Sequencing to test models of maize domestication
-- Robert McK. Bird

Fifteen years ago (MNL 53:53-54, 1979; Biotropica 12:30-41, 1980) I proposed the outline of what can be called the Intersectional Introgression Model, that (1) early gardeners in Mexico began cultivating a wild maize with miniature maize-like ears, a form of which was found in caves near Tehuacán, SE of Mexico City, (2) the early domesticate was spread to SE Guatemala and by chance hybridized with and was introgressed by Zea luxurians, and (3) various allelic combinations in the hybrid swarm allowed both the rapid cultural selection of widely varying maize races and the natural selection of weedy populations which became Mexican annual teosinte (MAT). Tehuacan Early Domesticated (TED) differed from the wild form perhaps only by having more adherent kernels. The oldest date measured directly on TED cobs is 2750 B.C., 2300 years more recent than originally estimated (uncalibrated; Long et al., Radiocarbon 31:1035-1040, 1989), so the hybridization of (2) may have occurred shortly before 2000 B.C., later than I earlier proposed because of the new dates. Wilkes presented a very similar model, also in 1979, but he proposed that the introgression came from Z. diploperennis. Archaeobotanical evidence closely fits the model: the oldest well dated maize samples from North, Central and South America are all fully distinct morphologically from teosinte, and there are still no archaeological specimens of teosinte or maize-teosinte hybrids which date securely before 1100 B.C. (a few teosinte rachids and some Mexican cobs which appear to be introgressed by teosinte and which are thought to be earlier need to be directly radiocarbon dated). The traits which maize and MAT share are also explained -- both derive from one hybrid swarm.

Since 1976 several well-known models have been defined wherein no maize was present until man selected mutant or transmutant forms of MAT, the latter without basic genetic change. Now that John Doebley, Adrian Stec and associates are isolating key genes responsible for some of the morphological differences between maize and teosinte, we are closer to being able to choose among evolutionary models by using nucleotide sequencing to define quantitatively the differences between individual key alleles of species in section Luxuriantes and of various maizes and Mexican annual teosintes in section Zea.

If maize were a domesticated version of MAT, with no introgression from Z. luxurians, the nucleotide sequences of alleles of the key differentiating loci within sect. Zea should reflect somewhat over five millennia of evolution since domestication. For four such loci, a to d, alleles should fit a pattern like a1 b1 c1 d1, a1 b1 c1 d3, a2 b2 c2 d2, a2 b2 c2 d4, etc. in maize, and a3 b3 c3 d3, a4 b4 c4 d4, a4 b4 c1 d3, etc. in MAT. The numerals indicate two maize and two teosinte lineages out of many, as well as some recent introgression between maize and MAT (e.g., a1 b1 c1 d3). The ancestor of domesticated maize and present MAT would have had alleles a0 b0 c0 d0. Alleles in sect. Luxuriantes would be separated by several millions of years of evolution from all comparable alleles in sect. Zea.

On the other hand, if the more complex Intersectional Introgression Model is correct, many alleles at the key loci in the MAT races and some of those in present-day maize should have been introgressed from Z. luxurians. Such alleles (capitalized) would now reveal a pattern like A5 B5 C5 D5 in Z. luxurians, a3 B7 C7 d3, a4 B8 C8 D8, etc. in MAT, and a1 b1 c1 d1, a1 b1 c1 D6, A8 b2 c2 d1, etc. in maize. The 5, 6, 7 and 8 indicate lineages of Z. luxurians alleles which evolved relatively independently after the hybridization, as in maize of highland South America and northeastern U.S. or in teosinte of west Mexico. There may even be some A5 B5 C5 d1 teosinte somewhere in Central America, a maize-introgressed Z. luxurians. Remember that Z. diploperennis might substitute for Z. luxurians in these discussions.

There could be small sets of Z. luxurians alleles in most maize, those alleles which had been "teosinte-negative" in the teosinte and selected in the hybrid progeny because they contributed positively to early maize races. Doebley and Stec (Genetics 134:566, 1993) found some alleles with unexpected effects which might be examples -- e.g., the teosinte allele at a QTL which they locate on arm 7L has a positive effect on CUPR, the number of cupules per rank. Such alleles could explain much about the explosive evolution of maize since 1100 B.C. with extraordinary increases in the size and number of various organs. Moreover, observations by Donald Robertson (MNL 57:6, 1983) and others, that progeny from the maize x Z. luxurians cross demonstrate a notable increase in mutation rate, suggest that activation of transposons by the proposed wide hybridization could have increased the potential for novel traits. Perhaps footprints left during various episodes of transposon insertion/exsertion will be revealed by sequencing, and these could help define phylogeny.

The Intersectional Introgression Model provides a parsimonious explanation of Sarah Hake's 1980 results using DNA hybridization (MNL 56:90-92, 1982). In full-genome hybridizations involving Tripsacum laxum, U.S. dent inbreds W64A and B37, Ladyfinger popcorn (LP) and several teosintes, she found a divergence of 8.8% (% divergence = ?Tm celsius) between W64A and T. laxum (probably a reflection of many millions of years of separation), 3.7-4.2% divergence between maize (LP and 2 inbreds) and sect. Luxuriantes (Z. diploperennis and Z. luxurians), and 2.7-3.3% divergence between Balsas teosinte and maize (B37 and LP). The 2.7-3.3% is not well explained if maize and MAT separated only several thousand years ago -- perhaps 0.002-0.008% would be expected. If Balsas teosinte contains many Z. luxurians alleles, the 2.7-3.3% figures make sense, although Balsas-Z. luxurians divergence is 3.6%. Maize (W64A and B37)-Chalco teosinte divergence was only 0.9%, perhaps reflecting the long-term introgression between maize and teosinte that many think has been occurring in the Chalco area; Balsas-Chalco divergence was 0.2%. 

Please Note: Notes submitted to the Maize Genetics Cooperation Newsletter may be cited only with consent of the authors

Return to the MNL 69 On-Line Index
Return to the Maize Newsletter Index
Return to the Maize Genome Database Page