Mammalian diversification bursts and biotic turnovers are synchronous with Cenozoic geoclimatic events in Asia

Significance

We provide a temporal framework on the relative roles of in situ speciation and historical colonization events in shaping mammal assemblages across Asia. About one-quarter of all colonization events initiated in the tropical forests of Southern Asia, which represents the primary source of mammal lineages. In contrast, our fauna-wide approach reveals that mountain hotspots surrounding the Qinghai–Tibetan plateau acted mainly as accumulation centers rather than as centers of diversification. We further show that the evolution of Asia’s mammal faunas was temporally congruent with other groups and was triggered by common geoclimatic events such as tectonic continental collisions and mountain uplift.

Abstract

Asia’s rich species diversity has been linked to its Cenozoic geodiversity, including active mountain building and dramatic climatic changes. However, prior studies on the diversification and assembly of Asian faunas have been derived mainly from analyses at taxonomic or geographic scales too limited to offer a comprehensive view of this complex region’s biotic evolution. Here, using the class Mammalia, we built historical biogeographic models drawn on phylogenies of 1,543 species occurring across Asia to investigate how and when the mammal diversity in Asian regions and mountain hotspots was assembled. We explore the roles of in situ speciation, colonization, and vicariance and geoclimatic events to explain the buildup of Asia’s regional mammal diversity through time. We found that southern Asia has served as the main cradle of Asia’s mammal diversity. Present-day species richness in other regions is mainly derived from colonization, but by the Miocene, in situ speciation increased in importance. The high biodiversity present in the mountain hotspots (Himalayas and Hengduan) that flank the Qinghai–Tibetan plateau is a product of high colonization instead of in situ speciation, making them important centers of lineage accumulation. Overall, Neogene was marked by great diversification and migrations across Asia and surrounding continents but Paleogene environments already hosted rich mammal assemblages. Our study revealed that synchronous diversification bursts and biotic turnovers are temporally associated with tectonic events (mountain building, continental collisions) and drastic reorganization of climate (aridification of Asian interior, intensification of Asian monsoons, sea retreat) that took place throughout the Cenozoic in Asia.

Areas with exceptional diversity and a high proportion of endemic species are scattered across the globe (1). While numerous evolutionary mechanisms (e.g., adaptive radiation, allopatric speciation, niche conservatism/evolution, sympatric speciation) may explain uneven spatial diversity, regional biodiversity ultimately represents the interaction between in situ speciation, extinction, and colonization (2). These three facets mold the species pool in any given region. Quantifying their relative importance informs us whether a region has mainly served as a center of species accumulation, where the biota assemblage is chiefly derived from immigration of lineages that originated in other regions, or a cradle of diversification, where in situ speciation has been the primary mode of faunal assembly. The relative strength of these processes may vary over time, so a center formed initially by species accumulation might turn into a cradle of diversification, as on large islands or when triggered by marked geo-climatic changes (3, 4). Uncovering the dynamics of speciation–colonization over time and relating them to paleoenvironmental events can elucidate patterns of biotic assembly and explain the formation of present-day biodiversity hotspots (2).

Asia is home to a great many species and it includes 14 of the globe’s 36 biodiversity hotspots (1). It is also the birthplace of many extant mammalian orders (5, 6). Throughout the Cenozoic, Asia has sustained active mountain building and dramatic climatic change, resulting both from intracontinental geodynamic events (7–9) and continental collisions, such as the collision of Indian-Asian tectonic plates during the Eocene (7, 10). Together these events prompted the formation of the highest plateau on earth, the Qinghai–Tibetan plateau (QTP), and the surrounding mountains (hereafter QTP region), including two of the world’s biodiversity hotspots –the Himalayas along the south and the Hengduan Mountains to the east. These large mountain systems have distinct geological histories, with idiosyncratic uplift phases (7), and are surrounded by different vegetative formations. The Himalayas and the Hengduan harbor high diversity even in comparison with surrounding regions (Fig. 1) and have been recognized as important centers of species accumulation and diversification (11, 12).

