AN ANALYSTS OF A NEGATNE RESPONSE 0R HTGH YTELD TN T0 SELECTION F WTNTER BARLEY, HORDEUM VULGARE. Thesis for the Degree of Ph. D. MTCHTGAN STATE UNNERSTTY CECTL D. NTCKELL 1967 11111111111111111111111 31293 01101 2915 This is to certify that the thesis entitled AN ANALYSIS OF A NEGATIVE RESPONSE To SELECTION FOR HIGH YIELD IN WINTER BARLEY, HORDEUM VULGAEE presented by Cecil D. Nickell has been accepted towards fulfillment of the requirements for PhoDo degree in C292 SCienCe / DateJfl 27/ /;>u/: 7 0.169 LIBRARY 1 Michigan State A Univcraity I... 1" BINDING av “a "MB & SUNS 300K BINDERY INC. “IRA“ Inn-true ABSTRACT AN ANALYSIS OF A NEGATIVE RESPONSE TO SELECTION FOR HIGH YIELD IN WINTER BARLEY, HORDEUM VULGARE by Cecil D. Nickell A winter barley population composed of 387 lines in the F5 generation was grown in 1966 from which 136 strains were se- lected with a mean yield of 123. 3 per cent of the check variety. The same selected population was grown in 1967 producing a mean yield of 90. 7 per cent of the check variety. The ”apparent" negative response to selection was ana- lyzed with reSpect to yield components; heads per unit area (X), seeds per head (Y), and seed weight (Z). Twice as many heads were produced in 1967 as compared to 1966, but fewer seeds per head and smaller seeds were produced in 1967 than in 1966. The inverse re- lationship between the components is explained by "component com- pensation. " The inter-annual correlations for the yield components- were slightly positive, while the value for yield was slightly negative. Cecil D. Nickell The overriding effect of the environment on the genetic processes and extreme compensation between yield components were postulated to cause the low inter—annual correlations. An ”universal" surface was created by plotting seed weight, seed number per area, and yield in three-dimensions. Mathematically and biologically, only one surface can be formed up- on which all seed bearing crops can be placed. Each seed crop has an unique position upon the surface. The swarm of points on the "universal" surface created by the data collected in both 1966 and 1967 can be thought of as representing a portion of the total evolu- tionary highway for barley. AN ANALYSIS OF A NEGATIVE RESPONSE TO SELECTION FOR HIGH YIELD IN WINTER BARLEY, HORDEUM VULGARE By Cecil D. 'Nickell A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Crop Science 1967 COpyright by CECIL DEAN NICKELL I967 ACKNOWLEDGMENTS The author wishes to express his sincere appreciation to Dr. J. E. Grafius for his help and guidance throughout this study. Thanks are due especially to Dr. M. W. Adams and Dr. C. M. Harrison for their assistance in preparation of the manuscript. The author also wishes to acknowledge the many hours spent by his wife assisting with the many chores associated with the study. Appreciation is expressed to the Malting Barley Improve- ment Association, Milwaukee, Wisconsin, for their financial support during the course of the investigation. ii TABLE OF CONTENTS ACKNOWLEDGMENTS LIST OF TABLES LIST OF FIGURES INTRODUCTION LITERATURE REVIEW . MATERIAL AND METHODS . RESULTS DISCUSSION . Component Inter-relationships . Universal Surface SUMMARY . LITERATURE CITED . iii Page ii iv 30 30 34 40 42 LIST OF TABLES Table Page 1. Comparison of yield component means between the selected and unselected population, and between the selected population grown in 1966and1967........,.......13 2. Correlations between yield and yield components calculated from the data collected in 1966 and 1967 . . . . . . . . . . . . . . . . . . 14 3. Inter-annual correlations of yield components and yield based upon data collected in 1966 and 1967 . . . . . . . . . . . . . . . . . . 15 4. Expected and observed values of the regression equations, seed number, and angles in each plane of seed number in Figure 7 . . . . . 27 iv Figure LIST OF FIGURES The distribution of 387 winter barley lines against yield categories for 1966 The distribution of the selected lines grown in 1967 against yield categories The distribution of 387 winter barley lines against seed number categories for 1966 The distribution of the selected lines in 1967 against seed number categories, with Hudson as a check The distribution of 387 lines against seed weight categories for 1966 The distribution of the 136 selected lines grown in 1967 against seed weight categories Regression of yield upon seed weight at various levels of seed number per . 9 meter in winter barley The regression of quantum change between the data of 1966 and that of 1967 for seed number per . 9 meter for 136 lines of winter barley, with 95% confidence bands . The regression of quantum change between the data of 1966 and 1967 for seed number per . 