UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Breeding system, genetic variability, and response to selection in Plectritis (Valerianaceae) Carey, Charles Kenneth 1981

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
UBC_1981_A1 C37.pdf [ 7.6MB ]
Metadata
JSON: 1.0095380.json
JSON-LD: 1.0095380+ld.json
RDF/XML (Pretty): 1.0095380.xml
RDF/JSON: 1.0095380+rdf.json
Turtle: 1.0095380+rdf-turtle.txt
N-Triples: 1.0095380+rdf-ntriples.txt
Original Record: 1.0095380 +original-record.json
Full Text
1.0095380.txt
Citation
1.0095380.ris

Full Text

BREEDING SYSTEM, GENETIC VARIABILITY, AND RESPONSE TO SELECTION IN PLECTRITIS (VALERIANACEAE) by CHARLES KENNETH CAREY B.Sc.,.The University of British Columbia, 1976 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Botany) We accept this thesis as conforming . to the required standard THE UNIVERSITY OF BRITISH COLUMBIA April, 1981 -(c) Charles Kenneth Carey, 1981 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Botany  The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Date July 20, 1981 - ii Abstract Plectritis congesta and P. brachystemon are two very closely related species which grow sympatrlcally, and differ in their: breeding system, some associated morphological.(floral) characters, and isozyme phenotypes. Plectritis  congesta is approximately 70% outcrossed in nature, while P. brachystemon is less than 3% outcrossed in natural populations. Theory would predict that, all other things being equal, the outcrossed species would be more variable genetically than the selfed species. Since selection acts on genetic variability, the two species could be expected to respond differently to it. Six generations of plants of both species were grown under controlled conditions, and measured for a number of characters. Control and treatment (selection for tall and short height, and for early and late anthesis) populations were maintained. Two sets of P. congesta populations were maintained, one outcrossed (approximately 65%) and one selfed (outcrossed approximately 15%); the P^ brachystemon populations were naturally self-pollinating. Selection pressure in the experiment was approximately 90%; 20 of the 200 plants in any population were selected to form the next -generation, on the basis of height or flowering time in the treatment populations, and at random in the control populations. The P^ congesta populations responded to divergent selection for height at anthesis, indicating that genetic variability for this character was present in the populations. The outcrossed lines, PCO, diverged 66% or 148 mm from the control line; the selfed lines, PCS, diverged 78% or 175 mm. There were no significant differences between the outcrossed and selfed P. congesta lines over the course of the experiment. Two estimates of narrow sense heritability - realised heritability (b^) and parent-offspring 2 : regression (h ) - quantified this genetic variability: in PCO b^ = 0.53, - iii 2 2 h = 0.45: in PCS b = 0.58, h = 0.44. There was a decline in the c phenotypic variance for height at anthesis in the 'P_ corigesta lines selected for this character. In contrast, the P. brachystemon populations did not respond to selection for height at anthesis, and appear to have no detectable genetic variability for this character. Both species appear to have significant genetic variability for flowering time, as both responded to divergent selection for this character. The PCO lines diverged 33.5% or 31.8 days from the control line, the PCS lines diverged 28.7% or 27.3 days, and the P. brachystemon lines>"PBS,-diverged 18.5% or 21.5 days. According to the heritability estimates, P. congesta is more variable genetically: in the PCO lines bc =0.77, h2 = 0.60; in PCS b = 0.75, h2 = 0.72; while in PBS b = 0.49, and h2 = 0.42. There c ' c was a decline in the phenotypic variance for flowering time in all three species groups. Of the other measured but unselected characters - number of days to emergence, number of nodes at anthesis, number of primary branches at anthesis, and fruit production - some responded to the selection pressure with divergence, notably those characters which were correlated with the selected characters (for example, number of nodes at anthesis, correlated with flowering time). With others there was no change which could be attributed to the selection procedure. There was no evidence from two qualitative characters - fruit wing phenotype and fruit pubescence pattern phenotype - for any response to selection; dispersion in both characters was not significantly different from that expected to result from random drift. The relatively high increase in aberrant characters in the P. congesta lines compared to the P.  brachystemon lines is probably indicative of inbreeding depression in the normally outcrossed P. congesta. <•,.' - iv It appears that despite the difference in breeding system, the two Plectritis species are able to maintain variability by similar processes (genetic) in some characters, as in flowering time, and by different processes (genetic in P. congesta, phenotypic in P_. brachystemon) in other characters, as in height at anthesis. Thus one quantitative character, height at anthesis, follows the pattern predicted by the breeding system difference, with the outcrossed P. congesta being much more variable genetically than the selfed P_ brachystemon. This agrees with the levels of variability observed by Layton (1980) in electrophoretically detectable isozymes, and observed by Ganders and Maze (unpublished) in metrical fruit characters. The other quantitative character, flowering time, shows considerable genetic variance in the populations of the selfed P^ brachystemon, though less than in the populations of P_ congesta. The maintenance of such relatively high levels of genetic variability in the face of the strong inbreeding pressures which must be present in P_^_ brachystemon populations is certainly adaptive, and probably comes about through occasional outcrossing and multiniche selection for variability among the segregating lines. - V Table of Contents Abstract ii List of Tables x List of Figures xAcknowledgements v Introduction 1 Breeding system and genetic variability 2 Theoretical considerationsExperimental evidence ..... 6 Monogenic traits 7 Quantitative traits 9 Genetic variability and the response to selection ..... 12 Theoretical considerations 1Experimental evidence 4 Breeding system and the response to selection 16 Theoretical considerations 1Experimental evidence 7 Outcrossed taxa 18 Inbreeding taxa ..... 19 Breeding system, genetic variability, and the response to selection in Plectritis 21 Materials and methods 23 Source populationsGrowing conditions 25 Measurements .. 25 - vi Breeding procedure 29 Selection procedure ..... 32 Base population ..... 33 First cycle of selection 3Subsequent cycles of selection 33 Progeny test and outcrossing rates in P_j_ congesta 35 Data treatment and analyses 3Metrical characters 6 Descriptive statistics 3Comparisons between distributions 38 Correlations 3Heritability estimates 38 Variance within populations 39 Other characters 3Results 40 Breeding systems in Plectritis congesta and P. brachystemon ..... 40 Outcrossing rates in the source populations (Mill Hill Pk. , 1977) 40 Outcrossing rates in the experimental populations 40 Characteristics of the base populations 41 Descriptive statistics of the metrical (quantitative) characters 41 Frequencies of qualitative characters 44 Response to selection of the selected characters 44 Height at anthesis 4Means- vii Estimates of variability 57 Variances 5Heritabilities ..... 60 Components of variance 63 Other changes in distribution ..... 64 Days to anthesis (flowering time) 6Means 6Estimates of variability 67 Variances 6Heritabilities 7 Components of variance 70 Other changes in distribution ..... 70 Changes in the unselected characters during the experiment 71 MeansDays to emergence 71 Height at anthesis (in lines selected for flowering time) 71 Nodes at anthesis 74 Primary branches at anthesis 7Flowering time (in lines selected for height at anthesis) 79 Fruit production 7Estimates of variability 84 Variances 8Heritabilities 4 Components of variance 89 Other changes in distribution- viii The effects of selection on correlations among the measured characters 92 The correlation between height at anthesis and flowering time, the characters under selection ..... 92 Other correlations 9Changes in qualitative characters ..... 124 Winged and wingless plant frequencies 124 Pubescence patterns 125 Aberrant charactersComparisons between the internal control populations and the source populations ..... 125 Quantitative characters 127 Plectritis congestaPlectritis brachystemon 127 Correlations 130 Summary of results 2 Discussion ..... 136 The experimental species 13Genetic variability and the response to selection ..... 136 Direct responses 13Plectritis congesta outcrossed versus P. congesta selfed 136 Plectritis congesta versus P^ brachystemon ..... 137 Height at anthesis 13Flowering time ..... 139 Confounding phenomena 140 Indirect responses to selection ..... 142 - ix Unselected characters 142 Other selection studies 4 Independent estimates of genetic variability in Plectritis 146 The effects of breeding system on the population genetic structure of Plectritis 149 Plectritis brachystemon 150 Plectritis congesta 6 Literature cited 163 Appendix 1: Between family/within family variance ratios in experimental populations 169 Appendix 2: Coefficients of variation, unselected characters ..... 172 - X List of Tables Table Title Page I Growing conditions 26 II Estimates of outcrossing rates in the experimental 42 populations III Measured characters, base populations 43 IV Realised heritability, calculated using the method 61 of Hill (1972) V Heritability, from parent-offspring regressions 62 VI Heritabilities from parent-offspring regressions, 90 unselected characters VII Frequencies of aberrant individuals 126 VIII Measured characters: G,. source populations 128 compared with G^ control populations IX Correlations in the G,. source populations 131 - xi List of Figures Figure Title Page 1. Source Plectritis populations: Mill Hill Pk., 24 June 1977 2 Morphology of Plectritis 28 3 Plectritis fruit wing phenotypes 30 4 Fruit pubescence phenotypes in winged Plectritis 31 congesta fruits 5 Experimental populations maintained through 5 34 generations of selection, G^ to G,. 6 An example of data transformation procedure used 37 on metrical characters 7 Frequency of various pubescence types in the 46 experimental populations a. PCO populations, type 0 b. PCO populations, type 1 c. PCO populations, type 2 8 Frequency of various pubescence types in the 48 experimental populations a. PCO populations, type 3 b. PCO populations, type 4 c. PCO populations, type 5 9 Frequency of various pubescence types in the 50 experimental populations a. PCS populations, type 0 b. PCS populations, type 1 c. PCS populations, type 2 10 Frequency of various pubescence types in the 52 experimental populations a. PCS populations, type 3 b. PCS populations, type 4 c. PCS populations, type 5 11 Frequency of wingless fruited plants in the 54 experimental populations a. PCO populations b. PCS populations 12 Mean height at anthesis in populations selected for height at anthesis. 56 - xii Figure Title Page 13 Coefficients of variation for height at anthesis 59 in populations selected for height at anthesis 14 Mean number of days to anthesis (flowering time) 66 in populations selected for flowering time 15 Coefficients of variation for, days to anthesis in 69 populations selected for flowering time 16 Mean number of days to emergence in various 73 populations a. Populations selected for flowering time b. Populations selected for height at anthesis 17 a. Mean height at anthesis in populations 76 selected for flowering time b. Mean number of days to anthesis in populations selected for height at anthesis 18 Mean number of nodes at anthesis in various 78 populations a. Populations selected for flowering time b. Populations selected for height at anthesis 19 Mean number of primary branches at anthesis 81 in various populations a. Populations selected for flowering time b. Populations selected for height at anthesis 20 Mean fruit production in various populations 83 a. Populations selected for flowering time b. Populations selected for height at anthesis 21 Coefficients of variation for number of days to 86 emergence in the experimental populations a. PCO b. PCS c. 'PBS 22 Coefficients of variation for number of nodes at 88 anthesis in the experimental populations a. PCO b. PCS c. PBS 23 Correlations between height at anthesis and flowering 94 time a. PCO populations b. PCS populations c. PBS populations - xiii Figure Title Page 24 Correlations between height at anthesis and 96 number of primary branches at anthesis a. PCO populations b. PCS populations c. PBS populations 25 Correlations between number of nodes at anthesis 98 and number of primary branches at anthesis a. PCO populations b. PCS populations c. PBS populations 26 Correlations between height at anthesis and fruit 100 production a. PCO populations b. PCS populations c. PBS populations 27 Correlations between number of days to emergence 103 and height at anthesis a. PCO populations b. PCS populations c. PBS populations 28 Correlations between number of days to emergence 105 and number of nodes at anthesis a. PCO populations b. PCS populations c. PBS populations 29 Correlations between number of nodes at anthesis 107 and fruit production a. PCO populations b. PCS populations c. PBS populations 30 Correlations between number of primary branches at 109 anthesis and flowering time a. PCO populations b. PCS populations c. PBS populations 31 Correlations between number of days to emergence 111 and number of primary branches at anthesis a. PCO populations b. PCS populations c. PBS populations 32 Correlations between number of days to emergence 113 and fruit production a. PCO populations b. PCS populations c. PBS populations Title Correlations between flowering time and fruit production a. PCO populations b. PCS populations c. PBS populations Correlations between days to emergence and flowering time a. PCO populations b. PCS populations c. PBS populations Correlations between number of nodes at anthesis and flowering time a. PCO populations b. PCS populations c. PBS populations Correlations between number of primary branches at anthesis and fruit production a. PCO populations b. PCS populations c. PBS populations Correlations between height at anthesis and number of nodes at anthesis a. PCO populations b. PCS populations c. PBS populations - XV Acknowledgements 1 am deeply grateful to my supervisor, Dr. Fred Ganders, for introducing me to Plectritis, a genus of truly endless interest. His knowledge and advice, interest and encouragement, and his willingness to discuss any aspect of the project, large or small, have added immeasurably to the quality of the result. I thank the members of my thesis committee, Dr. Tony Griffiths and Dr. Gary Bradfield, for their patience and advice. Their careful and constructive criticism throughout have helped me to direct the research to best advantage, and to clarify and round out its presentation. I am grateful to the members of the Department of .^Botany at U.B.C., whose combined excellence has instructed and inspired me. I thank the Natural Sciences and Engineering Research Council of Canada for financial support in the form of Post-graduate Scholarships and through grants to Dr. Ganders. This work would not have been the same without the patience and support of my wife, Alice, who kept me cheerful and optimistic through the more difficult moments, and shared my happiness when.things went well. - 1 Breeding system, genetic variability,, and response to selection in Plectritis (Valerianaceae). Introduction Genetic variability in various organisms has been a major object of study since the rediscovery of Mendel's laws of inheritance and their synthesis with Darwin's theory of evolution by natural selection. The kind of questions to which answers are sought include questions about the extent to which variability is present in individuals, populations,.or taxa; about the ways in which variability is generated, maintained, or lost; and about the effect which variability may. have on the fitness or survival of an individual, population, or taxon. This study deals with the following questions about.genetic variability. First, how has the amount of genetic variability underlying the expression of quantitatively inherited characters in a population of plants been affected by the breeding system of that population? Second, if the breeding system has affected the amount or nature of genetic variability, can this effect be detected by observing the response to selection in two taxa between which the major biological difference is in their breeding system? Finally, how does the genetic variability present in quantitatively inherited characters compare to that of other (monogenic) characters, both within and between taxa with different breeding systems? To combine the questions in more concrete terms., is a population of inbreeding plants more or less variable genetically than a population of "otherwise identical" outbreeding plants with respect to quantitatively inherited characters, does it respond more or less quickly to selection pressures on these - 2 characters, and does the genetic variability, .in multigenic characters follow the same patterns as variability in monogenic characters? Both theoretical predictions and experimental evidence have provided some answers to all three of these questions.. Differences in genetic variability between plant species of.varied.breeding systems have been studied by population biologists; in most cases the characters studied have been monogenic rather than quantitatively inherited. Differences in response to selection are of vital interest to plant breeders;, in most of their studies the goals have not included assessing the effects of the breeding systems of the plants involved, or more particularly comparing species of differing breeding system. There is a place (not to say a gap) to be filled by selection studies of quantitatively inherited characters in natural populations; it is to be.hoped that such studies will add to what is known about breeding systems, genetic.variability, response to selection, and the interactions among the three. Breeding system and genetic variability Theoretical considerations A first step in answering questions about genetic variability is necessarily the definition of some of the terms. Genetic.variability is a broad term which can lead to some confusion if loosely.applied. It encompasses a number of parameters in any.population, including-the numbers and frequencies of alleles at various gene loci, the numbers and frequencies of various genotypes, and the distribution of genotypic components,of the total phenotypic variance for various characters. Although these parameters are not independent, they can be divided into two groups on the basis of - 3 whether or not they are subject -to selection directly. The first group I will call the potential components of .the genetic variability in a population. In the simplest genetic sense,.a gene locus is variable (polymorphic) if more than one allele occurs at it. For any population the number and frequencies of alleles at various loci can theoretically be observed. These alleles will be combined to form the genotype of an individual. The number and frequencies of various genotypes in any population can also be