maize domestication and gene interaction · maize domestication and gene interaction michelle c....

Maize domestication and gene interactionMichelle C. Stitzer1,2,� and Jeffrey Ross-Ibarra1,2,3

1Center for Population Biology, University of California, Davis, Davis, CA 95616, USA2Department of Plant Sciences, University of California, Davis, Davis, CA 95616, USA

3Genome Center, University of California, Davis, Davis, CA 95616, USA

Domestication is a tractable system for following evolutionarychange. Under domestication, wild populations respond to shift-ing selective pressures, resulting in adaptation to the new eco-logical niche of cultivation. Due to the important role of do-mesticated crops in human nutrition and agriculture, the an-cestry and selection pressures transforming a wild plant into adomesticate have been extensively studied. In Zea mays, mor-phological, genetic, and genomic studies have elucidated how awild plant, the teosinte Zea mays subsp. parviglumis, was trans-formed into the domesticate Zea mays subsp. mays. Five majormorphological differences distinguish these two subspecies, andcareful genetic dissection has pinpointed the molecular changesresponsible for several of these traits. But maize domesticationwas a consequence of more than just five genes, and regionsthroughout the genome contribute. The impacts of these ad-ditional regions are contingent on genetic background, both theinteractions between alleles of a single gene and among alleles ofthe multiple genes that modulate phenotypes. Key genetic inter-actions include dominance relationships, epistatic interactions,and pleiotropic constraint, including how these variants are con-nected in gene networks. Here, we review the role of gene inter-actions in generating the dramatic phenotypic evolution seen inthe transition from teosinte to maize.

IntroductionThe advent of agriculture generated dramatic departures fromthe way humans had been interacting with their world. Ten totwelve thousand years ago, in independent locations aroundthe world, cereal grains began to be consumed in larger quan-tities (Larson et al., 2014). Agriculture had drastic impacts onhuman culture, societal structure, and human health (Larsen,2006). As a coevolutionary process, domestication also dras-tically altered the evolution of plants in novel agroenviron-ments. Although the temporal nature of this trajectory andtransformation has been thoroughly investigated in the ar-chaeological record (Smith, 2001), genetic and genomic ap-proaches using extant crop and wild relative diversity haveprovided additional evidence about the process of domesti-cation (Zeder et al., 2006), especially in those regions of theworld where environmental conditions are not conducive toarchaeological preservation.The initial stages of domestication are largely analogous toa plant encountering a new ecological niche. By grow-ing plants and then planting their seeds in the next gener-ation, selective breeding and seed saving allow for adapta-tion to agronomic environments. The domestication processmay shift selection pressures from biotic interactions suchas competition and colonization to traits of value for humanconsumption, including large non-dispersing seeds, reduced

branching, and other nutritional and harvest-related pheno-types (Doebley et al., 2006). Byproducts of cultivation canalso alter biotic interactions, for example by allowing peststo specialize on a domesticate (Bernal et al., 2017; Gaillardet al., 2018), or alter phenotypes less visible to conscious hu-man selection, such as root architecture (Burton et al., 2013).Among modern domesticated plants, maize (Zea mays subsp.mays) is perhaps the crop most often used as an example ofthe changes wrought by domestication, as the radical restruc-turing of the female inflorescence of maize generated a plantreliant on human cultivation for its survival. For example,a productive cob of corn contains hundreds of seeds, but allare constrained to the cob. If seeds germinate without re-moval by humans, the seedlings remain in close proximity,and competition inevitably impacts fitness. Maize seeds arealso apparent and unprotected from bird and mammal preda-tors and their harsh digestive tracts. Despite these character-istics that limit survival in the wild, the maize plant — oncenoted as a human-generated ‘monstrosity’ (Beadle, 1972) —is well-adapted to its new ecological niche.Maize was domesticated from the wild grass (Zea mayssubsp. parviglumis) about 9000 years ago in the Balsas re-gion of southwest Mexico (Figure 1) (Matsuoka et al., 2002;Piperno et al., 2009). The term ‘teosinte’ is used broadly torefer to wild taxa from any of five species within the genusZea (Doebley and Iltis, 1980; Iltis and Doebley, 1980). Thesespecies are highly adapted to their distinctive, local environ-ments (Hufford et al., 2012a,b) and form large populationsacross much of Central America (Wilkes, 1967).Domesticated maize did not arise as a result of selection ononly a few genes, although major effects enabled by alleles ata few genes were essential. Although we may never be able toperfectly reconstruct how selective forces gave rise to modernmaize, genetic and genomic tools can allow some insight intothe allelic trajectories and phenotypic shifts underlying theseprocesses. Maize domestication and comparison to teosintethus provide a temporal and phenotypic context in which tounderstand both the origin and aftermath of selection.Here we review major domestication genes, highlight the roleof gene interactions in the domestication of maize, and showhow these interactions have complicated attempts to achievecongruent results in understanding the inheritance of domes-tication phenotypes.

The genetic basis of maize domesticationMaize domestication altered plant morphology. Maizehas a female inflorescence so different from any wild plant

Stitzer et al. | June 12, 2018 | 1–14

PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26502v2 | CC BY 4.0 Open Access | rec: 12 Jun 2018, publ: 12 Jun 2018

●

●

15

20

25

30

35

−120 −110 −100 −90

Longitude

Latit

ude Taxon

mexicanaparviglumismays

Fig. 1. Collection locations of species within Zea mays. Subspecies mays in yellow, parviglumis in red, and mexicana in blue. The location of archaeological starch grains(Piperno et al., 2009) in the Balsas river valley are shown in a circled orange point, and location of the oldest cobs (Benz, 2001) are shown in a circled light blue point.Occurrence data from http://www.biodiversidad.gob.mx/genes/proyectoMaices.html and http://seedsofdiscovery.org/.

that its origin was debated throughout the 20th century. Al-though early experimentalists observing and crossing maizeand teosinte were assured of the ancestral state of teosinte(Harshberger, 1896; Collins and Kempton, 1920; Weather-wax, 1924), vocal criticism planted doubt in the minds ofbotanists throughout the middle of the 20th century (Man-gelsdorf and Reeves, 1939; Mangelsdorf, 1974). At oneextreme, the tripartite hypothesis proposed the ancestor ofmaize was an extinct popcorn, and that teosinte arose fromcrosses between corn and the related genus Tripsacum, withfurther crosses giving rise to the diversity of maize we ob-serve today (Mangelsdorf and Reeves, 1939). The alternativeteosinte hypothesis argued that teosinte was the direct ances-tral form of maize (Beadle, 1939). These conflicting originsof maize triggered a 50-year debate, resolved when molecu-lar methods and archaeological evidence vindicated teosinte,specifically Zea mays subsp. parviglumis, as the ancestor ofmaize (Bennetzen et al., 2001; Matsuoka et al., 2002; Pipernoet al., 2009).The gross vegetative morphology of teosinte is largely indis-tinguishable from maize, with the major evolutionary inno-vation of maize being its infructescence, or ear. Hence, earlydefinitions of major phenotypic differences between maizeand teosinte focused on ear phenotypes, putatively controlledby 4-5 genes or blocks of genes (Beadle, 1939; Mangelsdorfand Reeves, 1939).These distinguishing phenotypes are:

1. Maize has paired mature spikelets, while in teosinte,only a single spikelet matures. In grasses, the spikeletis a short branch on which flowers are borne. Mostgrasses form many single spikelets along the inflores-cence. But in the Andropogonae, a tribe of grassesthat includes maize and teosinte, spikelets are paired inboth female and male inflorescences (Kellogg, 2000;Wu et al., 2009). In the teosinte ear, only one spikeletdevelops into a seed, and remnants of the developmen-tally arrested pedicellate spikelet is found within the

a b

og ig

rank

cup

Fig. 2. (a) Teosinte infructescence showing alternate initiation of each set ofspikelet pairs, with each kernel in a single rank (rank). Husk leaves are removed.(b) Each teosinte fruitcase contains a single kernel, as the pedicellate spikelet doesnot mature. The outer glume (og) of the sessile spikelet (ss) forms a closed outersurface of the fruitcase, while the inner glume (ig) is found within the fruitcase. Thecupule (cup) forms the other sides of the fruitcase.

fruitcase of teosinte seeds (Figure 2b) (Weatherwax,1935; Sundberg and Orr, 1990; Doebley et al., 1995b).In maize, each internode of the cob (the female inflo-rescence) contains two spikelets (Figure 3a), and eachspikelet matures into a kernel. The consequence ofmaturation of both spikelets in maize is more pistillateflowers formed on the ear, and hence twice as manyseeds. In teosinte, the mature spikelet’s outer glumesharden and form a secure fruitcase, protecting the ker-nel from predators.

2. Maize has at least four ranks to its ear (‘polystichous’),while teosinte only has two (‘distichous’). This is aconsequence of phyllotaxy, in the initiation of new pri-

2 Stitzer et al. | Maize domestication and gene interaction


http://www.biodiversidad.gob.mx/genes/proyectoMaices.html

http://seedsofdiscovery.org/

mordia on the inflorescence meristem. The vegetativephyllotaxy of maize and teosinte is distochous — onlyone leaf is initiated per node — leading to the alter-nating leaves characteristic of the adult plant (Jacksonand Hake, 1999). In teosinte, this alternate initiationoccurs from the inflorescence meristem as well, visiblein the mature alternate triangular infructescence (Sund-berg and Orr, 1990) (Figure 2a). In maize, the flo-ral meristem exhibits whorled phyllotaxy, with multi-ple spikelet pairs initiated simultaneously (Bartlett andThompson, 2014), each pair representing a rank of theear (Figure 3a). Multiple ranks of kernels also giverise to more kernels per ear in maize than teosinte.

3. Maize has a non-disarticulating rachis, while theteosinte rachis disarticulates upon maturity. Therachis is the inflorescence, representing the entiretyof the ear in maize and teosinte. In teosinte, abscis-sion layers form at nodes between internodes of the ear(Figure 2a), and divide the rachis into individual fruit-cases which fall apart and can then disperse. In maize,these abscission layers do not develop, and the rachisremains intact upon maturity, with kernels attached tothe cob even after drying (Iltis, 2000; Chavez et al.,2012) (Figure 3). This intact cob eases harvest, andmakes the maize plant reliant on human interventionfor seed dispersal.