9 meter paired at random, with 95% con- fidence bands Page 10 10 11 11 26 29 29 Figure Page 10. A representation of development over the growing season of the winter barley plantinMichigan 30 vi INTRODUC TION Selection, practiced under a dynamic environment, pro- vides the plant breeder with many frustrations. Qualitative charac- ters are usually affected only in a minor way by the environment, allowing the plant breeder to make selection progress. On the other hand, quantitative traits are polygenic and are subject to many alter- ations by the environment. During recent years, there has been a tendency to partition the more complex quantitative traits into sim- pler components. Certain complex characters often may be thought of as artifacts made up on actions and interactions of their compo- nents. Presumably, the heritability of these complex characters is lower, in general than for the components. Yield in winter barley is defined as the product of, in order of development, heads per unit area (X), seeds per head (Y), and seed weight (Z). Genetically, these traits are assumed to be in- dependent, but they do interact with each other and with the environ- ment. Even though the heritability is frequently high for the compo- nents, they are assumed to be quantitative characters. Selection, I even for components of the complex trait, yield, may produce some very unusual and undesirable results. A population of winter barley lines was created by making nineteen crosses with the goal of producing high yielding progeny of good malting quality. Selection pressure was placed on yield with special emphasis on larger seeds and higher tiller number. The expected selected population mean in 1967 should have fallen between the selected and unselected population means of 1966, (Figure 1 and 2) when the check variety mean is used as a reference point. If the check variety is used as a constant point, the "ap- parent" genetic gain is negative. To the plant breeder, a negative response to selection is quite undesirable. It is the intent of this thesis to analyze and explain why such results should have been experienced for yield under two envi— ronm ents. 120 , . I ' l ‘ Check mean 100 - I I : (E I Unselected pop. mean (0 g 80 7 I <-—- Selected pop. mean «'5 ' = ,, 60 - I Q’ l “E 2 40 I— : W ' ' I 20 F ~ : I I , 1 4 I 7 ”III _ 1714 2143 2572 3001 3429 3857 4286 4715 5143 5562 6004 Kg/ha FIGURE 1. -- The distribution of 387 winter barley lines against yield categories for 1966. The cross-hatched area represents the distri- bution of the 136 selected lines in 1966. The broken line represents the yield of the check variety, Hudson, in 1966. Selected o . mean 100 — p p I Check mean 80— 60- 40- Number of lines 20.. r—J 1 1 I I-——-1 l I ’ l l 1714 2143 2572 3001 3429 3857 4286 4715 5143 5572 6004 Kg/ha FIGURE 2. -- The distribution of selected lines grown in 1967 against yield categories. LITERATURE Yield in oats has been defined by Grafius (7) as a geo- metrical construct, namely, a parallelopiped with the yield compo- nents as the edges: panicles per unit area (X), kernels per panicle (Y), and kernel weight (Z). The longest edge is the most subject to change and the changes in edges or components tend to counterbal- ance. In cotton, Hutchinson (9) defined bolls per plant, seed cotton per boll, seeds per boll, and lint per seed as yield compo- nents. Selection was more effective for certain components than others and some of the characters were affected greatly by the envi- ronment. He was one of the first individuals to show that the in- crease of one character could be associated with the decrease of another, which .was called ”physiological incompatabilities. " Whitehouse fill. (16) described the yield components of wheat as weight per kernel, kernel per Spike, spikelets per ear, and ears per plant. Complete independence between these components was indicated by correlation analysis. Olsson (14) reported a tendency toward a decrease in seed weight in Brassica and Sinapis with an increase in the number of seeds per pod, and selection for low seed number produced a de— crease in fertility. Archibong (2) found that competitive stresses in navy beans caused significant changes in the yield component interrela- tionships. Pod number per plant was more subject to change than seed weight or seed number per pod. Adams (1) found negative correlations between yield com- ponents in field beans. He found that within these components low to zero correlations would occur in space plantings and negative corre- lations would occur in close-interval plantings. Inter-plant compe- tition for metabolites was suggested as the mechanism for component compensation. Johnson 5111. (10) found a highly negative association be- tween seed number per plant and seed weight both in phenotypic and genotypic correlations in soybeans. A negative association was also found between pod number per plant and seed weight. Bal _e_t_al_. (3) reported an association of heavy seeds in barley with small heads, after natural selection, again presenting ‘ the idea of negative association between yield components. Kambal M. (11) reported negative associations between yield