observed-theoretically. Various indices can be derived from these types of observations; some examples which are commonly used in studies of real populations are the number or percentage of loci which are observed to be polymorphic, and the number of alleles per polymorphic locus. In addition,.the frequencies of genotypes expected in a population behaving according to particular assumptions can.be .calculated. Most often the assumptions are those leading to Hardy-Weinberg ..equilibrium in a theoretical population. The populations under study may not be behaving according to the.assumptions by which the genotype frequencies have been derived. For example, if at a particular locus a population has two alleles, and A^, then theoretically there are three genotypes -possible. An actual population may in fact be missing any one of the three genotypes (or even both'homozygous genotypes), and be less variable in fact than it appears in theory. Many studies of this type of genetic variability calculate genotype frequencies rather than observing them in populations. Since alleles and genotypes are not selected directly but rather through their expression in phenotypes, it should be kept in mind that indices which compare populations on this basis may not be assessing the genetic variability available for selection. This has been a particular problem in studies dealing with isoenzymes; the selective values of - 4 particular isoenzyme phenotypes are for the most part unknown.. I .will refer to-these estimates.of potential.genetic variability .as. estimates of genetic.diversity (a term.which.has a more specific application in some of the literature of population genetics). If the contribution of various genotypes to the genotypic component of the total phenotypic variance for a particular character can be assigned arbitrarily or determined by experiment, then one is dealing with the second group of parameters, which I will call the realised components of the genetic variability. For any character a particular phenotype can be assigned a numerical value, be it a fitness value or an actual measurement (height, weight, etc.), and the distribution of these values in a population will have.a mean and a variance. The genotypic component of this total phenotypic variance is referred to as the genetic variance, and it is upon this variance that selection may act. The effect of breeding system on the potential genetic variability, or genetic diversity, can be predicted in.theory if a number of assumptions are invoked. If two populations are initially identical in all respects, that is, contain equal numbers and frequencies of genotypes, and are otherwise in Hardy-Weinberg equilibrium, then breeding system differences will affect them in the following way. .The frequency of heterozygous individuals at a particular locus, and in consequence the total number of genotypes, will quickly decline in the inbreeding or autogamous population compared to the random mating population. However, the total number and frequencies of alleles, the percentage of loci which are polymorphic, and the number of alleles per. polymorphic, locus will remain the same, as will, the genotype frequencies expected under Hardy-Weinberg, equilibrium. If the equilibrium assumptions are relaxed, and realistic and finite - 5 population'size and selection, are affecting ..the populations, random drift and selection will.reduce..the.total number:of alleles in the selfing population faster than in the outcrossed one,, and.genetic diversity as measured by the percentage of loci polymorphic, the number of alleles per polymorphic locus, and.the expected.frequency of heterozygotes at equilibrium will thus be reduced in the selfer relative to the outcrossed population. The degree of.difference in genetic diversity between the two populations of different breeding system will depend on the difference in their rates of inbreeding, the effective population size, and rates of selection. The effect of the breeding system on the realised genetic.variability or genetic variance can also be predicted in theory, again subject to a number of simplifying assumptions.. If we start with the same two populations as mentioned above, with,identical genotypic structure under Hardy-Weinberg equilibrium, then breeding system differences will have the following effects. If • the environmental component of the .total phenotypic variance in this case is taken to be zero, then the genotypic values are contributing all of the phenotypic variability. With selfing, as heterozygous genotypes are lost from.the population the variance of the phenotypic values will increase relative to a random mating population. Thus inbreeding on its own will increase genetic variance. . If we again relax the equilibrium assumptions by assuming realistic finite populations on which selection is acting, then the theoretical prediction becomes rather problematical. The relatively larger loss of alleles due'to random drift and selection in finite populations of selfers will tend.to reduce the genetic variance. Simultaneously, inbreeding.will increase the genetic variance relative to a random mating population. The - 6 effects on genetic variance of inbreeding and random drift / selection will be in opposite directions, and the net effect cannot be generalized. Lande Q.977) has proposed a model which indicates that if the populations are large Cinfinite), but with selection and mutation acting, then the breeding system will have no effect on the amount of additive genetic variance maintained. To summarize, such theoretical treatments of realistic populations (without extensive simplifying assumptions) as are available should be generalized with caution. Nevertheless, compared to random mating, selfing in finite populations is likely, to lead to a loss of genetic diversity, and whether this is accompanied by a net reduction in genetic variance will depend on factors unrelated to the breeding system, such as population size and selection pressures. Experimental evidence Variability has been studied in populations of many plant species, both natural and domesticated. Discontinuities in variation patterns between taxa form the basis for taxonomic studies; genetic variability is the source of improvement by plant breeders in economically important species; and the interaction of genetic variation and natural selection is the'object of evolutionary studies. Studies specifically isolating breeding system and genetic variability are limited, and comparisons between closely related species differing in breeding system are as yet very few. The experimental evidence relating breeding system and genetic variability can be divided into two groups in which the estimates of genetic variability roughly parallel those outlined earlier as genetic diversity and genetic variance. The first group consists of evidence from monogenic or single locus traits, whose qualitative nature makes assignment of specific phenotypic values, and thus population parameters such as mean value and variance, difficult. In some cases fitness values have been estimated experimentally, but for the most part the experimental evidence is in the form of estimates of genetic diversity such as allele and genotype frequencies, and indices such as the percentage of loci polymorphic and number of alleles per polymorphic locus. The second group consists of evidence from characters which are known or assumed to be multigenic, or quantitatively inherited. For these characters the estimates are approximately estimates of genetic variance, although as will be seen they may require fairly elaborate experimental designs to eliminate environmental components of the variance and obtain more exact estimates of the genetic variance. Monogenic traits Because of the ease with which relatively large numbers of characters can be studied, the bulk, of studies allowing a comparison of breeding system and genetic variability in monogenic characters involve electrophor-etically detectable enzyme variation. Hamrick et al. (1979) have reviewed the isozyme data recently, relating levels of electrophoretic variation Cgenetic diversity) and life history characteristics in a large number of plant species. They report that for three indices of diversity -percentage of loci polymorphic, number of alleles per polymorphic locus, and a polymorphic index Cfrequency of heterozygotes expected under Hardy- Weinberg equilibrium) - 36 primarily outcrossed species showed more diversity than 33 primarily selfed species. For the most part these data combine - 8 relatively unrelated taxa in each breeding system group, but there are some comparisons which may be noted. For example, groups of congeneric species in which levels of genetic diversity have been confirmed to be higher in the outcrossed species than in the selfed species include: Limnanthes  alba (outcrossed) and L. floccosa (selfed) CBrown and Jain," 1979); Clarkia  rubicund a, C. amoena (outcrossed) , and C_. f rariciscana (selfed) (Gottlieb, 19 73) ; Gaura suffulta (outcrossed) and Gy triarigulata (selfed) (Levin, 1975) ; Phlox drumiiiondii, P_.^ roemarjana (outcrossed) , and P_j_ cuspidata (selfed) (Le-yin, 1978); Leavenworthia alabarnica, L. crassa, L. stylosa (outcrossed) , L. uriiflora, \'h. exjgua, "and L. torulosa (selfed) (Solbrig, 1972) ; Lycopersicon pimpinelljfolium (outcrossed) (Rick et al., 1977) and L. parviflorum (selfed) (Rick and Fobes, 1975). At the conspecific level, populations of Lycopersicon pimpinelljfolium vary in breeding system from relatively outcrossed to relatively selfed, and the more highly outcrossed populations have higher levels of genetic diversity (Rick et al., 1977). •Ln Oenothera, Ellstrand and Levin (1980) found that there was a significant difference in gene diversity between the more diverse, outcrossed 0. grandis and the selfed 0. mexicana. A third species, 0. 1aciniata, which is highly inbred but a permanent translocation heterozygote, has relatively high diversity, not significantly different from 0. grandis. Of particular interest is the study of the two closely related species Plectritis congesta and P_j_ brachystemon, which indicated that the outcrossed P_^_ congesta has much higher levels of genetic diversity as indicated by isozyme data than the selfed P. brachystemon (Layton, 1980). There are fewer studies of other types of monogenic characters, particularly in closely related taxa of different breeding systems. Jain and Marshall (1967) found that Avena fatua, which has a slighfly higher - 9 outcrossing rate than its relative A. bafbata, was polymorphic at three morphologically expressed loci, while A. barbata was monomorphic at the same loci. However, while there is a slight difference in breeding system between them, both Avena species are relatively highly selfed. In outcrossed Plectritis congesta, populations polymorphic for a monogenic fruit wing character are much more common C30 populations polymorphic of 32 studied) than in selfed P. brachystemon C3 of 11 populations) (Ganders et al., 1977b; Carey and Ganders, 1980). In Lycopersicon pimpinellifolium, Rick et al. CL977) found that two monogenic morphological characters showed their highest levels of polymorphism in populations with the highest outcrossing rates and were monomorphic in populations which were relatively highly selfed. All of the evidence from monogenic traits suggests that selfing reduces the genetic diversity relative to a comparable outcrossed population (most of this evidence comes from isozyme loci, which may or may not be representative of all loci). This is in agreement with the theoretical predictions. Quantitative traits Data on the genetic variability of multigenic or quantitative characters in selfed or outcrossed species are less straightforward than those on monogenic traits. First, many of these characters are assumed, rather than known, to be multigenic. They might more accurately be termed metrical or continuously distributed traits. Second, these characters are invariably confounded by an environmental component which may be difficult to remove except in large scale, carefully designed experiments. Third, estimates of genetic variability from these characters are further removed from the genome than those from monogenic characters; that is, allele frequencies, r 10 polymorphic loci, and rates of heterozygosity are rarely discernible directly from measurements of these characters unless extensive genetic study has been done. For these reasons a review which brings together data on various taxa from many studies, such as that of Hamrick et al.• C1979) for isozyme data, is not feasible, as there is little assurance that the measurements of genetic variability from a wide variety of metrical characters under different experimental designs can validly be compared. Finally, as with monogenic traits, there have been only a limited number of studies of closely related taxa with.'dif ferent-breeding.- systems. I will examine a few of these at this point. In most cases the variability that has been measured has not been partitioned into genetic and environmental components, and the actual measurements are of phenotypic variance, from which estimates of the genetic variability have been extrapolated, I will refer to it as genetic variability, although one may hope that genetic variance is being estimated approximately. In some cases no significant difference between taxa with different breeding systems was found. Brown and Jain (1979) studied 15 quantitative characters in Limrianthes alba (outcrossed) and L. floccosa (selfed) and concluded that both the total amount of genetic variability and the partitioning of the variability within the taxa (that is, within and between populations of the taxa) were not significantly different between them. This is in contrast to the situation found in the isozymes in these two species, described above (p.9). Studying the Lupinus nanus group, which included four L. nanus subspecies and two other species with outcrossing rates between 0 and 100%, Harding et al. (1974) found no correlation between the outcrossing rate and the amount of genetic variability (in this case estimated genetic variance) for six quantitative characters. In three - 11 grass species j Fe.sttj.ca micf ostachys Ccompletely • self ed) , Averia f atua (highly selfed) , and Lolium mtiltiflorum (.outcrossed) , Kannenberg and Allard (1967) found no difference among the three species in genetic variance for three quantitative characters. In some cases the more highly outcrossed species appears to be more variable. In Lycopersicon pimpinellifolium Rick et al. (1977) noted that several quantitative characters showed maximum variability in geographic areas which coincided with maximum isozyme diversity and maximum outcrossing rate, and minimum variability in areas with minimum outcrossing rates; these observations were unfortunately not based on actual measurements of the characters concerned. Strid.'(19 70) noted that populations of Nigella  degenii. an outcrossed species, showed more variability in flowering time and percentage of good pollen than N. dOerfleri, a selfed species, but again no measurements of the characters have been reported. In Stephanomeria  exjgua ssp. coronaria, an outcrossed species, and its obligately selfed derivative S. malheurensis, Gottlieb (1977) measured 33 quantitative traits and found 90% of them to be more variable in the outcrosser. These data must, however, be viewed keeping in mind that malheurensis is a recently evolved taxon whose population genetic structure is probably still extensively subject to phenomena such as the founder effect, which also limit genetic variability. In Ayena fatua and A. b arb at a J ain and Marshall (1967) studied 3 quantitative characters and found more genetic variance in the "outcrossed" A. fatua than in A. barbata, but as noted above, both species are actually highly selfed, and the comparison is less useful . in this instance. Rogers(1971) found in Papayer rhoeas (outcrossed), P. dubium, and P. lecoqli (selfed) that the outcrossed species showed more within population variability than between population variability for two quantitative characters, and that the selfed species showed the reverse, that is less - 12 within population variability and more between populations. It is unclear whether the overall variability was greater within the outcrossed than within the selfed species. Baker (1953) studied a similar situation in Armeria, looking at several quantitative characters in A. maritima ssp. maritima (self incompatible) and'A: maritima ssp. californica (self compatible but more or less outcrossed) where he found that the former, more highly outcrossed .taxon was more variable within than between populations, and the latter, more highly selfed taxon more variable between than within populations. In addition, he found that A..maritima ssp. maritima showed more genetic variability in toto than A. maritima ssp. californica. In Plectritis congesta and P. brachystemon, a study of a number of morphometric fruit characters by Ganders and Maze (unpublished) showed the outcrossed P. congesta to be more variable than the selfed P. brachystemon. Finally there are cases where the selfed taxon is apparently more variable than its outcrossed relative. Hillel et al. (1973) studied 36 quantitative characters in Triticum speltoides, (outcrossed) and T. longissimum (selfed) and concluded that the selfed species showed greater genetic variability within families and within and between populations than the outcrossed species. The overall impression from the evidence of genetic variability in quantitative characters is that the effect of a particular breeding system is in fact difficult to predict. Since in most of the cases presented here the estimate of genetic variance is not an accurate one, it is possible that differences in the environmental component of the total variance may be affecting the comparison. Genetic variability and the response to directional selection Theoretical considerations - 13 Can the response to selection for a particular character be predicted by an independent estimate of genetic variability? Again, as with the first question, it is necessary at the outset to define some terms. Selection is any process in a population which divides those individuals surviving to reproduce from those which do not survive to reproduce in Some way other than randomly. Directional selection for a character is the differential survival of individuals whose phenotypic expression for that character is different from the population mean in one direction. Response to selection is fundamentally any change in the genetic structure of a population which can be attributed to selection pressure. Selection acts on phenotypes, but the direct response to selection, if any, takes place in the genotypes in a population. It is because genotypic changes in a population are reflected in the total population phenotype that selection responses can be observed, and in the case of long term selection, that the selection process can continue. These responses are usually observed as changes in population mean and variance for those characters which are quantitatively inherited, or as changes in allele or genotype frequencies for characters whose genotype is directly observable in the phenotype of an individual. Fisher's fundamental theorem of natural selection states that the response to selection for a character in a population is directly proportional to the genetic variance for that character in the population. Of interest at this point is the prediction of response to selection for a character on the basis of estimates of genetic variability in independent characters. This is not possible in theory unless some assumptions are made about the extent to which particular characters are representative of the genome as a whole. We could assume that characters for which we have estimates of genetic V variability are completely representative of characters which might be - 14 subject to selection. The success- of the prediction in this case will depend once again on whether our estimate is of genetic diversity or of genetic variance. ' As outlined earlier, estimates of genetic diversity do not necessarily measure variability which is available for selection to act upon, whereas genetic variance estimates do. > Even genetic variance estimates neglect the environmental component, which may be so large a component of the total phenotypic variability upon which selection acts directly as to confound any prediction. A rigorous theoretical treatment of the question has not yet been produced. Nevertheless, all other.