocene (Pirie and Doyle 2012). The stem node of Anaxagorea dates to the Cretaceous and the genus has a Neotropical/Asia-Pacific distribution unique in the family, but not uncommon amongst angiosperms (Pirie and Doyle 2012). Although Xylopia is the most speciose Annonaceae genus in subhumid habitats, especially in Africa, it is notable that all eight inter-area distributed genera have species in either subhumid woodland or xeric environments (Maas et al. 2003; Johnson and Murray 2022; Erkens et al. 2023). Amongst these, Artabotrys and Uvaria, along with Xylopia, are about evenly distributed between Africa and Asia-Pacific. Colonisation of islands also distinguishes Xylopia, including its presence on ultramafic sites. In the Asia-Pacific area, Huberantha extends beyond the Solomon Islands and Australia and only the Asia-Pacific genus Goniothalamus compares to Xylopia in its breadth of island distribution (Thomas et al. 2017). In the Neotropics, Annona also speciated in the Caribbean. Diaspore types of genera dispersed overwater vary: fleshy pseudo-syncarps, berry-like monocarps and dehiscent monocarps with a seed reward (see Onstein et al. (2019) for a review; Chen et al. (2020)). Transitions to inundated habitats do not show a connection to later intercontinental dispersal in Xylopia or other genera. In conclusion, intercontinental genera are well-defined monophyletic groups that did not change significantly during movement between tropical areas, but show some ecological flexibility. Xylopia stands out because it underwent more dispersals across a broader area and overall exhibits greater ecological breadth. Endemism and conservation implications Xylopia showed a high level of in situ speciation (89.8%) in the BSM analyses compared to many plant taxa (Dupin et al. 2016; Nge et al. 2021, 2022, 2025a). In Xylopia, this manifested in diffusely distributed narrow endemics throughout its phylogeny and range; a similar level of in situ speciation was found for Asian rattan palms (Kuhnhäuser et al. 2025). This pattern contributes to the fact that almost half of the 123 Xylopia species evaluated by 2019 for the IUCN Red List (iucnredlist.org) were assessed as Near Threatened (NT) to Critically Endangered (CR); subsequently described species will augment that number (e.g. 24 spp., Johnson and Murray (2020); Johnson and Murray (2023)). Although the genus is present in many biodiversity hotspots (e.g. Myers et al. (2000); Mittermeier et al. (2011); Sandel et al. (2019); Tietje et al. (2023)), the distribution of Xylopia species presents problems for the hotspot approach to conservation. We used spatial metrics to help us assess this issue. Areas of high PD are limited (Fig. 10, Suppl. material 2: figs S7, S8), but were generally highest in continental rain forests and often due to the presence of a few widespread species, especially those of the early diverging clades. For example, the Pasoh Reserve (Malaysia), a refugial area with the highest known local species richness in Xylopia (Johnson and Murray 2015), falls in an area of high PD, but significant endemism is low to absent, as is also the case in West African refugia identified by Ernst et al. (2025). The greater conservation challenge lies in protecting diffusely distributed endemic species. PE was low globally, with only one high PE grid cell each on Jamaica and Madagascar. CE (Suppl. material 2: fig. S14) provided better identification of endemism, but important examples were still missed: obligate ultramafic endemics on New Guinea, five of the 12 spp. with EOO < 20,000 km2 in Africa (Johnson and Murray 2018) and many species of Sundaland. Even on Madagascar, which uniquely had elevated levels of all three metrics, narrowly endemic species fell outside of significant grid cells. Challenges for conserving plant phylogenetic diversity were summarised by Tietje et al. (2023). For Xylopia, this includes protection of vulnerable dispersers, such as the pigeons Alectroenas spp. in the