components in hybrid grain sorghum but these were absent in the parent population and, in fact, a highly positive correlation exis- ted between seeds per head and seed weight. Other authors have published on the negative interrela— tionship of yield components: corn (Leng (12) ), several grass spe— cies (Van Keuren (18) ), tomatoes (Williams (17) ), and crested wheatgrass (Dewey and Lu (4) ). METHODS AND MATERIALS In 1962, twelve winter barley lines were selected as parents to start the second cycle of a recurrent selection program. These lines were presumed to be essentially homozygous since they were in the F8 generation. Selection was based upon several char- acteristics which included yield, seed weight, seeds per head, heads per unit area, disease reaction and nine malting quality traits. Combinations of parental lines were selected using the vector ap- proach as described by Grafius (8). Nineteen crosses were made in 1962 and each cross maintained in bulk through the F3 generation, and agronomic and malting data were collected. Eleven crosses were selected based on superior malting and agronomic performance. Five hundred heads from each cross were picked at random and planted in head hills in the fall of 1964. Severe cold injury occurred during the winter reducing the number of head hills from 5, 500 to approximately 2, 500. In the summer of 1965, 387 hills were selected using the criteria of plant vigor, disease resistance, seed size and tiller number. These selections were planted in the fall of 1965 in- single rod-row plots with check plots of either Hudson, Wong, or Dicktoo winter barley planted every eleventh plot. Winter kill occurred early in 1966 resulting in an average survival of 56%. In the summer of 1966, 136 lines were chosen, with selection being based upon yield data, winter survival, seed weight, protein analy- sis, and seed number per unit area. Replicated plots were planted in the fall of 1966 with Hudson and Wong as checks. Data were collected in 1967 on yield, seed weight, seed number per unit area, and disease reaction. No winter kill was observed in 1967. RESULTS Yield data from the population of 387 winter barley lines harvested in 1966 are summarized in Figure 1. The mean of the total population of 387 lines was 3903. 6 Kg/ha (91. 08 bu/acre). The selected population mean yield was 4490. 0 Kg/ha (104. 78 bu/acre). The same 136 selected lines were grown in 1967 with Hudson as the check variety. The mean yield (Figure 2) of the population was 3206. 0 Kg/ha (74. 8 bu/acre) and the Hudson mean was 3535.0 Kg/ha (82. 5 bu/acre). A decrease of 1284. 9 Kg/ha (30 bu/acre) occurred in the selected population mean comparing the 1966 and 1967 results. The check variety change was only 107. 2 Kg/ha (2. 5 bu/acre) lower in 1967. Two check varieties were used in both years, Hudson and Wong, and each responded the same in the two years; therefore, Hudson will be used as the check variety throughout the results. Figure 3 and 4 provide a summary of seeds per 0. 9 meter of linear row obtained from both the selected and unselected p0pula- tion in 1966 and 1967. The mean for the unselected p0pulation was 3850 seedsl. 9m compared to 42 92 seeds]. 9m for the selected p0pu- lation in 1966. The check variety produced 4015 seeds/. 9m which Number of lines 120 100 00 O O) O .p. O 20 10 I 11—,— Unselected pop. mean ' Check mean ‘<— Selected pop. mean \§______W \\\ ///m Seeds/. 9 meter 1800 2200 2600 3000 3400 3800 4200 4600 5000 5400 5800 6200 FIGURE 3. -- The distribution of 387 winter barley lines against seed number categories for 1966. The cross-hatched area represents the distribution of the 136 selected lines grown in 1966. The broken line represents the yield of a check, Hudson, 120 Number of lines 100 80 60 40 20 95 in 1966. é—h— Pop. mean I <— Check mean I | l I I I Seeds/. 9 meter 1 r——‘—‘ 1 1% I 1800 2200 2600 3000 3400 3800 4200 4600 5000 5400 5800 6200 FIGURE 4. —- The distribution of the selected lines in 1967 against seed number categories, with Hudson as a check. 44 120 r I I I < Check mean I 100 h | : Unselected pop. mean 3 80 — I l<——-Se1ected mean c: :2: ' . E | . $4 60 - | I m '2 l :s 40 - I Z 20 _ : r . // //// /,%I 22 24 26 28 30 32 34 36 38 40 42 Mg/seed FIGURE 5. -- The distribution of 387 lines against seed weight categories for 1966. The cross-hatched area represents the distribution of the 136 selected lines. The broken line represents the seed weight of the check, Hudson, in 1966. 120 — l I '1 100 _ l‘ 1 Check mean I m I '4 Pop. mean E 80 A | l I; ' . 60 — E I I l I g 40 - Z I 20 - I : . . I l L I II I n l I l l I 22 24 2 28 30 32 34 36 38 40 42 44 Mg/seed FIGURE 6. -- The distribution of the 136 selected lines grown in 1967 against seed weight categories. Hudson, in 1 967. The broken line represents the check, 12 was above the unselected population mean and below the selected population mean. In 1967, the 136 lines produced an average of 4101. 3 seeds]. 