<things Being equal (rate .of selection, population size, breeding system), a population which has more genetic diversity would be expected in general to also have greater genetic variance, and to respond faster to selection than one with less genetic diversity. Similarly, a population which has greater genetic variance in some characters could be predicted to have greater genetic variance in otherfcharacters, which when selected would show a greater response. Experimental evidence There is ample evidence for response to artificial selection in a broad sense in many organisms. The domestication of hundreds of plant and animal species for human purposes has in almost every case involved changes in what are now the domestic taxa, sometimes to the point where feral relatives are unknown or so different from their domesticated derivatives as to make the origins of the latter extremely difficult to trace. Unfortunately, the selection involved in these, often prehistoric, domestications has not been documented in a manner to make them.useful in - 15 this study. Artificial selection, that is selection under the control of humans, should be considered as two separate kinds of process. The first is an attempt to duplicate the processes of natural selection, usually in order to understand what is going on in nature, and involves selecting individuals by their phenotype in one generation (mass selection), breeding them in some natural mating system, and forming the subsequent generation from their progeny. The second type of artificial selection is economically motivated and aimed at producing particular superior genotypes or populations of genotypes in crop plants and animals in the fastest and most economical way. Individual or mass selection is only one of many selection schemes which may be used, in combination with careful breeding programs, to isolate that portion of the available genetic variability which represents the superior genotype(s). Special breeding techniques (diallel crosses, inbreeding, sib matings, back crosses, etc.) and selection regimes (recurrent selection, progeny testing, etc.) are usually employed to increase selectable variation and speed up selection for economic reasons -artificial selection which approximates natural selection is often a relatively slower process. The two types of artificial selection are not mutually exclusive, that is valuable information about natural selection can be obtained from plant and animal breeding studies, and relatively natural selection schemes may produce economically valuable results in some cases. To help answer the questions about selection, genetic variability, and breeding system as posed, the evidence we require should ideally come from selection experiments where there is an independent estimate of genetic variability for the organism, where the organism studied has a known breeding system, and where the initial populations under selection have not been - 16 bred to change their natural levels of yarlability, for example by Inbreeding an outcrossed species or crossing a selfed species. In addition, for the purposes of comparison to the study undertaken here, the selection method should be comparable (mass selection). I have been able to find only one experiment where selection for a character has been performed on two taxa for which estimates of genetic diversity or variance from an independent source are available. Jain and Marshall (1970) selected for two extremes of heading date and seed size in Avena fatua and A. barbata. These are two species for which measurements of genetic diversity from isozymes (Marshall and Jain, 1969) and genetic variance in other characters (Jain and Marshall, 1967) are available, and in both cases A. fatua has been shown to be the more variable species. A.  fatua responded better to selection for both characters, and this result agrees with the prediction based on the independent estimate of genetic variability. Selection response is sufficiently ubiquitous that careful comparisons such as this are the only ones of real value in answering this question. Breeding system and the response to selection Theoretical considerations Can the response to selection for a particular character be predicted by the breeding system of the organism being selected? Breeding system can only affect the response to selection through its effect on the genetic variance present in the population. As we have seen in the discussion outlined earlier, the effect of breeding system on genetic variance is difficult to predict in theory, but the evidence indicates that selfed taxa contain less genetic diversity than outcrossed taxa. The evidence for differences in genetic variance between populations with different breeding systems is equivocal. " Inbreeding in an infinitely large population will increase the genetic variance relative to an outcrossed population, and response to directional selection will thus theoretically be faster initially. However, genetic variance will theoretically be depleted more quickly in the inbred population, and selection response will cease earlier ( the selection limit having been reached). If in a finite population the selection pressure is heavy enough, it is possible that the selection limit may be farther from the mean of the original population in an outcrossed population than in an inbred one, as loss of alleles through homozygosis and random drift in the inbreeder in early generations may prevent selection of the optimal genotype. Experimental evidence The evidence required to distinguish the effect of breeding systems on the response to selection should ideally come from the same type of experiments as outlined earlier under genetic variability and response to selection (p. 15). The available evidence comes mostly from plant breeding studies, and has the attendant shortcomings in this context. Most of the "populations" being selected are not representative of a natural population In terms of levels of genetic variability; that is, even species which have an outcrossed breeding system have usually been inbred, often highly inbred, and naturally selfed species may have been outcrossed a number of ways to increase variability. Most well documented evidence of response to selection in plants comes from plant breeding studies which use some regime of selection other than mass selection. And, of course, comparisons between r- 18 closely related taxa of different breeding systems- are scarce, so one is forced to deal with experiments on plants with particular breeding systems as a group. Outcrossed taxa The best example of long term mass selection in an outcrossed taxon is the 70 generation experiment selecting for oil and protein concentration in Zea mays, summarized by Dudley et al. (1974). The four populations selected for extremes in concentration have continued to show a significant response for 70 generations, with the means of the high protein, low protein, high oil, and low oil strains in generation 70 being respectively 215%, 23%, 341%, and 14% of the means of the original population. Zea mays has also been successfully selected for increased proflicacy (24% in 6 cycles of selection) and grain yield (18% in 6 cycles) (Arboleda-Rivera and Compton, 1974), increased yield (20% in 4 cycles) (Gardner, 1961), increased earworm resistance C28% in 10 cycles) (Zuber et al., 1971), increased cold germination (36% in 4 cycles) (McConnell and Gardner, 1979), and increased and decreased ear length (Cortez-Mendoza and Hallauer, 1979). Another outcrossed species which has been the subject of successful mass selection experiments is Medicago sativa. Response has been observed to selection for increased resistance to bacterial wilt (38% in 4 cycles) (Barnes et al.," 1971), increased self-sterility (77% in 2 cycles) and self-fertility (103% in 2 cycles) (Busbice et al., 1975), (73% in 3 cycles) (Villegas et al./ 1971), resistance to anthracnose (67% in 3 cycles) (Devine et al., 1971), and increased (300% in 2-3 cycles) and decreased (66% in 2^5 cycles) saponin levels (Pedersen et al.j 1973). - 19 In Ipomoea batatas mass selection has produced response in terms of decreased oxidation of the root fles-h (Jones, 1972) and changes in a complex composed of a number of economically valuable characters Clones et al., 1976). In Agropyroh deseftorum Schaaf (1968) has successfully selected for extremes of seed size (+7% in one cycle) and increased seed yield. Brassica hjrta (Sinapis alba) has been selected for extremes of oil content (+16%, -14% in 8 cycles) (Olsson and Andersson, 1963). Extremes of flowering time were successfully selected in Brassica campestris var. brown sarson by directional selection (+0.3%, -10% in 3 cycles) (Murty et al., 1972). Beta vulgaris has been selected for extremes of dry matter content in the root (+35%, -40% in 13 cycles) (Josefsson, 1963). Limnanthes  alba has responded to two cycles of selection for flowering time (+13%, ?.11%) (Jain, 1979). This is one situation where a closely related inbreeding species (L. floccosa) exists and has been studied for levels of genetic variability, but unfortunately no selection experiments have been done on it yet. Inbreeding taxa There are a number of studies of response to mass selection in inbreeding taxa. In Avena sativa, response has been observed to selection for increased panicle weight (15% in 2 cycles) (Chandhanamutta and Frey, 1973), reduced plant height (2 inches in 4 cycles) (Romero and Frey, 1966) , and to one cycle of divergent selection for heading date (+22%), plant height (+5%, -4%), grain yield (+9%), width; of seed (+5%, -3%), seed weight (+5%, -3%), and number of spikelets per panicle (+5%, -1%) (Geadelmann and Frey, 1975). In Avena fatua successful divergent selection has produced changes in growth habit (+13%, -31%), flowering time (+13%, -28%), and height t+10%, -15%) in - 20 one cycle (Imam and Allard, 19_65) . As noted earlier, both A. f atua and its more highly se 1 fed relatlve' A, B.arbata have been selected for increased and decreased seed size and heading date (Jain and Marshall, 19 70). Glycine max has been selected for extremes of seed size (+10%, -4% in 3 cycles) and specific gravity of seeds (Fehr and Weber, 1968). Divergent selection for 10 cycles in Sorghum bicolor has changed the mean seed weight (+34%, -18%), plant height (+31%, -26%), and flowering time (+10%, -2%) (Foster et al., 1980). Four cycles of mass selection for increased green weight of leaves in Njcotiana tabacum resulted in an increase of 18% (Matzinger and Wernsman; 1968). Allard et al. (1968) in their review of the genetics of inbreeding species report successful selection for intensity of coat colour in seeds of Phaseolus lunatus, and extremes of seed size in P. lunatus (+ 7.5% in 4 cycles), P. vulgaris (+4.5% in 4 cycles) and barley. In Eleusine, Hilu and deWet (1980) were able to increase germination rates by 20-100% in 4 cycles of selection in three species, E. indica, B. coracana, and E. tristachya. As has been the case with selection experiments in nearly every organism, plant or animal, and for nearly every character studied, enough genetic variability is present in both breeding system groups for a response to selection to occur. One potential problem which should be mentioned is that negative results (in this case, lack of response to artificial selection) may not be reported in - the literature.. Given the investment in time and effort involved in most careful selection experiments, this is probably not in fact a problem, and it appears from the evidence examined here that both outcrossed and selfed taxa show considerable response to selection for a number of characters. It is only from studies such as that with Avena fatua and A. barbata, which comes close to meeting the ideal conditions of relative ly natural levels of genetic variability, natural breeding systems, and - 21 mass selection, that the most useful comparisons may be drawn. Breeding system, genetic variability, and the response to selection in Plectritis The objective of this study was to examine the responses to divergent artificial selection of two natural populations of plants which differ mainly in their breeding system. If differences in selection response were observed, this could reflect differences in the amount and / or organization of the underlying genetic variability. The plants chosen for this study were the two species of Plectritis, P. congesta and P. brachystemon. There are a number of features of the two'species which make them ideal for this purpose. The two species are very closely related. Morey (1962), in the most recent treatment of the genus, considered them subspecies of P_^ congesta. Hitchcock and others have not even given them that rank, considering them one species (Hitchcock and Cronqui^t, 19 73). The species are very difficult to distinguish before they have flowered, since they have nearly identical vegetative habit. Populations of the two grow sympatrically in a number ofvlocations, and even when allopatric they occupy the same type of habitats, that is, thin, edaphically dry substrates on rocky coastal bluffs and headlands, and open slopes and clearings inland, with the same community of associated annual and perennial herbs. It is safe to assume that the large scale selective pressures that they have encountered in terms of habitat have not differed significantly between the two species for many generations. Nevertheless, the populations of the two species which occur in British Columbia are quite distinct, differing in a' number of floral morphological characters (flower colour and size, degree of"ptotandry, nectar - 22 production) which also reflect their basic breeding system difference. Plectritis congesta is largely crutcrossed, with a mean outcrossing rate measured in a number of populations and over a number of seasons of 70%; P. brachystemon is highly selfed, with a mean outcrossing rate of less than 2% (Carey and Ganders, 1980; Ganders et al., 1977a., 1977b.; Layton, 1980). In addition, the two species have proved to be intersterile in laboratory crosses, and no intermediate forms have been observed in a number of locations where the two grow sympatrically. In effect, the only major biological difference between the two species appears to be their breeding system. Both species are small, relatively easy to grow in crowded conditions in greenhouses or controlled environment chambers, and complete their life cycle, seed to seed, within 5 months under suitable conditions. In addition, independent estimates of genetic variability are available for the two species from two studies. Isozyme data has been analysed to determine levels of within; and between population diversity in a number of populations of both species (Layton, 1980) i Morphometric characters of the fruits have also been examined in populations of each species (Ganders and Maze,.unpublished). - 23 Materials and Methods Source populations The seed for the base populations of the two species came from two populations Cone of each species) growing sympatrically in Mill Hill Park near Victoria on southern Vancouver Island, British Columbia, Canada. Both local populations are part of more extensive populations covering the open hillsides in the park, more or less continuously in the case of P. congesta, but in isolated pockets in the case of P. brachystemon. The habitat is typical for the species: open, rocky hillsides with patches of shallow soil, wet in winter and edaphically dry by early summer, with a community of grasses, bryophytes, and herbaceous winter annuals and peren nials under scattered' Quercus garryana,'Arbutus, and Cytisus. Nine hundred sixty three plants of P. congesta and 590 plants of P_ brachystemon were collected in late fruit in June 1977. These numbers probably represent 50 - 75% of the total numbers in the local population. The extent of the two local populations collected and their overlap is diagrammed in Figure 1. The numbers of fruits per plant in the populations varied from 1 or 2 to many; all fruits were collected from each plant. I made no effort to collect equal numbers of fruits from each plant, and fruits from each species were lumped in bulk samples. Frequencies of winged and. wingless fruited plants were recorded, in ^ P. congesta for use in a progeny test to determine outcrossing rate. The winged and wingless fruits were bulked separately. All plants of P.. brachystemon at this locality are wingless fruited. - 25 Growing conditions All plants in the experiment were grown in standard 25 x 50 cm plastic flats. Fruits were planted 1 cm deep in approximately 4 cm of steam treated soil. The fruits were planted 200 to a flat in a grid of 10 rows and 20 columns spaced 2.5 cm apart. Fertilizer (Hi-Sol 20-20-20).. was added to the soil prior to planting, to remove some possible sources of heterogeneity within and between flats in nutrient levels. The amount of fertilizer added varied from generation to generation, as higher levels in the early generations caused excessive growth, which made the plants difficult to handle (Table I). All experimental populations were grown in a single Conviron walk-in controlled environment chamber. The conditions of light and temperature were set as much as possible to simulate natural conditions; the plants were germinated in a cold chamber, and then light and temperature were increased as the plants matured. The light and temperature conditions varied slightly from generation to generation as I attempted to find the best compromise between a short generation time and a plant habit best suited for manipulation (short, stocky plants with a strong root system) (Table I) . Positions of the flats in the growth chamber were assigned at random and the flats were shuffled several times during each generation to remove position effects. All flats were watered to saturation daily with tap water until fruit set was complete, and the plants were then allowed to die as the soil dried. Measurements All plants in each treatment population and every generation were Table I. Growing conditions Generation Fertilizer Temperature (gm/flat) ( C night/ C day) G 8 7/12 - 39 days 0 10/15 - balance 8 7/12 - 39 days 10/15 - balance 4 7/12 42 days 10/15 - balance 4 7/12 33 days 10/15 - 38 days 12/20 - 64 days 12/23 balance 2 7/12 _ 28 days 10/15 — 12 days 10/18 18 days 11/20 balance 2 7/12 28 days 10/15 —* 8 days 11/20 balance Light Notes (hr dark/hr light) 8/16 -growth chamber out of operation day 82 day 89, plants at room temperature -sprayed for aphids (Isotox) day 103 8/16 8/16 -sprayed for mildew (Benomyl) day 39 8/16 8/16 -intensity of light during day increased day 79 8/16 - 27 measured for the following characters; 1, Days to emergence The number of days- between planting and the complete emergence of the cotyledons above the soil surface was recorded. 