
Who Should Keep a Family Genealogy? A District Archive in Zaozhuang Now Holds a Hundred
2026年11月17日
Tropical Ground Keeps Little Ancient DNA: A 5,000-Year Ceiling Measured from 300 Caribbean Skeletons
2026年11月18日146 Regions Became 302: What AncestryDNA’s 2026 Update Actually Changed
On 8 October 2026, AncestryDNA released its 2026 ancestral regions update. Same saliva sample, same sequencing run, yet many customers opened their reports to find new regions, moved percentages, and low-percentage regions that had quietly vanished. The company’s explanation is blunt: your DNA did not change; our ability to read it did.
The headline number is the region count, which rose from 146 to 302. Two developments made that possible: the reference panel grew from 185,000 to more than 295,000 individuals, an increase of nearly 60%, and a new region-assignment algorithm replaced the old one.
The reference panel is the foundation of any ancestry estimate. It consists of people with well-documented family histories from specific places, and a customer’s DNA is measured against it. A larger and more finely divided panel makes it possible to separate neighbouring populations, which is why the number of regions could double.
The Americas changed most dramatically. The single broad Indigenous Americas-Mexico region became 14 specific ones, including the Centro-Oriente Highlands and Southern Sierra and Tehuantepec. The former Eastern South America region split into six, covering parts of Paraguay, Uruguay, Brazil and Argentina, together with the Maya Highlands of Guatemala and the Orinoco Delta in Venezuela. Because the algorithm now detects Iberian patterns more sharply, some customers will see Spanish ancestry resolve to the Canary Islands, reflecting the historical role of islanders in settling the Americas.
In Africa, continental catch-all labels gave way to ethnolinguistic ones. The total rose to 50 regions linked to 93 ethnic groups, among them Hausa and Fulani Peoples, Mali-Mande Peoples and Southern Benin-Yoruba Peoples. Broad descriptions such as ‘somewhere in West Africa’ were replaced by boundaries that follow language and community lines more closely.
South Asia saw the largest jump: India and Pakistan went from 6 regions to 36, including seven new ones across southern India covering Kerala, Tamil Nadu, Telangana and Andhra Pradesh. Europe gained finer Balkan divisions such as Dalmatia, Central Croatia, the Western Balkans, the Sava Plains, the Central Balkans and Bulgaria, plus new regions for Switzerland, Belgium and Northeastern Germany. French ancestry and French-Canadian lineages, including Quebec and Acadia, are now separated far more clearly.
More important than the count is the change in method. The old algorithm worked like compiling a dictionary: it first catalogued DNA patterns shared within each reference population, producing a kind of codebook, and then looked up a customer’s data in it.
The new algorithm compares directly. It measures a customer’s DNA segments against each of those 295,000 reference individuals and looks for the closest matches. The authors compare it to abandoning the dictionary and reading through an entire library, which is why the system can now tell a Southern accent from a Midwestern one instead of merely concluding that the language is English.
A second step cleans up the result. Once each segment has been assigned, an AI tool built on a transformer architecture weighs information from across the genome, smooths out questionable individual calls and returns the highest-probability combination of regions. This is the technical reason the same DNA can produce a different report: the update improves the stability of the assignment, not the biological data.
Customers should therefore expect their reports to move. New regions appear, percentages shift and low-percentage regions drop away, all of which is normal. Ancestry also notes that SideView 2.0, a tool that splits ancestry by parent, went live around the same date and independently changes regional precision. Two updates arriving together can easily look like a change in ancestry itself.
For East Asian customers there is a less conspicuous fact: the refinement concentrates on the Americas, Africa, South Asia and Europe. East Asia received no comparable subdivision, so a Chinese-language report may appear untouched. The algorithm is still running; what did not rise is the resolution of East Asian populations inside the reference panel.
That points to the right way to read any ancestry report. A percentage is a statistical estimate, not genealogical proof. How finely it can resolve depends on how many well-documented people from that region sit in the panel. Different companies use different panels and different algorithms, so the same phrase, 32% East Asian, is not comparable across services. Tracing a specific ancestor still requires genealogy, local archives and markers such as Y-chromosome and mitochondrial haplogroups, checked against one another.
Old photographs under a magnifying glass. Ancestry research usually begins with family memory, but it ends with documents and genetic markers confirming each other. Source: Tuchong Creative.





