TEXAS AGRICULTURAL EXPERIMENT STATION B. D. LEWIS, Director, College Station, Texas M‘ 76/ Comparative Morphology oi tho American Maydoao A -<‘».\ w- f; The TEXAS AGRICULTURAL AND MECHANICAL COLLEGE SYSTEM V’ ’ cuss GILCHHIST, Chancellor DIGEST The results of a comparative study 0f maize and its Amer- ican relatives indicate that teosinte is within the combin- ed range of variation of maize and Tripsacum in all of the 55 characters studied thus far, with two doubtful exceptions which should be disregarded. These results agree with a re- vised hypothesis that teosinte originated as a hybrid between primitive types of maize and of Tripsacum, whose plant char- acters were similar to those of extant forms but whose chrom- osomes were more closely homologous than those of forms which have been tested. In all the characters by which Mexican and Guatemalan teosintes were compared, the Mexican type was found to be the more maize-like; also, the North American varieties of maize were found to be more teosinte-like than those typical of the Andean region. Although a relatively small number of characters were used in comparing the various forms of maize and of teosinte, the results are in agreement with a current hypothesis that the teosinte of Mexico is further mod- ified by introgression from maize, and that North American maize is likewise modified by introgression from teosinte. CONTENTS v Page Digest ................................................................................................................. .. 2 Introduction ...................................................................................................... .. 3 Materials and Methods .................................................................................. .. 3 Results ............................................................................................................... .. 5 Paired and Single Spikelets ................................................................. .. 5 Compactness of Ear ................................................................................ .. 7 Depth of Alveolus of the Ear ............................................................... .. 9 Branching Habit ...................................................................................... ._ 9 Falling and Rooting of Culms ............................................................. .. 12 Adaptation to Poorly-drained Habitats .............................................. .. 12 Anatomical Characters of Leaves ....................................................... .. 13 Toughness of Culms ............................................................................... .. 17 Discussion .......................................................................................................... .. 20 Acknowledgments ............................................................................................ .. 25 Literature Cited .............................................................................................. .. 26 Comparative Morphology oi the American Maydoao R. G. Reeves, Professor, Departments of Agronomy and Genetics FOR MORE THAN A DECADE, students of the evolution of maize and its relatives have been interested in the apparent fact that most of the characters of annual teosinte are such that they could have been inherited from either Tripsacum or maize. Thirty four characters were tabulated by Mangelsdorf and Reeves (1939), in which teosinte was intermediate between the other two or indistinguishable from one of them. This was regarded as evidence for the hypothesis that teosinte originated as a segregate from a hybrid between maize and Tripsacum, after the original hybrid had backcrossed to maize an indefinite number of times. The main objective of the work reported in this bulletin was to collect and interpret additional data on this question. A secondary objective was to test the hypothesis that, in gen- eral, Mexican teosinte has more characters inherited from maize than has Guatemalan teosinte, and that Central and North American maize have more characters from teosinte, or indirectly from Tripsacum, than has Andean maize. If both of these hypothesis are valid, the five forms to be dis- cussed—Tripsacum, Guatemalan teosinte, Mexican teosinte, North American maize and Andean maize—should fall into a series showing gradations from Tripsacum to Andean maize. MATERIALS AND METHODS Except when the contrary is stated in the text, the maize used as material was an open-pollinated North American va- riety and the teosinte was the Florida variety, the latter hav- ing come originally from Guatemala. The Tripsacum used included both diploid and tetraploid forms of T. dactyloides, but diploids were used consistently and tetraploids only as supplementary material. When the type of Tripsacum is un- "designated in the text, it should be understood to be a diploid form of T. dactyloides. It is desirable to delineate in some detail the premises adopted in selecting these particular forms and, for the study 358i?3 _4_ of certain characters, only a few plants from each form. If it be assumed that maize and Tripsacum are parents of teo- sinte, only one form of maize and of Tripsacum would have been required t0 hybridize successfully t0 produce teosinte. For the results of this morphological study to be positive, therefore, it is only necessary for one form of maize and of Tripsacum to satisfy the requirements for the parents, and for one form of teosinte to satisfy the requirements for the offspring. If it were possible to choose the exact forms of teosinte, maize and Tripsacum, from all of those in existence, which would serve best on the basis of interfertility and cyto- genetical relationships as the original teosinte and as possible parents of teosinte, it seems that such forms would also serve best as materials for the morphological study. But the in- formation available does not permit such a choice to be made. A possibility exists, therefore, that combinations of forms other than those used might fulfill the theoretical require- ments even better. On the other hand, the possibility is fully recognized that combinations of forms might be found which would give results much less positive than those obtained. It is obviously impracticable to make a detailed morphological analysis of a large proportion of the thousands of forms in- cluded in these three plant groups. Therefore, an analysis of a few forms must suffice for the present to give an indi- cation of the morphological relationships. Thus, the writer disclaims the assumption that the forms of maize, teosinte and Tripsacum included in the study are the most appropriate, except for