Out of India, thrice: diversification of Asian forest scorpions reveals three colonizations of Southeast Asia

This study investigates whether the diversification and present-day distributions of Asian forest scorpions (Scorpionidae: Heterometrinae) are best explained by the Out of India hypothesis and, if so, when and how many times dispersal into Southeast Asia occurred. The context is a long-standing debate over the timing and mechanics of India’s northward drift and collision with Eurasia and the role of these geotectonic events in shaping South and Southeast Asian biotas. Prior work across plants, fishes, amphibians, reptiles, birds, and arthropods has often supported an Indian Gondwanan origin followed by colonization of Southeast Asia after collision, though some taxa show Into India patterns. Geological reconstructions variously posit a single early collision around 55–50 Ma or multiple contacts culminating in a final collision between ca. 35–20 Ma. Additional complexity arises from end-Cretaceous Deccan Traps volcanism and the Chicxulub impact, with proposed refugia in India’s Western Ghats and Sri Lanka. Within Scorpionidae, previous qualitative hypotheses suggested a Gondwanan origin and dispersal via India or Indochina, but lacked quantitative temporal and spatial analyses. The purpose of this work is to provide the first time-calibrated phylogeny, ancestral range estimation, and diversification analyses for Heterometrinae to reconstruct their origins, dispersal routes and timing, and subsequent diversification across South and Southeast Asia, thereby clarifying the applicability and complexity of the Out of India scenario.

Multiple lines of literature inform the study: (1) Biogeography of South/Southeast Asian taxa—numerous studies across plants, fishes, amphibians (e.g., rhacophorids), reptiles, birds, and arthropods have invoked an Out of India origin, with fewer examples supporting Into India colonization. (2) India–Eurasia collision timing—debates contrast an early, single collision at ca. 55–50 Ma versus models of repeated contacts from ~55 Ma with final collision 35–20 Ma. Proposed scenarios include India brushing Sumatra and contacting the Dazhuqu intra-oceanic arc (Aitchison & Ali) versus an initial collision of a Tibetan Himalayan microcontinent with Eurasia and later Indian collision (van Hinsbergen et al.). (3) End-Cretaceous events—Deccan Traps volcanism (ca. 67–66 Ma) and the Chicxulub impact likely drove environmental stress and extinctions; Western Ghats and Sri Lanka have been posited as refugia where several lineages persisted. (4) Scorpionidae systematics/biogeography—earlier hypotheses by Couzijn posited Gondwanan origins with lineages carried on Indian and Indochinese blocks and later exchanges via the Assam gateway; subsequent morphological and molecular studies recovered Scorpionidae monophyly with Asian Heterometrinae sister to African Pandininae, implying divergence linked to India–Africa separation. However, prior work lacked rigorous time calibration and quantitative ancestral range modeling. These bodies of literature frame alternative expectations for the timing and number of dispersal events and refugial dynamics tested here.