2, Days to anthesis (flowering time) The number of days between planting and the opening of the first flower on each plant was recorded. 3, Height at anthesis The height of the plant in mm, from the soil surface to the top of the main inflorescence (Figure 2) was recorded on the first day of anthesis. 4, Number of nodes at anthesis The number of nodes between the soil surface and the base of the main inflorescence (inclusive) was recorded for each plant on the first day of anthesis. In P. brachysteiiton, in which the upper nodes are still highly compressed at anthesis, the nodes were counted by identifying pairs of leaves at each node. 5, Number of primary branches at anthesis The number of branches or branch buds visible in the axils of leaves on the main axis on the first day of anthesis was recorded. This is an underestimate of the total amount of branching, as there are some primary branches, as well as secondary, tertiary, and higher order branches, which do not begin to develop until later in the flowering period (Figure 2). 6, Fruit production The ripe fruits were collected from the main inflorescence of each plant and counted. This figure is subject to a large amount of experimental error; fruits, when ripe, are easily dislodged from the plant and lost, and P. congesta requires artificial pollination to produce fruit well in the laboratory, so unavoidable variation in pollination levels will have Figure 2. Morphology of Plectritis, - 29 affected fruit set in this- species:, 7. Fruit phenotypic characters The fruit from each plant was scored for a number of phenotypic characters. Fruits of P. brachystemon were all monomorphic (in this case) for these characters. In P. congesta, fruits could be scored for wing phenotype (winged or wingless, see Figure 3), pubescence pattern (Figure 4), fruit colour (body and wings scored separately on the basis of an arbitrary colour classification using 4 colour classes), and the presence in winged fruits of a characteristic indentation in the margin of the fruit'wing. I attempted to record the shape of the fruit wing, which varied considerably between plants and relatively little within plants. The variation, however, was too continuous and complex to allow for an adequate scoring system and not amenable to any simple -measurement. In addition to these characters, I recorded various other sporadic and anomalous characters including: aberrations in the number of cotyledons, presence of more or less fused cotyledons, chlorotic seedlings, excessively darkly pigmented seedlings, abnormal branching patterns and other abnormalities in the adult growth habit, and abnormalities in flowering characteristics, most commonly aborted anthers and lack of good pollen, as well as flowers with abnormal numbers of parts (for example, more than 5 corolla lobes, more than three stamens, etc.). Breeding procedure Three groups of populations were involved in the experiment, based on a combination of species and breeding procedure. Plectritis congesta required manual pollination because it is protandrous and does not Figure 3. Plectritis- fruit wing phenotypes. Wingless ventral view- dorsal view - 32 automatically self-poll£nate successfully in the growth chamber, I took advantage of this to set up two groups of P^.-"congesta populations, one outcrossed and one selfed. The plants in each population were either self-pollinated or crossed to relatively unrelated plants (not siblings). The pollination was done by removing newly opened anthers with fine forceps from the pollen parent, and using them to pollinate appropriate stigmas. The success of this breeding procedure was evaluated by examining particular progenies in subsequent generations. Plectritis brachystemon is not protandrous, self-pollinates automatically, and sets fruit very successfully in the growth chamber; it is not easy to outcross, because of the small size of the flowers. For these reasons only selfed populations of P_.  brachystemon were involved in the experiment. Selection procedure The selection method was simple individual or mass selection, in which certain individuals of one generation were selected on the basis of their phenotype to produce seed for the next generation. The selection pressure was approximately 90%, that is 20 plants were selected from the 200 in a particular population to form the next generation. Lines were selected separately for short height at anthesis, tall height at anthesis, early anthesis, and late anthesis. An unselected control line was also maintained for each species group. One flat was planted with the last generation, containing 100 individuals of each species from the source populations; this served as an external control to changes which might have affected the internal, unselected control lines. These populations are designated P. congesta G,. source and P. brachystemon G,. source. - 33 Base population The first generation, base populations (GQ) consisted of 9. populations of 200 individuals, 3 each for the 3 species groups: P. congesta outcrossed (PCO) , -P, congesta selfed (PCS), and P. brachystemon selfed (PBS). For each group there was a control population, an anthesis population, and a height population. The fruits from which the base populations were grown were selected at random from the bulk sample from the source populations. Winged and wingless fruits in the PCO and PCS populations were planted in frequencies equal to their frequencies in the source population (12.5% wingless, 87.5% winged). First cycle of selection Prom the base populations, 20 plants were selected as parents for the next generation, G^. Ten fruits from each were taken to form a population of 200. Selection lines for early and late anthesis were begun by taking the 20 earliest and latest flowering plants in the G^ anthesis population as- parents. Similarly, short and tall lines were selected from the GQ height population. Twenty plants were selected at random from the GQ control population to form the G^ control population. Thus there were 15 treatment populations in G^ and subsequent generations as indicated in Figure 5. Subsequent cycles of selection In each generation from G^ on, the 20 earliest, latest, shortest, and . /- • tallest plants were selected in the respective populations as parents for - 34 Figure 5. Experimental populations-maintained through 5 generations of selection, G.. to G . P. congesta P. brachystemon PCO PCS PBS Control NF=200 N=200 N==200 Early anthesis N=200 N=200 N=200 Late anthesis N=200 N=200 N=200 Short height N-200 N=200 N-200 Tall height N=200 N=200 N=200 the subsequent generation. ' Again, the control lines were continued by- selecting plants at random from the control populations, in all selections in PCO the pollinations were made, as far as possible, between selected individuals of different families. The requirement that 10 fruits be produced before a plant qualified as a parent for selection meant that some individuals which would otherwise have been selected on the basis of their flowering time or height were disqualified. In effect, there was selection for a minimum level of fecundity in addition to selection for height and flowering time. In G^ the PCS and PBS populations selected for short height produced too few fruits per individual to even reach the minimum fecundity level, so I had ,to reduce the family size of 10 in this case. Thirty-seven individuals from PCS short and 35 from PBS short were selected to contribute families of varying size (1 - 10 progeny) to the next generation G^. Progeny test and outcrossing rate in P. congesta The winged and wingless phenotype frequencies in the source population, together with the observed frequencies of winged and wingless morphs in their progenies (GQ) allowed a progeny test which gave an estimate of the outcrossing rate, t, in the source population (Ganders et al., 1977a.). Data treatment and analyses All measured or scored characters in every line and generation were punched on computer cards and stored in files in the University of British Columbia computer system for analysis. Most statistical analyses were performed using the MIDAS statistical package (Fox and Guire, 1976). Metrical characters I used the six metrical characters - days to emergence, days to anthesis, height at anthesis, number of nodes at anthesis, number of primary branches at anthesis, and fruit production - to compute a further set of six transformed characters as follows. The grand mean and standard deviation of each were computed for all six P. cOrigesta populations together, and for all three P., brachystemon G^ populations together. In data for subsequent generations, all populations within a particular species group were transformed by a multiplicative and an additive factor, so that the distribution of the transformed character in the control (unselected) population had the same mean and standard deviation as the GQ standard. An example is given in Figure 6. This transformation effectively removes the following two sources of variation from the data, which would otherwise interfere with the interpretation of the experimental results: 1. the common effects of generation to generation fluctuations in growing conditions and other environmental factors, and 2. the effects of any uncontrolled selection pressures (for example, selection for growth under growth chamber conditions) and to some extent the effects of inbreeding which could be presumed to be acting equally on selected and unselected lines. Descriptive statistics The distributions of the metrical characters, raw. and transformed, were described in all populations in terms of number of individuals measured, maximum and minimum values observed, population mean, standard Figure 6. An example of data transformation procedure used on metrical characters Days to emergence PCO control PCO early PCO late PCO short PCO tall raw data x s ,d. 20.77 5,55 21.03 4.47 21.81 5.17 19.43 4.08 21.22 5.53 correction X. x 0.7522 + 2.776 transformed x' s.d! 18.4 4.17 18.6 3.36 19.2 3.89 17.4 3.07 18.7. 4.16 "0 standard s. d 18.4 4.1 x - 38 deviation, coefficient of variation, skewness, and kurtosis. In addition, frequency histograms of all distributions- were generated to depict them graphically. Comparisons between distributions For every metrical character the distributions of the selected populations were compared to those of the control in the same generation (for example, PCO early G^vs.. PCO control G^) by means of Kruskall-Wallis tests (non-parametric analysis of variance). The transformed data were also compared between generations within lines by means of Kruskall-Wallis tests (for example, PCS short G^ vs. PCS short G,_) . Correlations Correlations between all pairs of metrical characters within each population were calculated by Spearman's rank correlation procedure. Heritability estimates 2 Estimates of narrow sense heritability (h , the proportion of the total phenotypic variance in a population which is attributable to additive genetic effects) were calculated in the experimental populations for the selected characters, flowering time and height at anthesis, by two methods. Realized heritability was calculated after the method of Hill (1972 ). This estimate is based on the ratios of selection differential (selection pressure) to. response in lines under divergent selection. Heritabilities were also calculated for all metrical characters- by the -method of parent-offspring regressions- in the' control lines (Falconer, I960), Variance within populations The components of variance within populations were analyzed by univariate ANOVA for the metrical characters. The variance was partitioned into components between families and within families. Other characters Frequencies of fruit wing phenotypes, pubescence patterns, fruit wing and body colours, presence of wing indentation, and aberrant characters were tabulated for every population. Results Breeding systems in Plectritis- congesta • and P__ Brachystemon Outcrossing rates in the source populations (Mill Hill Pk., 1977) The outcrossing rate in the P. congest a s our ce population in 1977 was estimated by the progeny test method (Ganders et al., 1977a.) to be 61.6%. This is based on a total sample of 1175 individuals grown from seed. This estimate compares well with other estimates of outcrossing rates in populations of the species, which have averaged around 70% (Ganders et al., 1977a*, Carey and Ganders, 1980; Laytonj 1980). Since the source population of P^ brachystemon was monomorphic for all of the morphological markers which might have been used in a progeny test, no estimate of the outcrossing rate is yet available for it. There is no reason to expect this population to differ substantially from others measured throughout the range of the species in British Columbia. Ganders et al. (1977b,) and Layton (1980), using the fruit wing polymorphism and allozyme polymorphisms respectively, estimate that the average outcrossing rate in P^ brachystemon is 2%, and no populations were found exceeding 5%. Outcrossing rates in the experimental populations The effect of the experimental breeding system as practised can be roughly, estimated in P. congesta by examining particular progenies as follows. In the selfed group, PCS, the progenies of wingless fruited plants (homozygous recessives) can be scored to obtain the frequency of winged - 41 fruited progeny, which are necessarily the result of outcrossing events. This frequency will underestimate the actual rate of outcrossing, as a (small) proportion of the wingless fruited progeny are likely also to be the result of outcrosses to other homozygous recessive or heterozygous plants. In the outcrossed group, PCO, progenies of wingless fruited plants which have been crossed to homozygous dominant winged fruited plants can be scored to obtain the frequency, x, of wingless fruited offspring. If these were all the result of accidental selfing, the outcrossing rate would Be.l - :x. In practice this estimate is again an underestimate of the actual outcrossing rate, as some of the wingless fruited progeny are again likely, to be the result of accidental outcrosses, in this case to some plant other than the pollen parent of record. The error in these estimates is likely to be large, and mostly depends on the allele frequencies in the populations Cthe higher the frequency of the recessive allele, the larger will be the error); neyertheless, the estimates are of interest,,and are given in Table II. Characteristics of the base populations Descriptive statistics of the metrical (quantitative) characters The initial distributions of the various measured characters in the base populations, Gg,,of the two species are given in Table III. Under the experimental conditions in the growth chamber, P. brachystemon emerges later, grows taller, produces more nodes and more primary branches, flowers later, and produces more fruit than P. congesta. Examining the coefficients of variation, which are scale free estimates of the phenotypic variability, - 42 Table II. Estimates of outcrossing rates in the - experimental populations. P. congesta selfed Population Number of progenies Estimated outcrossing rate PCS control Go 3 0.25 PCS late Go 3 0.19 PCS short Go 5 0.35 PCS tall Go 3 0.08 PCS control Gl 4 0.18 PCS early Gl 1 0.00 PCS late Gl 6 0.10 PCS short Gl 5 0.12 PCS tall Gl 4 0.04 PCS control G2 4 0.05 PCS late G2 6 0.22 PCS short G2 6 0.06 PCS tall G2 6 0.04 PCS control G3 4 0.28 PCS late G3 7 0.22 PCS short G3 17 0.15 PCS tall G3 8 mean 0.14 0.15 P. congesta outcrossed PCO early Go 1 0.80 PCO late Go 1 0.60 PCO late'. Gl 3 0.59 PCO early G2 2 0.68 PCO late G2 1 0.29 PCO short G2 2 0.94 PCO late G3 1 0.67 PCO tall G3 1 mean 0.63 0.65 Table III. Measured characters:, base populations. N "Mean Standard Coefficient of deviation variation P_. congesta Days to emergence 9Q5 18.40 4.177 22.70 **a Chumber] Height at anthesis 855 223,87 45.121 20.15 ** (mm) Nodes at anthesis 855 8.91 1.108 12.43 ** (number) • Primary branches 855 2.71 3.417 126.09 ** (number) Days to anthesis 855 95.02 7.379 7.76 * (number) Fruit production 861 25.10 18.820 74.99 (number) P. brachystemon Days to emergence 304 21.08 5.729 27.17 (number) Height at anthesis 293 302.91 72.618 23.97 (mm) Nodes at anthesis 293 11.80 1.656 14.03 (number) Primary branches 293 6.43 6.841 106.40 (number) Days to anthesis 292 116.76 8.088 6.93 (number) Fruit production 291 61.26 28.48 68.80 Asterisks indicate coefficients of variation significantly different from those in Pbrachystemon; ** at the 1% level, * at the 5% level, according to modified F - tests (Lewontin; 1966). it appears that P. brachystemon is more variable than P. congesta for days to emergence, height, and number of nodes at anthesis, and less variable for number of primary branches at anthesis and flowering time. There was no significant difference in the coefficient of variation for fruit production. Frequencies of qualitative characters Plectritis brachystemon is monomorphic for all the fruit characters scored; the scores for the base populations of P. congesta for pubescence pattern are presented in Figures 7 to 10, and will be discussed later. The phenotype frequencies of winged and wingless fruits planted in GQ were the same as in the source population (12.5% wingless, 87.5% winged); the phenotype frequencies of the adult plants in GQ are presented in Figure 11 Response to selection of the selected characters Height at anthesis Means ' The mean heights of plants in the populations selected for height at anthesis departed significantly from the control (unselected) population in most generations, the exceptions being PCO short G^, PCO tall G^ and G^ PCS" short G^ PCS tall G2, and PBS short &2 (Figure 12). In the case of the P. .congesta populations, the means in the selected lines diverged over the, course of the experiment with, in most cases, the tall lines being - 45 Figure 7. Frequency' of various pubescence types (see Figure 4) In the experimental populations, a. Pubescence type 0 in PCO populations b. Pubescence typel in PCO populations c. Pubescence type 2 in PCO populations Figure 7. a. PCO type 0 .06 .04 .02 A / \ / v / \ / / \ v b. PCO type 1 o e 0) fi cr-cu M «w fi M 0) 4-> 4-1 cd cu a) o fi a) o to a) •8 P4. .06 .04 .02 \ \j_ate \ '\tall \ Control Height Anthesis >4 late c. PCO type 2 early short Generations of selection < -47 Figure 8. Frequency of various pubes-cence types Csee Figure 4) in the experimental populations-. a. Pubescence type 3 b. . Pubescence type 4 c. Pubescence type 5 in PCO populations in PCO populations in PCO populations Figure 8. a. PCO type 3 .6 h ^ short early b. PCO type 4 o cu cr a) u <4-l 0 cu a) o C cu o CD co ,3 r .2 r short early •\ tall ^ late 4 Control Height Anthesis c. PCO type 5 early tall Generations of selection - 49 Figure 9. Frequency of various ...pubescence types, (see Figure 4) in the experimental populations. a. Pubescence type 0 b. Pubescence type 1 c. Pubescence type 2 in PCS populations in PCS populations in PCS populations gure 9. PCS type. 0 PCS type 1 PCS type 2 .04 r Generations of selection - 51 Figure 10... Frequency of various pubescence .types (see Figure 4) in £he. experimental populations. a. -Pubescence type b. Pubescence type c. Pubescence type 3 in PCS populations 4 in PCS populations 5 in PCS populations Figure 10. - 52 a. PCS type 3 short 0 1 2 3 4 Generations of selection Figure U.., Frequency,.of? wingless ; fruited . plants in the experimental . populations: a. . PCO populations b. PCS populations / Control — Height —•— Anthesis - 55 Figure 12. Mean height,.at anthesi,swin,,.populatjons selected for . > height,.at .anthesis.