western Indian Ocean (Heinen 2023) and Ducula goliath in New Caledonia (Ibanez et al. 2025) and native pollinators on Mauritius (Kaiser-Bunbury et al. 2009), which may be limiting factors for conservation. With high in situ speciation and non-overlapping distributions, Xylopia presents a conservation paradox: its distribution corresponds to sites of conservation concern, yet its endemism is diffuse and seldom entailed either radiation or overdispersion in hotspot areas. Conclusions Xylopia has dispersed to all major tropical rain forests of the world, as well as to remote islands. Originating in a continental rain forest family, Xylopia repeatedly transitioned to novel environments, demonstrating ecological breadth. These environments acted largely as sinks, but together they encompass about half the species within the broad distribution of the genus. Niche conservatism continued in continental rain forests, which served as the major source areas for dispersal. Transitions occurred throughout the phylogeny and across geographic regions whenever climatic, edaphic and geological conditions created ecological opportunities. The distribution of Xylopia on islands provides clear evidence of long-distance dispersal. Over its evolutionary history, the genus retained a suite of fruit and seed traits variable in the combinations and sizes of attractants and rewards. This flexibility allowed the genus to adapt to a range of dispersers, likely contributing to its repeated dispersals and capacity to adapt to resident dispersers. However, dispersals were infrequent enough that loss of genetic connectivity drove high in situ speciation and endemism in the genus; the rare intercontinental movements of Xylopia suggest the geographic limits of its dispersal capacity. Traits promoting long-distance dispersal, together with the capacity to repeatedly transition to novel environments, drove biogeographic expansion of Xylopia. These drivers both operated repeatedly across time and place; Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 22 idiosyncratic historical drivers determined the opportunities for expansion. For Xylopia, interactions of these drivers led to the breadth and complexity of its pantropical distribution. Acknowledgements DMJ and NAM acknowledge funding support from the Jason Swallen Herbarium (OWU), Ohio Wesleyan University. This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 Research and Innovation Programme (GLOBAL project; 1035 grant agreement No. 865787) to TLPC. KF received funding support from the Summer Science Research Program of Ohio Wesleyan University and Dr. Bruce Roberts. At Ohio Wesleyan University, Scott Linder advised on correlation statistics and Douglas Thompson prepared the plates of Xylopia photographs. For photographs, we thank David Harris, Joseph Lai, Mervyn Lötter, Christopher Kaiser-Bunbury and Wayne Takeuchi. Permission to use photographs from Tropicos, the botanical information system of the Missouri Botanical Garden (https://www.tropicos.org) and iNaturalist (https://www.iNaturalist.org), under the terms of their Creative Commons licences, is acknowledged, with thanks to the individual photographers: A. Boupoya, E. Henríquez, J. B. Mba, O. M. Montiel, G. E. Schatz (Tropicos), D. Bernardes, H. Galliffet, N. Helme, R. Hoyer and “theophile18” (iNaturalist). We acknowledge the ISO 9001 certified IRD i-Trop HPC (South Green Platform) at IRD Montpellier for providing HPC resources that have contributed to the phylogenetic results reported within this paper. Our thanks to the reviewers and the handling editor for insightful comments that led to improvements in the manuscript. Author contributions DMJ and NAM conceived this study. DMJ, NAM, TLPC, CR-V, VRCS, and SS provided samples and data. FJN, KF, GS, DMJ, NAM, and TLPC analysed the data, with input from other authors. All authors, but mainly DMJ, NAM, FJN, GS, and TLPC contributed to writing and revising the manuscript. Data accessibility statement