Taxon sampling included a reduced dataset of 35 terminals (one per species) representing all seven genera and 31 (76%) species of Heterometrinae, plus outgroups: Pandinus (Pandininae), Scorpio (Scorpioninae), Opistophthalmus (Opistophthalminae), with Nebo hierichonticus (Diplocentridae) as root. DNA was extracted from ethanol-preserved leg muscle using Qiagen DNEasy kits. Five loci were sequenced: mitochondrial COI, 16S rDNA, 12S rDNA, and nuclear 18S rDNA, 28S rDNA (COI and 18S in fragments; all loci bidirectionally). Sequences were aligned in MAFFT v7.429 using the ‘leavegappyregion’ option and L-INS-i; aligned lengths: 18S 1761 bp; 28S 515 bp; 12S 341 bp; 16S 493 bp; COI 1078 bp (total 4188 bp). Model selection by BIC in jModelTest: 18S K80; 28S K80+I+G; 12S HKY+I+G; 16S HKY+I+G; COI GTR+I+G. Divergence-time estimation was conducted in BEAST v1.10.4 with an uncorrelated lognormal relaxed clock, birth–death tree prior, loci partitioned with unlinked clocks and models, and Heterometrinae+Pandinus constrained monophyletic; ingroup topology constrained to a more extensive prior phylogeny. Normal clock priors were applied: COI μ=0.007 (σ=0.00146), 16S μ=0.005 (σ=0.00270) based on published scorpion rates; other loci used uniform priors (12S ucld.min=0.002, ucld.max=0.5; 18S/28S ucld.min=0.0001, ucld.max=0.01). Two independent MCMC runs of 500 million generations each, sampling every 10,000, were combined with 25% burn-in; ESS > 200 checked in Tracer v1.7.1; summary tree via TreeAnnotator. Ancestral range estimation used BioGeoBEARS v1.1.1 in R on the dated tree (Pandinus retained as sister; other outgroups removed). Six regions were defined: A Africa; B Western Ghats + Sri Lanka; C Greater Indian Subcontinent; D Indochina; E Sundaland; F Philippines. Maximum areas per species = 4. A time-stratified analysis used five intervals (0–35, 35–45, 45–57, 57–68, 68–120 Ma) with adjacency and dispersal multiplier matrices (adjacent=1; intermediate=0.5; remote≈1e-07). Models compared: DEC, DIVALIKE, BAYAREALIKE, each with/without founder-event speciation (j). Phylogenetic diversity (Faith’s PD) per region was computed with 10,000 randomizations (picante), and evolutionary distinctiveness (ED) per species using fair proportions and equal splits. Diversification analyses were run on the ultrametric ingroup tree using the R package laser: models included pure birth, birth–death, density-dependent exponential (DDX), density-dependent linear (DDL), and two-rate Yule; variable speciation (SPVAR) and variable extinction (EXVAR) also tested. AIC identified best-fit model. Net diversification rates were estimated under low extinction (ε=0) and high extinction (ε=0.9), assuming low (n=5) and high (n=50) missing species. The gamma statistic (Pybus & Harvey) tested temporal constancy, and an MCCR test assessed sampling effects across clade sizes. A lineage-through-time plot was generated (phytools) with 10,000 Yule simulations for comparison. BAMM v2.5.0 estimated speciation/extinction and detected rate shifts (4 chains, 10 million generations; priors via BAMMtools), with Bayes factors for shift significance.

• Divergence-time phylogeny: Fully dichotomous with high posterior support (most genera ≥0.99, intergeneric relationships ≥0.92) and congruent with broader taxon sampling.
• Ancestral range model selection: DEC+j best fit (InL −53.45; AIC 112.9; AICc 113.8), outperforming DEC, DIVALIKE, BAYAREALIKE and their +j variants.
• Ancestral areas per genus (DEC+j): Western Ghats + Sri Lanka (B) for Gigantometrus, Sahyadrimetrus, Srilankametrus; Greater Indian Subcontinent (C) for Chersonesometrus, Deccanometrus, Heterometrus, Javanimetrus. Oldest–youngest divergences (Ma): Chersonesometrus 21–10; Deccanometrus 26–24; Gigantometrus 10–; Heterometrus 56–17; Sahyadrimetrus 32–12; Srilankametrus 21–.
• Regional phylogenetic diversity (PD): Highest in Western Ghats + Sri Lanka (B) and Greater Indian Subcontinent (C), with B showing high observed PD (458.4) and C significantly higher than random expectation (z=4.539, p=0.001). Philippines (F) lowest PD.
• Evolutionary distinctiveness: Heterometrus glaucus exhibited the highest ED among species.
• Diversification dynamics: Two-rate Yule model best (AIC 97.274) over pure birth, birth–death, DDX, DDL, SPVAR, EXVAR. Net diversification rates ranged from 0.016 (ε=0.9, n=5 missing) to 0.040 (ε=0, n=50 missing). Gamma statistic γ = −2.08 (p=0.019) indicates early concentrated branching; MCCR suggests γ becomes non-significant with ≥28% missing taxa. BAMM identified fast early speciation and one distinct rate shift; LTT plot showed declining speciation through time.
• Temporal-biogeographic timeline: Asian Heterometrinae diverged from African Pandininae ca. 113 Ma; major split within Heterometrinae ca. 92 Ma into a largely Indian clade and a Southeast Asian Heterometrus clade. Evidence supports range contraction to refugia (Western Ghats and Sri Lanka) around the KT boundary. Three independent Out of India dispersals into Southeast Asia aligned with multistage India–Eurasia contacts: (1) ca. 56 Ma, ancestor of Heterometrus glaucus into Sundaland during India’s approach/brush against Sumatra; (2) ca. 44–45 Ma, ancestor of Javanimetrus into Sundaland; (3) ca. 37 Ma, ancestor of the remaining Heterometrus clade into Indochina and Sundaland. Post-collision, Heterometrus dispersed NW–SE (Indochina basal to Sundaland/Philippines lineages). Subsequently, Indian genera (Chersonesometrus, Deccanometrus, Gigantometrus) spread north across India, diversifying into increasingly arid, open habitats during Miocene climatic reorganization and monsoon intensification.