- 'Means: are-expressed as a percentage of the ...means, in the control lines..:. Populations-which..intersect the , vertical lines .did .not differ significantly from the control ... - population in. the. same generation (the vertical lines do not represent standard deviations). The following pairs of consecutive populations within lines did not differ significantly: PCO short GQ and G^ G^ and G^ PCO tall GQ and G1 PCS tall G2 and G3 PCS short G. and GC 4 5 PBS tall G. and GC 4 5 taller than .the control, and the:,sh.ort-: lines shorter that. the. control. In PCO the divergence, hy,'the.::fifthi"cycle of. selection" amounted to 66% of the control height (+41%, -25%) or 148 mm (+92 mm, -56mm). In PCS, the divergence by.G amounted .to .78% of the control height. (+27%,.-51%) or 175 mm (+61 mm, -114 mm).. The means, in. the .,P. .brachystemon selected lines departed significantly from the control lines, but did not diverge, rather fluctuating erratically with both lines being shorter than the control in G^, and G,., and taller than the control in G^. In addition, the tall line ;was shorter than the short line in G.. and G, . 1 4 The mean heights in the various selected lines were also compared generation to generation.and proved to be significantly different in most cases, the following being the exceptions: PCO short G^ vs. GQ, G^ VS. G„, PCO tall G~ vs. GN, PCS short.GC vs. G., PCS tall G„ vs. G_, and PBS 3 I 0 5 4 3 2 tall GC vs. G.. 5 4 Estimates of variability •Variances The variances for both the raw and transformed values of height were compared among all populations. The variances in this case are phenotypic, although some.of the environmental component has hopefully been eliminated within generations by the use of a common environment, and between generations by correcting the selected population values against.an . . unselected control. Since.the variance of a character is dependent on the mean (in a population, with a larger mean, the variance will.also tend to be larger) , the ...selected lines, were compared, by means of the coefficients of -variation (that is„..,the . standardwdeviation .of.:;the, mean, .as . a ..percentage of the mean) in a modified F - test (Lewontin; 1966) (Figure 13). The v - 58 Figure 13.Gpef fi.ci.ents,,.o.f .'-yari.a.t'ion.^0,): height ,at anthesis;~ifi .populations . , .selected for,. height.-at.;anthesls.-.. Coefficients of, variation are expressed as a percentage of :.the coefficients:.in. the control .lines. Populations which intersect the vertical lines did not .differ significantly from the ..control population in the same .. -generation (the vertical lines do not represent standard deviations). coefficients of ..variation. ±n..tbe-. selected. ..populations .-were significantly.. different from the control in 14 out:of 30 cases, namely PCO short G^, G^, and G. ,._PCO tall -G-. ,..G„ , G' , andr Gcy., PCS' short G " and ,G. , PCS tall G-,., PB&; 4 124 5 3 4 1 short G^» and.. PBS. tall. G^., G^, and .G^..... Of.. these, six populations showed. an increase in variation (PCS and PBS lines) and eight showed a decrease in variation (PCO lines, PCS tall). Heritabilities The narrow sense heritability for.height at anthesis was.estimated two 2 ways. Narrow.sense~heritability.(h ) is the portion of the total phenotypic variability in a population which can be attributed to.additive genetic effects, (that is, genetic .effects excluding dominance, epistasis, and other interactive effects)..-Realised heritabilities were calculated.. after the method .of. Hill (1972) (Table IV) . In the P. ..congesta :lines. the heritability of height is significant and approximately equal between the 2 PCS (b or h = 0.58) and PCO (b .= 0.53).lines. Plectritis brachystemon c c * has essentially no heritability for height .under the conditions of this experiment. The estimates for PBS in Table IV are bracketed because the method is not meant to be applied.in cases where the selected lines diverge in the direction opposite to the direction of selection, as is the case here (see Figure. 12., PBS - lines.. ih.G^.. and G^) . . The estimated standard deviations for the heritability estimates .are quite large. The estimated, .heritabilities,:from...parent-off spring-regressions . (Table . V) are in reasonable,,agreement with the-realised heritability estimates. 2 The P. congesta .lines, have a fairly large h and there is little difference 2 2 ' between the PCS (h = 0.44) and PCO (h = 0.45) lines. Plectritis Table IV. Realised heritability, calculated using the method of Hill (1972). P. congesta PCO PCS Selection for early b 0.77 0.75 c or late anthesis sdu 0.12 0.14 b c Selection for short b 0.53 0.58 c or tall height sd. 0.61 0.64 D C P. brachystemon PBS 0.49 0.22 (0.06) (0.34) Table V. Heritability, from parent-offspring regressions* P. congesta PCO PCS 2 Days to anthesis h 0.60 0.72 r2 0.53 0.55 2 Height at anthesis h 0.45 0.44 r2 0.36 0.30 * Means of four generations of control populations. PBS 0.42 0.48 -0.06 0.23 i ON - 63 b r achystemon .agaitu.has essentially a heritability of zero for this 2 character. The r . .valuesaljstech estimate .the. proportion, of,.the..total variance explained .by the regression, ori.in other words the goodness of 2 fit of the •regression line to the parents-offspring points. The r values for height are fairly low, and quite a bit lower than those for flowering time. •Components of variance An analysis of variance-was performed for all populations between and Gj., .partitioning the observed (phenotypic). variance into components within and between families in each.population. It is only in particular cases, such as in progenies in pure breeding lines or in the F^ of a cross between pure breeding lines,.. that the variances , so partitioned can be considered precise .estimates of. environmental or additive genetic components. (Falconer., I960).. . The within .family variance -in a-.pure-breeding': -line is a-precise estimate of the environmental variance, as there is no genetic variance present. The populations in this experiment do not represent pure breeding lines, although P. brachystemon is certain to be highly inbred. The information that can be obtained.from.an ANOVA.comes, therefore more from any changes which might be observed.over the course of the experiment in the partitioning of the phenotypic variance between and within families. In theory ..inbreeding...will tend, .to .reduce, the.genetic .component of the within, family variance .and.-increase.-the .genetic .component of the. between .family variance. Selection, and random-drift..will, tend to decrease . the genetic components in both within and.between family estimates. In the present - 64 situation y;, .where... the., genet ic,Mstrucfcu re „of ..thei,populations.,, of,.;the two..: . . species -is unknown:,,:..,there, is .^,no';':S'$mple-^•p^.e4,i:e.ti.on'..of.^ the ..results- of ..the ANOVA; in..fact., the.. results showed very ..little. ..For. height:-at anthesis all three species groups had a significant between family component of variance in most populations (Appendix 1 ). There were, no obvious trends which might have been expected,...particularly the decrease..in between family variance which might have been expected in response to selection. Other changes in distribution There was no evidence from the frequency histograms of height at anthesis in the.various.populations to suggest that there had Beenchanges in the distribution .other :than ..the .changes in .mean and variance. That is, there was no evidence of changes in skewness -or kurtosis, or development of bimodality in the distributions. Days to anthesis (flowering time) Means As-with height.at anthesis, the.mean flowering times for the selected populations differed significantly from.the.controls In most cases, the sole exception in this case.being PCS early G^ (Figure 14'). High and.low selection lines in all three, species..groups-sdiverged,.,with-.:the^early. lines flowering earlier, than the controls, and the late lines flowering later. The divergence by the .fifth, cycle ..of ^selection.was,,, .in -the ...case .of .PCO.,.. 33.5%. of.the murnber . . of days -to. .anthesis,;.in.„the,..control.\.(«fc20%,,^13.. 5%.)... or .31..8 ... days (+19..days, -12.8 days). . In PCS the .divergence was 28.7% of the control (+16.3%, -12.4) - 65 Figure 14 Mean .number of . days to anthesis... (flowering time) in populations-selected .for flowering time. . Cleans are expressed as a percentage of the means in the control lines.. Populations which .intersect the vertical lines did not differ significantly from the control .population in the same generation (the vertical lines do not represent standard deviations). The following pairs of consecutive populations within lines did not differ significantly: PCO early•G3 and PCO late G., and G„; G. and G. 1 2 3 4 PCS early GQ,and G^ PCS late G1 and G2; .G2 and G3 PBS late G1 and G2; G2 and G3; G^ and Figure 14. PBS - 67 or 27.3 days (+15.5 .days, -11"•.8.. days J . In PBS the divergence was 18.5% of the control (+12.2%, -6.3%) or 21.5 days.(+14.2 days, -7.3 days). The mean.flowering.times'were also compared generation to generation within the..selected, lines,, with the following populations proving nofr'to be significantly different: PCS early G^ vs. GQ, PCS late G^ vs. G^, G^ vs. G^, PCO late G2 vs. G1, and PBS late G2 vs. G^ G3 vs. G2. Estimates of variability Variances The variances in the populations were converted to coefficients of variation to remove scale effects. The coefficients of variation are presented graphically in Figure 15. The selected lines were compared to the controls by means of the modified F - test.for coefficients of variation,. and were found to be significantly different in 16 of the 30 populations, namely PCO early G2 through G^, PCO late G^ and G5> PCS early G2 through G5, PCS late G2, G^, and G5> PBS early G3 and G^, and PBS late G5> Of these, in all cases except PCS late G2 and PBS late G,. the variation was less in the selected lines than in the controls. The general trend in all six selected lines was towards a decrease in the coefficient of variation. Heritabilities The realised heritabilities.(Table IV) again reveal the basic agreement ? between the.PCO. (b .or h =0.77) and PCS (b =0.75) lines. Plectritis c c brachystemon also has a reasonably large heritability.for flowering time (b = 0.49). In all .cases the standard deviation for the estimates is - 68 Figure 15. ..Coefficients , of -variation-for. :days to anthesis; in, populations selected for flowering ..time. ..Coefficients of./variation are expressed as a percentage of the coefficients in the control lines. .Populations which intersect the vertical lines did not . differ significantly from the control population in the same generation, (the vertical lines do not represent standard deviations). PBS — - 70 considerably •; smaller than inthe heritability. estimates for height at anthesis. The estimated, heritabilities from parent-offspring regressions for flowering time (Table y). are.comparable to the realised heritability 2 estimates, with the two P. congesta control lines similar (PCO h = 0.60, 2 2 PCS h = 0.72) and the P. brachystemon line slightly less (h = 0.42), but still appreciable. Again, the reliability of the estimates is indicated 2 by the goodness of fit of the regression lines to the data (r ), which 2 in this case shows a more reliable estimate of h than did the regression for height at anthesis. Components of variance As with height at anthesis in lines selected for height, all the lines selected for flowering time showed a significant between family component of variance for flowering time in most populations. Once again, there was no significant trend in these parameters over the course of the experiment (Appendix 1). Other changes in distribution There was no evidence from.the frequency histograms of flowering time in the populations selected for flowering time to suggest that there had been changes in the distribution .other than changes in mean and variance. That is, there was no evidence of changes in skewness or kurtosis, or development of bimodality in the distributions. Changes in the unselected characters -during the experiment - 71 The characters not under selection - days to emergence, height at anthesis in. the lines selected for flowering time, number of nodes at anthesis, number of-primary branches at anthesis, flowering time in lines selected for height at anthesis, and fruit production - were analysed in the same manner as the selected characters. Means Days to emergence A number of populations departed significantly from the controls in terms of the me.an- number of days to emergence: PCO early , PCO tall G^, PCS early G2> G3> PCS late G.^, G^ PBS early G^ PBS tall G^ and PBS short G^ (Figure 16). In all groups selected for either height.or flowering time, the G,. means for days to emergence in the plus selected lines were greater than the means in the minus selected lines, with the exception of PCS selected for flowering time. However, a number of lines experienced reversals, with the plus selected line falling, below, the minus ..selected line: PCS anthesis and PCO height G^.PCS height G^, and PBS height G^. There appears to be no regular trend in the changes in emergence date. Height at anthesis (in lines selected for flowering time) In the lines selected for .flowering time, the .mean heights differed significantly from the controls in PCO early G_, G , PCO late -G. , G,, PCS -72 Figure 16. Mean .number ...of. daya^to emergence .in various ..populations. Means are . expressed-..as a.pereentagecof-the/means in the control lines. Populations which, intersect .the vertical, lines did. not differ significantly from the control population in the same generation, (.the vertical lines do hot represent standard deviations). a. Mean number of days to ..emergence in populations selected for flowering time. b. Mean number of days to emergence in populations selected for height at anthesis. PBS - 74 early -G^. .PCS late. G^, •,PBS,.early » G3» and PBS late G3 (Figure 17). The plus selected lines^vere..allvtalle.r ,,than.;.the:minus selected lines by G,., but.all lines were shorter.than the control with the exception,of PBS late. Both P. congesta, lines ..experienced,^reversals in G^. . Again, there appear to be no long, term trends.in changes in height at anthesis in the lines selected for flowering time. Nodes at anthesis The mean number of nodes at anthesis departed significantly f.rom the controls in all populations except PCO late G^, PCO short G^jG^, PCO-tall. •G^, PCS short G^, PCS tall G^, and.PBS tall G^ (Figure 18). There was a strong trend toward divergence in lines selected for both height and flowering time, with all plus populations except PBS tall G^ above the control and all minus populations except PBS short G^ below it. The strong trend toward divergence is undoubtedly due in part to the same factors which lead to a strong correlation between the number of nodes at anthesis and flowering time (see Correlations below) as the divergence is more marked in lines selected for flowering time. Primary branches at anthesis The number of primary branches at anthesis in the selected lines showed large and somewhat erratic departures from the controls, the means being significantly.different in all cases except PCO early G^> PCO late G^, G3, PCO short G2, PCO tall. G^., G^, PCS early G2„ PCS late G^,, G2, G3,. PCS short G , PCS tall G , PBS early G , PBS late G,, PBS short G , .and PBS tall - 75 Figure 17. a. Mean height at anthesis in populations selected for flowering time. Means are expressed as a percentage of the means in the control lines. Populations which intersect the vertical lines did not differ significantly from the control population in the same generation ( the vertical lines do not represent standard deviations). b. Mean number of days to anthesis in populations selected for height at anthesis. Means are expressed as a percentage of the means in the control lines. Populations which intersect the vertical lines did not differ significantly from the control population in the same generation (the vertical lines do not represent standard deviations). Figure 17. - .76 o u CO 4-1 •H G CO o cu a 4J CH C O 4-1 60 rt rt 4-1 4-1 C cu 60 o •H H cu a) ^! ft C cn rt rt a) v—' 140 100 60 r 20. r b. CO •1-1 CO a) .—* 4-1 O rt M 4J O c 4-1 o u CO >•> CH rt o T3 cu 4-1 60 O rt 4J H « a) CU ,g o u cu ci ft C CO rt rt cu s 110 h 100 90 tall short short short Generations of selection PCO PCS PBS - 77 Figure 18. Mean number of nodes at anthesis in various populations. Means are expressed as a percentage of the means in the control lines. Populations which intersect the vertical lines did not differ significantly from the control population in the same generation (the vertical lines do not represent standard deviations). a. Mean number of nodes at anthesis in populations selected for flowering time. b. Mean number of nodes at anthesis in populations selected for height at anthesis. PBS — - 79 G^ (Figure 19). In most cases there was divergence, with the plus selected lines having more primary branches than the minus lines. A number of reversals were observed in the lines selected for height at anthesis (PBS G^ and G^, PCO G^) and most lines fluctuated erratically, often above and below the controls in different generations. There may have been some effect of selection in the lines selected for flowering time, but otherwise there were no trends in the changes of the means. The fact that the selected lines seem to cycle up and down relative to the controls from generation to generation reflects changes in the controls rather than in the selected lines, and indicates the sensitivity of this character to changes in the environment from generation to generation. Flowering time (in lines selected for height at anthesis) Selection for height at anthesis appears to have had some effect on flowering time. All the selected lines have diverged somewhat, with all being significantly different from the controls except PCO short G^, G^, PCO tall Gl9 PCS short G^ PBS short G2> and PBS tall Gl (Figure 17). The strong correlation between flowering time and number of nodes at anthesis can be seen in the similarity between the changes in the means for the two in lines selected for height at anthesis (compare Figure 17 b with Figure 18 b). Fruit production Fruit production was a character whose measurement was subject to a great deal of error. As can be seen in Figure 20, the means in the selected lines departed significantly from the controls in many cases, the exceptions - 80 Figure 19. Mean number of primary branches at anthesis in various populations. Means are expressed as a percentage of the means in the control lines. Populations which intersect the vertical lines did not differ significantly from the control population in the same generation (the vertical lines do not represent standard deviations). a. Mean number of primary branches at anthesis in populations selected for flowering time. b. Mean number of primary branches at anthesis in populations selected for height at anthesis. - 82 Figure 20.. Mean fruit production in various populations. Means are expressed as a percentage of the means in the control lines. a. Mean fruit production in populations selected for flowering time. The following populations did not differ from the