the fact that those used con- sistently could be grown locally in the field with little risk of abnormal development due to poor adaptation. Although North American maize, the prinicpal type stud- ied, is reputed to be tripsacoid, and therefore more similar to teosinte and Tripsacum than certain other types, it is suffi- ciently different from teosinte and Tripsacum to justify its use when the comparison includes only one maize variety. If North American maize really is tripsacoid, the results ob- tained probably are a little less positive than if only non-trip- sacoid maize had been included. According to conclusions of some investigators, Florida teosinte is one of the more “typical” teosintes known, the va- rieties from Northern Mexico being the results of recent in- trogression from maize. If this postulate is valid, Florida teosinte is a good choice of material for this study; if not, it might be regarded as having been taken at random. »<§L,C0ncerning the Tripsacum used as material, recognition "rbe made that according to Cutler and Anderson (1941), p of T. dactyloides is known to occur naturally in Mex- Central America, where teosinte is usually believed to originated. For this reason, there is some doubt as to ~ fer T. dactyloides should be considered as a possible par- ibf teosinte. However, the possibility is not completely ded, because (a) our present conception of the place of in. of teosinte may be in error, and (b) the distribution gydactyloides at the time of origin of teosinte may have it very different from that known at the present time. i,_it may be noteworthy that published results of attempts vtain fertile hybrids between maize and Tripsacum indi- _;_that diploid T. dactyloides has given the most positive ‘ ts of all forms tested thus far. n the original plan of the work, no statistical treatment Q ' results was. intended, because in the study of micro- crcharacters only 7 to 11 different clones of Tripsacztm jlbeen included. But later it appeared that such a treat- ‘of data on characters listed in Table 1 might be useful. éflach of them, analysis of variance was made, and the re- obtained are indicated later. RESULTS and Single Spikelets Although the pistillate spikelets of teosinte and Tfipsa- ~are borne singly, it is generally agreed that they origi- § were in pairs, as in maize, and became single by the '0n of one member of each pair. This is shown by the ce of the vestigial remains of a second spikelet accom- 'ng each normal one. It seems at first thought, therefore, that the three spe- “epresent two, and only two, distinct morphological classes ltespect to this character. But evidence is available that Tajte actually constitutes a class intermediate between and Tripsacum, at least genetically. Maize-Tripsacunz brids have single pistillate spikelets, while m.aize-teo- ;F1 hybrids have mainly paired spikelets; in fact, F1 hy- Ilbetween Nobogame and New teosinte were found to have uniformly paired spikelets (Figures 1, 2). The facts \here constitute fair evidence that teosinte is inherently ‘n maize and Tripsacum in this character. As Nobo- ‘ New hybrids have almost uniformly paired spikelets, Fig. 1-11.—Fig. 1-2. Spike of F1 hybrid of Nobogame X New teo- sinte.—Fig. 1. Edge vieW.—Fig. 2. Lateral view.—Fig 3-7. Rachis segments. Fig. 3. Diploid Tripsacum dactyloides; nodal parenchyma present.—-Fig. 4. Florida teosinte.--Fig. 5. Mexican teosinte.--Fig. 6. Maize, lateral vieW.—Fig. 7. Maize, dorsal vieW.—Fig. 8-11. Spikes.— Fig. 8. Maize X diploid T. dactyloides, distichous.-Fig. 9. Maize X Jutiapa (Guatemalan) teosinte, distichous.—Fig. 10. Maize X Jutiapa teosinte, polystichous, the less common type.—Fig. 11. Maize X (maize X diploid T. dactyloides), distichous. - brids of Florida teosinte with other varieties have few ' e, a further suggestion is plausible that the Mexican s Nobogame and New, are intermediate between Flor- sinte and maize. Thus, a series of formsis indicated, based on single and a spikelets, the sequence being Tripsacum, Florida teo- exican teosinte and maize. ctness of Ear fiegments of the rachises of the plants studied here are logous with the internodes of the vegetative culms, and be compared with them. The internodes of the culms of them are commonly rectangular in lateral view but may various other forms. ,-he rachis-segments of Tripsacum are slightly trape- i, lfin lateral view (Figure 3) ; those of Florida teosinte are “jstrongly so, in that their angles are more pronounced re 4). In the Mexican teosintes and in the variety from Antonio Huixta, Guatemala, the segments are triangular e 5), an exaggeration of the trapezoidal form. Each i - of specialization from the rectangular form through pezoidal to the triangular gives an increase in compact- _ the spike. The rachis with triangular segments nor- "produces twice as many grains per unit of length as if ' ents were rectangular, and the number of its alicoles t of rachis-length is the maximum for the distichous However, the number may be greater in the polysti- pike. The distichous spike with triangular segments, re, approaches the polystichous condition in compact- Ylfv zThe segments of the polystichous maize rachis (Figures may be interpreted as being specialized along the same _al pattern as those of Tripsacum and teosinte, but more ly so. Here the segment is the unit of the cob to which r of spikelets is attached and which normally produces r of grains. It is so highly specialized in form and ar- ent that it is scarcely recognizable as an internode, fflsometimes is interpreted as a node with no accompanying _f7 ode. The segments are so shortened that they are wedge- ZQJ» units never extending any farther inwards than the [a1 axis of the cob, and therefore some of them may stand opposite others. This specialization in arrangement j_ integral part of the polystichous character, and it con- fites toward a more compact ear. __g_ Several types of observations give further support to these interpretations. The trapezoidal and triangular form of internode, as found in the rachises of Tripsacum and teo- sinte, are found also in the ear-shanks of maize and are espec- ially pronounced in the culms of a brachytic form described by Kempton (1921). Distichous branches of maize ears have segments with approximately the same degree of compactness as those of teosinte spikes with paired spikelets (Figures 1, 2). Since the organs cited here are less compact than the maize cob but homologous with it, or in one instance a branch of it, the form of the ordinary cob-segment of maize is corre- lated with compactness of the ear. It follows also that the cob-segment is the result of a high specialization of the same kind observed, though