Fig. 1.

Biogeographic units and mammal richness in Asia. (A) Map showing the limits of the five continents (Inset) and the seven Asian regions used in this study. (B) Species-accumulation curves for each Asian region based on 1,000 rarefaction samples. (C) Temporal accumulation of mammal species (in a log scale) per region. Lines represent the averaged distribution over 2,000 biogeographic scenarios. Pli., Pliocene; Q., Quaternary; QTP, the Qinghai–Tibetan Plateau.

Asia’s modern biodiversity should have responded to its dynamic Cenozoic geodiversity either directly as novel habitats appeared and dispersal became limited or indirectly in response to shifting climates (8, 13, 14). Well-dated Cenozoic fossil records offer important but fragmentary pieces of information, insufficient themselves to draw a full panorama of historic events that shaped present-day faunas across Asia (15, 16). Part of this limitation has been recently overcome with broad phylogenetic-based range reconstruction models that shed light on the evolution of Asian floras (3, 11, 17). Clear signatures of diversification bursts linked to orogenic events and climate shifts have been uncovered and reveal a Paleogene rather than Neogene origination of the modern Asian flora (18). By contrast, the origin and timing of Asian faunas have been derived mainly from taxonomically restricted analyses and limited to one or few biodiversity hotspots (e.g., refs. 12, 19, and 20), which lack the taxonomic and geographic scale needed for a comprehensive synthesis. Because of their fossil records, mammals provide the means to test and refine reconstructions of biota assembly.

Here, we assess the tempo and mode of range evolution of Asian mammals to infer how and when the diversity across regions and mountain hotspots assembled. In particular, we investigate whether periods of intense tectonism and marked climatic changes are related to peaks in diversification and major dispersal events shaping the present-day regional mammal diversity in Asia. We base our conclusions on a suite of 2,000 historical biogeographic models drawn on the phylogenies of 1,543 species distributed across Asia and that account for both phylogenetic and range estimation uncertainty. The inferred biogeographic events were used to estimate the timing and amount of in situ speciation, vicariance, colonization, and directionality of dispersal across regions, which allow us to identify those acting as an accumulation center, a cradle of diversification, or both. Our findings were connected with geological (tectonic collision, mountain building) and climatological (global cooling, Asian monsoons, aridification, sea retreat) events and integrated with the available paleontological evidence.

Results

Formation of Asian Assemblages.

Our historical biogeographic reconstructions recover in situ speciation as the main event (53.1% of all events; SI Appendix, Table S1) that shaped present-day mammal diversity in Asia. However, 85% of all in situ speciation occurred within southern Asia (SA) (Table 1 and Fig. 2), reflecting that region’s high richness and endemism. We uncovered an earlier and steady rise in species assembly in SA that began in the late Cretaceous to the mid-Eocene followed by a slowing trend until late Paleogene, and then another rapid rise (Fig. 1C). Colonization also played an important role in the formation of Asian fauna assemblages (21% of all biogeographic events), and was highly asymmetric across regions (Fig. 3 and SI Appendix, Figs. S1–S13). Overall dispersal was limited during the Paleogene but greatly increased from Miocene onward (Fig. 3). SA was the primary source of mammal lineages, initiating 27% of all dispersals, at least three times higher than any other Asian region (Fig. 3 and Table 1). We documented the onset of mammal dispersal out of SA around early Eocene mainly toward Africa (Fig. 3 and SI Appendix, Figs. S1 and S2). From Miocene onward, emigration from SA greatly increased, especially toward the Himalayas and the Hengduan. On the other hand, dispersal into SA was very limited (13.3%, the lowest after QTP and northern Asia (NA)) and increases mainly in the last 10 Ma (Fig. 2B). Interestingly, we found that most of the dispersal toward SA derived from Africa (25%), Oceania (20%), the Himalayas (13%), and the Hengduan (11%) (Table 1).