9m and the check variety 4843. 1 seeds/. 9m. Between years the selected population mean was reduced only 191 seeds/. 9m but the check variety increased 842. 8 seeds/. 9m. Figure 5 summarizes the distribution of seed weight in 1966 for the total population. The unselected population mean was 35. 7 mg/seed compared to 36. 6 mg/seed for the selected group. Figure 6 presents the distribution for seed weight of the selected population in 1967. The mean seed weight decreased 9 milligrams per seed below the 1966 mean. The check variety was depressed from 32. 0 to 26. 8 milligrams per seed in 1967. Seeds-per-head was determined on a limited number of lines by selecting 5 heads at random from each plot and counting the seeds. Table 1 provides a summary of these results obtained on 139 lines in the unselected population and 51 of the selected group. The selected group mean did exceed the unselected population mean for all the components and also for yield. The number of heads per . 9m doubled in 1967 over 1966. At the same time seed weight and seed number per head was reduced. Seeds per . 9m decreased ap- proximately 390 seeds. Yield was greatly depressed in this limited set of lines as it was in the total selected population. 13 TABLE 1. -- Comparison of yield component means between the selected and unselected population, and between the selected popu- lation grown in 1966 and 1967. The unselected p0pu1ation contains 139 lines, and the selected group is made up of 51 lines. Unselected Selected Trait 1966 1966 1967 Heads per . 9 meter (X) 91. 3 100. 9 199. 5 Seeds per head (Y) 42. 6 43. 8 20. 2 Seed weight mg/seed (Z) 37. 1 37. 4 28. 5 Seeds per . 9 meter (XY) 3889. 4 4419. 4 4029. 9 Yield (W = XYZ) Kg/ha 4122. 7 4721.0 3266. 6 Bu/acre 96.2 110. 2 76. 5 Phenotypic correlations were calculated between yield components for the unselected and selected populations (Table 2). Both the pooled and unpooled values between tiller number and seed number per head were highly negative in both populations and in both years. A negative relationship was found between tiller number and seed weight in both years and p0pu1ations, but was statistically sig- nificant only in the selected population for 1966, when based upon the unpooled data. The correlation of seed weight with seed number per head was zero in 1966. However, in 1967, a highly significant nega- tive relationship was found. Seed number per unit area was highly l4 negatively correlated with seed weight in both years and in both p0pu- lations. TABLE 2. -- Correlations‘between yield and yield components cal- culated from the data collected in 1966 and 1967. (The t0p line rep- resents the pooled coefficients over 11 crosses and the bottom line the unpooled correlation coefficients.) r r r lJr xy xz zy nz Unselected population 387 lines _. 34>I<>1< -.28** 139 lines -, 44** -, 32M .12 -0 34.29.. -.72** -.16 -,02 _,41** Selected population 1966 136 lines -. 55:32: -.58** 51 lines , 77M: . 35* .16 .56** .71** ,38* .05 .59** 1967 136 lines , 60** .13»: 2239.012 .45“: 0 46** . 12 -. 37** . 50*:{z 51 lines . 562101: . 07 _. 27 . 45*): .34* .04 -.35* ,49** 1] . n = xy (seeds per . 9 meter of linear row) 15 Correlations were calculated between years for each of the yield components and yield (Table 3). The between-years asso— ciation for yield was zero, as was the case for tiller number and seeds per. head. A positive correlation was found for seed weight in the 136 lines in the selected population but it was low in magnitude. TABLE 3. -- Inter-annual correlations of yield components and yield based upon data collected in 1966 and 1967. U r r r r r xx .yy zz nn ww Selected population 136 lines . 24* . 16 . 05 . 36** .20* -. 12 51 lines .10 .22 .29* . 05 . 03 . 14 . 22 . 06 lIn = xy (seeds per . 9 meter linear row) Figure 7 is a multi-page representation of seed weight versus yield where each page represents one particular level of seed number. In order to obtain this graph, the data were stratified ac- cording to seed number using levels of seeds per . 9 meter, with levels differing by 400 seeds. The equation for the regression line on each plane and the corresponding correlation coefficient are presented in Table 4. I"! comm f I I . I I I ._.1.|.vI.JIuIII.tII.III I. All- IJ.I . l! . GoomRSV £935 comm . [km 2 A 82 Q . o o o o ., A 88 O o 0000 DO . ' . 3.. . “a 0 ol- I 0 O A 88 .u 5 a o _. o D . o... . o A coca o 53 5 don con—030m 0 $3 5 don vouoouom 4 . coon $92 5 ion H.309 4 an o o . ooom coon coou SNN 8w. , $3.23 .3553 E. gmwwfi @ . 29 .M 55:: noon mo 315a 9639» am Emma? comm con: 33.» mo nowmmoumom I .p mmDUHh A 26 288295 3985 8mm .3 h m. oooH o ooom ' O O 0 H o. O 0000 D D D J Doom .4. ooov pm: 5 dog poaoofiom 0 mmmfi 5 dog anoofiom d coon mama 5 don #309 4 d o ooom SNN com. 5 255 a. an Essa Ema .5 282 2.2:; E ESE... 8.... Swing hops?» cog: Born mo :ofimmopwmm I. .N. HMDUFH 26 833mg EmSB emmm. om ow om om OH H con: 0 O O u coon ’ u ooom h ooov 5:: E don Uwaoofiom O 3:: 5 dog popoofiom 4 ._ ooom $3 5 dog 388 d a. . m ooom oom— Soimn 93E? 5 .355 w. awn 598:3 comm mo £6.23 9.63m; «a £995 comm com: 32% mo :oflmmopmom I .N. mmDUHm 26 smegma: Ems? comm. cm 3. on om S _ . a _ . $2 9.. .aoa 883% 0 $2 5 dog 883% 4 .. 