not so highly developed, in both Trip- sacum and teosinte. Thus in form of rachis-segment, a com- ponent of compactness of ear, teosinte is intermediate between maize and Tripsacum. In number of rows of alicoles of the ear, teosinte and Trip- sacum appear on casual examination to comprise one class and maize another, because spikes of both teosinte and Trip- sacum are distichous and those of maize are polystichous. However, through studies of appropriate hybrids between the species, it was found that this character is multifactorial, that teosinte has more factors than Tripsacum tending to make it polystichous, and that certain of the varieties of teosinte dif- fer from one another in genotype. F1 hybrids of maize with Tripsacum (Figure 8) and with Florida teosinte (Figures 9, 10) had almost uniformly dis- tichous spikes, but the similarity ends here. The backcross progeny of (maize X'Tripsacum) X maize also had almost uniformly distichous spikes (Figure 11), but the correspond- ing backcross progeny of (maize X Florida teosinte) X maize was variable, the mean number of rows of alicoles being 4.49. Only one plant in the population of 79 had ears with two rows of alicoles. It must be pointed out that the proportion of maize germplasm in the two backcross progenies is only approxi- mately similar. The Tripsacum backcross progeny, which may be regarded as having the genomic composition MMT, is estimated to have approximately 67 percent of its genes from maize, and the teosinte backcross progeny an average of 75 percent. But the difference in number of rows of alicoles is out of proportion to the difference in percentage of maize germplasm. The comparable backcross progeny of maize with Nobogame teosinte, a Mexican variety, gave a mean of 4.80 rows of alicoles, and that with Durango teosinte, also a Mexican variety, a mean of 4.91. The differences between __9__ the maize backcross progenies involving the Guatemalan va- riety and each of the two Mexican varieties were highly sig- nificant, but the backcross progenies involving the Mexican varieties were not significantly different from one another. The results given here 0n hybrids of maize with teosinte are in agreement with those of Rogers (1950b). It is of incidental interest that the rachis-segments of Tripsacum have nodal parenchyma (Figures 3, 12), as de- scribed by Weatherwax (1926) in the Oriental genera Poly- tocrl, Scleraclzne and Chicnachne, of the Maydeae. Depth 0f Alveolus of the Ear This character is unique among those included in this study in that teosinte does not seem to come within the com- bined range of maize and Tripsacum. This conclusion Was reached without the taking of quantitative data, but it seems to be jusified on the basis of the examination of many speci- mens. The alveoli of teosinte are deeper in proportion to the diameter of the rachis-segment than those of Tripsacum (Fig- ures 12, 13) and deeper than those of most varieties of maize. Apparently, therefore, teosinte could not have inherited its deep alveolus from either of the other two species. However, it is possible that the teosinte alveolus represents a combina- tion of depth and compactness, because the alveolus of pod corn may vary greatly on a single ear. At the tip, where the rachis is elongated, the alveolus is a flat shield-like structure; but at the base, where the rachis is more compact, the alveolus may be a deep depression. An alternative explanation may have transgressive segregation as its basis, provided teosinte originated as a hybrid between maize and Tripsacum. Branching Habit The profuse tillering of Tripsacum, teosinte and certain varieties of maize have been observed and described many times previously. To make a fair comparison of the three groups, however, a study of the types and relative amounts of branching in general, rather than merely the amount of tiller- ing, is the more instructive. In a study of this kind, the un- derground branching associated with the rhizomes of Trip- sacum should not be overlooked, since this is fundamentally the same type of phenomenon as that associated with the pro- duction of tillers or of aerial branches. Bews (1929) has pointed out that it is but a step from the type of aerial culm which takes root at the nodes to the rhizome which pushes its _1()_ way through the soil. According to Bews, rhizome produc- tion is usually associated with the presence of extra-vaginal buds, which may be the principal character that initiates rhi- zome production. It seems pertinent, therefore, to consider the various types of branching, and to determine, if possible,’ A whether teosinte occupies a position between maize and Trip- sacum, both in the profusion of its branches and in the posi- tion of the branches on the culm. The Tflpsacum plant branches profusely at its base (Fig- ure 14) ; probably most of its branching occurs beneath the :5 n4 ‘S/ X '6 4*‘ 1s 2o l8 Fig. 12-20.-—Fig. 12. Rachis segment of diploid Tripsacum dacty- loides, showing depth of alveolus and nodal parenchyma. X 1.75—Fig. 13. Rachis segment of Florida teosinte, showing depth of alveolus. X 1.75.—Fig. 14-17.—Branching habits, diagrammatic.—Fig. 14. Diploid T. dactyloides.—Fig. 15. Florida teosinte—Fig. 16. A prolific type of North American maize.—Fig. 17. Typical Andean maize.—Fig. 18-20. Large leaf hairs. X 50.—Fig. 18. Maize.--Fig 19. Teosinte.—Fig. 20. Diploid T. dactyloides. _11_ of the soil. Also its aerial parts often give rise t0 es, but these are relatively few. The inflorescence ter- f» each culm is either unbranched or sparingly branch- 1.1» refore, two characters of the branching of Tripsacum noteworthy: (a) the branches are inclined to be basal; 1W they are extremely, abundant. rida toesinte (Figure 15), a Guatemalan variety, is M her low-branching and- m.ore or less diffusely so. But its pistillate spikelets are borne on branches arising Qhat above the base, and; its terminal inflorescences are ifmore elaborately branched than those of Tripsacum. In the branches of Florida teosinte depart at higher po- ‘than those of Tripsacum. If only aerial branches were it is possible that those of Florida teosinte would be QIIIIHGPOUS than those of Tripsacum. However, the dif- would evidently be small, and in the Mexican teosintes f‘ ‘al branches apparently are decidely less numerous than - sacum. erefore, Mexican varieties of teosinte are a step far- om Tripsacum than is Florida teosinte. Some of them maize-like, in that they have relatively few tillers new branches of the terminal inflorescence. However, ‘o-form tillers and more branches of other kinds than g-irarieties of maize. north American maize supplies the fourth stage in the (Figure 16). Some of its varieties actually have less Vitely branched tassels than the most maize-like varie- ‘fw teosinte, but this is not a common occurrence. This expected in some fraction of the varieties, certain Mex- ‘varieties for example, if they have a tendency not only f1.»