Table 1.

Summary of in situ speciation and exchange events of mammals involving the seven Asian regions and continents

Source	Recipient											Total	%
	SA	EA	NA	WA	Him	Heng	QTP	Afr	Ame	Oce	Eur
Southern Asia	1670.3	38.5	3.2	16.3	65.3	55.2	3.3	17.8	4.4	25.1	5.7			234.8	27.2
Eastern Asia	10.9	69.5	11.3	2.6	5.1	15.7	2.2	0.7	2.9	0.2	21.9			73.7	8.5
Northern Asia	1.5	11.7	51.1	18.8	4.2	3.6	7.1	1.2	0.9	0.04	20.7			69.7	8.1
Western Asia	5.2	3.1	25.1	104.5	6.2	1.7	2.6	11.6	0.7	0.1	22.2			78.6	9.1
Himalayas	12.8	4.6	4.5	4.1	34	12.9	9.7	1.4	1.2	0.1	2.3			53.9	6.2
Hengduan	11.4	10.7	2.9	1.9	10.2	17.1	6.3	0.7	0.8	0.05	1.2			46.3	5.4
QTP	0.8	1.4	5.1	1.2	7.5	6.7	10.8	0.2	0.6	0.1	0.95			24.4	2.8
Africa	25.7	3.7	4.1	33.5	3.8	2	1.3	.	1.1	0.9	6.02			82.1	9.5
America	7.8	8.5	4.8	3.4	3.1	2	1.5	1.3	.	0.1	5.7			38.3	4.4
Oceania	20.1	1.5	0.3	0.6	0.9	0.2	0.1	0.6	0.1	.	0.11			24.5	2.8
Europe	4.9	29.4	38.2	46	5.2	4.1	3.8	2.4	3.6	0.1	.			137.8	15.9
Total	101.1	113.1	99.5	128.4	111.5	104.1	37.9	37.9	16.3	26.8	86.8
%	13.3	14.8	13.1	16.8	14.6	13.7	5	5	2.1	3.5	11.4

Numbers averaged over 2,000 biogeographic scenarios account for phylogenetic and range reconstruction uncertainty. The source areas (origin of dispersal events) are given as rows and the recipient areas as columns. Bold numbers along the diagonal represent the estimated average number of in situ speciation per Asian region. Afr, Africa; Ame, America; EA, eastern Asia; Eur, Europe; Heng, Hengduan; Him, Himalayas; NA, northern Asia; SA, southern Asia; WA, western Asia;

Note that numbers representing non-Asian intercontinental dispersal events are likely underestimated given our biogeographical analyses focused on lineages from Asia.

Fig. 2.

Evolution of regional mammal diversity in Asia. (A) Schematic representation of the main geoclimatic events in Asia related to the regional assembly of mammal species. Timing of the events based on (3, 15, 21-24). The profiles of the QTP (blue), the Himalayas (orange) and the Hengduan (purple) at the bottom are snapshots of their approximate elevation at mid-Eocene and mid-Miocene, respectively. India-Asia plate tectonic map was obtained from ODSN Plate Tectonic Reconstruction Service (http://www.odsn.de/odsn). (B) Temporal in situ speciation, immigration, and emigration trend averaged over 2,000 biogeographic scenarios and clustered by 5-Ma bins per region. Shaded region indicates 5–95% quantile intervals. Cret., Cretaceous; Pli., Pliocene; Q., Quaternary; QTP, the Qinghai–Tibetan Plateau.

Fig. 3.