92: E 303909 d s . I4 oooH coon ooom ooov ooom coom $02.3 noun?» 5 .838 m . nod .8985: comm mo 39:: msoEw> Hm “swam? comm. com: 33% .Ho :ofimmopmom I .p HMDUHM 27 HASH S. a 3685555885.: H853 mo. 2... 568.55.... e833: .. H .mmcH o .c .cH. ovm .qu cow H. .vaomH o .oomHmH coco .o coco .H H. .pccc coco m comm Hfl .om o .o .pH com .mm com m .cmoovH o .HémwvH cHoc .o coco .H *H .msHm comm cH ooom c .3. o .o .mHV com .vm 0H5 m .momomH c .occcmH mmom .c coco .H *m .85 coov on come H .Em c .o .cH on L. omm w .33: m .mwmmmH. mcwc .o coco .H *m .32. ooHTHV 5. come mem o.o Em cum .qu ocm *chHocH o.omHHAHH comm .o coco.H Rocco ooqu HoH coon o .vv c .o .3 com .mH ohm o .mmmmcH o .ompmoH mmvc .c oooo .H c .ccmm occm coH oocm “woman o.o .m OHm .mm ovm $.3qu m.mcch Hmocc coco.H c.mmmm comm cw ocom o .HH o.o .cH omm .on OHm H..mcomo o.ococs momm .c coco.H *Anwocom oocm om oocm m .ocm- o.o .mw OHm .wm ocH H .moocv o.oovoc mono .o coco.H o.omHVm covm mH comm o.omu o.o .cH OHH .cm omH wcmvcm moocmm smooc coco.H o.ocom ocom HV oooH .20 .5me .20 9.: .20 9.: .80 9.: .30 .95 o : o:mHnH 9 on§ .300 :onmopmom .HooU:oHHmHo::oU 5.5852 comm .H. 0:5mHh 5 509.85: comm. .Ho mo:mHQ m50Hpm> 05 :H monSm Ham £0385: comm .m:0HHm5Hom :onmonwo: 9: H0 mo5Hm> Uozomno 5:5 couooaxm I .HV WAmHHwH 28 The correlation coefficients all closely approach unity. The regres- sion coefficient gradually changes from plane to plane, consequently the angle (6) also changes. Theta, Q , is the angle between the re- gression line and the axis upon which seed weight is plotted. The greater the seed numberthe greater the slope. Only one regression equation deviated significantly from the expected value, zero, of be. The correlations within each plane were all above 0. 88, with the majority above 0. 93. By plotting the data for 1966 and for the selec- ted population in 1967, across levels of seed number, a surface is formed. Figure 8 provides a graphic representation showing rela- tionships of the position of the progeny on a certain plane in 1966 and their position in 1967. A quantum as used in Figure 8 represents 400 seeds per . 9 meter. The change in quanta represents the increase or decrease in seed number between years for the population. Figure 9 provides the same type of graph presented in Figure 8, but represents the position of the progeny on a certain plane in 1966 and their position in 1967 when paired at random. 29 A Y = 8.2713 - 0.00214X r = -.7478** xy Quantum change o l -6 L l 1 l 1 1 1 l 1 1 2600 3000 3400 3800 4200 4600 5000 5400 5800 6200 6600 Seed number per . 9 meter FIGURE 8. -- The regression of quantum change between the data of 1966 and that of 1967 for seed number per . 9 meter for 136 lines of wintgr barley, with 95% confidence bands. + .. +4 +3 - +2 A Y = 9. 9864 - O. 00256X r = -.7762*>!< xy Quantum change .1. r _7 l l l 4 J l 1 I 1 1 2600 3000 3400 3800 4200 4600 5000 5400 5800 6200 6600 Seed number per . 9 meter FIGURE 9. -- The regression of quantum change between the data of 1966 and 1967 for seed number per .9 meter paired at random, with 95% confidence bands. D ISC U SSION Component Inter-relationships Yield is the economic end product of a barley plant' 3 life cycle, and may be represented as a geometric construct of three com- ponents: X, heads per unit area, Y, seeds per head, and Z, seed weight. These components deve10p in a sequential manner during the growing season. Winter barley is planted in September in Michigan and matures the following July (Figure 10). A AI l I J I I L I L I I I I 4 Sept Oct Nov Dec Jan Feb Mar Apr May June July Aug FIGURE 10. -- A representation of development over the growing season of the winter barley plant in Michigan. Tiller initiation (A) and elongation (A! ) occurs in Septem- ber and October and from March to May. Head formation (B) occurs 30 31 from the first week of May to the last week of May. Fertilization (C) occurs the last week of May and the first week of June, with seed filling occurring the second and third weeks of June. The barley plant may produce several hundred tillers, but the number which are actually produced depends upon tempera- ture, moisture, light, and the genetics of the plant. With adequate water, nitrogen, and light, tillers may be produced to a point at which the plant may be unable to use added nutrients. The inability of the plant to use added nutrients may be genetic or due to a limit- ing endogenous nutritional resource required for their utilization. During April and early May, the tillering process depends upon the soil for water and minerals and the leaves for photosynthetic metab— olites and growth factors. A primordial spike may be developed at the apex of each tiller bud. The size of the primordial