4. Maize has softer, smaller glumes, while the teosintefruitcase is entirely enclosed by the outer glume of thespikelet. Glumes are leaves that subtend a flower. Inteosinte and maize, each glume is associated with asegment of the rachis, referred to as the cupule (Dor-weiler and Doebley, 1997). In teosinte, the outer glumeand cupule fully enclose each kernel, and harden exten-sively at maturity (Figure 2b). The teosinte cupulatefruitcase prevents predation, meaning teosinte seedscan pass unscathed through the digestive tract of birdsand mammals (Wilkes, 1967). Maize glumes are re-duced and papery, and kernels are exposed once huskleaves are removed from the ear (Figure 3).

5. Although the vegetative portion of the plant is largelyhomologous between maize and teosinte, maize hasreduced axillary branching compared to teosinte. Inteosinte, upper lateral branches arising from juvenileand adult nodes are elongated and tipped by tassels,while lower lateral branches elongate into basal tillers(Doebley et al., 1997) (Figure 4a). Maize, in contrast,typically has shortened lateral branches at nodes nearthe top of the plant, tipped by ears, but little branchingat lower nodes (Figure 4b), except for an occasionaltiller (a side branch arising from an embryonic nodeat the base of the plant). The phenotypic difference isaccomplished by both shortening of internodes and areduction of branch initiation in maize, particularly injuvenile and adult nodes. This branching differenceaffects the architecture of the plant, and in reducinglateral branches into ears, limits photosynthetic input

a

b

ps

ss

cupglu

rank

ogig

cup

Fig. 3. (a) Cross-section of a maize ear. Kernels are borne on the paired pedicellatespikelet (ps) and sessile spikelet (ss). Each spikelet pair forms within a cupule (cup)of the ear. In this cross-section, there are 7 ranks (rank) of spikelet pairs, each withina cupule. Each spikelet is associated with two glumes, one of which can be seenat the base of the kernel. (b) Longitudinal section of a maize ear . Kernels areborne within a cupule (cup) of the ear. Outer glumes (og) and inner glumes (ig) arereduced relative to teosinte.

and reproductive output to fewer, more concentratedlocations on the plant.

In total, these phenotypes represent the five key morphologi-cal differences between maize and teosinte, used in their mostrecent taxonomic revision (Doebley and Iltis, 1980; Iltis andDoebley, 1980). A number of other traits distinguish maizeand teosinte, many of which can be explained by the pre-mature cessation of growth in axillary branches, leaves, andinternodes (Doebley and Iltis, 1980). Other traits involvedin domestication, and improvement of maize allowing itsworldwide spread include a loss of seed dormancy (Aven-daño López et al., 2011), a loss of photoperiod sensitivity(Huang et al., 2018), nutritional alterations (Hanson et al.,1996; Whitt et al., 2002), and other phenotypes less apparentthan these morphological changes. Collectively, these phe-notypes represent radical departures of maize from Zea mayssubsp. parviglumis, yet both are classified as belonging tothe species Zea mays in recognition of the continued fertil-ity of these taxa, in the absence of large scale gametophyticincompatibility alleles (Kermicle, 2006).

Inheritance of phenotypes. Building on these observeddifferences between taxa, numerous studies attempted to un-derstand the genetic basis of this differentiation by cross-ing maize and teosinte to observe phenotypic distributions ofthese traits in their progeny. In large part these experimentsled to murky conflictions, likely reflecting the use of diversegermplasm, differing quantification of phenotypes, and widecrosses between species with reduced fertility. This confu-

Stitzer et al. | Maize domestication and gene interaction 3


a b

Fig. 4. (a) Teosinte plant architecture is branched, with multiple ears per plant. (b) Maize architecture is apically dominant, with side branches tipped by female inflorescences(ears).

sion as to the ancestor of maize would not be resolved untilmolecular methods supported Z. mays subsp. parviglumis asthe direct ancestor (Bennetzen et al., 2001), a taxon rarelyused in initial investigation.

Early research into maize origins made use of diversesamples of teosinte including Zea luxurians (Collins andKempton, 1920; Rogers, 1950a,b; Lambert and Leng, 1965)and Zea mays subsp. mexicana (Langham, 1940; Rogers,1950a,b), and a variety of maize lines such as Tom Thumbpopcorn (Collins and Kempton, 1920), a photoperiod insen-sitive maize inbred (Rogers, 1950a,b), maize ‘of mediummaturity’ (Langham, 1940), and Hy2 (Langham, 1940).Given our modern understanding of the divergence amonstthese teosinte taxa (Ross-Ibarra et al., 2009) and differencesin photoperiod sensitivity among these lines when grownat high latitudes (Emerson, 1924; Rogers, 1950b; Wilkes,1967), it is perhaps not surprising that crosses using differ-ent teosinte and maize samples produced confounding re-sults. For example, some studies showed the mature pairedspikelets of maize to be controlled by a single Mendelian lo-cus, but others considered the trait to be quantitative, con-trolled by many genes. While Rogers (1950a) found multiplegenes linked to spikelet pairing, he recovered a locus on thesame chromosome identified by Langham (1940) who con-sidered the trait to be controlled by a single gene. And whencrossing Zea mays subsp. mexicana to a Peruvian maize va-riety, Galinat (1959) found F1 ears to contain single maturespikelets, but when the same teosinte was crossed to NorthAmerican maize, spikelets were paired in the F1 (Mangels-dorf, 1974). This shows that a maize allele at one locus can-not be the only gene controlling paired spikelets, without al-tered dominance in different genetic backgrounds, environ-mental effects on the trait when grown in different years and

locations, or subjective phenotypic classification of the traitby different investigators.In total, about half of these studies classified the five ma-jor morphological differences between maize and teosinteas quantitative (Collins and Kempton, 1920; Mangelsdorf,1947; Rogers, 1950a,b), while the other half considered themto be controlled by a single Mendelian gene (Langham, 1940;Galinat, 1971, 1988). And while the original interpretation ofboth Mangelsdorf and Reeves (1939) and Beadle (1939) wasthat of four major genes or chromosomal regions of linkedgenes, many later investigations suggested almost every chro-mosome contributed to the domestication phenotype (Man-gelsdorf, 1947; Rogers, 1950a,b). Viewed from today’s per-spective, these investigations highlighted 1) that the key mor-phological differences between maize and teosinte are oftenoligogenic, 2) that substantial genetic variation exists withinboth maize and teosinte, 3) that consistent experimental de-sign and environment can alter interpretations, and 4) that thegenetic background a maize allele is found in can determineits effect on phenotype. But, even after observation of tensof thousands of plants across these experiments, the geneticbasis of differentiation between maize and teosinte was stillunclear, making it difficult to investigate how they evolvedand how domesticators selected on them.

QTL mapping beyond single genes. While connectingthe inheritance of individual phenotypes to their underlyinggenes was complicated by the features discussed above, an-other alternative was to identify progeny of a F2 populationresembling maize and teosinte parents (Beadle, 1972). Incontrast to earlier crosses between maize and teosinte, Beadle(1972) selected ‘primitive’ maize varieties to avoid confusinggenetic variation that was selected during the modern breed-



ing of maize with changes due to domestication. He grewapproximately 50,000 plants of a cross between the maizerace Chapalote and his most maize-like teosinte, the Chalcorace of Zea mays subsp. mexicana (Beadle, 1972, 1980). Ap-proximately 1 in 500 plants yielded ears looking like eitherthe maize or teosinte parent (Beadle, 1972, 1980). This re-duced the number of genes involved to between four and five,similar to that suggested by Langham (1940). It is notable,however that this observation inherently suggests some devi-ation from additive gene action — with four genes, 1 out of256 plants should have been similar to each parent, and withfive genes, 1 out of 1024.In order to understand the genetic basis of these traits, Doeb-ley and Stec (1991) repeated this exact cross, phenotyping ap-proximately 250 F2 progeny for domestication related traitsbut also genotyping each plant. They identified 58 genomicregions associated with their 12 phenotypes, spread across all10 chromosomes; most of these, however, were within 5 largeregions on chromosomes 1, 2, 3, 4, and 5. Some phenotypeswere controlled by large effect loci, like a single locus onchromosome 2 that explained 77.5% of the phenotypic vari-ance for the number of kernel rows per ear. But the majorityof associated regions explained less than 10% of variation,consistent with a more oligogenic or even polygenic archi-tecture. These researchers extended this work by crossingthe presumed direct ancestor of maize, the annual teosinteZea mays subsp. parviglumis and the maize race Reventador.Doebley and Stec (1993) largely recapitulated their previousresults, and identified 50 associations, including some locionly associated in one of their two populations. Clearly, locithat frequently show conditional associations are unlikely tobe the key differences between maize and teosinte. Bothstudies, however, agreed that these five genomic regions onchromosomes 1, 2, 3, 4, and 5 disproportionately control thephenotypic differences between maize and teosinte, includ-ing over 70% of the loci explaining more than 10% of phe-notypic variance in any trait (Doebley and Stec, 1991, 1993).This overrepresentation was later validated in a larger exper-iment using progeny of Zea mays subsp. parviglumis and theinbred line W22 backcrossed to W22 (Briggs et al., 2007), inwhich 64% of large effect loci were located to these regions.Foreseeably, these five regions contain major effect loci thatdifferentiate maize and teosinte. Further research has suc-ceeded in cloning the genes underlying some of these QTL,enabling in some cases identification of the specific muta-tion underlying phenotypic differences between maize andteosinte. These genomic regions are presented here in thesame order as the morphological traits were introduced.They are:

1. The paired spikelets of maize are associated with vari-ants on chromosome 1 and 3 across multiple crosses ofmaize and teosinte. Over half of phenotypic variationin paired spikelets can be explained by these two loci(Doebley and Stec, 1991, 1993), but the interval cov-ers most of chromosome 1 and may represent multipleQTL (Doebley and Stec, 1993). These QTL are bothepistatic and pleiotropic (Doebley et al., 1995a), with

altered allelic effects in maize versus teosinte back-grounds, and impacting many other traits like plant ar-chitecture. Other loci are associated with paired versussingle spikelets, notably those on chromosomes 2, 4,and 10 (Doebley and Stec, 1991, 1993). Some genesunderlying spikelet formation are known from devel-opmental genetic screens (ramosa1 (ra1), chromosome7; ramosa2 (ra2), chromosome 3; ramosa3 (ra3), chro-mosome 7) (Vollbrecht et al., 2005; McSteen, 2006),but these do not fall into the QTL intervals identifiedin these crosses. Presumably, these genes are mem-bers of pathways involving genes in these QTL, asra1 controls the switch to inflorescence determinacythat occurs with the production of spikelet pairs (Voll-brecht et al., 2005), and shows evidence of selectionduring domestication (Sigmon and Vollbrecht, 2010).Fine-mapping the loci distinguishing paired and singlespikelets is complicated by difficulty in phenotypingpaired spikelets, as their appearance can be difficult toidentify given the high number of kernels in each rowof the ear of modern maize parents (Galinat, 1988).