Historic mammal faunal exchanges between Asian regions and other continents. (A) Proportion of historical dispersions across regions during the Paleogene, Neogene, Quaternary, and total based on the average of 2,000 range reconstruction scenarios. Color of lines relates to the source region and line thickness is proportional to the number of dispersal events. Only connections involving Asian regions and above 50th quantiles are shown. See SI Appendix, Fig. S13 for the full connections. (B) Estimated number of immigration (Top bars) and emigration (Bottom bars) events per Asian region, revealing the main recipient and source regions, respectively. S, southern Asia; E, eastern Asia; N, northern Asia; W, western Asia; H, Himalayas; D, Hengduan; Q, Qinghai–Tibetan Plateau; F, Africa; M, America; O, Oceania; U, Europe.

Present-day mammalian assemblages in other Asian regions began to rise in the middle Paleogene (Fig. 1C). In contrast to SA, the species diversity of other regions were mainly derived from colonization events (Fig. 2B and Table 1). We inferred colonization events in NA, eastern Asia (EA), and western Asia (WA) to be 1.9, 1.6, and 1.2 times higher than their in situ speciation rates, respectively. Nevertheless, the importance of in situ speciation on faunal assembly varies across regions and over time (Fig. 2B).

After a steady mid-late Eocene accumulation of mammals in WA derived mainly from African and European immigrants (Figs. 1C and 3 and SI Appendix, Fig. S1), we detected the onset of in situ diversification after the Eocene–Oligocene transition (EOT). Diversification rose in the mid-Oligocene and burst in the early–mid Miocene, when it became as important as colonization as the primary mode of faunal assembly (Fig. 2B), indicating a shift from a strict accumulation center to a joint accumulation–diversification center. We also detected an increase in emigrations out of WA by mid-Miocene, pumping novel lineages mainly into NA (32%), Europe (28%), and Africa (15%) (Fig. 3, Table 1, and SI Appendix, Fig. S3). The origin and diversification of NA and EA mammals show somewhat distinct trajectories. Both in situ speciation and colonization played a similar role in building mammal assemblage in these areas, especially during the Paleogene, but from early to late Miocene, in situ speciation increased and slightly overtook colonization as the main mode of species accumulation (Fig. 2B). Interestingly, in the last 10 Ma, we detected a further burst of immigration derived mainly from Europe (39%) and West Asia (25%) toward NA and from SA (34%), Europe (26%), North Asia (10%), and the Hengduan (9%) to EA (Fig. 3).

The Build-Up of Mammal Faunas of the QTP Region.

The high species richness now observed in the Himalayas and the Hengduan is primarily related to high colonization, mainly from SA (Fig. 3 and Table 1). This indicates that these montane hotspots have served as important centers of accumulation. Remarkably considering their small geographic areas, the Himalayas and the Hengduan represent the biggest “sink” regions in Asia, exhibiting colonization rates three and six times higher than in situ speciation, respectively (Table 1). In situ speciation in the Himalayas began to rise from the early Miocene, reaching high levels by mid-late Miocene (Fig. 2B). In the Hengduan, a first small peak of in situ speciation appeared much earlier by late Eocene (~35–40 Mya) and a rapid rise of diversification occurred around the early and mid-late Miocene (Fig. 2B). Importantly, colonization has remained the dominant mode of species accumulation over time for these two montane hotspots (Fig. 2B).

Hosting the lowest present-day mammal diversity in Asia (Fig. 1B), the QTP has the lowest inferred in situ speciation and colonization events (Table 1). In situ speciation in the QTP is concentrated around the mid-Miocene (Fig. 2B), which coincides with the rise in emigrations (SI Appendix, Fig. S8). Dispersal routes out of the QTP were mainly toward the Himalayas (30.6%), the Hengduan (27%), and NA (21%) (Table 1 and SI Appendix, Fig. S8).

Continental Faunal Interchanges and Cladogenetic Events.