- their branches higher on the plant than teosinte ‘n to produce fewer total branches. It may be explain- ttheir tendency to produce tassel branches, which arise v positions on the plant, is in part vitiated by their other ‘cy to produce but few branches. In its ears, this "fng habit associated with the strong capacity for the tion of grains contributes to compactness. '% dean varieties of maize produce the fewest tillers of s studied. Their tassels are branched, usually with iaries. This is especially true of the tall varieties which ieved to be typical of that region (Figure 17). As pifred with North American maize, the ears have shorter and are borne higher on the culms. iperefore, the groups of plants under consideration here, Y‘: the two types of maize, may be arranged in a series __12_ showing progressively fewer total branches and progressively higher positions of the divergence of the branches. The se- quence is Tripsacum, Florida teosinte, Mexican teosinte and maize. Although Andean maize produces its branches at higher positions on the plant than the North American, the two types have about equal numbers of branches. Falling and Rooting of Culms Weatherwax (1918) described the branches of teosinte and Tripsacum as having a tendency to become prostrate un- der certain conditions, and those of teosinte as taking root after becoming prostrate. These phenomena seem to be re- lated to the branching habits already described. Tripsaicuwz has underground branches, rhizomes, that regularly take root and enable the plant to live indefinitely as a perennial. Its aerial shoots also show a tendency towards this behavior, but in these shoots the mechanism is weak and belated. Annual teosinte shows the same tendency to an even greater extent in its aerial shoots, but to a less extent in general. It does not have true rhizomes, and its aerial shoots are intermediate between the aerial shoots and the rhizomes of Tripsacum; but they are more like aerial shoots, since they usually do not take root and relatively few of them actually become prostrate. Perhaps maize exhibits the same character to a negligible de- gree, for its culm sometimes takes root far above the base, es- pecially if by chance it becomes prostrate. In this character we have a series of forms, the sequence being Tripsacum, teo- sinte and maize. Adaptation to Poorly-drained Habitats In so far as the writer has observed it in the wild condi- tion, Tripsacum dactyloides often occurs in poorly drained soil, although it can be grown as a mesophyte. When grown with a medium amount of moisture, it becomes somewhat dor- mant in dry seasons and renews its growth in rainy seasons. Teosinte probably thrives best in poorly-drained locations and is reported to be able to survive with the bases of its culms in water. Its prostrate branches frequently take root in mar- shy soil. But it also thrives naturally under dry conditions, for Kempton and Popenoe (1937) described thousands of acres of teosinte along the ridge separating the Camoja Val- ley from that of Rio Huixta in Guatemala; and there is was regarded as the dominant vegetation. It should be explained, therefore, that the observations in Tripsacum of tolerance to poorly-drained conditions apply only _13_. acum dactyloides. In addition, there are indications h teosinte and Tripsacum, or at least various types of re more tolerant to drouth than maize. Thus it may the character to be observed in teosinte and Tripsacum y, by which they differ from maize, is not simply tol- “to high humidity nor to drouth, but to a Wide range Jdity conditions. In either case, they are somewhat to one another and different from maize. ical Characters of Leaves “comparative study was made of seven measurable ana- characters of the leaves of Tripsacum, teosinte and These are enumerated, along with the means of their fment, in Table 1. Observations Were made on three l leaf characters, the results of which are summar- it Table 2 and described briefly. All of the maize plants " except those used for the study of thickness of leaf, .f White Surcropper, a North American variety. The i pf Tripsacum were diploids collected at Angleton, Texas aploids collected at Nacogdoches, Texas and New Ha- nnecticut. The forms of teosinte studied were the w-Anatomical characters of the leaf and stem of maize and its ' relatives; means of measurements. Tripsacum - North aracter dactyloides tlggzliftaé American M (2.n) maize "hairs on upper ; ace, number t.’ sq. mm. 6.9 7.5 17.6 3f of epidermal X . .095 .125 .116 f- ‘of hygroscopic r row f6 2.6 ‘ 2.5 apart, strips of " pic cells, mm. .248 .443 .767 - apart, 1n the row, mm. .083 .114 .133 apart, rows of mm. .082 .110 .121 of stomates, mm. .031 .045 .048 r of vascular l" 0f stem, mm. .217 .292 .431 I and San Antonio Huixta varieties. Only data fromc Angelton Tripsacum, Florida teosinte and White Surcropper maize were included in the statistical treatments. In each of these forms, leaf hairs are of two distinct sizes, which will be designated for convenience as “large” and “small.” The relative sizes of large hairs in maize, teosinte and Tripsacum are shown in Figures 18 to 20. For the study of frequency of large hairs, counts were made on units of leaf surface which were 0.45 square millimeter. This character was found to be more variable in Tripsacum than in either maize or teosinte, because of two exceptionally hairy plants of Tripsacum. They had a mean of 40.7 hairs per unit of leaf surface; whereas, nine other plants of this stock had a mean of only 0.13. When these two exceptionally hairy plants were Table 2.—Summary of characters of American Maydeae showing the plant groups in ascending order of magnitude of each charac- ter Paired spikelets—Tripsacum, Guatemalan teosinte, Mexican teo- sinte, maize. Length of rachis-segments—Maize, Mexican teosinte, Guatemalan teosinte, Tripsacum. Number of rows of alicoles—Tripsacum, Guatemalan teosinte, Mexi- can teosinte, maize. Depth of alveolus—Maize, Tripsacum, teosinte(?). Number of culm branches—Andean maize, North American maize, Mexican teosinte, Guatemalan teosinte, Tripsacum. Concentration of culm branches below—Andean maize, North Amer- ican maize, Mexican teosinte, Guatemalan teosinte, Tripsacum. Falling of culms—Maize, teosinte, Tripsacum. Adaptation to marshy habitat—Maize, (teosinte, Tripsacum).‘ Number of large hairs per unit of leaf surface—(Tripsacum, teo- sinte, maize.) 