control in the same generation: PCO early G^, PCO late G^, PCS early G2> G3> G^, PCS late G^ G3> PBS early G^, G3> and PBS late G^ Gy G^. b. Mean fruit production in populations selected for height at anthesis. The following populations did'not differ from the control in the same generation: PCO short G^, PCO tall G^, G PCS short G^, G3> PCS tall G^ G2> G3> G^, PBS short G , and PBS tall G,, G.. 1 4 PBS being PCO early G^, PCO short G^, PCO tall G^, PCS early G2, G3, PCS late.Gj^,. G , PCS short G^ G^, PCS tall G^, G^, G3> PBS early G^ PBS late G^, G , PBS short G^, and PBS tall G^. It can also be seen that pairs of lines, plus and minus, wander erratically but together, suggesting that most of the movement is due to chance or error, and strongly,influenced by the fruit production in the control lines. Estimates of variability Variances The variances in the unselected characters, expressed as coefficients of variation, are presented in Appendix 2, and graphically for days to emergence and number , of nodes at anthesis in Figures 21 and 22. Variation in days to emergence shows a general increase, with wide fluctuations, as does variation in the number of primary branches at anthesis. Height at anthesis in lines selected for flowering time is less variable in PCO and PCS early by G^_, and slightly more variable in PCS late and PBS. Variation in the number of nodes at anthesis tended to decrease, particularly in those lines selected for flowering; time. Variation in flowering time in lines selected for height at anthesis decreased in the P^ congesta lines, but fluctuated above and below the control in the P. brachystemon lines. Variation in fruit production also changed very erratically, reflecting the error inherent in the measurements. Heritabilities - 85 Figure 21'. Coefficients of variation for number of days to emergence in the experimental populations. Coefficients of variation are expressed as a percentage of the coefficients of variation in the control lines. a. Coefficients of variation for number of days to emergence in PCO populations. b. Coefficients of variation for number of days to emergence in PCS populations. c. Coefficients of variation for number of days to emergence in PBS populations. - 87 Figure 22.. Coefficients of variation for number of nodes at anthesis in the experimental populations. Coefficients of variation are expressed as a percentage of the coefficients of variation in the control lines. a. Coefficients of variation for number of nodes at anthesis in PCO populations. b. Coefficients of variation for number of nodes at anthesis in PCS populations. c. Coefficients of variation for number of nodes at anthesis in PBS populations. - 89 There was only one estimate of heritability for the unselected characters from the experiment, as the realised heritability procedure is not applicable. The heritability estimates from parent-offspring regressions in the control lines for these characters are presented in Table VI. 2 As indicated by the coefficients of determination, r , the fit of the 2 regression line is quite poor in most cases. If the cases where r is greater than 0.2 are considered alone, the following heritabilities may be 2 estimated. For days to emergence, h =0.49 in PCO. For nodes at anthesis, 2 h = 0.55 in PCO, 0.57 in PCS, and 0.28 in PBS.; these values are comparable and somewhat intermediate to the estimated heritabilities for the selected characters. Only PBS has reasonably precise estimates for the number of 2 . primary branches, h =0.29, and none of the lines provided a reliable estimate for the heritability of fruit production. Components of variance There were significant between family components of variance in all the unselected characters in all three species groups and most populations (Appendix 1). As with the selected characters, there were no consistent trends over the course of the experiment, and nothing to suggest changes due to selection in the partitioning of the total variance between and within families. Other changes in distribution As with the selected characters, there was no evidence from the frequency histograms of the unselected characters to suggest any changes in distributions other than the changes in mean and variance/ Table VI. Heritabilities from parent-offspring regressions, unselected characters. PCO control PCS control PBS control Days to emergence .68 .09 .31 .0007-.65 .0058 .20 .0000. .43 ,18 ,12 ,28 .14 .03 .018 .082 .16 ,21 ,32 ,14 .026 .056 .14 .10 .49 Nodes at anthesis .71 .48 .46 .17 .48 .48 .39 .059 .43 .73 .49 .65 .20 .53 .46 .80 .066 .13 .28 .13 .0084 .052 .25 .087 x .55 .57 .'28 Primary branches at anthesis .05 .27 .076 .12 .0043 .092 .02 .035-.081 .34 .24 .38 .02 .22 .22 .18 .22 .29 ,20 .13 .069 .14 .18 .053 .29 Table VI, continued. PCO control PCS control PBS control u2 2 .,2 2 ,2 ' 2 h r h r ' h r Fruit production G .05 .0052 -.015 .0009 .15 .14 G3 -.10 ' .055 .11 . .07 -.034 .0045 G. .15 .099 .11 .023 -.083 .011 4 G5 .079 , .02 -.11. .055 -.033 .0059 - 92 The effects of selection on correlations among the measured characters The correlation between height at anthesis and flowering time, the characters under selection. The correlation.between the selected characters, height at1,anthesis and flowering time, was initially significant and positive in all three species groups (Figure 23). It decreased more or less steadily towards no correlation in the PBS lines, even becoming significantly negative in two cases (PBS tall and PBS short G^). There was some decrease evident in the P_ congesta-. lines, particularly in the third cycle, G^, when several of the populations showed a significant negative correlation, but by G^ the correlations were mostly positive again, and it is difficult to discern any trend from the first five generations of selection. Other correlations The other correlations can be divided into three broad groups. In some cases, the correlations differed between the P. congesta lines and the P. brachystemon lines. In the correlations between height at anthesis and number of primary branches (Figure 24), number of nodes and number of primary branches (Figure 25), and height at anthesis and fruit production (Figure 26) the correlations in the PBS lines were strong and positive, while the correlations in the PCO and PCS lines were mostly not significant and not consistently positive or negative. In some cases the correlations were essentially similar in. all three species groups and mostly not significant, or when significant not - 93 Figure 23. Correlations between height at anthesis and flowering time. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. early Xv? : late S.\. tall V short N early late early ^ tall short Generations of selection Control Height Anthesis - 95 Figure 24. Correlations between height at anthesis and number of primary branches at anthesis. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 24. .5 a. PCO b. PCS c. PBS CO •H CO CU Xi 4-1 0 c0_ 4J n) co cu Xi a fl cd H >. H Cfl •S M ft CO > CO •rH. CO cu 4-1 c cc) 4-> n) 4-1 Xi 60 •H CU Xi 4-1 a cu •H O •H 4-1 m cu o o a o •H cu H M O I late 5 earl\ short - tall -.5 .5 r tall _ late short early -.5 1-late short — tall early -.Si-Generations of selection Control Height . Anthesis - 97 Figure 25. Correlations between number of nodes at anthesis and number of primary branches at anthesis. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 25. - 98 a. PCO CO •w CO cu Xi •U Ci cd .5 CO cu Xi a % -.5 XI >. u cd I a. b. PCS CO > CO •H CO <u Xi CO Q) O d .5 -.5 I-c. PBS cu •H a <4-( <-H (U O O c O •H J-l cd rH (U rJ o O I short .5 r early -.5 Generations of selection ..A Control Height Anthesis - 99 Figure 26. Correlations between height at anthesis and fruit production. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation: was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 26. a. PCO .5 - 100 short late early tall b. PCS o •rl 4-> a T) O u PH 4J •H 3 H 4-1 to > CO •H CO cu XI 4-1 a CO -.5 I-.5 early V" «short x late - ' tall xi 60 •rl 0) Xi -.5 c. PBS cs d) •rl CJ •rl •4H m CU o o c o •rl 4J n) rH cu M rl o CJ .5 -.51-0 -early y •==^^»-^_. late' short tall Generations of selection Control Height Anthesis - 101 consistently positive or negative within a species group. For example, correlations between days to emergence and height at anthesis (Figure 27), days to emergence and number of nodes at anthesis (Figure 28), number of nodes at anthesis and fruit production (Figure 29), and primary branches and flowering time (Figure 30) fluctuate between being significant and positive, non-significant, and significant and negative, often within the same line (for example, primary branches vs. flowering time in PBS short shows this kind of pattern), Finally, some of the correlations were for the most part significant in all three species groups, and remained so throughout the experiment. This is the case with negative correlations between days to emergence and number of primary branches (Figure 31), days to emergence and fruit production (Figure 32), and flowering time and fruit production (Figure 33). Significant positive correlations were observed consistently between days to emergence and flowering time, with the exception of and G,. in the PBS lines, in which the correlation disappeared (Figure 34), between number of nodes and flowering time (Figure 35), number of primary branches and fruit production (Figure 36), and height at anthesis and number of nodes, with the exception of some of the populations in G^, in which the correlation dropped (Figure 37). It must be remembered that these are phenotypic, rather than genotypic, correlations, and as such are subject to environmental effects. Nevertheless, there does not appear to have been any change in any of the correlations between the various characters which could be attributed to the effects of selection, except perhaps in the case of the correlations between the selected characters. - 102 Figure 27. Correlations between number of days to emergence and height at anthesis. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 27. - 103 a. PCO b. PCS c. PBS cn •H cn cu xi •U c cO ^3 60 •r4 CU X! CO > cu CJ C cu 60 H CU a cu cn >> CO . T3 4-1 a cu •H a •H 4-1 <4-l CU O CJ C o •H 4-1 cd rH CU H H O O early tall hort ate Generations of selection Control Height Anthesis - 104 Figure 28. Correlations between number of days to emergence and number of nodes at anthesis. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlatic was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 28. .5 - 105 Generations of selection Control Height Anthesis - 106 Figure 29. Correlations between number of nodes at anthesis and fruit production. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are. significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 29. .5 a. PCO " late' a^, . early :short tall b. PCS a o •rl 4J O 0 'U o u o. 4-> •rl P u 4-1 CO > CO •rl CO QJ XI 4-1 c cd -4-1 cd CO a) o a -.5 I late — : early -~ tall short c. PBS 4-> C CU •rl a •rl m <4H o CJ o •rl (U i-l O CJ 3 .5 early ' late -.5 r Generations of selection Control Height Anthesis - 108 Figure 30. Correlations between number of primary branches at anthesis and flowering time. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 30. - 109 .5i a. PCO b. PCS S •rl 60 •rl u CD O CO > CO CD x; o "S u >. M ec) & !-l -.51 0 short c. PBS a CD •rl o •rl m a) o o cl o •rl CD u o CJ Pi short f -. early '" tall late Generations of selection Control Height Anthesis - 110 Figure 31. Correlations between number of days to emergence and number of primary branches at anthesis. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 31. .5 a. PCO co QJ o a rt H rQ >, H rt s •rl H ft -.5 tall 44 b. PCS CO > a) o C cu 60 H a) a cu _ o 4J CO rt _____ x-^cr— \ — i : early JxC late -.5 —short tall e QJ •A U •H 4-1 4H CU O CJ (3 O •H .5 c. PBS cu u u o o u Pi -.5 + early tall short Generations of selection Control Height Anthesis - 112 Figure 32. Correlations between number of days to emergence and fruit production. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations Figure 32. - 113 a. PCO a o •H 4- 1 O 3 T) O 5- I PH 4J •H 3 U 4-1 -.5 • late - tall short early b. PCS > cu a 0) 60 cu § 4-1 CO >^ a) T3. -.5 short — -=~"= late early tall 4-1 Cl CU •H CJ •H <4-l m cu o a CJ o •H c. PBS (U !-l U O a Pi early :•• -r :• short tall —- late -.5 Generations of selection Control Height Anthesis - 114 Figure 33. Correlations between flowering time and fruit production. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by. a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 33. a. PCO o •H •U CJ s 13 O M ft U •H 3 H m -.5 late »—^ early *' short tall CO > b. PCS c M H QJ_ & O 4H S CD •H O •rl 4-1 HH <u o o (3 O •H .5 I-0 .5 i-c. PBS cu u u o o Ai u Pi -.5 I-short ~1 early — tall late Generations of selection Control Height — Anthesis — - 116 Figure 34. Correlations between number of days to emergence and flowering time. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 34-. a. PCO .5 h tall early short late Generations of selection Control Height Anthesis - 118 Figure 35. Correlations between number of nodes at anthesis and flowering time. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 35-. Control Height Anthesis - 120 Figure 36. Correlations between number of primary branches at anthesis and fruit production. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 36. Generations of selection Control — Height Anthesis - 122 Figure 37. Correlations between height at anthesis and number of nodes at anthesis. Spearman's rank correlation coefficient is graphed for every population. Correlation coefficients which are significant at the 5% level are joined by a continuous line; lines are broken at those generations in which the correlation was not significant at the 5% level. a. Correlations in PCO populations. b. Correlations in PCS populations. c. Correlations in PBS populations. Figure 37. - 1 a. PCO CO •rl CO cu xi 9 .5 -.5 b. PCS CO cu T3 o CO > CO •rl CO cu Xi 4-1 a cd Xi oo •rl CU late tall early short c. PBS a cu •rl o •H 14-1 cu o o s o •H CU M U o o Ai Pi .5 late short tall x early -.5 Generations of selection Control Height , Anthesis - 124 Changes in qualitative characters A number of qualitative fruit characters were recorded in the P^ congesta lines, namely numbers and frequencies of winged and wingless fruited plants from generation to generation, numbers and frequencies of the various pubescence patterns, and numbers and frequencies of various fruit colours. The winged and wingless frequncies in the experimental populations are presented in Figure 11, and the frequencies of the pubescence patterns in Figures 7 to 10. I decided that the fruit colour scoring was too arbitrary and likely to have been subject to systematic change over the six generation period, and so have not attempted to include an analysis of this character. . There are definitely different colour morphs in this species which, like the wing shape character are fairly constant within plants and variable between, but a more reliable and objective scoring procedure needs to be devised to work with them. Winged and wingless plant frequencies There is no evidence of anything other than random drift affecting the phenotype frequencies at the fruit wing locus. There is no consistent divergence or other relationship between the plus and minus selected lines, whether selected for flowering time of height at anthesis. The increase in variance in the observed frequencies among the lines (from 0 to 0.020 in the PCO lines and from 0 to 0.043 in the PCS lines) is not significantly different from the dispersion expected to result from random drift (from 0 to 0.022 in 4 generations) (Falconer, 1960). - 125 Pubescence patterns The situation with the pubescence patterns is similar to that with the fruit wing phenotypes. There is no apparent pattern or trend in the frequencies which might be explained by or attributed to the selection procedure. In this case the increase in variance among the PCO lines appears to be slightly larger than among the PCS lines, particularly in pubescence types 2, 3, and 4, but as the genetic mechanism controlling the character is unknown, the change in dispersion cannot be compared to any theoretical prediction, as it could in the case of the fruit wing locus. As may be expected in such relatively small, inbred populations, the rarer phenotypes 0 and 1 have been lost in many of the lines. Aberrant characters The frequencies of the various individuals or characters observed are presented in Table VII. The overall trend was towards an increase in frequency compared to the frequencies observed in the base populations. The PCO late and PCS short lines had particularly high frequencies of abnormal types, and the P. congesta populations produced significantly 2 more abnormals than the P. brachystemon populations (x^£^_= 67.95, p < 0.0001). The G,_ source populations had low frequencies of aberrant types, similar to the frequencies observed in the G^ base populations, to which they are 2 comparable (X(_£_1= 0.85, p = 0.4). Comparisons between the internal control populations and the G^ source population. Table VII.. Frequencies of aberrant individuals. Cotyledons Three Fused Seedlings Chlorotic Dark Pigment PCO PCS PBS PCO PCS PBS . PCO PCS PBS PCO PCS PBS 2 3 3 3 4 7 3 1 3 4 9 2 1 5 2 1 2 1 2 6 1 5 5 9 2 2 5 7 4 3 26 12 12 2 14 ' 5 15 1 1 3 6 2 1 2 Habit Others PCO •PCS PBS PCO PCS PBS 1 2 !0 •6 15 8 1 2 11 9 12 8 11 1 3 10 PCO 452 680 •806 837 919 932 G,. source PCS 453 588 785 799 858 909 75 PBS 304 796 765 872 974 915 93 Total 1209 2004 2356 2508 2751 2756 168 PCO .020 .028 .038 .017 .034 .033 Frequency PCS PBS .007 .029 .046 .024 .024 .053 .013 .003 .008 .021 .003 .006 .019 0 Total .011 .020 .035 .014 .021 .034 .006 - 127 The two source populations, one of P. congesta and one of P. brachystemon, were grown in order to investigate whether uncontrolled selection pressures, random drift, or other processes had affected the internal control lines. Quantitative characters The statistics for the measured characters in the PC source G^ and PB source G,. populations are presented in Table VIII, along with the statistics for the G^ controls, for comparison. Plectritis congesta The P. congesta control lines did not change significantly in mean height at anthesis, number of nodes at anthesis, or number of primary branches at anthesis over the course of the experiment. There was a significant decrease in mean number of days to emergence and in days to anthesis, which may well be the result of some selective pressure for a shortened life cycle under the crowded conditions in the experimental populations. The phenotypic variance as estimated by the coefficient of variance was unchanged in all characters except the number of primary branches at anthesis, for which both control lines became significantly more variable, and flowering time, for which the PCS control line became significantly more variable. Plectritis brachystemon The P_^ brachystemon control populations developed significant Table VIII. Measured characters: G5 source populations compared with control populations. N Mean Standard Coefficien deviation variation Days to emergence PCO control G5 195 15.344 2.4198 15.77 PCS control G5 192 15.620 2.6101 16.71 PC source G,_ 74 18.554* 2.9477 15.89 PBS control G5 187 15.829 2.7519 17.39 PB source G,. 93 17.946* 2.8028 15.62 Height at anthesis (mm) PCO control G5 157 331.19 68.334 20.63 PCS control G5 165 339.05 61.979 18.28 PC source G,- 73 315.92 53.351 16.89 PBS control G5 153 524.42 84.939 16.20 PB source G<- 89 444.07* 55.570 12.51** Number of nodes at anthesis PCO control G5 159 7.9686 0.9963 12.50 PCS control G5 164 8.1829 1.4197 17.35 PC source G^ 73 7,8219 1.1942 15.27 PBS control G5 153 10.209 1.0108 9.90 PB source G,. 