spike deter- mines the number of potential spikelets and, ultimately, seeds per head. If the amount of available nutrients were constant, the amount allocated to each tiller would be less with several tillers than with few. Because of the lower nutrient supply per tiller bud, smaller spike primordia would be formed, resulting in smaller heads. This appeared to be the case in 1967. Twice as many tillers were pro- duced in 1967 as in 1966, and smaller heads were formed in 1967 than in 1966. With so many tillers per unit area, the available 32 nutrients per tiller were reduced causing smaller heads. In 1966, tiller-number per area reduced because of winter injury, allowing for a larger nutrient supply per tiller and hence bigger heads. With few tillers, the vascular system of each tiller could possess greater capacity for carrying essential nutrients. These tillers would be larger in diameter, allowing the development of greater vascular transport capacity. In cereals, approximately 70 to 80 per cent of the total carbohydrates in the seed are produced by the flag leaf, stem, and awns. Therefore, with larger heads, the photosynthetic area of these parts would be larger. With the compensatory effect within years due to envi- ronmental effects, it is understandable that reduced relationships might exist between seasons for yield components. The environment in 1966 produced optimum conditions for that particular set of com- ponents for yield. Tiller number was extremely low, allowing the plant to over-extend Y, seed number/head, and with an unlimited supply of nutrients at seed filling time, yield per plant was high. In 1967, component X was extremely over-extended and a resource limit resulted, induced by within plant competition and the environ- ment, causing small heads and small seeds. 33 Selection for increased head production and increased seed weight was successful, but their expression was overshadowed by component compensation in one or the other seasons. The ge- netic potential of the selected population to produce big seeds was demonstrated in 1966, while the genetic potential to produce many heads was eXpressed in 1967. Although the selection advance for yield was apparently negative, the plants still possess the genetic potential of producing high yields if compensatory interactions were removed. In order to take advantage of the plant' s potential to produce many tillers, large heads and heavy seeds, some solutions are suggested. An optimum number of tillers could be determined at which varieties can be grown to insure high yields. For example, in the p0pu1ation grown in 1967: if in the Spring the plant p0pu1ation could have been thinned to 70% of the original stand, large heads and large seeds might have been formed as was experienced in 1966. The question arises as to how the environment could be bypassed to provide artificial man-made compensation. A solution might be the application of high levels of fertility to produce many tillers, big heads and/ or big seeds. Irrigation coupled with high fertility would provide an opportunity to select for varietal response to these factors and, presumably to improve yield. Also, selection of individuals 34 which may possess high levels of photosynthetic efficiency would, along with adequate water and fertility, provide high nutritional levels preventing metabolic limits from restricting component de- velopment in barley. These solutions may sound simple but are really complex in nature. Only a better understanding of the genetic interrelationship of the factors affecting yield and genetic-environment relationship will provide, hopefully, better guidelines for making lasting selection gains. Universal Surface Figure 7 was developed in order to visualize more clearly the relationship of seed weight with yield. The two-dimensional re- lationship of seed weight with yield produced a correlation of . 3162, but by stratifying according to the third dimension, seed number, a highly positive association, was seen at once within each seed— number plane. If the seed-number planes of Figure 7 are arranged in a three-dimensional spatial configuration, each plane stacked upon another, from low seed number to high seed number planes, some- what like the spacing of individual slats of a venetian shade, then the data will form a surface in three-dimensional space. The surface is a continuous belt extending from a region of low seed number to a 35 region of high seed number, marked along the way by strains occu- pying co-ordinate points determined by their seed number and seed weight values. The surface is sloping throughout and seemingly up— lifted at a middle region, reflecting the somewhat better performance of the adaptive intermediates. Mathematically and biologically only one surface exists with this relationship. Yield is the product of seed number and seed weight. The deltas represent the selected population grown in 1966, the dots represent the lines