2. The two ranks of the teosinte ear are largely controlledby the gene Zea floricaula leafy2 (zfl2). This geneis responsible for reproductive identity, forming mul-tiple ranks along the inflorescence meristem. zfl2 isfound within a QTL interval on chromosome 2 that ex-plains 36-77.5% of phenotypic variance for the num-ber of rows of cupules (Doebley and Stec, 1991, 1993),and the gene itself explains 9% of variation in ear ranknumber across multiple maize-teosinte mapping pop-ulations that have an intact or mutant version of themaize allele (Bomblies and Doebley, 2006). Thereare additional QTL on chromosomes 1, 3, 4, 5, 9,and 10 that modify this phenotype (Doebley and Stec,1991, 1993; Briggs et al., 2007), including a para-log, Zea floricaula leafy1 (zfl1), on chromosome 10(Briggs et al., 2007) which may alter the effect of zfl2(Bomblies and Doebley, 2006).

3. The disarticulating rachis of the teosinte ear are con-trolled by alleles at ZmSh1-1and ZmSh1-5.1+ZmSh1-5.2, although variation within teosinte identifies Zeaagamous-like1 (zagl1) to be a candidate. This trait wasascribed to various loci that explained high amounts ofphenotypic variation in crosses using different teosinteparents (Doebley and Stec, 1991, 1993), and associa-tion mapping within teosinte identifies zagl1 as a po-tential candidate (Weber et al., 2008). Within maize,tandem duplicates of a ortholog of shattering 1 (Sh1)on chromosme 5 explain 23.1% of variation in shatter-ing (Lin et al., 2012), and maize alleles at ZmSh1-1, atranscription factor associated with shattering in othergrasses (Lin et al., 2012), elongate internodes via anepistatic interaction with tb1(Yang et al., 2016). zagl1,a MADS box transcription factor, is associated withear size and has pleiotropic effects on flowering time(Wills et al., 2018), but is unlikely to be the only locus



involved in ear disarticulation.

4. Although many ear traits differ between maize andteosinte, the most dramatic one observable in crossesis that of hardened glumes, controlled by the geneteosinte glume architecture1 (tga1). The maize al-lele of tga1 inhibits secondary sexual traits in the fe-male flower, preventing glumes from hardening (Pre-ston et al., 2012). A nonsynonymous mutation in exon1 of tga1 alters dimerization of the protein, affecting itsstability and preventing activation of downstream tar-gets (Wang et al., 2015). The chromosome 4 QTL thatcontains tga1 explains between 27-62.4% of pheno-typic variation for glume hardness (Doebley and Stec,1991, 1993; Briggs et al., 2007). Additionally, thisgene appears to have pleiotropic impacts on ear dis-articulation, lateral branch length, and ear phyllotaxy(Wang et al., 2015).

5. Aside from ear traits, the clearest morphological differ-ence between maize and teosinte is plant architecture,for which Doebley et al. (1995a) first identified teosintebranched1 (tb1) as the major locus. The QTL regiontb1 is found within explains 35.9% of variation in thenumber of basal branches or tillers (Doebley and Stec,1991). Later efforts identified the precise causal mu-ation: rather than a change in the coding sequence ofthe gene, a transposable element insertion 65 kb up-stream of the gene appears to enhance expression oftb1 (Studer et al., 2011). This increased expressionrepresses lateral branching, allowing the primary lat-eral inflorescence to compress into a female structure.The locus is allelic to a maize mutant that generatesa branched, tillered phenotype (Burnham, 1959), andother loci within the QTL are pleiotropic for ear archi-tecture traits (Studer and Doebley, 2011). Other genesinvolved in plant architecture were selected during do-mestication, including grassy tillers 1 (gt1), where themaize allele reduces the numerous small ears found inteosinte to a limited number of large ears (Wills et al.,2018).

Larger studies focusing on QTL for similar morphologicaltraits identified 314 QTL for 22 traits involved in domestica-tion and improvement (Briggs et al., 2007). But only 14 ofthese explained more than 10% of variation in a trait, whichoverlap (almost) perfectly with QTL regions shown to be ma-jor domestication QTL (Doebley and Stec, 1993). Some traitscould be explained by only six QTL, while others required26, and low-oligogenic traits were mostly for improvementpost-domestication. Together, this suggests that while thesemajor regions are important for domestication, many othergenes in the genetic background contributed and modulatedthese phenotypes.

Maize domestication involved many loci. Mangelsdorf(1974) attempted to reconstruct a teosinte phenotype by mov-ing the four major segments he identified as important fora maize phenotype from teosinte into a single maize inbred

line. This did not work, instead generating a plant indis-tinguishable from maize. However, selective breeding of ateosinte plant with maize ancestry rapidly turned a teosinte-like phenotype to maize in only 18 years (Weatherwax, 1924;Collins, 1925), suggesting segregating alleles can be selectedto reconstitute the maize phenotype. While the handful ofregions discussed above can have large phenotypic effects,it clearly takes more than five genes to make a maize plant.Briggs et al. (2007), in large crossing experiments betweenmaize and Z. mays subsp. parviglumis identified 314 QTL, ofwhich only 14 explained more than 10% of variance for anytrait. Only approximately 50% of total phenotypic variationin all traits could be explained by the identified QTL (Briggset al., 2007), leaving a large amount of unexplained vari-ance, presumably attributable to environmental differences,epistatic relationships, or small QTL that may become sta-tistically observable with differing experimental designs (Yuet al., 2008).In contrast to QTL approaches focused on identifying thegenetic basis of specific phenotypes, a number of popula-tion genetic studies have scanned the genomes of maize andteosinte for signs of natural selection, such as reduced nu-cleotide diversity around selected genes. Selection scanscan identify loci important for fitness beyond those underly-ing morphological differences associated with domestication,like loci involved in response to biotic or abiotic environ-ments. Analyses of microsatellite diversity (Vigouroux et al.,2002, 2005) and sequence data from hundreds of individualloci in teosinte and inbred maize lines (Wright et al., 2005)both suggested that 2-5% of the genome had been targetedby selection. Whole-genome resequencing of teosinte andtraditional maize landraces found a similar proportion of thegenome affected by selection, identifying 484 regions as out-liers, each likely representing a gene under selection duringdomestication (Hufford et al., 2012c). Together, these studiessuggest that a substantial proportion of the maize genome hasbeen selected during domestication. As selection during do-mestication occurred during a relatively short period of time,compatible interactions between many loci were required forphenotypes to shift.

The tempo of maize domesticationKnowledge of individual alleles that were selected duringdomestication allows investigation into the time these alle-les arose and how selection acted to increase their frequen-cies. Archaeological and molecular methods agree that maizewas domesticated 9000 years ago in the seasonally wet low-lands of the Balsas Valley of Mexico (Figure 1) (Matsuokaet al., 2002; Piperno et al., 2009; Ranere et al., 2009). Al-though macrobotanical maize remains from the region havenot been found, likely due to the poor conditions for preser-vation, phytoliths from cobs support the emergence of Zeamays subsp. mays in this region (Piperno et al., 2009). Thefirst macrobotanical remains morphologically characterizedas maize come from the dry highlands 6,200 years beforepresent (Mangelsdorf et al., 1967; Benz, 2001; Piperno andFlannery, 2001). Morphological intermediates are rare in



the archaeological record, reflecting a rapid transition froma teosinte-like phenotype to a maize-like phenotype. Inthe earliest archaeological samples of maize, cobs are non-disarticulating, have softened glumes with a narrow rachis,and shallow cupules with evidence of one or two spikelets(Mangelsdorf, 1974; Benz, 2001; Kennett et al., 2017).

Early genetic (Jaenicke-Despres et al., 2003) and recent ge-nomic (da Fonseca et al., 2015; Ramos-Madrigal et al., 2016;Vallebueno-Estrada et al., 2016; Swarts et al., 2017) investi-gation of archaeological samples integrate morphology andgenotypes at known domestication loci. A reduction in diver-sity around tb1 in all archaeological samples indicates thatthe maize allele was present, suggesting this allele rapidly in-creased in frequency. The maize allele at tb1 arose 28,000years before present (Studer et al., 2011), and is found seg-regating at frequencies up to 44% in teosinte populations(Studer et al., 2011; Vann et al., 2015). That the allele waspresent before domestication, and does not always generatea maize-like phenotype when present in teosinte (Vann et al.,2015) shows that the presence of the maize allele at tb1 is notenough to generate a maize-like phenotype. In contrast, themaize allele at tga1 that restructures glumes and cupules ofthe ear was fixed within the last 10,000 years (Wang et al.,2005), likely a result of selection on a new non-synonymousmutation (Wang et al., 2015). Of the two oldest cob sam-ples sequenced, Ramos-Madrigal et al. (2016) find the maizeallele at tga1, while Vallebueno-Estrada et al. (2016) do notrecover the causal polymorphism, but find elevated diversityin the region, which they suggest reflects incomplete selec-tion on the maize allele. Consistent with rapid selection af-ter mutational origin, no phenotypes resembling the effect ofa mutation at tga1 were observed when George Beadle ex-amined more than 2 million teosinte seeds from a number ofwild populations (Beadle, 1980; Berg and Singer, 2005). Mu-tations that alter the fruitcase were thus likely rare or absentin teosinte populations, and presence of the maize allele in ar-chaeological samples is consistent with rapid fixation duringduring domestication.