10. Number of small hairs per unit of leaf surface—Tripsacum, teo- sinte, maize. 11. Size of large hairs on leaf—Tripsacum, teosinte, maize. 12. Size of small hairs on leaf—(Tripsacum, teosinte), maize. 13. Length of epidermal cells of leaf-—-Tripsacum, (teosinte, maize). 14. Number of hygroscopic cells per row—(Maize, teosinte), Tripsacum. 15. Distance apart, strips of hygroscopic cells—Tripsacum, teosinte, maize. 16. Distance apart, stomates in the row-Tripsacum, teosinte, maize. 17. Distance apart, rows of stomates—Tripsacum, teosinte, maize. '18. Length of stomates—Tripsacum, teosinte, maize. 19. Total thickness of leaf blade—Tripsacum, (teosinte, maize). 20. Relative thickness of “rind” of culm—Andean maize, Guatemalan maize, Guatemalan teosinte, Tripsacum. 121. Size of vascular bundles of culm—Tripsa-cum, Guatemalan teosinte, Guatemalan maize, Andean maize. '22. Number of vascular bundles per unit of cross section area of culm- Andean maize, Guatemalan maize, Guatemalan teosinte, Tripsacum. '23. Development of sclerotic bundles sheaths-—Andean maize, Guate- malan maize, Guatemalan teosinte, Tripsacum. $°9°‘~‘9‘9‘P9°!‘°!“ ‘Groups whose names are enclosed in parentheses were indistinguishable. _15_ luded among the total of 11, the mean for the species was No other extraordinary character of the two hairy ts was detected. In general, the form of Tripsacum from ' Haven, Connecticut was more pubescent than the other forms; but even in this form, no plant Was found to be ly so pubescent as the two exceptional ones from Angle- In all of the forms of Tripsaczam studied, hairs occurred the midrid more frequently than elsewhere. In all three the species included in the study, leaf hairs 0f all types m t0 be restricted to the upper surface. In reference to small leaf hairs, teosinte is intermediate h‘ een maize and Tripsacum in number per unit of surface ‘ probably also in size. They are very scarce in Tripsacum, so few were found that the study of their size was not pletely satisfactory. The size of this type of hair in maize ‘roaches that of the large type in Tripsacum. For length of epidemal cells, approximately equal num- of measurements were made on the upper and lower sur- in each of the species, but the mean lengths of cells on I two surfaces were almost identical and the data taken 1 them were therefore combined. Specialized types of ‘o'demal cells were not included in these measurments; such, i. example, as those near vascular bundles, guard cells, cells acent to stomates, hygroscopic cells and cells giving rise ilhairs. Mean length of epidermal cells are shown in Table In actual means, maize is intermediate between Tripsacuwz ‘a teosinte, but the difference between maize and teosinte ¢ not significant. The differences between Tripsacum and ' of the other two species were highly significant. ,_ues for each species, and these require but little comment. q hygroscopic cells in all plants studied were localized in ‘gitudinal strips usually several cells wide, and the measure- j: of width of the strips were recorded in number of cells. b differences found between Tripsacum and each of the two species were highly significant. but the difference o een maize and teosinte was not significant. In distance _ Art of the strips of hygroscopic cells, and distance apart of pf». ates in the row, each of the three species showed highly ificant differences. In distance apart of the rows of sto- es, the differences between Tripsacum and each of the er species were significant at the .01 level, and the differ- ‘e between maize and teosinte was significant at the .02 p In length of stomates the differences between Tripsa- and each of the other species were significant at the .01 and that between maize and teosinte at the .05 level. Other leaf characters are listed in Table 1 with the mean_ .__ 15__ The relative thickness of leaves of the three species is illustrated in Figures 21 t0 23. N0 quantitative data were taken on this character, but all observations indicate that the leaves of maize and teo-sinte were thicker than those of Trip- sacum. There was little or no difference between maize and teosinte, but according to most of the observations the leaves of maize are a little thicker than those of teosinte. Fig. 2‘1-27.—Fig. 21-23. Cross sections of leaves showing relative thickness.—Fig. 21. Maize.—Fig. 22. Florida teosinte. Fig. 23. Dip- loid Tripsacum dactyloides. Fig. 24-27. Segments of culms, showing relative thickness of “rind” and size of culm.—Fig. 24. Diploid T. dacty- loides.—Fig. 25. Florida teosinte.—-Fig. 26. Guatemalan maize—Fig. 27 . Typical Andean maize. Observations indicate that the culms of Tripsacum are . most tenacious, in proportion to their size, of all of the ies included in this study, and that those of Andean maize "f - the least so. Teosinte and North American maize are in- rmediate between the extremes. When plants of the various 5» ds of maize were grown under similar conditions in the yld and subjected to strong winds, Andean maize was ob- frved to break most readily. In one field under observation, Q1 than 65 percent of the plants in the plots of Andean were broken, while North American maize growing in jacent plots showed practically no broken plants. Guate- }= an maize usually resembled North American maize in this it ect. These observations refer to immature plants. Such observations on toughness of culm were made ough a period of 3 years, and a study of the anatomical racters accounting for the difference was finally under- glen. Tripsacum, Florida teosinte, Guatemalan maize and .-| dean maize were examined for four anatomical characters: i) thickness of the woody peripheral region of the culm in flation to total size, (b) size of vascular bundles, (c) num- ~= of bundles per unit of area in cross section and (d) rela- degree of development of the sclerenchymatous bundle eaths. Figures 24 to 27 show the relative thickness o-f the woody _ ipheral region, sometimes designated as “rind”. Before ling photographed, these specimens were allowed to dry so to shrink the parenchyma and show the rind as clearly as "wcticable. An accurate study of the thickness of the rind relation to size of culm is difficult to make, on account of lack of sharp distinction between the rind and the central f3» but the most satisfactory measurements indicate that approximate ratios of thickness of rind to total diameter culm in the various forms are Tripsacum 1 :6, Florida teo- ite 1:8 Guatemalan maize 1:11 and Andean maize 1:13. A In all of these forms, the epidermis and a few layers of bepidermal cells are small and thick-walled, and the num- of these cell layers varies widely depending on the group p; plants. The vascular bundles in the