89 9.618* 0.9944 10.34 Number of primary branches PCO control G5 158 5.5506 3.7204 67.03 PCS control G,. 165 5.5212 3.8407 69.56 PC source G^ 73 6.6301 3.5099 52.94** PBS control G5 153 10.993 5.0945 46.34 PB source G,. 89 11.775 7.3450 62.37** Table VIII, continued. N Mean Standard Coefficient of deviation variation Days to anthesis PCO control G5 163 73.736 5.1169 6.94 PCS control G5 174 75.167 6.4217 8.54 PC source Gc 73 75.616* 4.4648 5.90** (vs. PCS) PBS control G5 160 93.331 4.6415 4.97 PB source G5 91 88.835* 5.0514 5.69 * Mean values in the source populations indicated are significantly different from the control populations at the 5% level. ** Coefficients of variation in the source populations indicated are significantly different from the control populations at the 1% level. - 130 differences from the source population over the course of the experiment for all the characters except number of primary branches. The mean number of days to emergence decreased significantly in the control line;, and the mean height at anthesis, number of nodes at. anthesis, and days to anthesis all increased significantly with respect to the source population. The variance of the PBS control was significantly increased in height at anthesis, and significantly decreased in number of primary branches at anthesis; variances for the other characters were unchanged* Correlations There were no significant correlations in the G,. source populations which suggested that there had been any change in the control line over the course of the experiment (Table IX). Table IX. Correlations in the G,. source populations. P. congesta Height at anthesis -.2163 Nodes at anthesis -.2769* 0.1698 Primary branches -.3123* 0.2056* 0.3121* Days to anthesis 0.3443* 0.0174 0.6226* Days to Height at Nodes at emergence anthesis anthesis P. brachystemon Height at anthesis Nodes at anthesis Primary branches Days to anthesis -.4984* -.2756* -.2936* 0.1290 Days to emergence 0.2987* 0.3381* -.0137 Height at anthesis 0.3864* 0.4910* Nodes at anthesis * Correlation coefficients significant at the 5% level. -.0234 Primary branches 0.0558 Primary branches - 132 Summary of results Breeding system in Plectritis Source populations Plectritis congesta had an estimated outcrossing rate of 61.6%. In Plectritis brachystemon the outcrossing rate was not estimated, but assumed to be less than 5%. Experimental populations The P. congesta outcrossed populations (PCO) had an estimated outcrossing rate of 65%. The P. congesta selfed populations (PCS) had an estimated outcrossing rate of 15%. The P. brachystemon populations ;(PBS) had an assumed outcrossing rate of less than 5%. Characteristics of the base populations The mean values for days to emergence, height at anthesis, number of nodes at anthesis, number of primary branches at anthesis, days to emergence, and flowering time were greater in the P. brachystemon populations than in the P. congesta populations. P. brachystemon was more variable for days to emergence, height at anthesis and nodes at anthesis, less variable for primary branches and flowering time. Response to selection Selected characters Height at anthesis Means The PCO lines diverged 66% or 148 mm compared to the control (+ 41%, 92mm; - 25%, 56mm); the PCS lines diverged 78% or 175 mm (+ 27%, 61 mm; - 51%, 114 mm); the PBS lines showed no divergence, but erratic fluctuations relative to the control. - 133 Variances There were no trends in the changes in variance as estimated by coefficients of variation; some populations showed significant increases, some significant decreases relative to the control. Heritabilities In PCO the realised heritability (bc) was estimated as 2 0.53; heritability from parent-offspring regressions (h ) 2 was estimated as 0.45. In PCS b = 0.58, h =0.44. In c PBS neither estimate was significantly different from zero. Components of variance Significant between family variance components were observed in most populations. There were no trends in changes in the partitioning of between / within family components over the course of the experiment. to anthesis (flowering time) Means The PCO lines diverged 33.5% or 31.8 days compared to the control (+ 20%, 19 days; - 13.5%, 12.8 days); the PCS lines diverged 28.7% or 27.3 days (+ 16.3%, 15.5 days; - 12.4%, 11.8 days); and the PBS lines diverged 18.5% or 21.5 days (+ 12.2%, 14.2 days; - 6.3%, 7.3 days). Variances The trend in all six selected lines was towards a decrease in the variance of flowering time as estimated by the coefficient of variation. Heritabilities In PCO b = 0.77, h2 = 0.60; in PCS b =0.75, h2 = 0.72; c c - 134 2 and in PBS b =0.49, h = 0.42. c Components of variance Significant between family variance components were observed in most populations. There were no trends in changes in the partitioning of between / within family components over the course of the experiment. Unselected characters Days to emergence There were no strong trends in changes in means or variances. 2 Heritability (h ) was estimated at 0.49 in PCO control line; Height at anthesis (in lines selected for flowering time) There was some divergence in the means in the P___ congesta plus and minus lines, but no trend in the means in the PBS lines. There were no trends in the changes in variances. Number of nodes at anthesis There was marked divergence in the means for all lines except PBS selected for height at anthesis. Some trend toward a decrease in variance was observed, particularly in lines 2 selected for flowering time. Heritability (h ) was estimated as 0.55 in PCO, 0.57 in PCS and 0.28 in PBS. Number of primary branches at anthesis There was some divergence in the means except for PBS selected for height at anthesis. There were erratic, changes in the means from generation to generation relative to the controls. There were no trends in changes in the variances. Heritability 2 (h ) was estimated as 0.29 in PBS control. Days to anthesis (in lines selected for height at anthesis) - 135 There was marked divergence in means and a decrease in variances in the P.. congesta lines; there were no trends in changes in means, or variances in the PBS lines. Fruit.production There were no trends in changes in means or variances in any of the lines. Correlations There were many significant correlations, both positive and negative, among the measured characters. The only significant change in correlations over the course of the experiment was the disappearance of the strong positive correlation between the two selected characters, height at anthesis and flowering time. Qualitative characters There was no evidence of anything other than random drift affecting the frequencies of the fruit wing phenotypes and fruit pubescence pattern phenotypes over the course of the experiment. There was some evidence of an increase of aberrant types, particularly in the P. congesta populations, which might be attributed to inbreeding. - 136 Discussion The questions to which this study was addressed are: 1. Is a population of inbreeding plants more or less variable genetically than a population of otherwise identical outbreeding plants with respect to quantitatively inherited characters; 2. Does the response to selection for such characters in the two populations reflect the difference; and 3. How does the genetic variability as estimated by the response to selection compare to other estimates of genetic variability in the two populations? The experimental species Plectritis congesta and P_j_ brachystemon are as nearly identical as two sexually reproducing species with respectively outcrossing and selfing breeding systems are likely to be. In both vegetative habit and habitat they are nearly impossible to distinguish. They both have chromosome numbers reported of n = 16 (Morey, .1963; Taylor and Brockman, 1966). The Mill Hill populations have the required breeding system differences, with an outcrossing rate of 61.6% in the P. congesta population, and a rate of less than 5% in the P^ brachystemon population. Genetic variability and the response to selection Direct responses Plectritis congesta outcrossed versus P_j_ congesta selfed There is no evidence of any difference between the two s.ets of congesta - 137 populations in their direct response to selection for either height at anthesis or flowering.time,, despite an estimated difference in outcrossing rate under the experimental conditions of about 50% (PCO t = 0.65, PCS t = 0.15). There was approximately equal change in the means, that is, a divergence of 66% in PCO,.78% in PCS plus and minus lines selected for height at anthesis, and of.33.5% in PCO 28.7% in PCS plus and minus lines selected for flowering time. There was no difference between the outcrossed and selfed P. congesta groups in the trends in the coefficients of variation, with erratic changes in both for the variability of height at anthesis, and a general decrease in both for the variability of flowering time. And, finally, the estimates of heritability were essentially the same for height at anthesis (bc: PCO = 0.53, PCS.= 0.58; h2: PCO =0.45, PCS-= 0.44) and for flowering time (bc: PCO = 0.77, PCS = 0.75; h2: PCO = 0.60, PCS = 0.72). It is likely that the small population sizes in the experiment (N = 200) combined with the intensity of the selection (90%) to produce a rate of inbreeding which swamped any differences in inbreeding attributable to differences in the outcrossing rates. In comparisons with the P. brachystemon populations, therefore, I will treat the P. congesta lines essentially as duplicates, and refer to them together. Plectritis congesta versus P^ brachystemon Height at anthesis There was significant difference between P_. congesta and P___ brachystemon in response to selection for height at anthesis. The plus and minus lines diverged 66% in PCO and 78% in PCS, but no consistent divergence resulted in - 138 the PBS lines, which fluctuated erratically. There is no difference between P. congesta and P. brachystemon in the changes in the phenotypic variance, which were erratic in ail three species groups. The realised heritability estimates, however, reflected the significant response in P. congesta, which resulted in a realised heritability of 55% compared to an estimated heritability of zero for P. brachystemon, which showed no response. In effect there appeared to be no genetic variance in height at anthesis available for selection in the P. brachystemon population, but enough genetic variance present in the P_^_ congesta populations to result in a significant response. It is interesting .that•• while-the-estimated genetic variance for height at anthesis is greater in P. congesta, the phenotypic variance measured in.the base populations was significantly larger in P.  brachystemon.. This anomaly could be explained by an increased phenotypic variability or plasticity in the genetically less variable P. brachystemon. There is a body of evidence to suggest that highly homozygous organisms may be phenotypically more variable as a result of their homozygosity, or conversely that heterozygosity has abuffering effect on phenotypic variability (Allard and Bradshaw, 1964; Baker, 1974; Bradshaw, 1965*, Dobzhansky and Wallace, 1953,* Falconer; 1960; Lerner, 1954; Lewontin, 1957). In a study of P. congesta, six metrical characters - height, dry weight, degree of branching, number of primary branches, number of secondary branches, and number of nodes - were all more variable in plants homozygous for either the dominant or recessive allele at the fruit wing locus, than in plants heterozygous at the same locus. Whether the plants were grown in a warm, dry environment, or in a cool, \<ret environment, this buffering effect of heterozygosis was apparent (Carey and Ganders, 1980; Carey, unpublished). In Limrianthes, Brown and Jain (1979) found that the selfed floccosa - 139 showed more phenotypic plasticity than the outcrossed L. alba. Harding et al. (1974) found similar results working with the Lupinus nanus group of subspecies, in which the more highly selfed subspecies were more variable phenotypically than the outcrossed subspecies. Finally, Avena barbata, a more highly selfed species than its relative, A. fatua, is also* more variable phenotypically, but maintains less genetic variability and responds less well to selection (Jain and Marshallj 1967, 1970). Flowering time The direct response to selection for flowering time also differed between the two species, but.the differences were not marked, and only appeared in the fifth cycle of selection. Through the fourth cycle of selection there was no appreciable difference, with PCO, PCS, and PBS lines responding equally well to selection and diverging about 20% compared to the control (Figure 14). In the fifth cycle, the P. congesta lines continued to diverge, with a final divergence of. 33.5% (PCO) and 28.7% (PCS), but the P. brachystemon populations ceased to diverge, with the final divergence being only 18.5% of the control value. The phenotypic variances (Figure 15) followed a similar pattern, with a relatively steady decrease over the first four cycles for all three species groups. The decrease continued in the fifth cycle in the case of the P-. congesta, but not in the case of P.  brachystemon, in which the phenotypic variance increased again. Finally, the realised heritability estimates for flowering time showed approximately equal heritability in the PCO and PCS .lines (0.77 and 0.75) and slightly less, but still appreciable heritability in the PBS lines (0.49). All of the evidence indicates that there is substantial genetic variance for - 140 flowering time in both Plectritis species, with somewhat more in P. congesta than in P. brachystemon. The fact that.the .response of P.  brachystemon was similar to that of P. congesta for the.early generations, and then changed abruptly in the fifth generation, may indicate that the genetic variance is organized differently in the selfed species. This is not unexpebted; since there is so little outcrossing in P. brachystemon populations, the genetic variance in these populations could be expected to derive largely from differences between families, that is, between one or a number of relatively highly homozygous lines. The selection response would then represent selection of families rather than individuals, and the depletion of the variance and tailing off of the response would occur rapidly, particularly in small populations such as those maintained in this experiment. In contrast, P. congesta could continue to maintain genetic variance between and within families, in spite of the selection pressure, by means of the recombination coming directly from outcrossing, and from segregation in subsequent generations. Confounding phenomena 1 There are a number of phenomena which could potentially confound the effects of direct selection on flowering time and height at anthesis. The first is sampling error (genetic drift), which could operate against the direction of selection in very small populations. The selective pressures used in this experiment were large enough to reduce.to insignificance the possibilities that drift could affect the characters under direct selection. There is, however, evidence.that random processes may have affected some of the unselected characters. The fruit wing phenotype and fruit pubescence - 141 pattern frequencies increased in variance among the various P___ congesta lines over the course of the experiment, and the increase in dispersion was not inconsistent with that expected to result from random genetic drift; certainly there seemed to be no connection between the changes in particular lines and the selection pressure. A second factor which might have affected the response to selection is the effect of selective forces accompanying the response to the intended artificial selection, which act to counter or reduce that response. These could include, for example, reduced viability or reduced fecundity in the selected individuals or their progeny in proportion to the degree to which they depart from the control mean. There was some evidence of a decrease in viability and fecundity in some of the extreme individuals, especially those individuals which were very short and those which flowered very late in all three species groups. Since the criteria for selection included survival to produce at least ten apparently viable fruits (except in those cases noted, that is, PCS short G3 and PBS short G3) the effect of this counter selection was reduced somewhat. There was no obvious reduction in the total population rates of germination, survival, fitness, or fruit production in the later generations, and no notable difference between P. congesta and P. brachystemon in the deleterious effects of selection. A third factor which might have affected the direct response to selection is inbreeding depression. In particular, this might be expected to have affected the outcrossed P. congesta. Inbreeding in the experimental populations, which was unavoidable with such small populations and heavy selection, might result in an increase in homozygosity which could produce relatively unfit homozygous recessive genotypes. In a selfing species like P. brachystemon these unfit genotypes would presumably have been selected - 142 against and eliminated from natural populations. If the unfit homozygotes were involved in the characters being selected, their deleterious effect would be equivalent to the counter selection mentioned above, except that the same inbreeding depression would be expected to affect the control (unselected) lines more or less equally with the selected lines. There was some evidence of inbreeding effects in the increase in frequency of aberrant individuals over the course of the experiment. Plants with abnormal numbers of cotyledons or fused cotyledons, chlorotic seedlings, excessively pigmented seedlings, and plants with other abnormalities in their habit all increased in frequency, particularly in the P. congesta populations. There was no evidence that aberrant types increased in frequency in the treatment populations any more than in the control populations, though some lines had particularly high frequencies in most generations (PCO late and PCS short). The higher frequencies noted in P. congesta as compared to P. brachystemon were to be expected, since the selfed species would be less likely to retain deleterious recessive alleles in the population. The source populations had low frequencies of aberrant types, comparable to the frequencies in the GQ populations. Indirect responses to selection Unselected characters There was considerable change in the unselected characters, some of which could be attributed to selection. As mentioned above, the fruit phenotypic frequencies in the P. congesta. lines seemed to change more in response to random drift than in response to selection pressure. In most - 143 of the metrical characters, however, there were changes in the means that could be related to the selection pressure. This could be seen most often as a divergence between the plus and minus lines, with all three plus lines (PCO, PCS, and PBS) having higher mean values than the corresponding minus lines. In many cases, the divergence between the plus and minus lines was not accompanied by a divergence of both away from the control line, that is, either the plus or the minus line in these cases was not significantly different from the control, though both were significantly different from the other