No genetic evidence yet for hinnies at Mazongshan (400-160 BCE), northwestern China

A re-analysis of ancient DNA from the Mazongshan site refutes the recent claim of finding hinnies (horse-donkey hybrids) from 400–160 BCE, revealing that the four specimens previously identified as hybrids were actually donkeys, thereby necessitating new data to support the historical role of hinnies in trans-regional trade.

The Gist

Imagine you just read a headline in a science magazine that says: "Scientists found ancient mules and hinnies (horse-donkey hybrids) in China, proving that ancient traders were breeding them on a massive scale!"

That was the story published by a team led by Li and colleagues in 2026. They claimed to have found DNA from four "hinnies" (the offspring of a male horse and a female donkey) at an ancient jade mining site called Mazongshan. This was a big deal because hinnies are incredibly rare in history—like finding a four-leaf clover in a field of three-leaf ones.

But now, a new team of scientists (led by Ludovic Orlando) has come along, looked at the same data with a different set of glasses, and said: "Hold on a minute. Those aren't hinnies. They're just regular donkeys."

Here is the story of how they figured that out, explained simply.

1. The "Fake Out" in the Lab

Think of the original DNA data like a very old, crumpled, and slightly dirty letter. When the first team tried to read it, they saw patterns that looked like a hybrid.

However, the new team realized the "letter" had a weird smudge on it. It turns out the lab process used to prepare the DNA added a tiny, four-letter "tag" to the beginning and end of every DNA strand. It's like someone stuck a piece of tape on the front and back of every page of a book before scanning it.

Because the computer software didn't know about this tape, it got confused. It thought the tape was part of the story, creating a "ghost" pattern that looked like a hybrid. When the new team carefully peeled off that tape (by trimming four letters from the ends of the DNA strands), the picture cleared up.

2. The "Identity Crisis"

Once the data was cleaned, the team ran the DNA through a "genetic ID scanner."

• The Original Claim: The scanner said, "This looks like a mix of Horse and Donkey!"
• The New Reality: After cleaning the data, the scanner said, "Nope, this is 100% Donkey."

To prove this, they created a "control group." They took real horse DNA and real donkey DNA, mixed them together in a computer to simulate what a real hinny would look like. When they compared the ancient bones to this simulation, the ancient bones didn't sit in the middle (where a hybrid should be). Instead, they sat right next to the donkeys, looking exactly like them.

3. Why Does This Matter?

You might ask, "So what? It's just four donkeys instead of four hybrids."

Here is the big picture:

• The Rarity of Hinnies: Making a hinny is biologically very hard. It's like trying to bake a cake where the ingredients barely want to mix. In nature and history, hinnies are extremely rare (less than 1% of hybrids). Mules (the opposite mix: male donkey + female horse) are common because they are much easier to produce.
• The 40% Claim: The original study claimed that 40% of the animals at this site were hinnies. If that were true, it would mean ancient Chinese farmers were doing something incredibly difficult and expensive on purpose, just to make these rare hybrids. It would rewrite our understanding of how they managed animals.
• The Correction: By showing those four animals were just donkeys, the new study removes the only evidence that hinnies were ever common there. It means the ancient farmers were likely doing what everyone else did: raising horses and donkeys, and maybe making the occasional mule, but not a factory of hinnies.

The Bottom Line

This paper is a great example of science working exactly as it should. It's like a group of detectives reviewing a crime scene. The first team found a clue that seemed to point to a suspect (the hinny). The second team re-examined the clue, realized it was actually a trick of the light (a lab artifact), and corrected the record.

The Takeaway:
We still know that ancient people at Mazongshan raised both horses and donkeys, which is fascinating in itself. But the idea that they were mass-producing rare hinnies to help with trade? That story has been debunked. The "hinnies" were just donkeys all along.

Technical Summary

1. Problem Statement

The paper addresses a recent claim by Li et al. (2026) regarding the genetic identification of hinnies (first-generation hybrids of horse stallions and donkey jennets) at the Mazongshan jade mining site in northwestern China, dated to 400–160 BCE.

• The Claim: Li et al. reported finding four hinny specimens out of ten equine samples, suggesting a 40% prevalence of hinnies. This would imply highly unusual breeding practices, as hinnies are biologically difficult to produce (low pregnancy rates compared to mules) and are historically rare in ancient assemblages.
• The Issue: The authors of this re-analysis (Tressières et al.) identified methodological inconsistencies in the original data processing, specifically regarding DNA damage patterns and mapping artifacts, which cast doubt on the taxonomic identification of the four putative hybrids.