All of the data that support the findings of this study are available in the main text or Supplementary Information. References Abadi S, Azouri D, Pupko T, Mayrose I (2019) Model selection may not be a mandatory step for phylogeny reconstruction. Nature Communications 10: 934. https://doi.org/10.1038/s41467-019-08822-w Abrams MD (1998) The red maple paradox: What explains the widespread expansion of red maple in eastern forests? BioScience 48(5): 355–364. https://doi.org/10.2307/1313374 Akaike H (1974) A new look at the statistical model identification. IEEE Transactions on Automatic Control 19: 716–723. https://doi. org/10.1109/TAC.1974.1100705 Ali JR, Hedges SB (2022) A review of geological evidence bearing on proposed Cenozoic land connections between Madagascar and Africa and its relevance to biogeography. Earth-Science Reviews 232: 104103. https://doi.org/10.1016/j.earscirev.2022.104103 Alves-Ferreira G, Machado Mota FM, Talora DC, Oliveira CV, Solé M, Heming NM (2024) ‘phyloraster’: an R package to calculate measures of endemism and evolutionary diversity for rasters. Ecography 2024: e06902. https://doi.org/10.1111/ecog.06902 Antonelli A, Smith RJ, Perrigo AL et al. (2022) Madagascar’s extraordinary biodiversity: Evolution, distribution, and use. Science 378: eabf0869. https://doi.org/10.1126/science.abf0869 Armstrong KE, Stone GN, Nicholls JA, Valderrama E, Anderberg AA, Smedmark J, Gautier L, Naciri Y, Milne R, Richardson JE (2014) Patterns of diversification amongst tropical regions compared: a case study in Sapotaceae. Frontiers in Genetics 5: 362. https:// doi.org/10.3389/fgene.2014.00362 Aslanian D, Pellen P, Rabineau M, Moulin M, Leroux E, Delaunay A, Baby G, Courgeon S, Linol B, Suc J-P, Popescu S, Fauquette S, Mazza PPA, Couette S, Delpero M, Huck T, Penven P, Le Hir T, Zhang Y, Génin F, Masters J (2023) The postulation of intermittent land bridges as an explanation for reiterated colonization events of Madagascar by African vertebrates: an in-depth review and novel insights in honour of the late Judith Masters and Fabien Génin. Earth-Science Reviews 246: 104585. https://doi.org/10.1016/j. earscirev.2023.104585 Bacon CD, Silvestro D, Jaramillo C, Smith BT, Chakrabarty P, Antonelli A (2015) Biological evidence supports an early and complex emergence of the Isthmus of Panama. Proceedings of the National Academy of Sciences of the United States of America 112: 6110–6115. https://doi.org/10.1073/pnas.1423853112 Baker WJ, Couvreur TLP (2013) Global biogeography and diversification of palms sheds light on the evolution of tropical lineages. I. Historical biogeography. Journal of Biogeography 40: 274–285. https://doi.org/10.1111/j.1365-2699.2012.02795.x Bayas QEB, Salvador SASJ, Ragragio EM, Obico JJA (2018) Taxonomic survey of nickel hyperaccumulating plants in a mining site on Luzon island, Philippines. Philippine Journal of Systematic Biology 12(1): 103–108. https://doi.org/10.26757/ pjsb.2018a12001 Bello C, Galetti M, Montan D, Pizo MA, Mariguela TC, Culot L, Bufalo F, Labecca F, Pedrosa F, Constantini R, Emer C, Silva WR, Silva FR, Ovaskainen O, Jordano P (2017) ATLANTIC-FRUGIVORY: a plant-frugivore interaction dataset for the Atlantic forest. Ecology 98: 1729. https://doi.org/10.1002/ecy.1818 Blach-Overgaard A, Kissling WD, Dransfield J, Balslev H, Svenning JC (2013) Multimillion-year climatic effects on palm species diversity in Africa. Ecology 94: 2426–2435. https://doi.org/10.1890/121577.1 Bollback JP (2006) SIMMAP: Stochastic character mapping of discrete traits on phylogenies. BMC Bioinformatics 7: 88–94. https:// doi.org/10.1186/1471-2105-7-88 Bouckaert R, Heled J, Kühnert D, Vaughan T, Wu C-H, Xie D, Suchard MA, Rambaut A, Drummond AJ (2014) BEAST 2: a software platform for Bayesian evolutionary analysis. PLOS Computational Biology 10: e1003537. https://doi.org/10.1371/journal.pcbi.1003537 Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 23 Bracho-Estévanez CA, Mariano Cuadrado M, Sánchez I, Onrubia