which were not selected, and the circles represent the response of the selected p0pu1ation to the environment of 1967. This method of plotting data provides a way of observing the effect of the environment upon yield components and yield. The swarm of points for barley will expand and move it in the reference Space--on the universal surface-- depending upon the reSponse of the components to the environment. The dimensions and position of the barley swarm upon the surface will also change due to selection pressure placed upon the population by man. If genetic changes are made in seed weight or seed number which alter the boundaries, the swarm will assume a new position laterally or linearly on the surface. Theoretically, if the surface were extended to infinity for seed number, seed weight would approach zero because of the negative association of seed number and seed weight, hence yield 36 would approach zero. At the other extreme, when seed number ap- proaches zero, seed weight would become large, but yield would again approach zero. The array of points which form the surface could be thought of as part of the evolutionary surface for barley. If one as- sumes that Hordeum vulgare has arisen from the inter-crossing of Hordeum agriocrithon, a Six-row barley, and Hordeum spontaneum, a two-row barley, it would appear that directional selection by the environment and by man has been toward larger seeded and higher yielding varieties. The wild Species of barley have small seeds, which are produced in great numbers. Man has selected for heavy seeds and high yields, but due to the negative association of seed number with seed weight, a compensatory genetic shift has occurred toward fewer seeds. In a compensatory situation, the genotypes most likely to produce heavier seeds are those that produce on the average fewer tillers. Consequently, by deliberately selecting for heavier seeds, man has been selecting for fewer tillers and hence fewer seeds. If larger values of seed number were added as planes, they would contain the wild Species of barley such as Hordeum A jubatum, bulbosum, pusillum and murinum, to name a few. By 37 adding smaller seed numbered planes, the two-row barleys can be placed on the surface. This "universal" surface is the only possible one upon which all species of the seed cr0ps can be placed for this particular relationship. Each seed crop will produce itS own "swarm" on the surface. When the environment affects any one of the components it is reflected by movement of the points on the surface. Where upon the universal surface Should selection be practiced to make the most fruitful progress? If the phenotypic cor- relations between seasons are highly positive, selection should be practiced on the planes which have high seed number and large seeds, resulting in high yield. In this case, the lines would be ex- pected to repeat their relative performance year after year. On the other hand, if the environment overrides the genetic expression of the lines through inducing component compensation, then genetic gain by selection strictly on the component basis is more difficult. The inter-annual correlation is a measure of repeatability of performance between years and in the present case this value for seed number was quite low (. 2023). A method--a graphical plot—-is Shown in Figure 8 which demonstrates the inter-year relationship of seed number. The seed number scale for 1966 is marked on the horizontal axis. On the vertical axis is scaled the number of 38 "quanta" by which a line changed in 1967 (a "quantum, " in the sense used, refers to a 400-Seed shift, either in a plus or minus direction, from its seed number in 1966). For example, strain A produced 3, 000 seeds per . 9 meter in 1966. If in 1967, strain A produced 3, 400 seeds per . 9 meter, it would have changed upward by one "quantum. " AS may be observed in Figure 8, there is a marked ten- dency for lines of low seed number in 1966 to produce more seeds in 1967, and for lines of high seed number in 1966 to produce fewer seeds in 1967. The regression of "quanta" Shift in 1967 upon seed number position in 1966 was negative and highly significant. There were two possible interpretations of this result. In the first, it could be postulated that the genotypes are reSponding differentially to an environmental factor for which the two years are markedly distinct. This factor must cause the low seed-numbered line in 1966 to produce a high number of seeds in 1967, and vice versa. Genotype-environmental interactions of this type are rare in plant biology. In the present case, no Single environmental factor capable of affecting the observed responses is apparent. The lines of barley studied were previously unselected for any differential responses, and their very number is so large that numerous failures of differential reSponse would have been expected to show up. None 39 did. Furthermore, if an inversion of performance for 1966 and 