In order to ascertain the presence of a maize allele in ateosinte population, and identify when selection acted, thefunctionally relevant causal variant must be identified. Un-fortunately, tb1 and tga1 remain the only loci selected duringmaize domestication for which this level of detail is known.Outside of the alleles at tb1 and tga1, only the gt1 locushas been investigated in this manner, with a maize allele thatarose 13,000 years ago, and is found segregating at low fre-quencies in teosinte (Wills et al., 2013). The relative im-portance of these two extremes — from segregating variationfound today in teosinte (tb1 and gt1), and from a new de novomutation (tga1) — cannot be generalized without further re-search into the causal alleles of maize domestication.

The presence of standing variation can accelerate the speedof evolution when selection pressures shift. Rather than wait-ing for a mutation that generates a phenotype, the presence ofboth genetic and phenotypic variation allows rapid selection.Whether an allele will be found as standing variation dependson how deleterious a mutation is — which can be affected by

its dominance, epistatic interactions with other alleles, andits pleiotropic effects on other traits. Hence, it is important tointerpret the effect of any allele not independently, but in ag-gregate with consideration of other loci throughout the maizegenetic background.

Genetic interactions and selection duringmaize domesticationShifting dominance of domestication alleles in maize.Dominance, or interaction between alleles at a single locus,affects the exposure of an allele to selection. Although domi-nance modifiers can evolve, the recessivity of new mutationsseems to be a general feature (Orr, 1991). Dominance is in-formative as to how selection could act on new mutations,and relevant to thinking about the visibility of segregatingvariation to selection. As the alleles present at a locus varybetween individuals, dominance can alter the phenotypic out-come of different crosses.In Zea mays, the dominance of a given allele can differ basedon its genetic background. When QTL carrying a maize al-lele from chromosome 1 or 3 were introgressed into a teosintebackground, the maize allele was on average recessive tothe teosinte allele in its effect on seven different phenotypes(Doebley et al., 1995a). But when a teosinte allele at eitherlocus was introgressed into a maize background, the maizeallele was partially dominant to the teosinte allele, an ef-fect also seen when segregating in a F2 population (Doeb-ley et al., 1995a). Dominant maize-like alleles arising in ateosinte background may be rapidly selected against, as if amaize-like phenotype arises in teosinte, it will be detrimentalto plant fitness. And as the alleles we today denote as ‘maize’initially arose in a largely teosinte background, their domi-nance measured today in maize may not reflect that they firstexperienced. If this recessivity of new ‘maize’ alleles is gen-eral to domestication loci, alleles could survive with relaxedfitness costs until human selection increased their frequency(Doebley et al., 1995a; Lauter and Doebley, 2002).Although the dominance relationships of these two QTL havebeen most extensively studied, other evidence exists for therecessivity of maize alleles in a teosinte background and theirimpacts on human selection. The maize allele of tga1 is dom-inant to the teosinte allele, reducing deposition of silica inglumes (Dorweiler and Doebley, 1997). In a maize back-ground, the teosinte allele of tga1 decreases grain quality,because kernel growth is restricted by the hardened glumes,leading to cracking and susceptibility to pathogens (Dor-weiler et al., 1993). But the maize allele of tga1 likely arosein a teosinte genetic background and the effect of the maizetga1 allele in a teosinte background is less detrimental. Al-though ears of such a plant are shorter, they are still pro-tected within husks until harvest (Dorweiler et al., 1993),and although mature kernels are exposed to pests, cultiva-tion practices can ameliorate such danger. This phenotype ofthe maize allele in a teosinte background could allow visualidentification of heterozygotes (Wang et al., 2005), yet nonewere identified in a screen of more than 2 million teosinteseeds (Beadle, 1980; Wilkes, 2004; Berg and Singer, 2005).



Consistent with this, the maize allele appears to have arisende novo during domestication (Wang et al., 2015), suggestingthe necessity of human cultivation to select for this dramaticphenotype.Beyond single loci, the phenotypic means in F2’s of maize ×teosinte crosses and backcrosses of teosinte to maize deviatetowards the teosinte parent (Lambert and Leng, 1965; Doe-bley et al., 1990; Doebley and Stec, 1993; Doebley et al.,1995a). While some of these results could reflect epistasis, itnonetheless suggests that for a substantial portion of the ge-netic background, the teosinte allele is dominant in a teosintebackground — most similar to the genomic environment anallele would be selected in.Another way to consider dominance is through the molecu-lar phenotype of gene expression. Transcriptome-wide, 52-58% of genes have maize alleles that are more highly ex-pressed than teosinte alleles (Hufford et al., 2012c; Swanson-Wagner et al., 2012; Lemmon et al., 2014; Wang et al., 2018),which may result from selection for consistent expressionacross changing environments (Doebley et al., 1995a; Lo-rant et al., 2017). Further, this pattern is intensified in do-mestication candidate genes, with 60% of candidate genesshowing higher expression of the maize allele (Hufford et al.,2012c; Swanson-Wagner et al., 2012). In allele-specific ex-pression studies of crosses between maize and teosinte, themaize allele of genes with trans regulatory control are morecommonly dominant to the teosinte allele in ear and leaf tis-sue, consistent with a contribution of dominant transcriptionfactors and expression modifiers by the maize parent (Lem-mon et al., 2014). The effect is not subtle, with 70-80%of genes showing dominance of the maize allele (Lemmonet al., 2014). Both dominance of expression and higher ex-pression in maize may reflect selection for robustness of ex-pression in the face of differing environmental conditions, as-suring fitness under cultivation (Doebley et al., 1995a).Although maize alleles are dominant to teosinte alleles in amaize background, dominance relationships shift when alle-les are segregating in an F2, a teosinte backcross, or a maizebackcross (Doebley et al., 1995a; Briggs et al., 2007; Doustet al., 2014). This suggests that interactions among loci —perhaps many loci — can alter the efficacy of selection on aparticular trait.

Epistasis. Epistasis can be envisioned in numerous ways,but we highlight two extremes. Statistical epistasis refersto deviations from additive relationships in a model (Fisher,1918), while biological epistasis describes the interaction ofgene products in vivo (Bateson, 1909). The two can be diffi-cult to distinguish with experimental data, because, for exam-ple, it is impossible to identify stastistical epistasis for a locusexhibiting biological epistasis if the experimental populationlacks variation for one interacting gene partner.In maize, statistical epistasis is rarely observed in QTL anal-yses (Edwards et al., 1987; Stuber et al., 1992; Briggs et al.,2007) or in genome-wide scans in panels of inbred maizelines (Wallace et al., 2014) (but see Shi et al. (2017) forepistatic effects in leaf orientation), but is more commonlyfound when individual cloned QTL are placed into other-

wise isogenic maize or teosinte backgrounds (Doebley et al.,1995a; Weber et al., 2008; Studer and Doebley, 2011). Asvariation both in the genotype and phenotype of a causativelocus must occur in a mapping population, studies focusingon maize specifically may fail to observe differences. A lackof genetic variation may not be due solely to selection havingremoved it — the genetic bottleneck arising from maize do-mestication altered allele frequencies throughout the genome(Eyre-Walker et al., 1998; Tenaillon et al., 2004; Wrightet al., 2005; Beissinger et al., 2016), and modern maize oftenlacks phenotypic variation for relevant traits (Briggs et al.,2007; Xue et al., 2016; Xu et al., 2017). Statistical epistasisthus may not exist in most mapping populations and exper-iments. Consistent with this argument, statistical epistasisis identified with comparable ease in QTL populations thatinclude a teosinte parent (Weber et al., 2008), although thisdoes not always ameliorate the issue of limited allelic input.For example, zfl2, the locus controlling ear phyllotaxy, wasnot recovered in a backcross of Zea mays subsp. parviglu-mis to inbred maize (Briggs et al., 2007), complicating theinterpretation of this major effect locus.

While insufficient variation at the loci involved is likely atleast partly responsible for discrepancies among studies, thedesign of mapping populations can also dramatically impactthe power to detect different forms of epistasis. Because ofthe large number of potential combinations and the need tocontrol for genetic background, very large experimental pop-ulations are needed to test for statistical epistasis, and oftenonly the strongest effects can be identified. Indeed, by pheno-typing seven times more progeny than earlier mapping stud-ies of maize and teosinte, Briggs et al. (2007) identified 29two-locus epistatic interactions, although only one was foundin both environments studied.

In addition to sample size, the kinds of crosses made will de-termine what allelic variation is present that could be scoredas epistasis. For example, in an F2 between maize andteosinte, the combined additive and epistatic effects of twoQTL, on chromosomes 1 and 3, explain 60% of variation inpaired vs. single spikelets (Doebley et al., 1995a), but whenthese regions are introduced into a teosinte background viabackcrossing they explain only 7.3% of variation in this phe-notype. This suggests that numerous other genes in the ge-netic background interact to generate this phenotype and sup-ports earlier experiments that found different numbers of locicontrolling the trait in progeny from different crosses (Lang-ham, 1940; Szabó and Burr, 1996).

During domestication, epistatic variation may be convertedto additive variation, as alleles fix at one or more of a setof interacting loci. But during intermediate phases after anallele arises but before selection fixes it, epistasis may al-ter the efficacy of selection. This can be seen in the interac-tion between QTL on chromosomes 1 and 3. When the fre-quency of the maize allele of the chromosome 1 QTL is low,the chromosome 3 QTL has little effect on the proportionof branches terminated by male inflorescences, a teosinte-like trait (Doebley et al., 1995a). But when the chromosome1 allele containing tb1 increases in frequency, the ability to



select on its interacting partner on chromosome 3 increases,as this epistatic variance increases at intermediate allele fre-quencies (Goodnight, 2004). With both teosinte alleles ina maize background, terminal inflorescences are 90% male,but by simply substituting either QTL, this proportion is re-duced to 21% with a teosinte allele at chromosome 1 or 0.5%with a teosinte allele at chromosome 3 (Lukens and Doeb-ley, 1999). The recent characterization of the candidate genetassels replace upper ears1 (tru1) within the chromosome 3QTL (Dong et al., 2017) and additional candidates within thechromosome 1 QTL (Yang et al., 2016) may allow finer scaletemporal tracking of allele frequencies and the role of selec-tion on epistatic partners. It appears that both the phenotypepresented to selection and the response to selection are de-pendent on other loci in the genome. Although the statisticalpower to detect epistasis may be poor, such examples of bio-logical epistasis may be common. For example, much of theshade avoidance pathway downstream of tb1 contains genesshown to be targets of selection (Studer et al., 2017), but thesegenes are not detected in screens for statistical epistasis. To-gether, despite the fact that few loci have shown evidence ofstatistical epistasis in mapping studies, there is evidence forepistasis — both statistical and biological — contributing todomestication.In total, these epistatic effects and the importance of the ge-netic background may alter the course of selection on phe-notypes. If buffered by their interaction with other genes,maize alleles could have been maintained in wild populationsof teosinte with little effect on phenotype or fitness. Indeed, anumber of experiments in teosinte have demonstrated the ex-istence of such cryptic variation for maize-like traits (Lauterand Doebley, 2002; Weber et al., 2007, 2008; Vann et al.,2015). The introduction of new variation — via new muta-tions or hybridization between populations — could then re-lease cryptic epistatic variation, generating novel phenotypes(Doebley et al., 1995a).

Pleiotropy. Pleiotropy describes the effect of an allele of agene on seemingly unrelated phenotypes. It can constrain thepath of evolution, as directional selection on a phenotype canbe limited if the allele has deleterious effects on another phe-notype, or facilitated if the allele has beneficial effects on fit-ness in different life stages. Alleles that are more pleiotropicare more likely to negatively impact at least one other trait(Fisher, 1918). Understanding pleiotropy and the correlationof phenotypes can help to understand constraint on the out-come of selection.But pleiotropy can be confusing, as phenotypes that appearsuperficially different may not be truly distinct. Plants areconstructed of repeated units of leaf, stem, and bud, knownas phytomers. The genes involved in generating these phy-tomers thus can be readily pleiotropic via development, hav-ing an effect on phenotypes that may appear at first glancedistinct. In light of the phytomer, it is not entirely surpris-ing that pleiotropic loci explain correlation in developmentaltraits of ear and tassel (Brown et al., 2011), flowering time inmale and female flowers (Buckler et al., 2009), or leaf lengthand flower length (Tian et al., 2011). But pleiotropic loci

extend even beyond the phytomer, as QTL involved in tas-sel and ear development are also classified as flowering timegenes (Xu et al., 2017). In many such studies, it is not yetclear how many genes contribute to the observed pleiotropy,as efforts to fine-map individual QTL can split effects withinthe region into multiple heritable loci (Lemmon and Doebley,2014).

Understanding pleiotropy and the correlation of phenotypescan help to understand constraint on the outcome of selection.For example, the maize allele at zfl2, is implicated not onlyin the spiral ear phyllotaxy that generates increased kernelnumber but also in a number of traits including earlier flower-ing (Bomblies and Doebley, 2006). In an environment wheremaize is well-adapted, stabilizing selection on flowering timemight limit the response to directional selection for increasedkernel number. Hence, such pleiotropy may constrain domes-tication alleles when selection axes disallow variation. Forexample, if selecting for increased kernel row number wasaccompanied by an acceleration of flowering time to valuestoo early for current environmental conditions, this correlatedselection may limit response to selection for increased kernelnumber.

It has long been noted that many of the loci that differenti-ate maize and teosinte are pleiotropic (Collins and Kempton,1920; Beadle, 1939; Mangelsdorf and Reeves, 1939; Lang-ham, 1940). Recent dissection of the regulatory architec-ture of tb1 provides a detailed understanding of pleiotropyfor a single domestication gene. tb1 is pleiotropic acrossmany traits — apical dominance, length of lateral branches,growth of leaves on the lateral branches, pedicillate spikeletdevelopment, and root architecture (Hubbard et al., 2002;Gaudin et al., 2014). As a transcription factor, tb1 bindsto many regions of the genome. It directly regulates tga1by binding its promoter but is also intimately linked to thecell cycle as it represses two cell cycle genes (proliferatingcell nuclear antigen2(pcna2) and minichromosome mainte-nance2/prolifera (mcm2/prl)) (Studer et al., 2017). Beyondtb1, other loci within the QTL region on 1L found by mul-tiple studies (Doebley and Stec, 1991, 1993; Briggs et al.,2007) contribute to ear morphology (Studer and Doebley,2011; Yang et al., 2016), suggesting pleiotropy is common.

The presence of pleiotropy creates genetic correlationsamong traits, potentially constraining the action of selection.Perhaps because of this kind of constraint, the only muta-tion thought to have arisen de novo and rapidly fixed dur-ing domestication is the nonsynonymous substitution in tga1.While the tga1 ortholog in rice has pleiotropic effects on in-florescences and vegetative structures (Preston et al., 2012;Wang et al., 2015), tga1 is expressed only in the maize ear,likely a result of gene duplication of an ancestral single lo-cus, with subsequent subfunctionalization restricting expres-sion to the ear (Preston et al., 2012; Wang et al., 2015). Theparalogous locus, not tga1(not1), retains expression profilesas found in other grasses (Wang et al., 2011, 2012), suggest-ing that in maize, the effects of the maize allele of tga1 arelimited to the fruitcase itself, freeing it from constraints on se-lection imposed by pleiotropic effects elsewhere in the plant.



Gene networks of maize domestication alle-lesThe molecular basis of pleiotropy may occur when key genesin regulatory networks alter different downstream targets, af-fecting different pathways. Selection on correlated traits dur-ing domestication could act through a single mutation to anode in a regulatory network, if the trait correlation is dueto the network. One of the biggest alterations in the transi-tion from teosinte to maize is the floral morphology of theear, and the pleiotropic consequences of downstream targetsof key domestication genes seem to be involved. MADSbox transcription factors are overrepresented as showing ev-idence of selection during maize domestication as comparedto random genes (Zhao et al., 2011). In plants, MADS boxtranscription factors are disproportionately represented in thepathways specifying floral organs (Theißen, 2001), and sev-eral MADS box genes are directly regulated by either tga1or tb1 in maize (Wang et al., 2015; Studer et al., 2017). Bytargeting floral-organ genes, pleiotropic outcomes are limitedin scope to those affecting ear morphology, from the shape ofthe rachis to changes in lignification and silica deposition inthe glume and rachis (Doebley, 1996; Dorweiler and Doeb-ley, 1997). Indeed, with the exception of tb1, many allelesselected during maize domestication only altered ear mor-phology. Selection during domestication can amplify the im-pacts of alelles through pathways and networks, generatingmore extreme visible phenotypic outcomes, often intensifiedby dominance and epistasis.Efforts to unite domestication loci into pathways and net-works have elucidated the targets of selection. In contrast tothe largely background-specific effects of many maize alle-les, tb1 remains robust to genetic background — so much sothat tillering was not phenotyped in F2 crosses beyond initialwork by Doebley and Stec (1991). That tb1 was so routinelyimplicated in differences between maize and teosinte may betrue because it has an effect in every population tested thus farand because it is near the top of the shade avoidance pathway(Studer et al., 2017). Consequently, phenotypic effects can beamplified and fine-tuned through downstream targets. Addi-tionally, several downstream targets of tb1 show signatures ofselection (Studer et al., 2017), suggesting further constrainton the entire pathway.Nonallelic mutations in gene networks can result in pheno-types that look superficially similar to those selected duringdomestication. Although recovery of spontaneous maize mu-tants has been one of the most useful features of maize asa model genetic organism (Strable and Scanlon, 2009; Nan-nas and Dawe, 2015), relating phenotypes from such alle-les to natural variation can sometimes be misleading. Spon-taneous mutant phenotypes that make a maize plant lookmore like teosinte are common (e.g. suppressor of sessilespikelets1(sos1) (Doebley et al., 1995b), barren stalk1 (ba1)(Gallavotti et al., 2004), tru1 (Dong et al., 2017), tunicate1(tu1) (Wingen et al., 2012), corngrass1 (cg1) (Chuck et al.,2007)). Upon detailed analyses, however, alleles of these locioften do not show population genetic signatures of selectionduring domestication and lack functional differentiation be-

tween the maize and teosinte alleles. Since large effect hubsof gene networks — like tb1 — are frequently observed asthe target of selection, there is genetic redundancy in gen-erating similar phenotypes, by impacting different stages inpathways or positions in networks important for plant devel-opment (Doebley et al., 2006).

Implications of gene interactions on evolu-tion and selectionInvestigation into domestication is by definition retrospective— we are attempting to reconstruct how selection acted togenerate phenotypes observed today. But an understanding ofthe importance of genetic interactions, including dominance,epistasis, and pleiotropy — and how they have complicateddissection of even simple phenotypes may provide guidelinesfor future efforts seeking to identify the genetic basis of phe-notypic variation. That genetic background can impact phe-notype is not a surprise, for example, as textbook examplesof epistasis include anthocyanin genes in maize kernels (Coeet al., 1988). But it does suggest that in order to understandthe effect of an allele, researchers should test it in diversegenetic backgrounds. This is especially true in a genome asdiverse as maize. Sampling of a set of modern inbred lines— even excluding most of the vast diversity in landracesand teosinte — finds a single nucleotide variant every 25base pairs (Bukowski et al., 2017). In addition to single nu-cleotide variants, variation in gene copy number (Swanson-Wagner et al., 2010) or position (Liu et al., 2012) and on-going subfunctionalization (Pophaly and Tellier, 2015) high-light the extent of allelic diversity in maize, providing exten-sive opportunity across genetic backgrounds for a multitudeof genetic interactions. Finally, though we often try to under-stand the genetic basis of phenotypes through mapping genes,the majority (85%) of the maize genome is derived fromtransposable elements (Schnable et al., 2009), which showdramatic variation in presence and absence among geno-types (Wang and Dooner, 2006) and can generate substan-tial background-specific effects, including synthetic lethality(de la Luz Gutiérrez-Nava et al., 1998). In short, geneticbackground may have more pervasive effects than expectedwhen only looking at QTL from a single cross, and definingthe expected behavior of an allele based on a single back-ground likely provides a woefully incomplete picture of whatis likely a rich landscape of genetic interactions.

ConclusionHistorically, hypotheses about the genetic architecture ofmaize domestication have varied between the extremes of afew large-effect loci (Beadle, 1939; Mangelsdorf and Reeves,1939) to extremely polygenic (Iltis, 1983). Mapping of lociinvolved has tempered these two extremes, identifying hun-dreds of QTL (Briggs et al., 2007) or genes (Wright et al.,2005; Hufford et al., 2012c) but also identifying large ef-fect loci that explain the majority of variation for some traits.To understand the function of an allele, biologists often re-strict study to the genetic backgrounds in which the allele



is most penetrant and expressive. Termed ‘breeding dissec-tion’ (Wilkes, 2004), this essentially erases the backgroundnoise of polygenicity by isolating key loci in restricted ge-netic backgrounds. But careful genetic examination has alsoshown that dominance, epistasis, and pleiotropy play signif-icant roles in modulating phenotypes on which selection canact, and may help explain contrasting results from investiga-tions of single loci and those of broader mapping studies. Thenovel selective pressure of maize domestication generatedconditions amenable to understanding how evolution workswhen selective optima shift. Through careful genetic anal-yses of these phenotypes, genic interactions — at the levelof dominance, epistasis, and pleiotropy — have been discov-ered to play an important role in the evolution of the maizephenotype.

AcknowledgmentsWe are grateful to Mitchell Provance for his illustrations inFigure 2, 3, and 4. We thank John Doebley, Virginia Wal-bot, and Lynda Delph for helpful comments and perspectives,Jean-Philippe Vielle-Calzada and two anonymous reviewersfor fantastic reviews, and Wenbin Mei, Markus Stetter, AnneLorant and others in the Ross-Ibarra lab for comments, all ofwhich improved the clarity and message of this manuscript.Finally, we are grateful to Jeffry Mitton for providing tinycobs from Beadle’s F2 experiment.M.C.S. acknowledges support from the National ScienceFoundation Graduate Research Fellowship under Grant No.1650042; J.R.-I. would like to acknowledge support from theUSDA Hatch project CA-D-PLS-2066-H.

Author ContributionsM.C.S and J.R.-I. planned and designed the research andwrote the manuscript.

Avendaño López, A. N., J. de Jesús Sánchez González, J. A. Ruíz Corral, L. De La Cruz Larios,F. Santacruz-Ruvalcaba, et al., 2011 Seed dormancy in mexican teosinte. Crop Science 51:2056–2066.

Bartlett, M. E. and B. Thompson, 2014 Meristem identity and phyllotaxis in inflorescence develop-ment. Frontiers in Plant Science 5: 508.

Bateson, W., 1909 Heredity and variation in modern lights. Darwin and Modern Science 85: 101.Beadle, G. W., 1939 Teosinte and the origin of maize. Journal of Heredity 30: 245–247.Beadle, G. W., 1972 Mystery of maize. Field Museum Natural History Bulletin 43: 2–11.Beadle, G. W., 1980 The ancestry of corn. Scientific American 242: 112–119.Beissinger, T. M., L. Wang, K. Crosby, A. Durvasula, M. B. Hufford, et al., 2016 Recent demogra-

phy drives changes in linked selection across the maize genome. Nature Plants 2: 16084.Bennetzen, J., E. Buckler, V. Chandler, J. Doebley, J. Dorweiler, et al., 2001 Genetic evidence and

the origin of maize. Latin American Antiquity 12: 84–86.Benz, B. F., 2001 Archaeological evidence of teosinte domestication from Guilá Naquitz, Oaxaca.

Proceedings of the National Academy of Sciences 98: 2104–2106.Berg, P. and M. Singer, 2005 George Beadle, an Uncommon Farmer: The Emergence of Genetics

in the 20th Century . Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.Bernal, J. S., A. M. Dávila-Flores, R. F. Medina, Y. H. Chen, K. E. Harrison, et al., 2017 Did maize

domestication and early spread mediate the population genetics of corn leafhopper? InsectScience doi: 10.1111/1744-7917.12555.

Bomblies, K. and J. F. Doebley, 2006 Pleiotropic effects of the duplicate maize floricaula/leafygenes zfl1 and zfl2 on traits under selection during maize domestication. Genetics 172: 519–531.

Briggs, W. H., M. D. McMullen, B. S. Gaut, and J. Doebley, 2007 Linkage mapping of domesticationloci in a large maize–teosinte backcross resource. Genetics 177: 1915–1928.

Brown, P. J., N. Upadyayula, G. S. Mahone, F. Tian, P. J. Bradbury, et al., 2011 Distinct geneticarchitectures for male and female inflorescence traits of maize. PLoS Genetics 7: e1002383.

Buckler, E. S., J. B. Holland, P. J. Bradbury, C. B. Acharya, P. J. Brown, et al., 2009 The geneticarchitecture of maize flowering time. Science 325: 714–718.

Bukowski, R., X. Guo, Y. Lu, C. Zou, B. He, et al., 2017 Construction of the third-generation Zeamays haplotype map. GigaScience 7: gix134.

Burnham, C., 1959 New characters. Maize Newsletter 33: 74.

Burton, A. L., K. M. Brown, and J. P. Lynch, 2013 Phenotypic diversity of root anatomical andarchitectural traits in Zea species. Crop Science 53: 1042–1055.

Chavez, N. B., J. J. Flores, J. Martin, N. C. Ellstrand, R. Guadagnuolo, et al., 2012 Maize xteosinte hybrid cobs do not prevent crop gene introgression. Economic Botany 66: 132–137.

Chuck, G., R. Meeley, E. Irish, H. Sakai, and S. Hake, 2007 The maize tasselseed4 micrornacontrols sex determination and meristem cell fate by targeting tasselseed6/indeterminatespikelet1. Nature Genetics 39: 1517–1521.

Coe, E., M. Neuffer, and D. Hoisington, 1988 The genetics of corn. In Corn and Corn Improve-ment , edited by G. Sprague and J. Dudley, pp. 81–258, American Society of Agronomy, CropScience Society of America, Soil Science Society of America, Madison, WI, USA.

Collins, G., 1925 The “metamorphosis” of Euchlaena into maize. Journal of Heredity 16: 378–380.Collins, G. N. and J. H. Kempton, 1920 A teosinte-maize hybrid. Journal of Agricultural Research

19: 1–37.da Fonseca, R. R., B. D. Smith, N. Wales, E. Cappellini, P. Skoglund, et al., 2015 The origin and

evolution of maize in the southwestern united states. Nature Plants 1: 14003.de la Luz Gutiérrez-Nava, M., C. A. Warren, P. León, and V. Walbot, 1998 Transcriptionally active

MuDR, the regulatory element of the Mutator transposable element family of Zea mays, ispresent in some accessions of the Mexican land race Zapalote chico. Genetics 149: 329–346.

Doebley, J., 1996 Genetic dissection of the morphological evolution of maize. Aliso: A Journal ofSystematic and Evolutionary Botany 14: 297–304.

Doebley, J. and A. Stec, 1991 Genetic analysis of the morphological differences between maizeand teosinte. Genetics 129: 285–295.

Doebley, J. and A. Stec, 1993 Inheritance of the morphological differences between maize andteosinte: comparison of results for two F2 populations. Genetics 134: 559–570.

Doebley, J., A. Stec, and C. Gustus, 1995a teosinte branched1 and the origin of maize: evidencefor epistasis and the evolution of dominance. Genetics 141: 333–346.

Doebley, J., A. Stec, and L. Hubbard, 1997 The evolution of apical dominance in maize. Nature386: 485.

Doebley, J., A. Stec, and B. Kent, 1995b Suppressor of sessile spikelets 1 (sos1): a dominantmutant affecting inflorescence development in maize. American Journal of Botany 82: 571–577.

Doebley, J., A. Stec, J. Wendel, and M. Edwards, 1990 Genetic and morphological analysis of amaize-teosinte F2 population: implications for the origin of maize. Proceedings of the NationalAcademy of Sciences 87: 9888–9892.

Doebley, J. F., B. S. Gaut, and B. D. Smith, 2006 The molecular genetics of crop domestication.Cell 127: 1309–1321.

Doebley, J. F. and H. H. Iltis, 1980 Taxonomy of Zea (Gramineae). I. a subgeneric classificationwith key to taxa. American Journal of Botany 67: 982–993.

Dong, Z., W. Li, E. Unger-Wallace, J. Yang, E. Vollbrecht, et al., 2017 Ideal crop plant architectureis mediated by tassels replace upper ears1, a btb/poz ankyrin repeat gene directly targeted byteosinte branched1. Proceedings of the National Academy of Sciences 114: E8656–E8664.

Dorweiler, J. and J. Doebley, 1997 Developmental analysis of teosinte glume architecture1: A keylocus in the evolution of maize (Poaceae). American Journal of Botany 84: 1313–1313.

Dorweiler, J., A. Stec, J. Kermicle, and J. Doebley, 1993 Teosinte glume architecture 1: a geneticlocus controlling a key step in maize evolution. Science 262: 233–235.

Doust, A. N., L. Lukens, K. M. Olsen, M. Mauro-Herrera, A. Meyer, et al., 2014 Beyond thesingle gene: How epistasis and gene-by-environment effects influence crop domestication.Proceedings of the National Academy of Sciences 111: 6178–6183.

Edwards, M., C. Stuber, and J. Wendel, 1987 Molecular-marker-facilitated investigations ofquantitative-trait loci in maize. I. numbers, genomic distribution and types of gene action.Genetics 116: 113–125.

Emerson, R., 1924 Control of flowering in teosinte: Short-day treatment brings early flowers.Journal of Heredity 15: 41–48.

Eyre-Walker, A., R. L. Gaut, H. Hilton, D. L. Feldman, and B. S. Gaut, 1998 Investigation of thebottleneck leading to the domestication of maize. Proceedings of the National Academy ofSciences 95: 4441–4446.

Fisher, R. A., 1918 The correlation between relatives on the supposition of mendelian inheritance.Transactions of the Royal Society of Edinburgh 52: 399–433.

Gaillard, M. D., G. Glauser, C. A. Robert, and T. C. Turlings, 2018 Fine-tuning the plantdomestication-reduced defense hypothesis: specialist vs generalist herbivores. New Phytolo-gist 217: 355–366.

Galinat, W. C., 1959 The phytomer in relation to floral homologies in the American Maydeae.Botanical Museum Leaflets, Harvard University 19: 1–xvi.

Galinat, W. C., 1971 The origin of maize. Annual Review of Genetics 5: 447–478.Galinat, W. C., 1988 The origin of corn. In Corn and Corn Improvement , edited by G. Sprague

and J. Dudley, pp. 1–31, Madison, WI, USA, American Society of Agronomy, Crop ScienceSociety of America, Soil Science Society of America.

Gallavotti, A., Q. Zhao, J. Kyozuka, R. B. Meeley, M. K. Ritter, et al., 2004 The role of barren stalk1in the architecture of maize. Nature 432: 630–635.

Gaudin, A. C., S. A. McClymont, S. S. Soliman, and M. N. Raizada, 2014 The effect of altereddosage of a mutant allele of teosinte branched 1 (tb1-ref) on the root system of modern maize.BMC Genetics 15: 23.

Goodnight, C. J., 2004 Metapopulation quantitative genetics: the quantitative genetics of popula-tion differentiation. In Ecology, Genetics and Evolution of Metapopulations, edited by I. Hanskiand O. Gaggiotti, pp. 199–223, Elsevier Academic Press, Burlington, MA, USA.

Hanson, M. A., B. S. Gaut, A. O. Stec, S. I. Fuerstenberg, M. M. Goodman, et al., 1996 Evolution ofanthocyanin biosynthesis in maize kernels: the role of regulatory and enzymatic loci. Genetics143: 1395–1407.

Harshberger, J., 1896 Fertile crosses of teosinte and maize. Garden and Forest 9: 398–402.Huang, C., H. Sun, D. Xu, Q. Chen, Y. Liang, et al., 2018 ZmCCT9 enhances maize adaptation to

higher latitudes. Proceedings of the National Academy of Sciences 115: E334–E341.Hubbard, L., P. McSteen, J. Doebley, and S. Hake, 2002 Expression patterns and mutant pheno-

type of teosinte branched1 correlate with growth suppression in maize and teosinte. Genetics162: 1927–1935.

Hufford, M. B., P. Bilinski, T. Pyhäjärvi, and J. Ross-Ibarra, 2012a Teosinte as a model system for



10.1111/1744-7917.12555

population and ecological genomics. Trends in Genetics 28: 606–615.Hufford, M. B., E. Martínez-Meyer, B. S. Gaut, L. E. Eguiarte, and M. I. Tenaillon, 2012b Infer-

ences from the historical distribution of wild and domesticated maize provide ecological andevolutionary insight. PLoS One 7: e47659.

Hufford, M. B., X. Xu, J. Van Heerwaarden, T. Pyhäjärvi, J.-M. Chia, et al., 2012c Comparativepopulation genomics of maize domestication and improvement. Nature Genetics 44: 808–811.

Iltis, H. H., 1983 From teosinte to maize: the catastrophic sexual transmutation. Science 222:886–894.

Iltis, H. H., 2000 Homeotic sexual translocations and the origin of maize (Zea mays, Poaceae): Anew look at an old problem. Economic Botany 54: 7–42.

Iltis, H. H. and J. F. Doebley, 1980 Taxonomy of Zea (Gramineae). II. subspecific categories in theZea mays complex and a generic synopsis. American Journal of Botany 67: 994–1004.

Jackson, D. and S. Hake, 1999 Control of phyllotaxy in maize by the abphyl1 gene. Development126: 315–323.

Jaenicke-Despres, V., E. S. Buckler, B. D. Smith, M. T. P. Gilbert, A. Cooper, et al., 2003 Earlyallelic selection in maize as revealed by ancient DNA. Science 302: 1206–1208.

Kellogg, E. A., 2000 A model of inflorescence development. In Monocots: Systematics and Evolu-tion, edited by K. Wilson and D. Morrison, pp. 84–88, CSIRO Publishing, Melbourne, Australia.

Kennett, D. J., H. B. Thakar, A. M. VanDerwarker, D. L. Webster, B. J. Culleton, et al., 2017 High-precision chronology for Central American maize diversification from El Gigante rockshelter,Honduras. Proceedings of the National Academy of Sciences 114: 9026–9031.

Kermicle, J. L., 2006 A selfish gene governing pollen-pistil compatibility confers reproductive iso-lation between maize relatives. Genetics 172: 499–506.

Lambert, R. and E. Leng, 1965 Backcross response of two mature plant traits for certain corn-teosinte hybrids. Crop Science 5: 239–241.

Langham, D. G., 1940 The inheritance of intergeneric differences in Zea-Euchlaena hybrids. Ge-netics 25: 88.

Larsen, C. S., 2006 The agricultural revolution as environmental catastrophe: Implications forhealth and lifestyle in the Holocene. Quaternary International 150: 12–20.

Larson, G., D. R. Piperno, R. G. Allaby, M. D. Purugganan, L. Andersson, et al., 2014 Currentperspectives and the future of domestication studies. Proceedings of the National Academyof Sciences 111: 6139–6146.

Lauter, N. and J. Doebley, 2002 Genetic variation for phenotypically invariant traits detected inteosinte: implications for the evolution of novel forms. Genetics 160: 333–342.

Lemmon, Z. H., R. Bukowski, Q. Sun, and J. F. Doebley, 2014 The role of cis regulatory evolutionin maize domestication. PLoS Genetics 10: e1004745.

Lemmon, Z. H. and J. F. Doebley, 2014 Genetic dissection of a genomic region with pleiotropiceffects on domestication traits in maize reveals multiple linked QTL. Genetics 198: 345–353.

Lin, Z., X. Li, L. M. Shannon, C.-T. Yeh, M. L. Wang, et al., 2012 Parallel domestication of theshattering1 genes in cereals. Nature Genetics 44: 720.

Liu, S., K. Ying, C.-T. Yeh, J. Yang, R. Swanson-Wagner, et al., 2012 Changes in genome contentgenerated via segregation of non-allelic homologs. The Plant Journal 72: 390–399.

Lorant, A., S. Pedersen, I. Holst, M. B. Hufford, K. Winter, et al., 2017 The potential role of geneticassimilation during maize domestication. PLoS One 12: 1–16.

Lukens, L. N. and J. Doebley, 1999 Epistatic and environmental interactions for quantitative traitloci involved in maize evolution. Genetics Research 74: 291–302.

Mangelsdorf, P., R. MacNeish, and W. Galinat, 1967 Prehistoric wild and cultivated maize. InPrehistory of the Tehuacan Valley. Vol. I. Environment and Subsistence, edited by D. Byers,pp. 178–200, Austin, TX, USA, University of Texas Press.

Mangelsdorf, P. C., 1947 The origin and evolution of maize. Advances in Genetics 1: 161–207.Mangelsdorf, P. C., 1974 Corn, Its Origin, Evolution and Improvement . Belknap Press of Harvard

University Press.Mangelsdorf, P. C. and R. G. Reeves, 1939 The origin of Indian corn and its relatives. Texas

Farmer Collection 574: 1–315.Matsuoka, Y., Y. Vigouroux, M. M. Goodman, J. Sanchez, E. Buckler, et al., 2002 A single domes-

tication for maize shown by multilocus microsatellite genotyping. Proceedings of the NationalAcademy of Sciences 99: 6080–6084.

McSteen, P., 2006 Branching out: the ramosa pathway and the evolution of grass inflorescencemorphology. The Plant Cell 18: 518–522.

Nannas, N. J. and R. K. Dawe, 2015 Genetic and genomic toolbox of Zea mays. Genetics 199:655–669.

Orr, H. A., 1991 A test of Fisher’s theory of dominance. Proceedings of the National Academy ofSciences 88: 11413–11415.

Piperno, D. R. and K. V. Flannery, 2001 The earliest archaeological maize (Zea mays l.) from high-land Mexico: new accelerator mass spectrometry dates and their implications. Proceedingsof the National Academy of Sciences 98: 2101–2103.

Piperno, D. R., A. J. Ranere, I. Holst, J. Iriarte, and R. Dickau, 2009 Starch grain and phytolithevidence for early ninth millennium bp maize from the Central Balsas River Valley, Mexico.Proceedings of the National Academy of Sciences 106: 5019–5024.

Pophaly, S. D. and A. Tellier, 2015 Population level purifying selection and gene expression shapesubgenome evolution in maize. Molecular Biology and Evolution 32: 3226–3235.

Preston, J. C., H. Wang, L. Kursel, J. Doebley, and E. A. Kellogg, 2012 The role of teosinte glumearchitecture (tga1) in coordinated regulation and evolution of grass glumes and inflorescenceaxes. New Phytologist 193: 204–215.

Ramos-Madrigal, J., B. D. Smith, J. V. Moreno-Mayar, S. Gopalakrishnan, J. Ross-Ibarra, et al.,2016 Genome sequence of a 5,310-year-old maize cob provides insights into the early stagesof maize domestication. Current Biology 26: 3195–3201.

Ranere, A. J., D. R. Piperno, I. Holst, R. Dickau, and J. Iriarte, 2009 The cultural and chronologicalcontext of early Holocene maize and squash domestication in the Central Balsas River Valley,Mexico. Proceedings of the National Academy of Sciences 106: 5014–5018.

Rogers, J. S., 1950a The inheritance of inflorescence characters in maize-teosinte hybrids. Ge-netics 35: 541.

Rogers, J. S., 1950b The inheritance of photoperiodic response and tillering in maize-teosintehybrids. Genetics 35: 513.

Ross-Ibarra, J., M. Tenaillon, and B. S. Gaut, 2009 Historical divergence and gene flow in the

genus Zea. Genetics 181: 1399–1413.Schnable, P. S., D. Ware, R. S. Fulton, J. C. Stein, F. Wei, et al., 2009 The B73 maize genome:

complexity, diversity, and dynamics. Science 326: 1112–1115.Shi, Y., X. Wang, S. Guo, Z. Ren, L. Ku, et al., 2017 Detection of epistatic and environmental

interaction QTLs for leaf orientation-related traits in maize. Plant Breeding 136: 33–40.Sigmon, B. and E. Vollbrecht, 2010 Evidence of selection at the ramosa1 locus during maize

domestication. Molecular Ecology 19: 1296–1311.Smith, B. D., 2001 Documenting plant domestication: the consilience of biological and archaeo-

logical approaches. Proceedings of the National Academy of Sciences 98: 1324–1326.Strable, J. and M. J. Scanlon, 2009 Maize (Zea mays): a model organism for basic and applied

research in plant biology. Cold Spring Harbor Protocols 2009: pdb–emo132.Stuber, C. W., S. E. Lincoln, D. Wolff, T. Helentjaris, and E. Lander, 1992 Identification of genetic

factors contributing to heterosis in a hybrid from two elite maize inbred lines using molecularmarkers. Genetics 132: 823–839.

Studer, A., Q. Zhao, J. Ross-Ibarra, and J. Doebley, 2011 Identification of a functional transposoninsertion in the maize domestication gene tb1. Nature Genetics 43: 1160–1163.

Studer, A. J. and J. F. Doebley, 2011 Do large effect QTL fractionate? a case study at the maizedomestication QTL teosinte branched1. Genetics 188: 673–681.

Studer, A. J., H. Wang, and J. F. Doebley, 2017 Selection during maize domestication targeted agene network controlling plant and inflorescence architecture. Genetics 207: 755–765.

Sundberg, M. D. and A. R. Orr, 1990 Inflorescence development in two annual teosintes: Zeamays subsp. mexicana and Z. mays subsp. parviglumis. American Journal of Botany 77: 141–152.

Swanson-Wagner, R., R. Briskine, R. Schaefer, M. B. Hufford, J. Ross-Ibarra, et al., 2012 Re-shaping of the maize transcriptome by domestication. Proceedings of the National Academyof Sciences 109: 11878–11883.

Swanson-Wagner, R. A., S. R. Eichten, S. Kumari, P. Tiffin, J. C. Stein, et al., 2010 Pervasivegene content variation and copy number variation in maize and its undomesticated progenitor.Genome Research 20: 1689–1699.

Swarts, K., R. M. Gutaker, B. Benz, M. Blake, R. Bukowski, et al., 2017 Genomic estimation ofcomplex traits reveals ancient maize adaptation to temperate North America. Science 357:512–515.

Szabó, V. M. and B. Burr, 1996 Simple inheritance of key traits distinguishing maize and teosinte.Molecular and General Genetics 252: 33–41.

Tenaillon, M. I., J. U’ren, O. Tenaillon, and B. S. Gaut, 2004 Selection versus demography: a mul-tilocus investigation of the domestication process in maize. Molecular Biology and Evolution21: 1214–1225.

Theißen, G., 2001 Development of floral organ identity: stories from the MADS house. CurrentOpinion in Plant Biology 4: 75–85.

Tian, F., P. J. Bradbury, P. J. Brown, H. Hung, Q. Sun, et al., 2011 Genome-wide association studyof leaf architecture in the maize nested association mapping population. Nature Genetics 43:159–162.

Vallebueno-Estrada, M., I. Rodríguez-Arévalo, A. Rougon-Cardoso, J. M. González, A. G. Cook,et al., 2016 The earliest maize from San Marcos Tehuacán is a partial domesticate with ge-nomic evidence of inbreeding. Proceedings of the National Academy of Sciences 113: 14151–14156.

Vann, L., T. Kono, T. Pyhäjärvi, M. B. Hufford, and J. Ross-Ibarra, 2015 Natural variation in teosinteat the domestication locus teosinte branched1 (tb1). PeerJ 3: e900.

Vigouroux, Y., M. McMullen, C. Hittinger, K. Houchins, L. Schulz, et al., 2002 Identifying genes ofagronomic importance in maize by screening microsatellites for evidence of selection duringdomestication. Proceedings of the National Academy of Sciences 99: 9650–9655.

Vigouroux, Y., S. Mitchell, Y. Matsuoka, M. Hamblin, S. Kresovich, et al., 2005 An analysis ofgenetic diversity across the maize genome using microsatellites. Genetics 169: 1617–1630.

Vollbrecht, E., P. S. Springer, L. Goh, E. S. Buckler IV, and R. Martienssen, 2005 Architecture offloral branch systems in maize and related grasses. Nature 436: 1119–1126.

Wallace, J., S. Larsson, and E. Buckler, 2014 Entering the second century of maize quantitativegenetics. Heredity 112: 30–38.

Wang, H., T. Nussbaum-Wagler, B. Li, Q. Zhao, Y. Vigouroux, et al., 2005 The origin of the nakedgrains of maize. Nature 436: 714–719.

Wang, H., A. J. Studer, Q. Zhao, R. Meeley, and J. F. Doebley, 2015 Evidence that the origin ofnaked kernels during maize domestication was caused by a single amino acid substitution intga1. Genetics 200: 965–974.

Wang, Q. and H. K. Dooner, 2006 Remarkable variation in maize genome structure inferred fromhaplotype diversity at the bz locus. Proceedings of the National Academy of Sciences 103:17644–17649.

Wang, S., K. Wu, Q. Yuan, X. Liu, Z. Liu, et al., 2012 Control of grain size, shape and quality byosspl16 in rice. Nature Genetics 44: 950–954.

Wang, S.-S., C.-S. Wang, T.-H. Tseng, Y.-L. Hou, and K.-Y. Chen, 2011 High-resolution geneticmapping and candidate gene identification of the slp1 locus that controls glume developmentin rice. Theoretical and Applied Genetics 122: 1489–1496.

Wang, X., Q. Chen, Y. Wu, Z. H. Lemmon, G. Xu, et al., 2018 Genome-wide analysis of transcrip-tional variability in a large maize-teosinte population. Molecular Plant 11: 443–459.

Weatherwax, P., 1924 The reported origin of indian corn from teosinte. In Proceedings of theIndiana Academy of Science, volume 34, pp. 225–227.

Weatherwax, P., 1935 The phylogeny of Zea mays. The American Midland Naturalist 16: 1–71.Weber, A., R. M. Clark, L. Vaughn, J. de Jesus Sánchez-Gonzalez, J. Yu, et al., 2007 Major regu-

latory genes in maize contribute to standing variation in teosinte (Zea mays ssp. parviglumis).Genetics 177: 2349–2359.

Weber, A. L., W. H. Briggs, J. Rucker, B. M. Baltazar, J. de Jesús Sánchez-Gonzalez, et al.,2008 The genetic architecture of complex traits in teosinte (Zea mays ssp. parviglumis): newevidence from association mapping. Genetics 180: 1221–1232.

Whitt, S. R., L. M. Wilson, M. I. Tenaillon, B. S. Gaut, and E. S. Buckler, 2002 Genetic diversityand selection in the maize starch pathway. Proceedings of the National Academy of Sciences99: 12959–12962.

Wilkes, G., 2004 Corn, strange and marvelous: But is a definitive origin known. In Corn: Origin,History, Technology, and Production, edited by C. Smith, J. Bertran, and E. Runge, pp. 3–63,



Hoboken, NJ, USA, John Wiley and Sons.Wilkes, H. G., 1967 Teosinte: The Closest Relative of Maize.. The Bussey Institution of Harvard

University, Boston, MA, USA.Wills, D. M., Z. Fang, A. M. York, J. B. Holland, and J. F. Doebley, 2018 Defining the role of

the MADS-box gene, Zea Agamous-like1, a target of selection during maize domestication.Journal of Heredity 109: 333–338.

Wills, D. M., C. J. Whipple, S. Takuno, L. E. Kursel, L. M. Shannon, et al., 2013 From many, one:genetic control of prolificacy during maize domestication. PLoS Genetics 9: e1003604.

Wingen, L. U., T. Münster, W. Faigl, W. Deleu, H. Sommer, et al., 2012 Molecular genetic basisof pod corn (tunicate maize). Proceedings of the National Academy of Sciences 109: 7115–7120.

Wright, S. I., I. V. Bi, S. G. Schroeder, M. Yamasaki, J. F. Doebley, et al., 2005 The effects ofartificial selection on the maize genome. Science 308: 1310–1314.

Wu, X., A. Skirpan, and P. McSteen, 2009 Suppressor of sessile spikelets1 functions in the ramosapathway controlling meristem determinacy in maize. Plant Physiology 149: 205–219.

Xu, G., X. Wang, C. Huang, D. Xu, D. Li, et al., 2017 Complex genetic architecture underliesmaize tassel domestication. New Phytologist 214: 852–864.

Xue, S., P. J. Bradbury, T. Casstevens, and J. B. Holland, 2016 Genetic architecture ofdomestication-related traits in maize. Genetics 204: 99–113.

Yang, C. J., L. E. Kursel, A. J. Studer, M. E. Bartlett, C. J. Whipple, et al., 2016 A gene for geneticbackground in Zea mays: Fine-mapping enhancer of teosinte branched1. 2 to a yabby classtranscription factor. Genetics 204: 1573–1585.

Yu, J., J. B. Holland, M. D. McMullen, and E. S. Buckler, 2008 Genetic design and statistical powerof nested association mapping in maize. Genetics 178: 539–551.

Zeder, M. A., E. Emshwiller, B. D. Smith, and D. G. Bradley, 2006 Documenting domestication:the intersection of genetics and archaeology. Trends in Genetics 22: 139–155.

Zhao, Q., A. L. Weber, M. D. McMullen, K. Guill, and J. Doebley, 2011 MADS-box genes of maize:frequent targets of selection during domestication. Genetics Research 93: 65–75.



maize domestication and gene interaction · maize domestication and gene interaction michelle c....

Documents