A, González-Varo JP (2024) Plant traits determine seed retention times in frugivorous birds: Implications for longdistance seed dispersal. Functional Ecology 38: 2247–2260. https://doi. org/10.1111/1365-2435.14642 Brady KU, Kruckeberg AR, Bradshaw Jr. HD (2005) Evolutionary ecology of plant adaptation to serpentine soils. Annual Review of Ecology, Evolution, and Systematics 36: 243–266. https://doi. org/10.1146/annurev.ecolsys.35.021103.105730 Brosset A, Erard C (1986) Les oiseaux des régions forestières du nord-est du Gabon, Vol. 1: Écologie et comportement des espèces. Société Nationale de Protection de la Nature, Paris, 289 pp. https://doi.org/10.3406/revec.1986.5179 Brown ED, Hopkins MJG (2002) Tests of disperser specificity between frugivorous birds and rainforest fruits in New Guinea. Emu 102: 137–146. https://doi.org/10.1071/MU00082 Butzin M, Lohmann G, Bickert T (2011) Miocene ocean circulation inferred from marine carbon cycle modeling combined with benthic isotope records. Paleoceanography 26: PA1203. https://doi. org/10.1029/2009PA001901 Chen J, Thomas DC, Saunders RMK (2019) Geographic range and habitat reconstructions shed light on palaeotropical intercontinental disjunction and regional diversification patterns in Artabotrys (Annonaceae). Journal of Biogeography 46: 2690–2705. https:// doi.org/10.1111/jbi.13703 Chen J, Thomas DC, Saunders RMK (2020) Correlated evolution of diaspore traits and potential frugivore-mediated selection in a fleshy-fruited tropical lineage (Artabotrys, Annonaceae). Evolution 74(9): 2020–2032. https://doi.org/10.1111/evo.14046 Christianini AV, Oliveira PS (2010) Birds and ants provide complementary seed dispersal in a neotropical savanna. Journal of Ecology 98: 573–582. https://doi.org/10.1111/j.1365-2745.2010.01653.x Corlett RT (2017) Frugivory and seed dispersal by vertebrates in tropical and subtropical Asia: an update. Global Ecology and Conservation 11: 1–22. https://doi.org/10.1016/j.gecco.2017.04.007 Corner EJH (1978) The freshwater swamp-forest of South Johore and Singapore. Gardens’ Bulletin Supplement No. 1. Singapore Botanic Gardens, Singapore, 266 pp. [+ 40 plates] https://doi. org/10.5962/bhl.title.77526 Costa MF, Checon CT, Silva WR (2024) Shifting from seed predator to seed disperser in the Picazuro Pigeon: Insights from an urbanized area. https://doi.org/10.1590/SciELOPreprints.9707 Couvreur TLP (2015) Odd man out: Why are there fewer plant species in African rain forests? Plant Systematics and Evolution 301: 1299–1313. https://doi.org/10.1007/s00606-014-1180-z Couvreur TLP, Chatrou LW, Sosef MS, Richardson JE (2008) Molecular phylogenetics reveal multiple tertiary vicariance origins of the African rain forest trees. BMC Biology 6: 54. https:// doi.org/10.1186/1741-7007-6-54 Couvreur TLP, Dauby G, Blach-Overgaard A, Deblauwe V, Dessein S, Droissart V, Hardy OJ, Harris DJ, Janssens SB, Ley AC, Mackinder BA, Sonké B, Sosef MSM, Stévart T, Svenning J-C, Wieringa JJ, Faye A, Missoup AD, Tolley KA, Nicolas V, Ntie S, Fluteau F, Robin C, Guillocheau F, Barboni D,, Sepulchre P (2021) Tectonics, climate and the diversification of the tropical African terrestrial flora and fauna. Biological Reviews 96: 16–51. https://doi.org/10.1111/brv.12644 Couvreur TLP, Helmstetter AJ, Koenen EJM, Bethune K, Brandão RD, Little SA, Sauquet H, Erkens RHJ (2019) Phylogenomics of the major tropical plant family Annonaceae Frontiers in Plant Science 9: 1941. https://doi.org/10.3389/fpls.2018.01941 Couvreur TLP, Pirie MD, Chatrou LW, Saunders RMK, Su YCF, Richardson JE, Erkens RHJ (2011a) Early evolutionary history of the flowering plant family Annonaceae: Steady diversification and boreotropical geodispersal. Journal of Biogeography 38: 664–680. https://doi.org/10.1111/j.1365-2699.2010.02434.x Couvreur TLP, Porter-Morgan H, Wieringa JJ, Chatrou LW (2011b) Little ecological divergence associated with speciation in two African rain forest tree genera. BMC Evolutionary Biology 11: 296. https://doi.org/10.1186/1471-2148-11-296 