We think this is the “correct” graph. Shows Z. diploperennis (DIP), Z. mays mexicana (MEX), Z. mays parviglumis from the Balsas (‘PARV’), ancient Paredones Peruvian sample (N16), and modern landrace diversity panel (L340).
This model should also hold for other individual populations. Replace L340 with S. American lowlands (SAM), Mexican lowlands (MEXL), Mexican highlands (’MEXH), ancient S. American sample (Z6), modern maize diversity panel (507), and ancient Tehuacan sample (TH162`).
This graph has no admixture and should be wrong.
This model only has admixture from MEX without subsequent admixutre from PARV postulated by Kistler 2020.
This model only postulates admixture with PARV and ancient maize, without MEX.
I think Kistler 2020 suggests Andean maize (ANDES) should have no secondary PARV gene flow?
Subsequent admixture with MEX in MEXH compared to MEXL.
Admixture not due to PARV-MEX admixture.
And additional PARV admixture with modern maize doesn’t solve this: