Molecular Evidence for Gene

Flow among Zea Species Genes transformed into maize through genetic engineering could

be transferred to its wild relatives, the Teosintes

John Doebley

mong the concerns associated with the "release" of trans- genic plants into the environ-

ment is the possibility that genes transformed into a crop plant by re- combinant DNA methods could spread into its wild relatives. The process of concern is introgression, the transfer of genes from one popu- lation into another via hybridization and subsequent backcrossing. If these genes confer some selective advantage on the wild relatives, then they could contribute to the development of new or more troublesome weedy forms. For example, genes for disease, drought, and herbicide resistance could all produce problems by en- hancing the fitness of the wild rela- tives. The available evidence for gene flow between crops and their relatives provides a context for beginning to make realistic assessments of the risk of introgression of engineered genes into wild crop relatives.

In this article, I review the evidence for gene flow between maize and its nearest wild relatives, the teosintes. Because these plants are relatively well-studied, the evidence for gene flow between them can provide in- sight into the potential for the trans- fer of engineered genes from crops to their wild relatives. However, it is important to note that conclusions drawn from maize and teosinte may not be applicable to all other crop-

John Doebley is an assistant professor in the Department of Plant Biology, Univer- sity of Minnesota, St. Paul, MN 55108. ? 1990 American Institute of Biological Sciences.

The risk of engineered genes in maize

spreading into teosinte would seem easily

controlled

wild relative complexes. Because much of the evidence for gene flow between maize and teosinte is circum- stantial, I focus on studies employing molecular markers, which provide some of the best available evidence.

Documenting introgression Whereas it is quite common to find naturally occurring hybrids between a crop and its nearest wild relatives, the demonstration of introgression be- tween them is a much more difficult task (Heiser 1973, Small 1984). If one observes that a crop and its wild rela- tive in a particular geographic region both possess a particular morphologi- cal trait, it may seem reasonable to conclude that the trait was transferred between them. However, these popula- tions, which are growing under the same environmental regime, may sim- ply share the trait as a result of either convergent evolution or joint inheri- tance of the trait from their common ancestor.

The use of presumably neutral mo- lecular markers (allozymes or DNA markers) reduces the likelihood of convergence; however, the problem

of distinguishing introgressive molec- ular markers from ones jointly inher- ited from the common ancestor still exists (Heiser 1973). This problem is particularly troublesome because crops and their progenitors are sepa- rated by short evolutionary periods (10,000 years or less), and thus they are apt to possess similar genotypes. For this reason, convincing evidence for introgression is most likely to be obtained between crops and their more evolutionarily divergent wild relatives, which have more distinct genotypes.

An additional problem facing crop evolutionists is to infer the direction of gene flow (crop into wild versus wild into crop). One criterion em- ployed by some authors (Doebley et al. 1987a, Rick et al. 1974) is the relative frequencies of the presumed introgressive allele in the crop and wild species. If, where a crop is sym- patric (growing in the same region), with a wild relative, it possesses in low frequency a marker typical of the wild taxon and lacking in the crop elsewhere, then this evidence may be appropriately interpreted as intro- gression from the wild form into the cultigen. Introgression from the crop into the wild form could be docu- mented by the reverse situation.

Notwithstanding the difficulty of demonstrating introgression, intro- gressive hybridization between crops and their wild relatives seem proba- ble. Crops and their wild relatives often grow sympatrically, and the lat- ter frequently occur as weeds in cul- tivated fields. In some cases, as crops spread beyond their cradle of domes-

June 1990 443

Figure 1. Distribution map of teosinte in Mexico and Central America.

tication, they came into contact with other closely related wild species, cre- ating new opportunity for introgres- sion. Further, crops and their wild relatives often lack complete genetic barriers to hybridization, and natu- rally occurring hybrids are frequently reported by crop scientists involved in field studies (Small 1984). This de- scription does not deny the existence of temporal and genetic barriers to gene flow, but merely notes that such barriers, even in self-pollinating spe- cies, are usually imperfect.

Taxonomic and evolutionary background Maize (Zea mays ssp. mays) and the teosintes all belong to the genus Zea. Teosinte is a common name applying to several distinct wild taxa, all native to Mexico and Central America (Fig- ure 1). Zea diploperennis Iltis, Doeb- ley, and Guzman and Zea perennis (Hitchc.) Reeves and Mangelsdorf are respectively diploid and tetraploid pe- rennial teosintes with narrow distri- butions in Jalisco, Mexico. Zea luxu- rians (Durieu and Ascherson) Bird is an annual teosinte from southeastern Guatemala.

There are three additional types of annual teosinte that are genetically so similar to maize that they have been treated taxonomically as subspecies of Z. mays (Iltis and Doebley 1980,

Doebley in press b). Z. mays ssp. mexicana (Schrader) Iltis (central Mexican annual teosinte) is a large- flowered, mostly weedy teosinte with a broad distribution across the central highlands of Mexico. Z. mays ssp. parviglumis Iltis and Doebley is a small-flowered, mostly wild teosinte of southern and western Mexico. Z. mays ssp. huehuetenangensis (Iltis and Doebley) Doebley is a narrowly distributed teosinte of the western highlands of Guatemala.

Z. diploperennis, Z. perennis, and Z. luxurians are clearly distinguished from maize by a number of morpho- logical, biochemical, and cytogenetic characteristics (Doebley 1983, Doeb- ley et al. 1984, Kato 1976, Timothy et al. 1979). Of the three subspecies of teosinte, Z. mays ssp. huehue- tenangensis is clearly differentiated from maize by several independent lines of evidence (Doebley et al. 1984, Kato 1976), Z. mays ssp. mexicana shows clear allozymic differences from maize, but Z. mays ssp. parviglumis is essentially indistin- guishable from maize by all measures except morphology (Figure 2).

All teosintes can be crossed to maize, and they all form fertile hy- brids with maize except for the tetra- ploid, Z. perennis (Wilkes 1967). Nevertheless, field observations re- veal that some of these taxa form natural hybrids with maize much

more frequently than do others (Wilkes 1977).1 Z. luxurians hybrid- izes only rarely with maize, whereas Z. mays ssp. mexicana forms frequent hybrids with maize. Populations of Z. mays ssp. parviglumis are variable in this regard; a few hybridize fre- quently with maize but most fail to form hybrids often.

There are several types of barriers preventing or reducing gene flow be- tween maize and teosinte. Temporal and spatial factors effectively isolate Z. mays ssp. parviglumis from maize, and there appears to be some degree of genetic incompatibility between maize and both Z. luxurians and Z. mays ssp. mexicana (Ting 1958, Wilkes 1967). Even if F1 hybrids form (Figure 3), they are strongly selected against by both man and nature. The specific problem is that the pistillate inflorescences (ears) of maize and te- osinte are vastly different in structure, one adapted to the needs of human- kind for harvestability and the other to the demands of nature for natural seed dispersal (Galinat 1983, Iltis 1983, Weatherwax 1935). The result is that the influence of maize-teosinte hybrids on subsequent generations is minimized (Weatherwax 1935).

Reciprocal introgression between teosinte and maize has been the sub- ject of numerous articles since the beginning of this century (see Doebley 1984 for a review). Based on field observations of morphological traits, several authors have proposed exten- sive reciprocal introgression between Mexican annual teosinte and maize (Collins 1921, Wilkes 1977).

Of all the taxa, Z. mays ssp. mexi- cana has been most frequently de- scribed as possessing introgressive maize germ plasm. Some authors have argued against extensive intro- gression between maize and Mexican annual teosinte on the basis of cytol- ogy (Kato 1976, 1984) and field ob- servations (Doebley 1984). My prin- cipal objection to the argument that there is extensive introgression be- tween maize and teosinte is that alter- native hypotheses, such as conver- gence and joint retention of the ancestral condition, have not been convincingly eliminated by the field observations.

1J. Doebley, 1990, unpublished observations.

BioScience Vol. 40 No. 6 444

Molecular evidence

Using molecular markers, my col- leagues and I have examined the extent of introgression between maize and te- osinte (Doebley et al. 1984, 1987a). One approach used allozymes, which are allelic forms of enzymes that can be distinguished with gel electrophoresis. Our data showed that maize and Z. mays ssp. parviglumis are so similar in allozyme constitution that it would be difficult to apply allozymes to the study of introgression between them. How- ever, the remaining teosinte taxa pos- sessed allozymes distinct from those of maize at several loci. These allozymes represent molecular markers that po- tentially enable one to trace gene flow among these taxa.

A second approach examines the chloroplast genome, a circular DNA molecule housed in the chloroplast. Us- ing restriction endonucleases (enzymes that cleave DNA at specific sites), we are able to distinguish the chloroplast genomes of different species or popula- tions, and we can thereby detect cases in which the chloroplast genome of one species has become incorporated into another via introgression.

Allozyme analysis We first examined whether introgres- sion causes maize and teosinte plants collected from the same field to be homogeneous for their allozyme con- stitutions. Initially, five pairs of maize and teosinte populations were ana- lyzed at 21 loci (Doebley et al. 1984). Since then, we have examined one additional sympatric pair of popula- tions; this pair includes Z. mays ssp. mexicana from the Nobogame Valley (Doebley and Nabhan 1989).

Our hypothesis was that if introgres- sion is continual between sympatric maize and teosinte populations, then genetic distance calculated between such populations should be less than that between nonsympatric popula- tions. Our results (Doebley et al. 1984, Doebley and Nabhan 1989) were con- trary to this expectation. The popula- tions of Z. mays ssp. mexicana, ssp. huehuetenangensis, and Z. diploperen- nis all showed their closest relationship to other populations of their own taxon and not to the maize with which they were growing. Similarly, the six maize populations all formed a closely related

Figure 2. The teosinte, Zea mays ssp. mexicana, in the Valley of Mexico. a. A robust teosinte plant taken from a maize field and a smaller teosinte plant found growing along the edge of the highway. The small plant is held by Hugh Iltis of the University of Wisconsin, Madison. b. The ears or female inflorescences of the teosinte Zea mays ssp. mexicana. These ears differ remarkably from those of maize in appearance and structure despite the plants being members of the same biological species.

group and did not show a greater sim- ilarity to their sympatric teosinte pop- ulations. Thus our results demon- strated that sympatric maize and teosinte populations maintain distinct genetic constitutions despite the occa- sional formation of F1 hybrids.

One result of the allozyme analysis was particularly startling. This analy- sis revealed that Z. mays ssp. mexi- cana, which forms frequent hybrids with maize and has robust maize-like vegetative characteristics, is allozymi- cally quite different from maize; how- ever, Z. mays ssp. parviglumis, which rarely hybridizes with maize and ap- pears more slender and grasslike, is essentially indistinguishable from maize in its allozyme constitution (Doebley et al. 1984). This paradoxi- cal result underscores the importance of examining multiple sources of data

(allozymes, cytology, and morphol- ogy) when studying evolutionary re- lationships and introgression.

Although our results show that sympatric maize and teosinte popula- tions maintain distinct allozymic con- stitutions, there may be occasional cases of introgression. To address this possibility, we searched our raw data for allozymes with distributions that would suggest introgression (Doebley et al. 1984, 1987a). If an allozyme typical of maize should appear at low frequency in a sympatric teosinte that typically lacks that allozyme, this finding would constitute evidence for introgression from maize into te- osinte. The reverse situation would constitute evidence for introgression from teosinte into maize.

For example, Glul-7 is reasonably common in Z. mays but unknown in

June 1990 445

Figure 3. Ears of teosinte (Zea mays ssp. mexicana) and some naturally occurring hybrids of maize and teosinte. The ear on the left shows no indication of maize contamination, but the ear on the right appears to be an F1 hybrid between maize and teosinte. The ears in the center appear to be those of either F2 plants or back- crosses to teosinte. The hybrid ears tend not to fracture and disperse their seed as effectively as pure teosinte ears, and thus they are strongly selected against in na- ture. Photo by Hugh Iltis.

Z. diploperennis, Z. perennis, and Z. luxurians, with the exception of one plant of Z. luxurians (Table 1). The most parsimonious explanation for this distribution would be to infer introgression of this allozyme from maize into Z. luxurians.

A second more convincing case of introgression was found with Z. diploperennis. One plant of this wild species possessed two allozymes (Enpl-8 and Pgdl-3.8) that are oth- erwise unknown in this species but are common in maize (Table 1). Be- cause the two loci involved are tightly

linked (3 map units apart) on chro- mosome 6, the segment of chromo- some 6 that carries these two loci probably was transferred from maize into this single Z. diploperennis plant. These facts do not suggest that maize germ plasm forms a standard (typical) part of the Z. diploperennis genome, because the vast majority of Z. diplo- perennis plants possess typical allo- zymes for this species at these two loci.

Given that only 2 out of 219 Z. diploperennis and Z. luxurians plants showed signs of introgression, intro- gression appears to occur only at a low level. In both these cases, we suggest that the allozymes were de- rived from maize (Z. mays ssp. mays) and not another of the Z. mays sub- species because only maize is sympat- ric with these two species.

We also used allozyme distribu- tions to check for introgression be- tween maize and Z. mays ssp. mexi- cana (Doebley et al. 1987a). Here, the challenge is greater because these two taxa are closely related, and thus they have similar allozyme constitutions. Nevertheless, we identified distribu- tions of six allozymes at five loci that can be considered evidence of intro- gression (Table 2). The best evidence comes from Enpl-14 and Glul-8. Both these allozymes are common in Z. mays ssp. mexicana but much rarer in maize. More important, they are found exclusively in maize from the central highlands of Mexico, the region to which Z. mays ssp. mexi- cana is indigenous. Because these two allozymes are common in Z. mays ssp. mexicana, much rarer in maize, and found only in maize growing in the geographic range of Z. mays ssp. mexicana, it seems most likely that these allozymes originated in Z. mays ssp. mexicana and have been trans- ferred into maize.

Several additional allozymes (Glul- 11, Pgdl-1.8, Pgd2-8, and Pgm2-7.2) occur in both maize and Z. mays ssp.

mexicana in the central highlands of Mexico (Table 2). Because these alloz- ymes were not found in maize or te- osinte from other regions, it seems probable that their occurrence in maize and teosinte from the central highlands is the result of introgression. The joint inheritance of these allozymes from a common ancestor hypothesis seems less likely for several reasons. Available ev- idence suggests that maize was domes- ticated only once and that the ancestral teosinte was most closely related to Z. mays ssp. parviglumis (Doebley in press a,b). If the co-occurrence of these alloz- ymes in maize and teosinte (Z. mays ssp. mexicana) from the central high- lands is the result of joint inheritance from their common ancestor, then both maize from other regions and Z. mays ssp. parviglumis should also possess these allozymes, which they do not. Therefore, in this case the joint inheri- tance hypothesis would require that these allozymes must have been lost from both Z. mays ssp. parviglumis and maize from regions other than the central highlands. This collection of hypothetical events seems far less par- simonious than the events required by the introgression hypothesis.

Chloroplast DNA analysis Although allozymes provide good markers for the study of gene flow between maize and teosinte, chloro- plast DNA (cpDNA) data offer much different means of assessing introgres- sion. A feature of the chloroplast ge- nome is that it is maternally inherited in Zea and most other angiosperms (Corriveau and Coleman 1988). Thus the chloroplast genome of one taxon can be combined with the nuclear ge- nome of another taxon by continued backcrossing to the paternal parent after an initial hybridization.

The chloroplast genome typically has little effect on the phenotype of plants. A plant that contains a teosinte

Table 1. Frequencies for allozymes with distributions that suggest they have been transferred by introgression from maize to Zea diploperennis, Zea perennis, or Zea luxurians.

Taxa

Zea Za Zea Zea mays ssp. Locus-allele diploperennis perennis luxurians huehuetenangensis mexicana parviglumis mays

Enpl-8 0.01 0.00 0.00 0.02 0.04 0.02 0.07 Glul-7 0.00 0.00 0.01 0.00 0.22 0.06 0.17 Pgdl-3.8 0.01 0.00 0.00 0.02 0.63 0.73 0.66 Pgm2-7.2 0.00 0.02 0.00 0.00 0.03 0.00 0.01

BioScience Vol. 40 No. 6 446

Table 2. Frequencies for allozymes with distributions that suggest they have been transferred by introgression between maize and Zea mays ssp. mexicana.

Taxa

Zea Zea e a Zea mays ssp. Locus-allele diploperennis perennis luxurians huehuetenangensis mexicana parviglumis mays

Enpl-14 0.00 0.00 0.00 0.00 0.30 0.00 0.01 Glul-8 0.00 0.00 0.00 0.00 0.45 0.00* 0.01 Glul-11 0.00 0.00 0.00 0,00 0.01 0.00 0.01 Pgdl-1.8 0.00 0.00 0.00 0.00 0.01 0.00 0.01 Pgd2-8 0.00 0.00 0.00 0.00 0.03 0.00 0.01 Pgm2-7.2 0.00 0.02 0.00 0.00 0.03 0.00 0.01

*Allozyme present at a frequency of 0.002.

chloroplast genome but a maize nu- clear genome looks like typical maize.

There are a few restraints to the free introgression of the chloroplast genome between maize and teosinte. First, incompatibilities between the nuclear genes of one taxon and the chloroplast genes of another could reduce the fitness of the plant (Allen et al. 1989). Second, the F1 and first few backcross generations would be ill-adapted to the needs of both hu- mans and nature, as discussed above, and thus strongly selected against.

The available data provide some evidence for the introgression of the chloroplast genome between maize and teosinte (Doebley et al. 1987b, Doebley and Sisco 1989). The strong- est evidence is the detection of an uncommon cpDNA genotype (Type A; Table 3) in three accessions of Z. mays ssp. mexicana from Copandiro, Michoacan, Mexico, and one acces- sion of maize, cytoplasmic male ster- ile-S (CMS-S). CMS-S maize and

Copandiro teosinte also share a mito- chondrial genome that has unusual plasmid DNAs (called S-1 and S-2), so it is improbable that the similarity of their chloroplast genomes is due to convergence. Another possible inter- pretation is that CMS-S represents a separate domestication from Copand- iro teosinte. But there is no other evidence supporting a separate do- mestication (Doebley in press a, Doe- bley and Sisco 1989). Therefore, it would appear that the chloroplast and mitochondrial genomes of CMS- S maize were derived from Copandiro teosinte by introgression.

Further evidence of the introgres- sion of the chloroplast genome be- tween maize and teosinte is presented in Table 3. First, maize and both Z. mays ssp. mexicana and ssp. parviglu- mis share a number of cpDNA types. This situation could result from the introgression of these cpDNA types among these taxa. However, such gene flow could not have been com-

pletely promiscuous, because cpDNA type D is restricted to Z. mays ssp. parviglumis, and type E, although common in maize, is not found in Z. mays ssp. mexicana. Second, there is no evidence for the introgression of the chloroplast genome between maize and any of the teosinte species Z. luxurians, Z. diploperennis, or Z. perennis. None of the more than 30 teosinte plants analyzed contained maize cpDNA types, and none of the 80 accessions of maize examined con- tained chloroplast genome types found in the three teosinte species (Doebley et al. 1987b, Doebley in press a,b). This result suggests that the introgression of the chloroplast genome between these three teosinte species and maize is at least rare or perhaps nonexistent.

Conclusions

The earliest studies of gene flow be- tween maize and teosinte focused on

Table 3. Distribution of chloroplast DNA genotypes among accessions of Mexican annual teosintes and maize. The number of accessions in which each cpDNA genotype occurred is shown. (Data from Doebley et al. 1987b, in press a.)

cpDNA Genotypes* Taxon A B C D E Total

ssp. mexicana

Nobogame 2 2 Central Plateau 3 2 11 - 16 Chalco 2 9 - 11 Total 3 4 22 - 29

ssp. parviglumis Jalisco 2 5 7 Southern Guerrero 4 4 Central Balsas 1 9 7 3 20 Total 1 15 12 3 31

ssp. mays USA 1 4 7 12 Mexico 16 6 22 Guatemala 6 1 7 South America -10 8 - 21 39 Total 1 36 8 35 80

*Using Zea diploperennis as an outgroup, cpDNA type A would be judged most primitive. Type B is the next derived form, distinguished by a deletion mutation (LM-3; Doebley et al. 1987b). Types C, D, and E are the most advanced, each possessing LM-3 plus additional restriction site losses or gains. Type C has a unique Eco RI site; Type D has a unique Cfo I site; Type E has a unique Eco RI site.

June 1990 447

This publication is available in microform.

University Microfilms International reproduces this publication in microform: microfiche and 16mm or 35mm film. For information about this pub- lication or any of the more than 13,000 titles we offer, complete and mail the coupon to: University Microfilms International, 300 N. Zeeb Road, Ann Arbor, MI 48106. Call us toll-free for an immediate response: 800-521-3044. Or call collect in Michigan, Alaska and Hawaii: 313-761-4700.

:Please send information about these titles:

N am e .

_ .. ... . ... ....... ...

Name _

CompanyllInstitution

Ad .ress

.

. .. ... .. ....... ..... Address .

City

State Zip_

Phone I

University Microfilms

International

morphological traits. On the basis of that data, several authors suggested that there has been extensive gene flow between maize and teosinte and that the genetic constitution of te- osinte has been greatly altered by maize germ plasm. Recent molecular analyses confirm that there is gene flow between maize and teosinte. The molecular data suggest that this intro- gression goes in both directions, but that it occurs at low levels. Conse- quently, maize and teosinte maintain distinct genetic constitutions despite sporadic introgression.

From this evidence, one can deduce that genes placed in maize by genetic engineering could be transferred to teosinte and, under some circum- stances, spread throughout teosinte populations. Several conditions could facilitate the transfer of engineered genes from maize to teosinte. The engineered gene would not interfere with traits, such as natural seed dis- persal, that allow teosinte to survive as a wild plant. The engineered gene would confer some benefit on teosinte that makes it competitively superior to pure teosinte types. Maize contain- ing engineered genes would be grown in areas where teosinte is native.