Crayn DM, Costion C, Harrington MG (2015) The Sahul–Sunda floristic exchange: dated molecular phylogenies document Cenozoic intercontinental dispersal dynamics. Journal of Biogeography 42: 11–24. https://doi.org/10.1111/jbi.12405 Crisp MD, Laffan S, Linder HP, Monro A (2001) Endemism in the Australian flora. Journal of Biogeography 28: 183–198. https://doi. org/10.1046/j.1365-2699.2001.00524.x Dagallier L-PMJ, Janssens SB, Dauby G, Blach-Overgaard A, Mackinder BA, Droissart V, Svenning J-C, Sosef MSM, Stevart T, Harris DJ, Sonke B, Wieringa JJ, Hardy OJ, Couvreur TLP (2020) Cradles and museums of generic plant diversity across tropical Africa. New Phytologist 225: 2196–2213. https://doi. org/10.1111/nph.16293 Dagallier L-PMJ, Condamine FL, Couvreur TLP (2024) Sequential diversification with Miocene extinction and Pliocene speciation linked to mountain uplift explains the diversity of the African rain forest clade Monodoreae (Annonaceae). Annals of Botany 133: 677–696. https://doi.org/10.1093/aob/mcad130 Daru B, Farooq H, Antonelli A, Faurby S (2020) Endemism patterns are scale dependent. Nature Communications 11: 2115. https:// doi.org/10.1038/s41467-020-15921-6 Daru BH, Park DS, Primack RB, Willis CG, Barrington DS, Whitfeld TJS, Seidler TG, Sweeney PW, Foster DR, Ellison AM, Davis CC (2018) Widespread sampling biases in herbaria revealed from largescale digitization. New Phytologist 217: 939–955. https://doi. org/10.1111/nph.14855 Davis CC, Bell CD, Mathews S, Donoghue MJ (2002) Laurasian migration explains Gondwanan disjunctions: evidence from Malpighiaceae. Proceedings of the National Academy of Sciences of the United States of America 99: 6833–6837. https://doi. org/10.1073/pnas.102175899 Davis CC, Fritsch PW, Bell CD, Mathews S (2004) High-latitude Tertiary migrations of an exclusively tropical clade: evidence from Malpighiaceae. International Journal of Plant Sciences 165(suppl.): S107–S121. https://doi.org/10.1086/383337 Denk T, Grímsson F, Zetter R (2010) Episodic migration of oaks to Iceland: evidence for a North Atlantic ‘land bridge’ in the latest Miocene. American Journal of Botany 97: 276–287. https://doi. org/10.3732/ajb.0900195 Dick CW, Abdul-Salim K, Bermingham E (2003) Molecular systematic analysis reveals cryptic tertiary diversification of a widespread tropical rain forest tree. American Naturalist 162: 691–703. https://doi.org/10.1086/379795 Dick CW, Pennington RT (2019) History and geography of Neotropical tree diversity. Annual Review of Ecology, Evolution, and Systematics 50: 279–301. https://doi.org/10.1146/annurevecolsys-110617-062314 Frontiers of Biogeography 18, 2025, e159992 David M. Johnson et al. 24 DRYFLOR [Banda-R K et al.] (2016) Plant diversity patterns in neotropical dry forests and their conservation implications. Science 353: 1383–1387. https://doi.org/10.1126/science.aaf5080 Dupin J, Matzke NJ, Särkinen T, Knapp S, Olmstead RG, Bohs L, Smith SD (2016) Bayesian estimation of the global biogeographical history of the Solanaceae. Journal of Biogeography 44: 887–899. https://doi.org/10.1111/jbi.12898 Erkens RHJ, Blanpain LMP, Carrascosa JI, Runge K, Verspagen N, Cosiaux A, Couvreur TLP (2023) Spatial distribution of Annonaceae across biomes and anthromes: Knowledge gaps in spatial and ecological data. Plants, People, Planet 5: 520–535. https://doi. org/10.1002/ppp3.10321 Erkens RHJ, Chatrou LW, Maas JW, van der Niet T, Savolainen V (2007) A rapid diversification of rainforest trees (Guatteria; Annonaceae) following dispersal from Central into South America. Molecular Phylogenetics and Evolution 44: 399–411. https://doi. org/10.1016/j.ympev.2007.02.017 Erkens RHJ, Maas JW, Couvreur TLP (2009) From Africa via Europe to South America: migrational route of a species rich genus of Neotropical lowland rain forest trees (Guatteria: Annonaceae). Journal of Biogeography 36(12): 2338–2352. https://doi.org/10.1111/ j.1365-2699.2009.02162.x Ernst M, Rödel M-O, Blom MPK (2025) Towards a