Ancient genomes reveal insights into ritual life at Chichén Itzá

Chichén Itzá rose to prominence during the Late and Terminal Classic (AD 600–1000) as a dominant political and ritual center in the northern Maya lowlands. Subterranean features such as cenotes and chultúns were associated with water, ritual activity, and access to the Maya underworld, and prior work documented abundant ritual offerings and human remains, including sacrificed juveniles, at the Sacred Cenote. Historical accounts suggested a focus on female victims, but osteological studies indicate both sexes were sacrificed. Isotopic evidence implies some non-local individuals may have been involved. Key questions remain about who was selected for sacrifice, their kinship, sex, origins, and how ancient inhabitants relate genetically to present-day Maya. This study investigates 64 subadults from a chultún near the Sacred Cenote and compares them to 68 present-day Maya from Tixcacaltuyub to test for biological kinship, sex, population affinities, diet, genetic continuity, and signatures of selection, particularly in immunity genes potentially shaped by colonial-era epidemics.

Prior research at Chichén Itzá and other Maya sites has documented ritual killings in cenotes and caves, with iconography and historical sources highlighting sacrificial practices. Early colonial narratives emphasized female sacrifices, but bioarchaeology shows both male and female victims and predominantly juvenile remains at Classic Maya sites. Isotopic analyses suggested some sacrificial victims were non-local, potentially from distant regions. Twins occupy a central role in Maya mythology (for example, the Popol Vuh Hero Twins), yet twins had not been identified previously in ancient Maya mortuary contexts. Genetic studies in the Americas have shown long-term continuity among Indigenous groups and post-contact admixture with European and African lineages, with evidence for selection at immune loci, including HLA, following disease introductions during the colonial period.

• Sampling: From the chultún mass burial near the Sacred Cenote, only left petrous bones were sampled to avoid resampling individuals. Radiocarbon dating (n=26) constrained use of the chultún from early 7th century through mid-12th century AD. Ancient DNA was retrieved from all 64 subadults (YCH). Blood samples from 68 present-day Maya from Tixcacaltuyub (TIX) were collected for comparison.
• aDNA lab and sequencing: YCH libraries were UDG-treated; TIX libraries non-UDG-treated. Sequenced initially to ~5–11 million reads to assess preservation. For 11 YCH, single-stranded UDG-treated libraries were also prepared. Contamination was assessed (<5%) via nf-core/Eager. All TIX and 56 YCH had >0.1% human DNA; libraries were in-solution enriched for 1.24 million ancestry-informative SNPs, mtDNA, and an immune gene panel. Post-capture sequencing yielded ~40 million reads per library.
• Bioinformatics and analyses: Population genetic analyses and HLA typing were conducted after QC. Genetic sexing used X/Y SNP coverage comparisons. Kinship used pairwise mismatch rates (PMR). Stable isotope analysis of bone collagen (δ13C, δ15N) characterized diet. Population structure used worldwide PCA and unsupervised ADMIXTURE; outgroup f-statistics f3/fST assessed affinities; qpWave modeled shared ancestry and admixture proportions; a maximum-likelihood continuity test evaluated whether TIX descend from YCH. Uniparental markers (mtDNA haplogroups/haplotypes; Y-chromosome haplotypes) were determined. Selection scans used locus-specific branch lengths (LSBL) with comparisons YCH vs Iberians (IBS) and Han (CHB), and TIX vs IBS and YCH; gene ontology enrichment used GoWinda. HLA region selection was further probed via non-overlapping association patterns (fmet metric) across HLA loci. In silico HLA class II peptide-binding predictions to Salmonella-derived peptides employed NetMHCIIpan via IEDB.