subepidermal region ,- relatively small and their sheaths very strongly developed. e sheaths of two or more adjacent bundles sometimes con- ge, so as to form a continuous, thick layer of sclerenchyma. A gradual transition was found from subepidermal cells th extremely thick walls and no intercellular spaces to the __13_ typical parenchyma cells of the central region. However, it was usually possible to distinguish the typical parenchyma of the central region from the transitional region external to it, and frequently the transitional region could be reasonably Well distinguished from the subepidermal area of typical scler- enchyma. In Andean maize, a subepidermal layer three to six cells in thickness was found to be made up of elements with strong- ly thickened walls. Beneath this, the transitional region was 25 to 30 cells thick. In Guatemalan maize, the strongly scler- ified subepidermal layer was found to be of about the same thickness as that of Andean, but the transitional region was thicker, usually 35 to 50 cells. Florida teosinte has a sub- epidermal layer of sclerenchyma two to five cells thick and a transitional region 15 to 20 cells thick. However, the cells of the transitional region had thicker Walls in proportion to the size of their lumina than was found in the corresponding region of either type of“ maize. _ These cells usually had little more than half the diameter of the typical parenchyma cells of the central region, and their walls were three to five times as thick, but intercellular spaces were not uncommon. In Tripsacum, the sheaths of the most peripheral bundles were so strongly developed that they usually converged with one another and with the fibers just beneath the epidermis. In this way the external fibrous region was about the width of one or two vascular bundles, and no distinction was seen be- tween the subepidermal region of sclerenchyma and the fi- brous sheaths of the bundles. Areas occurred within this cir- cular zone, however, in which the bundles sheaths did not con- verge. Here the cells were thick-walled and small, but inter- cellular spaces sometimes occurred. Although this region is the homologue of the transitional region found in the other three forms, it is less genuinely transitional, because most of its cells have greatly thickened walls. For the study of size~of vascular bundles, measurements were made of the maximum diameter of bundles from plants of each of the four groups. Measurements were made only on bundles located in the central region of the culms, because these were differentiated from the tissue surrounding them. The measurements were recorded in tenths of a millimeter and the means from all except Andean maize are given in Table 1. The mean for Andean maize is 0.456mm. In this character, all the differences between North American maize, teosinte and Tripsacum were highly significant, and the difference be- tween Andean and North American maize was significant at the .05 level. Figures 28 to 31 illustrate the size relationship Tall"). A\__.A 1 ~r \_ <_ -I =o\1.f9~'»~*'~‘/@@>\ e 1- .3 .»>@.-»./~@~ - ..‘..'i-t~‘=»*;@~@2—=@' ._ f. Q ._ .44 l.‘ “hwy . %_ f‘ "‘ “ 1%).‘ I_ ‘(If p233," ‘5_Q\£Q"\.. Hag‘; ..|I’.O 0 , - I r 4'. i-n- f; \ "u Tv» n l. V!‘ ‘J ‘s9? " Q9 \ o -v I-lx \ ®;@§3‘§~/@@;2§¢§' wasigaafié’ L’ HP\%§QGQY§§ (a . .»/\(Q_,~2-=-.~a§2-a~' “Mlgfiv Qr r-lIT-fiflfi!‘ War?» ‘ Q!‘ l l ' '..§QI“'O~ / 2am»! ha’ wk. > > I .( ,~\_/ ' _ _ ‘ , >1\/'\.7' 1 2., :- l / > -. ," . . I ~ ') ’ $04.; @@)¢é>@4r¢)§""7 “ com Q 2 9 @© ...~@~a@g@§@,@§@g§§2@g@g L © ,@ o . . @@> Q’ 9% '9’ 905-“6. ® @ A, . ,7©@@e§'1~€’ a ~--. \ * . O_@‘L»"@§“§@@ 69%“ w Orb/a" “ow @ t; Qéa Q; ®5\ @052 xk w" ., Jot; \\)l§7’@ @ ‘f9 m ms.- ®-»,®@ ¥~Q Qqosl? ‘(.\~1.?§'1@ time‘ 3 gzswzaé. filvvjigf? 3° -c5>3@§@"*@§’ li->@@A Q=€>QY°QQEQ5€Q§ d3‘; o‘Z>“~Y@@@' @<.§aoa»~@@@; kfw». a; @(‘@@<-‘ © Q eec so roe” @¢"@@ C» f 3 dg§éigg~éwbs°° eh . ’ Q ~._'.Qa? - _ .,@@ ‘@353 00909 T) u! éQ-‘GQQQQQ i?» ‘a "'7 ' ‘?%‘s<> . ‘%§€@.i%2§1%»@ _'..->,: ‘gs-a. ESQ? <93»? a Peggy fin ‘rsa .92‘ . ~>;;@Q%9e=?‘ . OQQQQQ‘ ° ?§’o'~%° 0 =35‘ 00g 3% Fig. 28-35.—Fig. 28-31. Vascular bundles from the inner, parenchy- -matous region of culms. X 87.-—Fig. 28. Typical Andean maize.— ifljqFig. 29. Guatemalan maize.—Fig. 30. Florida teosinte.-—Fig. 31. Dip- loid Tripsacum dactyloides.—Fig. 32-35. Vascular bundles from the {peripheral zone, or “rind”, of culms. X 132.—Fig. 32. Typical Andean .l;maize.—Fig. 33. Guatemalan maize.—Fig. 34. Florida teosinte-Fig. V435. Diploid T. dactyloides. _2()__ in bundles from the inner region, and Figures 32 to 35 from v the peripheral. Observations on number of vascular bundles per square millimeter 0f cross section of culm indicate that the series in descending order is Tripsacum, Florida teosinte, Guatemalan maize, Andean maize. This is suggestive that the greater number of bundles per unit of area is associated with their minuteness, and such a relationship in itself would be expected to increase the mechanical strength of the culm. The structure of individual bundles also is interesting with respect to an explanation of mechanical strength. Trip- sacum usually has the greatest proportion of thick-walled cells and Andean maize the least. Guatemalan maize and teosinte are inclined to be between these extremes, this being es- pecially true of teosinte (Figures 28 to 35). The absence of protoxylem elements and xylem paren- chyma in the peripheral bundles is in agreement with a re- cent report of Sass (1951). The occurrence of bundles re- duced to strands of sclerenchyma, however, was not included in the study. DISCUSSION The 23 characters of maize, teosinte and Tripsacum an- alyzed on preceding pages and the conclusion drawn from each of them are summarized in Table 2. An attempt will now be made to interpret the entire list, along with related data recorded in the literature. Thirty-four characters of maize, teosinte and Tripsacum were previously compared by Mangelsdorf and Reeves (1939). Two of them, paired spikelets and number of rows of alicoles, are repeated here to give a more complete understanding of them. When the 23 characters listed in Table 2 are combined with the 34 previously reported and allowance made for the two repetitions, 55 characters are found to have been studied. In still two other characters mentioned in the previous report, total number of tassel branches and number of secondary tas- sel branches, maize appeared to be intermediate between Trip- sacum and teosinte. Since the report was published, how- ever, Reeves (1950, Figures 8 and 9) reported varieties of maize having numerous branches of both kinds; in fact, pro- fusely branched tassels of maize, including those of varieties