Different tests have different statistical properties — different sensitivities, different abilities to distinguish between closely related models — and the results they produce are not always directly comparable.

A study that uses a test with lower sensitivity to weak signals may not detect a Population Y signal that a more sensitive test in a different study would find.

The quality and quantity of ancient DNA from individual specimens is another source of variation.

Ancient genomes sequenced at low coverage — where the genome has been read only a small number of times on average, leaving many positions without reliable data — are less informative for the statistical tests that detect population structure than high-coverage genomes.

If the key ancient specimens from Lagoa Santa were sequenced at low coverage, the results will be less definitive than if they were sequenced at high coverage.