In addition to in situ speciation, sympatric-subset speciation (where one daughter inherits the original ancestral widespread range and the other daughter inherits only a subset of the ancestral range (21)) and vicariance have also played important roles in the biogeography of Asian mammals (SI Appendix, Table S1). Most of these cladogenetic events resulted from the division of one binary ancestral area into two, and the most common vicariance involved the split of lineages originally occurring in an Asian region and one of the surrounding continents. Lineage splits involving Africa and SA, Oceania and SA, and Africa and WA were the most common (SI Appendix, Tables S2–S4 and Figs. S14 and S15). We also recorded a high number of cladogenetic events between SA and the two montane hotspots. Interestingly, the first evidence of vicariance dates to middle Eocene involving Africa and SA, while those involving Oceania began only at early Miocene; both SA-Himalayas and SA-Hengduan splits rise from mid-Miocene onward (SI Appendix, Figs. S14 and S15). Our models show that periods of increasing intercontinental faunal dispersal were followed by rising cladogenetic events.

Discussion

Our study provides a comprehensive picture of the origin and range evolution of the Asian mammalian fauna and how it was assembled over time. We established a temporal framework on the relative roles of in situ speciation and historical colonization events in shaping mammal assemblages across Asia.

The inferred timing and direction of colonizations revealed SA has consistently served as a major center of diversification, being the main cradle of mammal diversity in Asia. Indeed, we found that by the early Paleogene SA tropical forests already hosted a rich mammal fauna, in agreement with available fossil records (5, 22, 23). These results mirror those observed in the Neotropics, where the Amazon tropical forest acted as the main source of vertebrate diversity across the continent (24). Generally speaking, hyperdiverse regions are likely to pump new lineages into less diverse regions (25). Accordingly, the least diverse region, the QTP, provided the fewest immigrants (Figs. 1 and 3 and Table 1). However, unlike the Neotropics, dispersal into SA is very limited. Such strong dispersal asymmetry can be explained by the earlier SA biota formation. As niches were already occupied, incumbency would limit the establishment of immigrants (25, 26). Under this scenario, increased dispersal into SA since the late Miocene may reflect the contraction of tropical forests caused by global cooling after the Mid-Miocene Climatic Optimum (MMCO) (27) and the increasing diversification of the surrounding regions and continents. Thus, the mammal fauna of SA has been largely derived from in situ diversification with limited influences from other Asian regions. However, the intraregional evolution of the SA biota, comprising a complex of subregions with their own asynchronous assemblies, clearly reflects distinct cycles of migrations within southeast Asian archipelagos related to global sea-level changes and intermittent land bridges, whose dynamics are a fitting subject for additional studies (8, 28).

Mammal assemblages in non-SA regions date from mid-late Eocene, which accords with the fossil evidence showing large extinctions of archaic mammal lineages and the appearance of most modern families by that time (29). Our findings are also highly congruent with geoclimatic events that took place in Asia through the Cenozoic. In the Eocene, forest mammal faunas across Asia were little differentiated (23), indicating widespread faunal continuity over large distances and reduced in situ speciation outside of SA. In contrast, Oligocene mammal assemblages from northern and western Asia had become associated with open arid environments and quite distinct from the forest fauna inhabiting southern Asia (6, 23). The turnover from widespread Eocene forest-associated fauna (dominated by large-sized Perissodactyla) to Oligocene arid-typical faunal elements (dominated by small-sized Glires) is known as the Mongolian Remodeling and is clearly marked in both floral and faunal fossil assemblages from west and north Asia (6, 23, 30–32). Such discontinuity in distribution is also imprinted in our range models, in which dispersal from SA toward NA and WA is limited over the Oligocene, while toward EA, the Hengduan and the Himalayas followed a continuous upward trend since the mid-late Paleogene (SI Appendix, Fig. S2). The EOT represents the transition from a greenhouse to an icehouse world associated with global cooling. Together with the Paratethys retreat, this led to aridification of Asia’s interior as documented in paleobiological records and climate models (15, 33–35) (Fig. 2A). From mid-Miocene onward, aridification strengthened in the Asian interior driven by global cooling (especially after MMCO), Tethys final closure, and the rapid orogeny of the QTP–Himalayas, blocking delivery of moist Indian air (rain shadow effect) (36–39). Aridification resulted in the burst diversification of desert-associated and grassland-associated mammals (15, 34) observed both in NA and WA assemblages in the mid-Miocene (Fig. 2B), and an increase in faunal interchanges between these areas and emigrations from them (Fig. 3, Table 1, and SI Appendix, Figs. S3 and S4).

