Molecular exploration of fossil eggshell uncovers hidden lineage of giant extinct bird

Elephant birds (Aves: Aepyornithidae) were giant, flightless ratites of Madagascar that went extinct roughly a millennium ago. Their relationship to other birds was unresolved until genetic studies identified them as sister to New Zealand’s kiwi, reshaping understanding of avian diversification. However, biodiversity and intragroup relationships among elephant birds have been unstable for over 150 years because most species are known from few, incomplete Pleistocene–Holocene post-cranial remains from southern and central Madagascar. Morphology-based taxonomy generally recognized about eight species across two genera, but a recent morphometric re-evaluation proposed four species in three genera (Aepyornis, Mullerornis, and Vorombe). These morphological revisions are debated due to potential homoplasy and convergent evolution affecting skeletal characters, which can obscure species boundaries and phylogenetic relationships in extinct taxa. Ancient DNA and associated biomolecular approaches provide an alternative means to resolve these uncertainties by directly assessing genetic relationships, biogeography, and ecology from well-preserved substrates such as fossil eggshell.

Previous work established a sister relationship between elephant birds and kiwi using ancient DNA, revising the ratite phylogeny. Traditional taxonomy relied on limited and fragmentary post-cranial bones, leading to shifting species and genus delimitations. Morphometric analyses have recently reclassified elephant bird diversity, including the erection of Vorombe, but such morphology-based revisions are susceptible to convergent traits and do not always reflect evolutionary history. Prior molecular efforts, mostly from bone, provided limited intra-Madagascar resolution due to preservation issues and sparse sampling. Eggshell has emerged as a robust substrate preserving ancient DNA and proteins, enabling reconstructive phylogenetics, species delimitation, and palaeoecology. Comparative datasets from other palaeognaths (moa, rhea, emu, cassowary, kiwi) offer benchmarks for intra- and interspecific genetic distance thresholds (e.g., COI K2P distances) to contextualize elephant bird divergences. Stable isotope studies of elephant bird bones and lemur faunas in Madagascar inform biogeographic and ecological interpretations relevant to diet and habitat use. Together, these strands underscore the need for an integrated palaeogenomic and palaeoecological approach to elephant bird systematics.

Specimen collection and dating: Over 960 elephant bird eggshell fragments were collected from 291 localities across southwestern, central, and—for the first time—northern Madagascar. Twenty-one new radiocarbon dates indicate eggshell deposits span ~1290 ± 20 to at least ~610 years BP, broadly contemporaneous with dated skeletal material nearby. Eggshell thickness was measured (digital calipers; mean of four slices) to characterize morphotypes across regions (north: 197; south: 312; other southern localities: 241). Morphometrics and mass estimation: Using phylogenetically informed regressions between eggshell thickness and egg/bird mass derived from 65 bird species (Supplementary Note 9), the authors estimated egg and body masses corresponding to thickness morphotypes. Ancient DNA: DNA was extracted from 63 eggshells in dedicated aDNA facilities (Curtin University TRACE) following modified published protocols. Single-stranded library preparation (Gansauge & Meyer) was used. Mitochondrial enrichment employed custom baits based on Aepyornis and Mullerornis reference mitogenomes. Size selection (140–300 bp) and sequencing were performed on Illumina NextSeq. Bioinformatics: Reads were trimmed and quality-filtered with USEARCH; chimeras removed. Iterative mapping to a consensus elephant bird mitochondrial reference was performed in Geneious, with BLAST assignments and MEGAN-based taxonomic assessment. Consensus genomes used ≥4× coverage; mapDamage2.0 assessed aDNA authenticity. Phylogenetics and species delimitation: Twenty elephant bird mitogenomes (including four previously published from bone) plus outgroups were aligned (MAFFT/MUSCLE). Genes were partitioned (protein-coding by codon; rRNA by loops/stems); models selected with Modeltest/jModelTest. Trees were inferred with RAxML (ML with rapid bootstrapping) and Bayesian methods; support assessed via bootstrap and posterior probabilities. COI (596 bp) K2P distances within/between clades were computed (MEGA) and compared to other ratites; additional species delimitation analyses used a Geneious plugin. Mantel tests examined isolation-by-distance. Molecular dating: MCMCTree (PAML) with nine fossil calibrations and representative samples estimated divergence times; previously published nuclear data were incorporated. Palaeoproteomics: Eggshell proteins (XCA1, XCA2 keratin-like proteins) were extracted, digested (trypsin/elastase), and analyzed by LC-MS/MS on an Orbitrap (HF-X). Searches (PEAKS) included variable deamidation/oxidation; contaminants were filtered (cRAP). Amino acid substitutions diagnostic of clades were assessed. Micro-CT: Twenty eggshells across morphotypes were micro-CT scanned (Bruker) to quantify porosity, pore density, and pore volume over defined ROIs; statistics included one-way ANOVA and t-tests with Bonferroni correction. Stable isotopes: δ15N, δ18O, and δ13C from eggshell carbonate/organic fractions were measured (EA-IRMS). Diet sources were interpreted relative to C3/C4/CAM plant baselines by biome; some mixing model tools (IsoSource/ISOCOR) were referenced for diet proportion estimation. Data access: Mitogenomes deposited in GenBank (e.g., OP141790–OP141810); proteomics in PRIDE (PXD35725); code and data in Dryad.