comprehensive view on evolutionary refugia in West African rainforests. Frontiers of Biogeography 18: e139537. https://doi.org/10.21425/ fob.18.139537 de la Estrella M, Forest F, Klitgård B, Lewis GP, Mackinder BA, de Queiroz LP, Wieringa JJ, Bruneau A (2018) A new phylogeny-based tribal classification of subfamily Detarioideae, an early branching clade of florally diverse tropical arborescent legumes. Scientific Reports 8: 6884. https://doi.org/10.1038/s41598-018-24687-3 Faith DP (1992) Conservation evaluation and phylogenetic diversity. Biological Conservation 61: 1–10. https://doi.org/10.1016/00063207(92)91201-3 Falcon JE, Schoereder JH, Ribeiro VS, Christianini AV, Camargo PH, Paolucci LN (2024) How do birds and ants contribute to the recruitment of a tropical tree? Biotropica 56: e13372. https://doi. org/10.1111/btp.13372 Federman S, Dornburg A, Downie A, Richard AF, Daly DC, Donoghue MJ (2015) The biogeographic origin of a radiation of trees in Madagascar: implications for the assembly of a tropical forest biome. BMC Evolutionary Biology 15: 216. https://doi.org/10.1186/ s12862-015-0483-1 Fick SE, Hijmans RJ (2017) WorldClim 2: new 1km spatial resolution climate surfaces for global land areas. International Journal of Climatology 37(12): 4302–4315. https://doi.org/10.1002/joc.5086 Foster GL, Lear CH, Rae JW (2012) The evolution of pCO2, ice volume and climate during the middle Miocene. Earth and Planetary Science Letters 341: 243–254. https://doi.org/10.1016/j.epsl.2012.06.007 Friis I, Weber O (2024) Xylopia nilotica (Annonaceae) in Ethiopia, Sudan, South Sudan, and Uganda, with a review of the genus in North Eastern Tropical Africa. Webbia, Journal of Plant Taxonomy and Geography 79(2): 239–254. https://doi.org/10.36253/jopt-16134 Frost L, Santamaría-Aguilar DA, Singletary D, Lagomarsino LP (2021) Neotropical niche evolution of Otoba trees in the context of global biogeography of the nutmeg family. Journal of Biogeography 49: 156–170. https://doi.org/10.1111/jbi.14290 Galey ML, van der Ent A, Iqbal MCM, Rajakaruna N (2017) Ultramafic geoecology of South and Southeast Asia. Botanical Studies 58: 18. [28 pp] https://doi.org/10.1186/s40529-017-0167-9 Gamisch A, Comes HP (2019) Clade-age-dependent diversification under high species turnover shapes species richness disparities among tropical rainforest lineages of Bulbophyllum (Orchidaceae). BMC Evolutionary Biology 19: 93. https://doi.org/10.1186/ s12862-019-1416-1 Garnica‐Díaz C, Berazaín Iturralde R, Cabrera B, Caldern‐Morales E, Felipe FL, García R, Gómez Hechavarría JL, Guimarães AF, Medina E, Paul ALD, Rajakaruna N, Restrepo C, Siebert SJ, van den Berg E, van der Ent A, Velasquez G, Hulshof CM (2022) Global plant ecology of tropical ultramafic ecosystems. The Botanical Review 89: 115–157. https://doi.org/10.1007/s12229-022-09278-2 Garwood NC (1995) Studies in Annonaceae. XX. Morphology and ecology of seedlings, fruits, and seeds of selected Panamanian species. Botanische Jahrbucher für Systematik, Pflanzengeschichte und Pflanzen-geographie 117: 1–152. Génin F, Mazza PPA, Pellen R, Rabineau M, Aslanian D, Masters JC (2022) Co-evolution assists geographic dispersal: the case of Madagascar. Biological Journal of the Linnean Society 137: 163– 182. https://doi.org/10.1093/biolinnean/blac090 Gillespie RG, Baldwin BG, Waters JM, Fraser CI, Nikula R, Roderick GK (2012) Long-distance dispersal: a framework for hypothesis testing. Trends in Ecology and Evolution 27(1): 47–56. https://doi. org/10.1016/j.tree.2011.08.009 Givnish TJ, Renner SS (2004) Tropical intercontinental disjunctions: Gondwana breakup, immigration from the boreotropics, and transoceanic dispersal. International Journal of Plant Sciences 165(4 Suppl.): S1–S6. https://doi.org/10.1086/424022 Glenn ME, Bensen KJ (2013) The Mona Monkeys of Grenada, São Tomé and Príncipe: long-term persistence of a guenon in permanent fragments and implications for the survival of forest primates in protected areas. In: Marsh LK, Chapman