The evolutionary trajectory of EA mammals mirrors the evolution of the temperate East Asia flora (EAF), in which most modern lineages appeared during the Miocene whereas its Paleogene-aged genera were mainly immigrants from other regions (40). Such resemblance points to common geologic events triggering the synchronous evolution of East Asian fauna and flora. By mid-Miocene, the Himalayas had reached an elevation similar to that of the present, projecting above the QTP (Fig. 2A) and forming an effective barrier to northward moist winds from the Indian Ocean (38). Consequently, the remodeling of atmospheric circulation coupled with the global cooling and the retreat of Tethys increased rainfall seasonality in East Asia (intensifying Asia’s monsoons) and favored the expansion of temperate forest environments and associated biota (37, 40). This is temporally congruent with the rise of vicariant events between clades originally distributed in SA and EA, indicating the formation of a distinct nontropical fauna in east Asia (SI Appendix, Fig. S14).

The mountain hotspots surrounding the QTP had up to six times higher immigration rates than in situ speciation, showing they have acted mainly as accumulation centers. This agrees with findings in birds (26) but challenges an earlier view that the Hengduan has served mainly as a cradle of biotic evolution and is contrary to floral studies, which show that in situ speciation became the primary mode of species accumulation in the Hengduan by mid-Miocene (11). At least for mammals, these hotspots have acted mainly as accumulation centers overtime. Nevertheless, we uncovered clear signatures of lineage diversification associated with uplift phases. The first evidence of in situ diversification in the Hengduan around ~35 Ma is consistent with the recent paleo-evidence that by the mid-Eocene, part of the Hengduan had already reached near-present elevations (41, 42). Similarly, the rise of in situ speciation in the Himalayas dates from the early Miocene when it had reached ~4-km elevation (3, 38). In both montane hotspots, mid-late Miocene is marked by high diversification rates, where in situ speciation approached colonization as the main mode of species accumulation (Fig. 2B) during a period of rapid mountain building when most of Himalayas and the Hengduan reached near present-day elevations (38, 42). Steep environmental changes and markedly heterogeneous vegetation zonation along mountain slopes would have promoted divergent selection and ultimately triggered speciation (13). This also coincides with the marked upward trend of vicariant events, splitting lineages previously distributed in the montane hotspots and in SA tropical forests (SI Appendix, Figs. S14 and S15).

In situ speciation in the QTP is concentrated around the mid-Miocene, which is also congruent with both fossil record and phylogenomic estimations of QTP-originated groups. For instance, pikas originated in the QTP ~15 Ma and rapidly diversified across the plateau before emigrating to lower elevations (43). Similarly, the earliest records of QTP endemic mammals are known from the mid-late Miocene, such as the Tibetan chiru and the muskox (16, 44). The formation of a high Tibetan flat plateau as observed today dates to mid-Miocene, resulting from the compression and sedimentation of a deep wet valley located among the Qiangtang and the Gangdese mountains (7) (Fig. 2A). In addition, in the mid-late Miocene, the >5-km Himalaya imposed a strong rain shadow effect on the core area of the QTP, resulting in drastic modification of its climate and vegetation (7, 39). Collectively, these events seem to have boosted the onset of modern mammal diversification within the QTP, leading to the appearance of taxa adapted to extremely high elevations (16). Our results offer support to the “out-of-Tibet” hypothesis, which postulates that cold-adapted lineages originated in the QTP and dispersed toward NA and from there colonized Europe and America (43, 44).