- Extensive sampling: >960 eggshell fragments from 291 localities; 21 new radiocarbon dates constrain samples to late Holocene (~1290 ± 20 to ≥610 BP). - Eggshell morphotypes and inferred sizes: A thin morphotype (<1.5 mm), a thick morphotype (>3 mm; mean ~3.32 mm), and an intermediate northern morphotype (~1.95 mm). Mass estimates via phylogenetic regressions indicate: medium eggs ~3.18 kg laid by ~230 kg birds; thickest eggs up to ~1.07 kg egg mass (with very large body mass estimates for southern aepyornithids); northern/intermediate eggs ~3 kg egg mass with ~20–30 kg body size estimate (text juxtaposes body mass numbers inconsistently, but relative size patterns hold). - Mitogenomes and phylogeny: One near-complete (>14,000 bp; ~27× coverage) and four partial (>500 bp; ~3×) mitogenomes were recovered from eggshells; together with four published bone-derived genomes, they reveal four well-supported mitochondrial clades aligned with eggshell thickness and geography (north, central, south). Nodes received high ML bootstrap and Bayesian support. - Family-level split and taxonomy: Thin eggshells (<1.5 mm) form a monophyletic Mullerornis clade distinct from thicker aepyornithid eggshells (>1.5 mm) clustering with Aepyornis and Vorombe. COI K2P distance between Mullerornis and Aepyornis/Vorombe clades is ~21.9%, >10× within-clade distances and >3× typical between-genus within-family distances in ratites. Micro-CT porosity supports family-level separation: aepyornithid porosity > mullerornithid (p=0.032; pore density p=0.031; pore volume p=0.198). Proteomics also shows consistent amino acid differences (e.g., XCA1 residue 74: histidine in Mullerornis-like vs tyrosine in aepyornithids; additional differences in XCA2). Molecular dating places the Aepyornithidae–Mullerornithidae split at ~30 Ma (95% HPD 20.6–40.3 Ma). Authors support resurrecting Mullerornithidae for Mullerornis. - Low diversity within southern taxa: Mullerornis shows very low within-clade COI divergence (mean 0.27% ± 0.05%), matching within-species variation in other ratites. In southern Aepyornithidae, two purported genera (Aepyornis, Vorombe) do not show sufficient mitochondrial divergence: COI distances between aepyornithid clades <1.10% (vs 2.3–5.1% for other ratite within-family between-group distances). Average within-clade COI distance in the southern aepyornithid clade is 0.102% (95% CI ~0.058%). No clear geographic genetic structure (Mantel’s p=0.093). No differences among southern aepyornithid eggshells in micro-CT metrics or stable isotope signatures (PERMANOVA p=0.1161; F=2.058; n=49). These findings question the validity of Vorombe as a separate genus and suggest fewer species than morphology implies. - Hidden northern lineage: Four eggshell mitogenomes from far northern Madagascar form a distinct Aepyornis lineage separate from central A. hilderbrandti and southern A. maximus-associated forms. Isolation-by-distance is significant across Madagascar (Mantel p=0.001; Z=0.31), but northern-central reciprocal monophyly suggests a cryptic taxon; its formal rank remains unresolved (species/subspecies/population). The northern lineage shows diet dominated by C3 shrubs, distinct δ15N values (mean ~6.6‰; ~2‰ lower than regional plants’ mean yet ~2.5‰ higher than A. hilderbrandti), suggesting ecological differentiation. - Divergence timing and gigantism: Divergence within Aepyornis coincides with early Pleistocene aridification and landscape fragmentation: central–northern split ~1.22 Ma (95% HPD 0.6–1.9 Ma); southern vs central/northern split ~1.4 Ma (95% HPD 0.8–2.1 Ma). Ancestral state reconstructions indicate extreme gigantism in A. maximus is derived, with body size roughly doubling from mid-to-late Pleistocene ancestors, aligning with Neogene/Pleistocene cooling-linked gigantism trends. - Ecological inference: Micro-CT shows family-level differences in eggshell pore structure; stable isotopes reveal regional dietary differences (C3 vs mixed C3/C4) and possible behavioral ecology (nocturnality hypotheses linked to δ13C/δ15N patterns). - Conservation/palaeo implications: Low genetic diversity in southern Aepyornithidae suggests fewer species and potentially reduced resilience to late Holocene anthropogenic environmental change, contributing to extinction.