• Context and chronology: The chultún was used from the early 7th to mid-12th centuries AD (radiocarbon n=26). aDNA recovered from 64 subadults.
• Sex and kinship: All 64 analysed chultún individuals were genetically male. PMR identified two pairs of monozygotic twins (YCH018–YCH019; YCH033–YCH054) and nine other close relative pairs. Overall, 25% (n=16) of children had a close relative in the assemblage.
• Diet: Bone collagen isotopes for 54 individuals showed δ13C from −13.9‰ to −7.6‰ (mean −9.9±1.5‰) and δ15N from 5.9‰ to 14.0‰ (mean 9.7±1.5‰), consistent with Classic Maya diets rich in C3 resources with variable C4 and aquatic inputs. Related individuals had more similar isotopic values, suggesting shared diets.
• Population structure and continuity: YCH cluster with unadmixed Indigenous Americans and present-day Maya on PCA; TIX show low-level European/African admixture. ADMIXTURE detected no non-Indigenous components in YCH and low European/African in TIX, with 18 TIX individuals showing no non-Indigenous signal. Outgroup f-statistics show YCH and TIX share high drift with Central/South American Indigenous groups; ancient Belize individuals (including a 9,300-year-old Mayahak Cab Pek individual) are genetically closest to YCH. qpWave indicates YCH and present-day Mayan groups share ancestry; TIX can be modeled as ~92% Indigenous American, ~7% European, ~0.03% African (P=0.11). Continuity tests support TIX as direct descendants of YCH (average drift ~0.5, P≈1.0 per TIX individual).
• Uniparental markers: mtDNA haplogroup distributions (A, B, C, D) are nearly identical in YCH and TIX, though YCH show higher haplotype diversity. All YCH Y-chromosome haplotypes (n=51) belong to Q; TIX Y-chromosomes include 47.37% European and 5.26% Middle Eastern lineages, indicating male-biased post-contact admixture.
• Selection signals: LSBL scans highlight lipid metabolism genes (e.g., FADS in YCH; FTO, TCF7L2, ADCY family in TIX). GO enrichment shows fertility-associated processes enriched in YCH and lipid/cholesterol metabolic processes in TIX.
• HLA region: 15 (YCH) and 7 (TIX) HLA-region SNPs are in top 0.5% LSBL, with no overlap between groups. Non-overlapping HLA associations are higher in present-day Maya (TIX and Lacandon) than YCH, consistent with pathogen-driven selection. Significant allele frequency shifts (pcorr<0.05) include decreases in HLA-B40:02 (0.2447→0.0821), HLA-DQA103:03 (0.1277→0.0224), HLA-DQB104:02 (0.1809→0.0299); increases in HLA-A68:03 (0.0532→0.2687), HLA-B39:05 (0.0532→0.2687), HLA-C07:02 (0.2021→0.3955), HLA-DQB103:02 (0.4894→0.7015), and HLA-DRB104:07 (0.2340→0.4627). In silico binding suggests HLA-DRB104:07 strongly binds Salmonella-derived peptides, whereas HLA-DQA103:03/DQB1*04:02 and related haplotypes are weak binders. These patterns support selection on HLA class II (notably DR4) following colonial-era epidemics (e.g., 1545 cocoliztli involving Salmonella enterica Paratyphi C).

Findings provide a detailed view of Classic-period ritual practice at Chichén Itzá: all analysed subadults in the chultún were male, indicating sex-specific selection for sacrifice in this context. The unusually high occurrence of close relatives, including two sets of identical twins, suggests that biological kinship influenced selection of sacrificial victims and resonates with the prominence of twins in Maya mythology and underworld rites associated with subterranean spaces. Isotopic similarity within related pairs, coupled with similar ages at death, supports the interpretation that some kin were sacrificed together during the same ritual events. Population genetic analyses reveal long-term genetic continuity between ancient Chichén Itzá inhabitants and present-day Maya from Tixcacaltuyub, while uniparental markers demonstrate strongly male-biased non-Indigenous ancestry inputs after European contact. Selection scans and HLA analyses indicate that immunity-related loci experienced directional shifts consistent with pathogen-driven selection following colonial-era epidemics, particularly involving Salmonella, aligning with historical records of catastrophic depopulation and previous ancient pathogen findings. Dietary isotopic variability may reflect non-local origins of some individuals or social-status-related dietary differences, but related individuals shared dietary signatures, implying shared upbringing within extended family networks. Together, the data address the study’s questions on kinship, sex, origins, and genetic legacy of sacrificial victims, and illuminate the evolutionary impact of colonial disease exposures on Maya immunity genes.

This study integrates archaeogenetics, isotopes, and population genomics to reconstruct ritual life and population history at Chichén Itzá. It documents a male-only subadult chultún assemblage with an unexpectedly high frequency of close relatives and two sets of monozygotic twins, linking ritual selection to kinship and Maya twin mythology. Genome-wide analyses demonstrate continuity between ancient inhabitants and present-day Maya, while uniparental markers record male-biased post-contact admixture. Signals of selection, especially in HLA class II alleles such as DRB1*04:07, together with non-overlapping HLA associations and peptide-binding predictions, support pathogen-driven selection during the colonial period, likely influenced by Salmonella epidemics. Future work could expand sampling across additional Maya sites and time periods, include broader present-day comparative datasets, generate local faunal isotopic baselines to refine dietary reconstructions, and recover ancient pathogen DNA from the chultún to directly link host selection signals to specific pathogens.

• Sampling constraints: Only left petrous bones were available, and 64 of an estimated 106 individuals had suitable preservation, potentially underestimating the number of relatives.
• Isotopic context: Lack of local faunal baselines limits precise dietary reconstruction; breastfeeding and status effects may confound δ15N variability.
• Population comparisons: The genetically closest present-day population to YCH may be unsampled or no longer extant; post-contact admixture complicates direct comparisons.
• Selection inference: HLA selection signals are inferred from allele frequency shifts, LD patterns, and in silico binding rather than direct pathogen detection in the same assemblage.