from the Andean region, now are commonly observed. It seems probable that maize should be regarded as essentially similar to teosinte in these characters. _2]__ _"‘;_;~ of the 55 characters should be disregarded until ; more completely, because the bearing they may have froblem is doubtful. In depth of alveolus, the evidence “y ear that the condition observed in teosinte could have Ad from either of its putative parents, maize or Trip- {jor from a hybrid between them. Frequency of large is extremely variable, and although the mean of is between that of maize and Tripsacum, the results - significant. According to the Writer’s observations ervations of colleagues, the other three characters of ‘l; rs, numbered 10 to 12 (Table 2), are more variable ch of the three groups, North American maize, Guate- eosinte and diploid T. dactyloides, than the data indi- owever, the samples were chosen without prior knowl- their minute characters, and, in fact, without regard zof their special characters. It is believed, therefore, samples are fairly representative of their respective in the remaining 53 characters, even though an occas- lant or population may depart rather widely from them. y, apparent correlation between certain of the charac- ‘y give rise to the impression that in some instances a “ation of correlated characters is controlled by a single ith pleiotropic effects. Examples of such characters use numbered 20 to 23 (Table 2), which contribute to news of culm. Martin and Hershey (1934-5) found that lulm in maize is correlated with large vascular bundles th few bundles per unit of cross section; and such cor- ns seem to exist in some measure in the plants studied jIt may be that the characters of the stomates, numbers V} 8, are similarly correlated. om the standpoint of the origin of these groups of ; the number of genes controlling the characters, as well Qnumber of characters themselves, is important, because inic differences indicate the number of mutations neces- f» the direct descent of one group from another. Even I’ true that some o-f these combinations of characters are led by genes with manifold effects, there are at least ‘i-reasons for a suggestion that some of the single char- f differentiating the plant groups are quantitative, or ent on two or more genes. lirst, it would be difficult to postulate three or more ‘genetic classes based on differences in the same char- * such as we have in many of these instances, as differ- a single mutation. Second,» the analysis of variance f: h of epidermal cells, distance apart of stomates in the i; d length of stomates indicates that the differences be- _22__ tween the three species are significant at the .01 level when tested with differences between plants within species. Third, the genetical work of Mangelsdorf (1947) and Rogers (1950a, b) on eight characters in which maize and teosinte differ show that each 0f those characters is controlled by genes on more than one chromosome. For paired vs. single spikelets, at least two chromosomes and possibly six others are involved; for tillering habit at least three. Other characters not included in the present study, such as glume development and disti- chous vs. polystichous ear, were found by Mangelsdorf and Rogers to be influenced by genes on at least 7 of the 10 chro- mosomes. In as much as the present study was not designed to determine the number of genes controlling the characters, there would be little value in further discussion of the ques- tion. It may be dismissed for the present with the assertion that in all probability the number of mutations necessary for the derivation of one of these species from another is much greater than the number of characters which separate them. In all the 55 characters mentioned earlier in this discus- sion, teosinte is dissimilar to maize in 19 characters of the previous report and 19 of the present, which make a total, af- ter subtracting the two repetitions, of 36. Indeed, this list is subject to revision and supplement by additional study, but it is indicative of the number of mutations necessary for the direct descent o-f maize from teosinte. Further reference should be made to the eight characters listed in Table 1, for which small samples from only one or a few forms from each species were studied. Although the interpretation of results applies only to the restricted samples, if we add the assumption that the small samples are fairly representative of the original teosinte and its parents, the re- sults take on a more positive meaning. Yet, the possibility remains that forms somewhat different from these might also fit the hypothesis of the hybrid origin of teosinte. It is un- necessary that the samples studied here be extremely similar to the original teosinte and its parents; it is of course neces- sary that any combinations of samples, from whatever source, representing the hypothetical parents, possess such characters that hybrids between them might be expected to bear a fairly close resemblance to some known form of teosinte. It is now noteworthy that teosinte is intermediate be- tween maize and Tripsacum or similar to one of them in all characters studied, with two doubtful exceptions. If the char- acters studied comprise a fair sample, the results impose a serious difficulty on the theory of Weatherwax (1918) that maize, teosinte and Tripsacum descended from a common sn- _23_ A V a series of mutations. Assuming that the origin of mutant characters is a matter of chance, that theory vmand a series of coincidences, occurring against great the results are in complete agreement with those ex- éif we adopt the hypothesis that teosinte originated as ,4: of a hybrid between maize and Tripsacum. The '_are not finally conclusive on the problem of the degree g rity of the maize and Tripsacum studied to the pre- arental types that gave rise to teosinte. They give no (on of the time when the hybridization might have oc- Y and therefore none on thf time of origin of teosinte. ,_er, the literature contains a few facts and hypothesis interrelated questions. M: a study of the literature on archeology and plant _phy, Mangelsdorf and Reeves (1938) estimated the origin of teosinte as about 600 A.D., and later (1939) l this estimate to 900 A.D. Either of these relatively dates implies that the maize and Tripsacum postulated ynts of teosinte were recent, perhaps present-day, forms. (1939), Weatherwax (1950) and Randolph (1952) red the difficulty in the hybridization of maize and m to be a ruinous weakness of the hypothesis of the _5origin of teosinte. In the meantime, Mangelsdorf and (1949) described archeological specimens of maize at Cave, New Mexico, and some of the specimens show- ence