2. Methodology

The authors performed a comprehensive re-analysis of the raw sequencing data (FASTQ files) from the National Genomics Data Centre (accession CRA027067).

• Data Processing & Artifact Correction:
  • Initial Observation: The original data showed an unusual base composition profile where reads consistently started with the motif GTCT and ended with AGAC. This suggested an exogenous 4-base tag added during library preparation that interfered with mapping.
  • Correction: The authors trimmed the first and last four bases from the reads.
  • Impact: This simple trimming step significantly increased the estimated endogenous DNA content (from ~0.2% to 0.24–5.80%) and improved mapping depth.
• Authentication of Ancient DNA (aDNA):
  • Damage Patterns: After trimming, the authors re-ran mapDamage and PMDtools. The corrected data showed the expected aDNA signatures:
    • Excess of cytosine residues at the position preceding read starts (indicating USER-enzyme treatment).
    • Characteristic C-to-T misincorporation rates at read starts, particularly in CpG contexts, confirming the data is authentic ancient DNA and not modern contamination.
• Taxonomic Identification:
  • Sex Determination: Used Zonkey to confirm sex (6 males, 4 females/undetermined).
  • Hybrid Detection:
    • PCA: Projected the ancient specimens onto a Principal Component Analysis space derived from a comparative panel of 24 modern and ancient equine genomes (including horses, donkeys, zebras, and wild asses).
    • ADMIXTURE: Performed supervised ancestry analysis (K=3: Horse, Donkey, Zebra) to quantify genetic ancestry.
    • Phylogenetics: Constructed Neighbor-Joining trees to assess clustering.
    • Simulation: Created in silico simulated F1-hybrids (50% horse/50% donkey) to establish the expected genetic position of a true hinny in the PCA space.

3. Key Results

• Reclassification of Specimens:
  • The three specimens originally identified as horses and the three identified as donkeys were confirmed as such.
  • Crucial Finding: The four specimens originally identified as hinnies (Ancient4–Ancient7) were reclassified as pure donkeys.
  • Evidence:
    • PCA: The four specimens clustered tightly with the donkey group, not in the intermediate space between horses and donkeys where the simulated hybrids were located.
    • ADMIXTURE: These specimens showed >99% donkey ancestry with no detectable horse (caballine) contribution.
    • Phylogeny: They grouped monophyletically with other donkeys, distinct from the horse cluster.
• Methodological Insight: The original misidentification was likely caused by the untrimmed 4-base exogenous tags, which created mapping artifacts that distorted the genetic signal, leading to erroneous hybrid calls.
• Statistical Context: While the study found no hybrids in this specific sample of 10, the authors note that the absence of evidence is not evidence of absence. Statistically, with a sample size of 10, hybrids could still represent up to ~26% of the population. To confidently rule out hybrids at a 5% frequency, ~59 specimens would be needed.

4. Key Contributions

1. Correction of Taxonomic Error: The paper definitively refutes the claim of 40% hinny prevalence at Mazongshan, correcting the record to show only horses and donkeys were present in this specific assemblage.
2. Methodological Demonstration: It highlights the critical importance of inspecting raw read composition (specifically for exogenous adapters or tags) before taxonomic assignment in ancient DNA studies. Failure to trim these artifacts can lead to false positives in hybrid detection.
3. Validation of aDNA Authenticity: Despite the initial misidentification, the re-analysis confirmed the data is authentic ancient DNA, validating the site's chronological and biological context.
4. Historical Insight: The study confirms the earliest genetic evidence for the presence of donkeys in China (400–160 BCE), indicating integrated management of horses and donkeys in the region during the Late Warring States/Early Han periods.

5. Significance

• Archaeological Implications: The claim of widespread hinny breeding challenged established models of equine husbandry, as hinnies are biologically inefficient to produce compared to mules. By correcting this to "donkeys," the paper restores the understanding that the site utilized standard equine species (horses and donkeys) rather than a specialized, rare hybrid breeding program.
• Trade Network Context: While the specific claim of hinnies as "integral components of trans-regional trade" is retracted, the confirmed co-existence of horses and donkeys still supports the idea of integrated animal management strategies in northwestern China, likely facilitating the Silk Road trade networks.
• Scientific Rigor: The paper serves as a vital example of the self-correcting nature of science, where re-analysis of raw data can overturn significant historical conclusions based on technical artifacts.

Conclusion: The Mazongshan site contains horses and donkeys, but no genetic evidence for hinnies exists in the current dataset. The original identification of hinnies was an artifact of untrimmed sequencing adapters. Future research requires a larger sample size to definitively assess the prevalence of equine hybrids in this region.