CA (Eds) Primates in fragments, Developments in Primatology: Progress and Prospects. Springer, Heidelberg, 413–422. https://doi. org/10.1007/978-1-4614-8839-2_27 Gomes AL, Revermann R, Meller P, Goncalves FMP, Aidar MPM, Lages F, Finckh M (2021) Functional traits and symbiotic associations of geoxyles and trees explain the dominance of detarioid legumes in miombo ecosystems. New Phytologist 230: 510–520. https://doi. org/10.1111/nph.17168 Gotty K, Kergoat GJ, Jouannais P, Invernon V, Merlot S, Pillon Y (2022) Relict lineages with extreme ecology and physiology: metal hyperaccumulation on ultramafic substrates in New Caledonian Alseuosmineae (Asterales). Plant Ecology & Diversity 15: 227–242. https://doi.org/10.1080/17550874.2022.2093290 Hanberry BB (2022) Confronting the issue of invasive native tree species due to land use change in the eastern United States. Land 11: 161. Heinen JH, Florens FBV, Baider C, Hume JP, Kissling WD, Whittaker RJ, Rahbek C, Borregaard MK (2023) Novel plant–frugivore network on Mauritius is unlikely to compensate for the extinction of seed dispersers. Nature Communications 14: 1019. Howe HF, De Steven D (1979) Fruit production, migrant bird visitation, and seed dispersal of Guarea glabra in Panama. Oecologia 39: 185–196. https://www.jstor.org/stable/4215810 Frontiers of Biogeography 18, 2025, e159992 Xylopia biogeographic expansion 25 Hu H, Scheben A, Verpaalen B, Tirnaz S, Bayer PE, Hodel RGJ, Batley J, Soltis DE, Soltis PS, Edwards DE (2022) Amborella gene presence/absence variation is associated with abiotic stress responses that may contribute to environmental adaptation. New Phytologist 233: 1548–1555. https://doi.org/10.1111/nph.17658 Huelsenbeck JP, Nielsen R, Bollback JP (2003) Stochastic mapping of morphological characters. Systematic Biology 52: 131–158. https://doi.org/10.1080/10635150390192780 Hughes CE, Pennington RT, Antonelli A (2013) Neotropical plant evolution: assembling the big picture. Botanical Journal of the Linnean Society 171: 1–18. https://doi.org/10.1111/boj.12006 Ibanez T, Goizé S, Munzinger J (2025) Revising the role of the endemic New Caledonian imperial pigeon (Ducula goliath Gray, 1859) as a key disperser of tropical rainforest trees. Journal of Tropical Ecology 41(e6): 1–10. https://doi.org/10.1017/S026646742500001X Isaac NJ, Turvey ST, Collen B, Waterman C, Baillie JE (2007) Mammals on the EDGE: conservation priorities based on threat and phylogeny. PLOS ONE 2(3): e296. https://doi.org/10.1371/ journal.pone.0000296 Iturralde RB (2001) The influence of ultramafic soils on plants in Cuba. South African Journal of Science 97: 510–512. https://hdl. handle.net/10520/EJC97258 IUCN (2025) IUCN Red List of Threatened Species. http://iucnredlist. org [Visited on 23/02/2025] Jacobs BF, Pan AP, Scotese CR (2010) A review of the Cenozoic vegetation history of Africa. In: Werdelin L, Sanders J (Eds) Cenozoic mammals of Africa. University of California Press, Berkeley, 55– 72. https://doi.org/10.1525/california/9780520257214.003.0005 Johnson DM, Munzinger J, Peterson JA, Murray NA (2013) Taxonomy and biogeography of the New Caledonian species of Xylopia L. (Annonaceae). Adansonia, sér. 3, 35(2): 207–226. https://doi. org/10.5252/a2013n2a3 Johnson DM, Murray NA (1995) Synopsis of the tribe Bocageeae (Annonaceae), with revisions of Cardiopetalum, Froesiodendron, Trigynaea, Bocagea, and Hornschuchia. Brittonia 47: 248–319. https://doi.org/10.2307/2807118 Johnson DM, Murray NA (1999) Four new species of Polyalthia (Annonaceae) from Borneo and their relationship to Polyalthia insignis. Contributions from the University of Michigan Herbarium 22: 95–104. https://biostor.org/reference/177646 Johnson DM, Murray NA (2015) A contribution to the systematics of Xylopia (Annonaceae) in Southeast Asia. Gardens’ Bulletin Singapore 67(2): 361–386. https://doi.org/10.3850/ S2382581215000307 Johnson DM, Murray NA (2018) A revision of Xylopia (Annonaceae): the species of tropical Africa. PhytoKeys 97: 