Of these three conditions, the third can be easily controlled because te- osinte has a limited distribution and has not shown any tendency to spread beyond its natural range in Mexico and Central America. Thus the risk of engineered genes in maize spreading into teosinte would seem easily con- trolled. The risks for crops with more broadly distributed wild relatives (e.g., sorghum) are potentially greater.

Acknowledgments This work was supported in part by NSF grant BSR-8508490 and by Pio- neer Hi-Bred International, Inc., of Johnston, IA. Greg Anderson and Glenn Furnier provided helpful com- ments on the manuscript.

References cited

Allen, J. O., G. K. Emenhiser, and J. L. Kermi- cle. 1989. Miniature kernel and plant: inter- action between teosinte cytoplasmic ge- nomes and maize nuclear genomes. Maydica 34: 277-290.

Collins, G. N. 192i. Teosinte in Mexico. J. Hered. 12: 339-350.

Corriveau, J. L., and A. W. Coleman. 1988. Rapid screening method to detect potential- biparental inheritance of plastid DNA and results for over 200 angiosperm species. Am. J. Bot. 75: 1443-1458.

Doebley, J. F. 1983. The maize and teosinte male inflorescence: a numerical taxonomic study. Ann. Mo. Bot. Gard. 70: 32-70.

. 1984. Maize introgression into te- osinte-a reappraisal. Ann. Mo. Bot. Gard. 71: 1100-1113.

In press a. Molecular evidence and the phylogeny of maize. Econ. Bot.

In press b. Molecular systematics of Zea (Gramineae). Maydica.

Doebley, J. F., M. M. Goodman, and C. W. Stuber. 1984. Isoenzymatic variation in Zea (Gramineae). Syst. Bot. 9: 203-218.

. 1987a. Patterns of isozyme variation between maize and Mexican annual teosinte. Econ. Bot. 41: 234-246.

Doebley, J. F., and G. Nabhan. 1989. Further evidence regarding gene flow between maize and teosinte. Maize Genetics Cooperation Newsletter 63: 107-108.

Doebley, J. F., W. Renfroe, and A. Blanton. 1987b. Restriction site variation in the Zea chloroplast genome. Genetics 117: 139-147.

Doebley, J. F., and P. Sisco. 1989. On the origin of the maize male sterile cytoplasms. Maize Genetics Cooperation Newsletter 63: 108-109.

Galinat, W. C. 1983. The origin of maize as shown by key morphological traits of its ancestor, teosinte. Maydica 28: 121-138.

Heiser, C. 1973. Introgression re-examined. Bot. Rev. 39: 347-366.

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

iltis, H. H., and J. F. Doebley. 1980. Taxon- omy of Zea (Gramineae). I. Subspecific cat- egories in the Zea mays complex and a genetic synopsis. Am. J. Bot. 67: 994-1004.

Kato, T. A. 1976. Cytological studies of maize. Mass. Agric. Exp. Stn. Bull. 635.

. 1984. Chromosome morphology and the origin of maize and its races. Evol. Biol. 17: 219-253.

Rick, R., R. W. Zobel, and J. F. Fobes. 1974. Four peroxidase loci in red-fruited tomato species: genetics and geographic distribution. Proc. Natl. Acad. Sci. 71: 835-839.

Small, E. 1984. Hybridization in the domesti- cated-weed-wild complex. Pages 195-210 in W. F. Grant, ed. Plant Biosystematics. Aca- demic Press, San Diego, CA.

Timothy, D. H., C. S. Levings, D. R. Pring, M. F. Conde, and J. L. Kermicle. 1979. Or- ganelle DNA variation and systematic rela- tionships in the genus Zea: teosinte. Proc. Natl. Acad. Sci. 76: 4220-4224.

Ting, Y. C. 1958. Inversions and other charac- teristics of teosinte chromosomes. Cytologia 23: 239-250.

Weatherwax, P. 1935. The phylogeny of Zea mays. Amer. Midi. Nat. 16: 1-71.

Wilkes, H. G. 1967. Teosinte: The Closest Rel- ative of Maize. Bussey Institute, Harvard University, Cambridge, MA.

. 1977. Hybridization of maize and teosinte in Mexico and Guatemala and the improvement of maize. Econ. Bot. 31: 254- 293.

BioScience Vol. 40 No. 6 I