Our study also revealed important faunal interchanges between Asia and other continents synchronously with tectonic and climatic events. The rise in dispersal during the mid-Eocene between Africa and both SA and WA (SI Appendix, Figs. S2, S3, and S9) coincides with the collision of Indian-Asian plates (10) which likely facilitated exchanges of African-originated lineages between the Indian subcontinent and Asia mainland (27). We also detected a further burst of dispersal between Africa and WA by mid-Miocene, which coincides with the Arabian–Asian continental collision that formed a land bridge connecting the two continents (45). Additionally, the initial Australian–Asian collision during the early Miocene (46) is consistent with a rapid increase in dispersal and vicariant events involving Oceania and SA (SI Appendix, Figs. S2, S11, and S16). In addition to tectonic events, changes in climatic, sea level, and vegetation configuration have also fostered intercontinental dispersals (29, 47). For example, mammal interchanges between Europe and Asia increased after the EOT (SI Appendix, Figs. S1–S5 and S12), as much of the European continent was isolated from Asia during the Eocene and after EOT, a marked drop in sea level allowed the establishment of land connections and greater terrestrial interchange. These events coincided with periods of marked faunal turnover between Asia and the surrounding continents, which are corroborated by the fossil record (29, 47). Furthermore, our historical biogeographic reconstructions show that elevated dispersal between continents is associated with bursts in cladogenetic events, suggesting that species diversify rapidly after invading a new continent. This same reasoning may explain the much higher diversification involving the Himalayas and the Hengduan mountains during the mid-Miocene, a period of active uplift and increasing immigration, where populations were subjected to strongly contrasting habitats over a short spatial scale. These findings reveal that periods of active tectonic events during the Cenozoic have amplified the regional diversity of Asian mammals.

In summary, the active tectonic movement and mountain building in the Cenozoic of Asia coupled with striking changes in the seasonal atmospheric circulation under global cooling left notable imprints on the evolutionary history of Asian biota as documented here for mammals and by previous studies on other groups (3, 8, 11, 15, 20). Synchronous diversification bursts and faunal/floral turnovers are temporally associated with tectonic events and drastic climate reorganizations that took place in Asia throughout the Cenozoic (8, 30). Our findings, in agreement with the available fossil record, reinforce that the Neogene was a period of marked diversification and migrations across Asia leading to the formation of the modern biota (47), but Paleogene environments were already populated by rich and diversified mammal assemblages.

Materials and Methods

Species Distribution and Phylogenetic Data.

Mammal distribution data were obtained from IUCN range maps and complemented with records available in public databases, museum collections, and literature (SI Appendix). The mammal phylogenies used in this study were based on (48) and randomly drawn from Vertlife (SI Appendix for further details). The pool of Asian mammals included in the analyses comprised 1,543 species (Dataset S1 and SI Appendix, Fig. S17).

Biogeographic Units.

We defined seven biogeographic units in Asia representing four geographic areas (SA, EA, NA, and WA), two montane hotspots (Himalayas and Hengduan), and the QTP (Fig. 1A). The SA unit comprises the modern tropical forests of Asia including the India subcontinent, south China, southeast Asia, and the Malay Archipelago. Despite its complex geological history (28), the subunits that composed SA still share important biotic characteristics. For example, tropical forests have been the dominant biome over extended timescales, and forest communities share remarkable similarities across this region (49, 50). It includes five of the world’s biodiversity hotspots (1): Indo-Burma, Sundaland, Philippine, Wallacea, and Western Ghats and Sri Lanka. The EA unit comprises most of the temperate broadleaf and coniferous forest of the eastern portion of the Palearctic realm including Japan’s biodiversity hotspot. The NA unit is delimited on the south by the QTP, at north by Europe, and its western limit coincides with western range of the mountains of central-Asia biodiversity hotspot. It is currently characterized by lowland deserts and temperate grasslands. The WA region includes southwest Asia and the Arabian Peninsula and is characterized mostly by deserts but also includes a small portion of the Mediterranean forests. It comprises five biodiversity hotspots: Irano-Anatolian, Caucasus, and part of the Mediterranean basin, Horn of Africa, and Eastern Afromontane. Limits of the two mountain hotspots were based on (51), and for the QTP limits we used the core area (i.e., reducing 50 km buffer) of the 3-km boundary defined by (52). All maps were projected to an equal-area Behrman projection. See SI Appendix for further information.