of contamination from teosinte. The specimens 5». in six strata, and Arnold and Libby (1951) esti- i; the ages of the various strata by the radiocarbon . Specimens from the three lower strata, 3500 to 2249 years old, showed only doubtful evidence of contami- from teosinte. Specimens from Stratum IV, age 2239 ,years, and all higher strata showed unmistakable evi- Yof contamination. From these data, the inference is 'ed that teosinte was present in the vicinity of Bat Cave ‘the beginning of the Christian Era, possibly many cen- “ before. Mangelsdorf and Smith's publication (1949), “rbefore the radiocarbon technique was applied to the ave specimens, states that the hybridization between Qand Tripsacum which produced teosinte must have oc- ' no later than 500 B.C. and perhaps much earlier. This Ste and that of Arnold and Libby are in satisfactory ‘- ent. bbins (1950), recognizing that the genes of teosinte ;»'ch it resembles Tripsacum rather than maize are not at random over the chromosomes but tend to be _24_ grouped in a few segments of certain chromosomes (Mangels- dorf 1947), states that it is difficult to see how such a situa- tion could have arisen except through hybridization. He also recognizes the weakness of that version of the hypothesis which requires the derivation of a fertile segregate from a hybrid between forms of maize and Tripsacum which thus far have been cross-pollinated experimentally. In an effort to explain these two apparently contradictory bodies of data, Stebbins adopts the view, which is well supported by facts, that the extant 18-chromosome (gametic number) forms of Tflpsacum ordinarily designated as diploids are themselves really allopolyploids, and therefore of hybrid origin. He sug- gests that one, o-r conceivably both, of the extinct parents of Tripsacum contained forms which were more closely related to maize and more interfertile with it than is Tripsacum it- self. Thus, he assumes that teosinte might have originated as a hybrid between ancient maize and a 9 or possibly a 10- chromosome parent of Tripsacum. It should be pointed out that the possibility is not ex- hausted of finding combinations of modern maize and Trip- sacum that are substantially more interfertile than those al- ready tested. Also, it may be that fertile segregates eventu- ally will be obtained from hybrids between the forms of maize and Tripsacum which thus far have shown only low interfer- tility. But in the absence of positive evidence on these pos- sibilities, Stebbins’ assumption of hybridization in a remote period must be viewed with favor as a provisional hypothesis. However, Stebbins’ explanation of teosinte as a hybrid between ancient maize and one of the 9-chromosome parents of Tripsacum apparently means that the genom brought into Tripsacum by its other parent accounts in large part for the strong barrier now observed between Tripsacum and maize. If this were true, we should expect more regular synapsis and gene exchange than has been reported between the modern maize genom and one of the 9-chro-mosome genoms of Trip- sacum. Cytogenetical studies of hybrids between various com- binations of modern maize and Tripsacum, reported by Man- gelsdorf and Reeves (1939) and verified by Randolph (1952), show very little synapsis or exchange of genes. It seems, therefore, that if teosinte did originate as a hybrid between some ancient form of maize and an interfertile 9 or 10-chro- mosome parent of Tripsacum, the chromosomes of Tripsacum must have been more closely homologous with those of maize, and that maize and Tripsacum were more interfertile at that ancient time than now. This removes at least part of the necessity of Stebbins’ assumption that the non-maize parent _25__ inte was also one of the parents of Tripsacum. Like- f1 nce Tripsacum is fairly satisfactory as a hypothetical p of teosinte, from the standpoint of plant characters, l on might well be raised as to whether one of its own would be equally as satisfactory. This, of. course, ain an openquestion. view of all these facts and deductions, the suggestion Qjustified that teosinte originated as a natural hybrid In a form of maize and an 18-chromosome Tripsacum, lbridization having occurred at a remote time when jzand Tripsacum possessed somewhat the same plant p ers as now but when they were more interfertile. ACKNOWLEDGMENTS Y} ring the preparation of the manuscript, many helpful fj_'tions were made by P. C. Mangelsdorf of Harvard Uni- ,, H. C. Cutler formerly of the Chicago Natural History _ , now of the Missouri Botanical Garden, and C. B. _v of the Agricultural and Mechanical College of Texas. j iter wishes to express his appreciation for this assist- deae. Indiana Univ. Studies 13, No. 73. LITERATURE CITED Arnold, J. R., and W. F. Libby. 1951. Radiocarbon dates. Science 113: 111-120. Beadle, G. W. 1939. Teosinte and the origin of maize. , J our. Heredity 30: 245-247. Bews, J. W.‘ 1929. The world's grasses. Longmans, London. Cutler, H. C., and Edgar Anderson. 1941. A preliminary survey of the genus Tripsacum. Annals Missouri Bot. Gard. 28: 249-269. Kempton, J. H. 1921. A brachytic variation in ‘maize. U.S.D.A. Bul.‘ 925. ‘ -————i-——, and W’. Popenoe. 1937. Teosinte in Guatemala. Car- negie Inst. Washington Pub. N0. 483: 199-218. Mangelsdorf, P. C. 1947. The origin and evolution of maize. In Demerec, M. Advances in Genetics 1: 161-207. Academic Press, New York. —-i—-—-——-—, and R. G. Reeves.‘ 1938. The origin of maize. Proc. Nat. Acad. Sci. 24: 303-312. . 1939. The origin of Indian corn and its relatives. Texas Agric. Expt. Sta. Bul. 574. , and C. E. Smith, Jr. 1949. New archeological evi- dence on evolution of maize. Harvard Univ. Bot. Mus. Leaflets. 13: 213-247. Martin, J. N., and A. L. Hershey. 1934-35. The ontogeny of the maize plant-—The early differentiation of stem and root structures and their morphological relationships. Iowa State College Jour. Sci. 9: 489-503. Randolph, L. F. 1952. New evidence on the origin of maize. Amer. Nat. 86: 193-202. i Reeves, R. G. 1950. Morphology of the ear and tassel of maize. Amer. Jour. Bot. 37: 697-704. Rogers, J. S. 1950a. The inheritance of photoperiodic response and: tillering in maize-teosinte hybrids. Genetics 35: 513-540. 1950b. The inheritance of inflorescence characters“ in maize-teosinte hybrids. Genetics 35: 541-558. Sass, J. E. 1951. “Reduced” vascular bundles in maize. Iowa State College Jour. Sci. 26: 95-98. Stebbins, G. L., Jr. 1950. Variation and evolution in plants. Columbia Univ. Press, New York. Weatherwax, P. 1918. The evolution of maize. Bul. Torrey Bot. Club 45: 309-342. 1926. Comparative morphology of the Oriental May- 1950. The history of corn. Scientific Monthly 71: a 50-60. <