The analyses support a nuanced Out of India scenario for Asian forest scorpions, resolving both the number (three) and timing (ca. 56, 44–45, 37 Ma) of dispersal events into Southeast Asia, congruent with multistage India–Eurasia interactions. The sister relationship of Asian Heterometrinae to African Pandininae and divergence around 113 Ma align with prolonged Late Cretaceous biotic connectivity between Africa and India prior to full isolation. Ancestral range reconstructions and phylogenetic placement of Western Ghats/Sri Lanka endemics imply contraction into southern refugia near the KT boundary, consistent with impacts from Deccan volcanism and/or the Chicxulub event. Following India’s collision and environmental stabilization, Indian lineages recolonized and diversified across the subcontinent, tracking Miocene aridification and monsoon onset, with associated ecomorphological shifts for open habitats. These findings refute earlier qualitative models (e.g., Couzijn) that posited different landmass histories and relationships, and they argue against Eurogondwana explanations for Heterometrinae. Collectively, the results reconcile geological debates by aligning scorpion dispersal pulses with proposed phases of India’s proximity to Sumatra/Indochina and demonstrate that Southeast Asian colonization was not a single event but a sequence of founder-driven dispersals, followed by in situ diversification.

This study provides the first quantitative, time-calibrated reconstruction of Heterometrinae biogeography, revealing: (1) divergence from African Scorpionidae during the Cretaceous (~113 Ma) as India separated from Africa; (2) KT-associated range contraction to refugia in the Western Ghats and Sri Lanka; (3) three independent Out of India colonizations of Southeast Asia at ~56 Ma (Heterometrus glaucus lineage), ~45 Ma (Javanimetrus), and ~37 Ma (remaining Heterometrus), coincident with episodic India–Eurasia contacts; and (4) subsequent northward dispersal and diversification across India during Miocene climatic reorganization, with adaptation to more arid, open habitats. These conclusions refine the Out of India hypothesis for a major arachnid clade and align with patterns observed in other South/Southeast Asian taxa. Future work should expand taxon sampling to full species diversity, incorporate additional genomic data, and seek fossil constraints to further refine divergence times and test alternative dispersal scenarios under updated tectonic models.

• Fossil calibration: No known fossils for Heterometrinae necessitated reliance on relaxed molecular clocks and secondary rate priors (COI, 16S) from other scorpion families, introducing uncertainty in absolute node ages.
• Taxon sampling: The dated analyses included 31 of ~41 described heterometrine species (76%); MCCR results indicate diversification inferences (gamma) are sensitive to missing taxa beyond ~28%.
• Biogeographic model assumptions: Area definitions, time stratification, and dispersal multipliers are model-based approximations tied to debated tectonic reconstructions; alternative paleogeographies could shift inferred timings/routes.
• Founder-event (j) processes: Although DEC+j was best-supported, interpretation of j can be contentious and sensitive to sampling and area delimitation.
• Geological uncertainty: Ongoing debates about the timing and mode of India–Eurasia contact (e.g., arc collisions, microcontinents) limit precision in correlating biotic events with specific tectonic episodes.