1–252. https://doi. org/10.3897/phytokeys.97.20975.figure10 Johnson DM, Murray NA (2019) Two new species of Xylopia (Annonaceae) from peninsular Thailand. Thai Forest Bulletin, Botany 47(2): 264–269. https://doi.org/10.20531/tfb.2019.47.2.16 Johnson DM, Murray NA (2020) A revision of Xylopia L. (Annonaceae): the species of Madagascar and the Mascarene islands. Adansonia, sér. 3, 42(1): 1–88. https://doi.org/10.5252/adansonia2020v42a1 Johnson DM, Murray NA (2022) Xylopia. In Johnson DM, Murray NA (Coords), Annonaceae. In: Chayamarit K, Balslev H (Eds) Flora of Thailand, Volume 16, Part 1. The Forest Herbarium, Department of National Parks, Wildlife and Plant Conservation, Bangkok, 324–332. Johnson DM, Murray NA (2023) A contribution to the systematics of Xylopia (Annonaceae) in the New Guinea region. Gardens’ Bulletin Singapore 75(2): 183–255. https://doi.org/10.26492/ gbs75(2).2023-02 Jordano P (2016) What is long-distance dispersal? And a taxonomy of dispersal events. Journal of Ecology 105: 75–84. https://doi. org/10.1111/1365-2745.12690 Joyce EM, Thiele KR, Ferry Slik JW, Crayn DM (2021) Plants will cross the lines: climate and available land mass are the major determinants of phytogeographical patterns in the Sunda–Sahul Convergence Zone. Biological Journal of the Linnean Society 132: 374–387. https://doi.org/10.1093/biolinnean/blaa194 Kaiser-Bunbury CN, Memmott J, Müller CB (2009) Community structure of pollination webs of Mauritian heathland habitats. Perspectives in Plant Ecology, Evolution and Systematics 11(4): 241–254. https://doi.org/10.1016/j.ppees.2009.04.001 Kanmegne G, Mbibong Fotso DA, Omokolo DN (2017) Effects of substrates, different pretreatment protocols and dehydration on the induction of seeds germination of Xylopia aethiopica (Dunal) A. Rich. International Journal of Biological and Chemical Sciences 11(2): 597–608. https://doi.org/10.4314/ijbcs.v11i2.6 Kays R, Jansen PA, Knecht EMH, Vohwinkel R, Wikelski M (2011) The effect of feeding time on dispersal of Virola seeds by toucans determined from GPS tracking and accelerometers. Acta Oecologica 37(2011): 625–631. https://doi.org/10.1016/j.actao.2011.06.007 Keppel G, Nge FJ, Ibanez T (2023) Slowing taxon cycle can explain biodiversity patterns on islands: Insights into the biogeography of the tropical South Pacific from molecular data. Journal of Systematics and Evolution 62: 201–214. https://doi.org/10.1111/ jse.13026 Kingston N, Waldren S, Bradley U (2003) The phytogeographical affinities of the Pitcairn Islands: A model for south-eastern Polynesia? Journal of Biogeography 30(9): 1311–1328. https://doi. org/10.1046/j.1365-2699.2003.00959.x Kitamura S, Thong-Aree S, Madsri S, Poonswad P (2011) Characteristics of hornbill-dispersed fruits in lowland dipterocarp forests of southern Thailand. The Raffles Bulletin of Zoology, Supplement No. 24: 137–147. Kitamura S, Poonswad P (2013) Nutmeg-vertebrate interactions in the Asia-Pacific region: importance of frugivores for seed dispersal in Myristicaceae. Tropical Conservation Science 6(5): 608– 636. https://doi.org/10.1177/194008291300600503 Kleyheeg E, Fiedler W, Safi K, Waldenström J, Wikelski M, van Toor ML (2019) A comprehensive model for the quantitative estimation of seed dispersal by migratory mallards. Frontiers in Ecology and Evolution 7: 40. https://doi.org/10.3389/fevo.2019.00040 Koné I, Lambert JE, Refisch J, Bakayoko A (2008) Primate seed dispersal and its potential role in maintaining useful tree species in the Ta region, Cte-d’Ivoire: implications for the conservation of forest fragments. Tropical Conservation Science 1(3): 293–306. https://doi.org/10.1177/194008290800100309 Kuhnhäuser BG, Bates CD, Dransfield J, Geri C, Henderson A, Julia S, Lim JY, Morley RJ, Rustiami H, Rowan J, Schley RJ, Bellot S, Chomicki G, Eiserhardt WL, Hiscock SJ, Baker WJ (2025) Island geography drives evolution of rattan palms in tropical Asian rainforests. Science 387: 1204–1209. https://doi.org/10.1126/ science.adp3437