The study addresses long-standing uncertainty in elephant bird systematics by leveraging fossil eggshell as a high-quality substrate for ancient DNA, proteins, microstructure, and isotopes. Mitochondrial phylogenies, congruent with eggshell thickness, micro-CT porosity, and proteomic signatures, reveal a deep family-level split between Mullerornis and Aepyornis/Vorombe, with a crown divergence ~30 Ma. Within Aepyornis, recent Pleistocene divergences likely reflect climatic aridification and topographic barriers (e.g., highland corridors and valleys) driving population isolation and differentiation. The discovery of a distinct northern Aepyornis lineage—despite the absence of corresponding skeletal finds—demonstrates cryptic diversity and expands the inferred historical range of A. hilderbrandti-like forms, while stable isotope differences suggest ecological niche differentiation. Extremely low mitochondrial diversity within southern aepyornithids and the lack of concordant morphological/structural or isotopic distinctions challenge morphology-based inflation of species/genera (e.g., Vorombe), pointing toward fewer southern taxa and possible intraspecific size variation (including potential sexual dimorphism). These integrated molecular and palaeoecological lines offer a revised phylogeographic framework, linking environmental change to diversification and the evolution of extreme gigantism over relatively short timescales.

This work demonstrates that fossil eggshells can resolve the phylogeography, taxonomy, and ecology of extinct megafauna. The data support elevating Mullerornis to its own family (Mullerornithidae) separate from Aepyornithidae, document surprisingly low mitochondrial diversity within southern Aepyornithidae, and recover a previously unrecognized northern Aepyornis lineage. Divergences within Aepyornis occurred during early Pleistocene environmental shifts, consistent with population fragmentation and subsequent evolution of extreme gigantism. The authors advocate revising elephant bird taxonomy to integrate palaeogenomic and palaeoecological evidence, likely consolidating southern forms into Aepyornis rather than a separate Vorombe genus. Future research should prioritize recovery of nuclear genomes to test for sex-linked dimorphism and refine species boundaries, expand geographic sampling (mid-to-northwest, central east, southeast Madagascar), and integrate additional pre-Holocene material to capture deeper temporal diversity.

Key limitations include reliance on mitochondrial DNA, which reflects maternal lineages and may not capture nuclear genomic diversity, sex-specific dispersal, or recent admixture. Isolation-by-distance and incomplete lineage sorting cannot be fully excluded without denser sampling between central and northern regions and nuclear markers. The far north lacks described skeletal specimens, limiting morphological corroboration. Sampling may be temporally and geographically biased with scant pre-Holocene records, and eggshell-only analyses prevent direct sex typing and some morphological inferences. Some proteomic residue differences are weakly supported, and micro-CT sample sizes are limited. Together, these factors may affect taxonomic resolution and generalizability.