1967 had occurred, the inter—annual correlation Should have been nega- tive. It was, in fact, Slightly positive. A second interpretation may also be given. If mostly random processes are Operating between years rather than directed (genetic) processes, then the data will conform to a graph like Fig- ure 8. Figure, 9 was obtained by taking the value of each line in 1966 and pairing it at random with a value from the 1967 data array, and plotting in the same manner aS was used in Figure 8. On the basis of probability alone, it iS obvious that a line on the low end of the 1966 data array would—-if random processes predominate--end up in 1967 at a point higher on the scale. And, Similarly, a line high in 1966 would end up lower in 1967. The second interpretation is supported by the low inter— annual correlation and, without other evidence supporting a Specific genotype-environment interaction, suggests an overriding effect on seed number genes of random varieties imposed by the environment. SUMMARY The best-planned selection procedures can produce some unusual and undesirable results. Even so, logical interpretations must exist for these responses. Negative associations between yield components exist due primarily to inter-plant and intra-plant competition which has been termed "component compensation” by Adams (1). The low inter-annual relationships of yield components, even thoughlarger in magnitude than for yield, are due to the over- riding effects of the environment upon the genetic systems, and to extremes in component compensation within the two environments. A pictorial representation of the evolution of barley was presented, based upon yield and its components. The surface that is formed can be called "universal" Since only one surface is pos- sible for all seed crop plants. The effect of the environment on the components can also be represented upon this surface. Evidence was presented that optima exist for yield com— ponents and their inter-relationship in order that maximum yields can be reached. Selections in one environment for these optima do 40 41 not mean necessarily that under another environment the same pro- duction performance will be obtained. Adams, LITERATURE CITED M. W. 1967. Basis of yield component compensation in crop plants with Special reference to the field bean, Phaseolus vulgariS. Crop Science 7:505-510. Archibong, D. 1963. On the correlation among yield compo- Bal, B. Dewey, Donald, Duarte, nentS in the navy bean, as influenced by spacing and interstrain competition. M. S. Thesis, Michigan State University. 8., C. A. Suneson, and R. T. Ramage. 1959. Genetic shift during thirty generations of natural selection in barley. Agronomy Jour. 51:555-557. D. R. , and K. H. Lu. 1959. A correlation and path coefficient analysis of components of crested wheatgrass production. Agronomy Jour. 51:515-518. C. M. 1962. In search of yield. Jour. of Aust. Inst. of Ag. Science 28:171-178. R. A. 1966. ReSponses of yield and yield components from recurrent selection practiced in a bean hybrid pop— ulation at three locations in North and South America. Ph.D. Thesis. Michigan State University. Grafius, J. E. 1956. Components of yield in oats: A geomet- rical interpretation. Agronomy Jour. 48:419-423. 1965. A geometry of plant breeding. Michigan State University. Agric. Exp. Station. Res. Bull. No. 7. Hutchinson, J. B. 1940. The application of genetics to plant breeding. I. The genetic interpretation of plant breed- ing problems. Jour. of Genetics 40:271-282. 42 10. 11. 12. 13. 14. 15. 16. 17. 18. 43 Johnson, H. W., H. F. Robinson, and R. E. Comstock. 1955. Genotypic and phenotypic correlations in soybeans and their implications in selection. Agronomy Jour. 47: 477-483. Kambal, A. E., and O. J. Webster. 1966. Manifestations of hybrid vigor in grain sorghum and the relations among the components of yield, weight per bushel and height. Crop Science 6:513-515. Leng, E. R. 1963. Component analysis in inheritance studies of grain yield in maize. Crop Science 3:187—190. Leopold, A. C. 1964. Plant Growth and Development. McGraw- Hill Company. New York. 466pp. Olsson, G. 1960. Some relations between number of seeds per pod, seed Sixe, and oil content and effect of selections of these characters in Brassica and Sinapis. Hereditas 46: 29-70. Steel, R. G. D., and J. H. Torrie. 1960. Principles and Pro- cedures of Statistics. McGraw-Hill Book Company, Inc. New York. 481pp. Whitehouse, R. N. H., J. B. Thompson, and M. A. M. DoValle Ribeiro. 1958. Studies on the breeding of self-pollinating cereals. 2. The use of a diallel cross analysis in yield prediction. Euphytica 7:147-169. Williams, W. 1959. Heterosis and the genetics of complex characters. Nature 184:527-530. Van Keuren, R. W., and C. L. Canode. 1963. Effects of spring and fall plantings on seed production of several graSS Species. Crop Science 3:122-124. HICHIGRN STRTE UNIV. LIBRQRIES 1|!1111111111111IHI111111111||H|||||||1Hl1| 31293011012915