Species Rarefaction Curve.

To assess present-day mammal richness across the seven Asian regions so defined while accounting for the difference in area, we built species rarefaction sample-based curves with 1,000 permutations and sites added in random order using the function accumcomp of Biodiversity R package (53). Because of the narrow and irregular shape of the Hengduan hotspot, we first created a buffer of 50 km around each hotspot and defined grids with 50-km resolution as the scaling factor. For this analysis, we used the IUCN range maps to calculate the total number of mammal species per grid.

Historical Range Reconstruction Analysis.

We estimated the ancestral range evolution of mammals in Asia using the BioGeoBEARS R package (54). Species were first assigned to the seven Asian regions so defined and four continents (Americas, Africa, Oceania, and Europe) (Fig. 1A). The continent limits were obtained from Natural Earth (http://naturalearthdata.com). Following this classification, Europe includes the Siberia region, but we considered Papua as part of Oceania and combined South and North America (see Fig. 1A). To avoid including species only marginally distributed in an Asian region, only species with at least 20% of its range in that region were deemed present. While this may reduce the regional pool, it provides a more realistic picture of the core mammal fauna present in each Asian region and avoids including species inhabiting only ecotones (Dataset S1).

To estimate ancestral range evolution, we first selected 21 clades composed mainly of Asian species using the speciesgeocodeR package (55). In total the 21 clades comprise 3,114 mammal species, of which 1,543 occur in Asia, representing ~75% of the continent’s mammal fauna (see SI Appendix, Table S5 and Fig. S17). For each of the clades, we compared the likelihood of four biogeographical models: i) a dispersal–extinction–cladogenesis (DEC) unconstrained model, in which dispersal among areas have similar probabilities; ii) DEC unconstrained model with founder-event speciation (DEC + j); iii) a likelihood-version of dispersal-vicariance analysis (DIVALIKE) model; and vi) a likelihood-version of BayArea (BAYERLIKE) model (21). Models were ranked by Akaike information criterion (AICc) and Akaike weight. We deliberately avoided models with distinct temporal constraint probabilities (SI Appendix for further details), and so our models require fewer assumptions and the results do not reflect any prior hypotheses on geodynamic effects on mammal evolution (11).

Using the parameters from the best-fit biogeographical model (SI Appendix, Table S6), we performed a set of 20 stochastic mappings over 100 randomly drawn phylogenies. This results in 2,000 possible range reconstruction histories for each clade that simultaneously account for both phylogeny and range estimation uncertainty (56, 57) (SI Appendix). The averaged biogeographical events were then used to estimate the number of in situ speciation, vicariance, and dispersal events, and to assess the directionality of dispersals between regions over time. Dispersals between widespread regions were quantified for each region involved (24). Temporal trends of biogeographic events were quantified based on 5-Ma bins.

Regional Temporal Richness Accumulation.

We explored temporal trends in species accumulation for each region using lineage-through-space-time plots (LTST) (58) based on the historical range models. LTST is an extension of traditional lineage-through-time plots and uses the estimated ancestral ranges derived from the previous analysis to reconstruct the temporal diversification associated with each region separately (58).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (CSV)

Data, Materials, and Softwre Availability

Phylogenies used in this study were obtained from https://vertlife.org/phylosubsets/ and are available on Zenodo (https://zenodo.org/record/6961471) together with input files to perform ancestral range evolution models (59). We used the script provided by ref. 57 to perform biogeographic stochastic maps over a sample of trees.