selected line. Thus, days to emergence measured in each of the plus lines was greater than in the minus lines by the fifth cycle of selection, except in the PBS lines selected for height at anthesis; height at anthesis in the lines selected for flowering time was greater in the plus lines than in the minus lines except in the PBS lines; in all cases the number of nodes at anthesis was greater in the plus lines than in the minus lines; number of primary branches at anthesis was greater in plus lines than in minus lines except in the PBS lines selected for height at anthesis; and flowering time in those lines selected for height at anthesis was greater in the plus lines than in the minus lines. The measurement of fruit production was subject to a large error, and I found no consistent trends which could be attributed to selection. The fact that in general the unselected characters tended to track the selected characters is reflected in the relatively strong correlations among the characters. Thus, flowering time and number of nodes at anthesis were strongly correlated, and the coincident changes in mean values reflect this. Height at anthesis and flowering time were strongly correlated in the first three generations, which probably explains the initial divergence in flowering time observed in the plus and minus lines selected for height at - 144 anthesis. Days to emergence and flowering time were strongly correlated, as were height at anthesis and number of nodes at anthesis. If the correlations do indicate some degree of underlying genetic linkage, then the response of the unselected characters may have been due in part either to a direct link with the selected characters which were themselves responding to selection (nodes at anthesis, height at anthesis, and days to emergence correlated with flowering time in lines selected for the latter; flowering time and number of nodes at anthesis correlated with height at anthesis in lines selected for the latter). Alternatively an indirect link through one of the unselected characters could conceivably have produced the response (days to emergence correlated with flowering time in lines selected for height at anthesis). Other selection studies The results I observed in this experiment are comparable to the results in such other selection experiments as were designed similarly, that is, with mass selection in a population that has not been radically altered in its genetic characteristics by inbreeding (in outcrossed taxa) or by outcrossing (in selfed taxa), and where generations are produced by a more or less natural breeding programme. There are few reports in the literature of a lack of response to selection, largely, I suppose, because selection is successful to some degree in most cases. In the studies mentioned in the introduction (pp. 17 -20),.the 21 cases in outcrossed taxa for which a per cycle response was recorded had an average change in the mean value of the selected character of 14.8% per cycle of selection. In the selfing taxa, the 15 cases where a - 145 per cycle response, was. recorded-.had .an. average,, change in the mean of 8.3% per cycle. If the two ..Plectritis species are treated in the same way, the average per cycle change in.the -mean value of the two.selected characters was approximately 5% iri P. congesta and approximately 1% in P. brachystemon. Some of the selection studies reported in the literature used the same characters as I.did, that, is, height and flowering time, although height was not always measured at the same stage of the life cycle. In Limnanthes alba, an outcrosser, selection.for height resulted in a 6% change per cycle of selection (Jain, 1979) . Three.studies.involved selection for height in selfed taxa, with responses respectively of 4.5% per cycle in Avena.sativa (Geadelmann and Frey, 1975), 12.5% per cycle in A. fatua (Imam and Allard, 1965), and 2.8%.per cycle in' Sorghum.bicolor (Foster et al., 1980); the mean response in selfed taxa was 6.6% per cycle. Both breeding system groups had a response to selection for this character similar in degree to the response of 7.2% per cycle (+ 66% in PCO, + 78% in PCS after 5 cycles) noted in P___ congesta. Selection for flowering time has been.reported in a number of species. Among selfing taxa, in Avena sativa Geadelmann and Frey (1975) found a response to selection of 22% per cycle;.Imam and Allard (1965) noted a ; response of 20.5% per cycle in A. fatua; and in Sorghum bicolor Foster et al. (1980) found a response of 0.6% per cycle. In Brassica campestris var. brown sarson Murty et al. (1972) observed a response of 1.7% per cycle. The mean response in these taxa is about 12%.per cycle, which is considerably higher than either the 3% per cycle response in P. congesta or the 2% per cycle response in P. brachystemon. The longer.the selection continues, the lower will be the per cycle response, as the total response will decrease with the depletion of genetic variance. The two experiments with Avena, in which the - 146 per cycle response was.relatively,high, involved only.one cycle of selection, whereas the experiments with Sorghum and Bfassica involved 10 and 3 cycles respectively. Allowing for these differences between experiments, the response to selection for the two characters, height and flowering time, in both Plectritis species would seem to fall well within the range of observed responses in other plant species. To answer the first two questions posed.at the beginning of the discussion, the Plectritis brachystemon population, which is highly inbred, has significantly less genetic variability for one quantitatively inherited character, height at anthesis, than the P. congesta population, which is highly outbred but otherwise nearly identical; the difference between the two species in levels of genetic variability is reflected in the response to selection for height. In contrast, however, there is considerable genetic variability for the second quantitatively inherited character, flowering time, in both populations, and this, too, is reflected in the response to selection for this character. The levels of genetic variability and selection response for both characters in Plectritis are similar to those found in other plant species. The third question was how independent estimates of genetic variability compare to the estimates derived from the response to selection for the two quantitative characters. Independent estimates of genetic variability in Plectritis There is ample evidence (based mainly on isozyme data) to indicate - 147 that outcrossed taxa are more diverse, that is, more highly heterozygous and more polymorphic on average, .than are selfed..taxa. The evidence from the studies of Layton (1980) in P. congesta and P. brachystemon is in agreement with this. He calculated Nei's index of gene diversity within populations for the two species to be 0.22 for P. congesta and.0.06 for P. brachystemon. Ganders and Maze (unpublished)'studied fruit wing characters measured in two populations of "P^ congesta and one of P_;_ brachystemon and concluded that the variability in the outcrossed species was significantly greater than that in the selfed species. The two particular populations chosen as sources for this experiment differ in the amount of phenotypic variability shown in certain characters. The P^ congesta population at Mill Hill is polymorphic for a number of fruit characters: presence or absence of fruit wings, pubescence pattern, wing shape, and fruit colour. The P. brachystemon.population, in contrast, is monomorphic: all plants produce medium brown wingless fruits with the same pubescence pattern. The base populations, GQ, which are presumed to be.a random or unselected sample of genotypes in the source populations, also provide some evidence of differences in variability. For the measured characters the coefficients of variation were significantly different between the two species in 5 of 6 cases, with the coefficient being larger in P. congesta for the number of primary branches at anthesis and flowering time, and larger in P. brachystemon for number of days to emergence, height at anthesis, and number of nodes at anthesis. These coefficients are based on variances which are essentially phenotypic, although the environmental component has been reduced by the use of a common controlled environment. It is possible that the greater variability shown by P^ brachystemon for three characters is due - 148 to phenotypic plasticity, about which I will say more later. The final evidence, independent of the selection.experiment, for differences between the two species in.variability comes from the estimates of heritability from the parent-offspring regressions in the three control lines. For the two selected characters . (Table V) P. brachystemon has considerably less genetic variance as estimated by heritability for height at anthesis (essentially none as compared to 45% for P___ congesta), and only slightly less genetic variance for flowering time (42% as compared to 60-70% for P. congesta). For the.unselected characters (Table VI) there are too few good estimates of heritability to make a comparison between P. congesta and P. brachystemon possible, except in the case of the number of nodes at* anthesis. In this case the heritabilities are comparable to those estimated 2 for flowering time, with P. brachystemon showing less genetic variance (h = 2 0.28) than P. congesta (h = 0.55), but both species having low to intermed iate heritabilities for the character. The similarities in the heritabilities for these two characters, flowering time and number of nodes at anthesis, is not surprising considering the strong and consistent positive correlation between them, which was evident in all experimental populations (Figure-35). It is interesting that the coefficient of variation for number of nodes at anthesis was greater in P. brachystemon G^than in-the two P. congesta GQ populations, while the heritability estimates indicate that genetic-variance' is greater in P. congesta. There are similar anomalies which were noted when comparing the results of the selection experiment with the GQ coefficients of variation, and they will be discussed in the following sections. The answer to the third question posed at the beginning of the discussion, then, is that the independent evidence of genetic variability in the two - 149 species is equivocal. 'Most of., the.characters are more variable in the outcrossed P. congesta than in.-the selfed P...brachystemon. . These include the isozyme and fruit. phenotype-. polymorphisms, and the phenotypic variances in number of primary branches'and flowering time in the base populations grown in controlled,and.identical environments. These characters agree with the levels of variability in height at anthesis as observed in the response to selection. The phenotypic variances of days to emergence, height at anthesis, and number of nodes at anthesis in the GQ populations are higher in the P. brachystemon populations than in the P. congesta populations. If we assume that the environmental components are constant in both species (perhaps a tenuous assumption) then this evidence would indicate that there was more genetic variability in the selfer than in the outcrosser. None of the independent evidence agrees closely with the observed levels of approximately equalgenetic variance in..both species in response to selection for flowering time. The effects of breeding system on the population genetic structure of Plectritis The two species of Plectritis are very similar in habit and habitat, but are well distinguished by their breeding biology and population genetic structure. Isozyme studies show that the genetic diversity is much greater in total and within populations in P. congesta than in P^ brachystemon; most of the genetic diversity in P. brachystemon is between populations (Laytori, 1980).. Similarly, the genetic diversity in the fruit phenotypic characters - presence of fruit wings (for which the genetics are known), pubescence pattern, wing shape, and fruit colour (for which the genetics are - 150 not known) - is greater in P. congesta populations than in P. brachystemon populations. With respect to the fruit phenotype characters, the Mill Hill populations of Plectritis are typical, that is, P. congesta is relatively highly polymorphic and P. brachystemon is monomorphic. The isozyme patterns in these particular populations have not yet been studied. A largely winged-fruited population of P__ brachystemon was less variable for fruit wing characters than two comparable winged populations of P. congesta; this comparison dealt with the phenotypic variability of the characters that were measured, but the particular characters are certain to have a small or negligible environmental component (Ganders and Maze, unpublished). The phenotypic variability of the two Mill Hill populations as estimated by the variances in the G^ populations was greater for P. brachystemon in some cases and for P. congesta in others. In the cases where the genotypic component of this variance could be estimated, either by parent-offspring regressions or by the response to selection, P___ congesta was the more variable taxon genetically, although significant genetic variance was indicated for P. brachystemon in number of nodes at anthesis and flowering time. In two cases, height at anthesis and number of nodes at anthesis, P. brachystemon was more variable phenotypically but less variable genetically than P. congesta, a result which can only be explained by postulating a relatively higher element of phenotypic plasticity in P___ brachystemon, at least under the experimental conditions. Plectritis brachystemon Plectritis brachystemon, a highly selfed species, appears to have lost - 151 genetic diversity as a consequence of having evolved a selfing breeding system. It has not lost all of its genetic variability for some characters, however, and there presumably has been some selection pressure to maintain genetic variance in flowering time and number of nodes at anthesis, while genetic variance for height at anthesis has been lost. The difference between the low genetic diversity as estimated from isozymes and fruit characters, and the presence of significant genetic variance in other characters can be explained in a number of ways. Occasional outcrossing events which occur in habitually selfing populations will continue to segregate heterozygous individuals for a number, of generations, subject to the forces of random drift, inbreeding, and selection. In this case the low genetic diversity in the isozymes and low genetic variance in height at ( anthesis could be the result of selection acting on these characters, reducing their variability. Conversely, the high genetic variance in flowering time and number of nodes could represent the level of variability maintained by occasional outcrossing and segregation. This explanation requires that the genetic determinants of flowering time and number of nodes be nearly neutral, and that the outcrossing rate be high enough and population sizes large enough to permit accumulation of these heterozygous types. None of these assumptions is reasonable for P. brachystemon populations. Alternatively, the low genetic diversity in isozymes and low genetic variance in height at anthesis could be mainly the result of the high levels of inbreeding, small population size, and random loss of alleles. The relatively high genetic variance in flowering time and number of nodes could be the result of selection for variability. The balance of selective and random forces acting on isozymes is still a subject of much debate, but there - 152 is probably selection acting on at least some of the loci, and there is likely to be selection affecting height at anthesis to some degree in natural populations. It is likely that the actual situation in P. brachystemon populations is somewhere between the two extremes. There is probably some selection acting on isozymes and height, but the low levels of genetic ; variability in these traits relative to P. congesta reflect largely the effects of breeding system and random loss of variability. The relatively high genetic variance in flowering time and nodes at anthesis is probably maintained by selection. Selective forces which maintain variability involve a number of mechan isms. , Those which involve mutation, migration, or disassortative mating as sources of new variability are not likely to be the major forces involved in the short term in P. brachystemon populations. A form of selection which is more plausible in this light is heterozygote advantage. Heterozygote advantage can be used in a narrow sense, referring to an absolute fitness advantage of a heterozygote over homozygotes under all conditions, or in a broad sense, where the net fitness advantage of the heterozygote only appears as the fitness values of the various.genotypes change in time or space. Heterozygote advantage has been demonstrated to operate in P. congesta under some conditions at the fruit wing locus (Carey and Ganders, 1980) and may be a factor in preserving some of the polymorphism in that species. In P. brachystemon, however, there are probably too few heterozygotes produced in natural populations to permit heterozygote advantage to be a factor in the maintenance of genetic variance. The frequency of heterozygotes at the fruit wing locus in one population was estimated to be less than 3% (Ganders et al., 1977$; the frequency of heterozygotes observed at polymorphic isozyme loci in P___ brachystemon - 153 populations averaged 0.45% (Layton, 1980). There is some evidence that heterozygote advantage may be operating in some selfing species to maintain variance (Allard, Jain and Workman, 1968; Jain and Allard, 1960). All of this evidence comes from characters in which the genotypes can be observed (monogenic or simply inherited characters). Maintenance of genetic variance in quantitatively inherited characters by heterozygote advantage in a selfing species requires even higher rates of production of the hetero-zygotes than those required to maintain monogenic polymorphisms. The most likely form of selection to maintain genetic variability in P_j_ brachystemon populations is some type of patterning of the selective pressures either in time or in space. If microhabitats and their associated selection pressures occur, patchily or form a mosaic, then the segregating lines which result from the rare outcrossing events in P^ brachystemon populations could be maintained, highly homozygous within each microhabitat but with different genotypes in different microhabitats. That genetic variance is divided among a number of homozygous lines in jP^ brachystemon is also suggested by the observed form of the response to selection for flowering time. A relatively large response in the early generations, followed by a rapid tailing off in the last generation, could occur as the selected lines were reduced in number and variance between the lines was no longer available for selection. It is possible that an examination of a number of the isozyme loci simultaneously in individuals from P. brachystemon populations would reveal a number of lines of different multilocus homozygotes; the levels of isozyme polymorphism discovered to date in populations would allow for a maximum of 11 different homozygous lines per population at 13 isozyme loci (Layton, 1980). The isozyme data in Plectritis have not:been analysed in this way to determine multilocus genotypes, but there is evidence from other - 154 predominantly selfing species for a population structure composed of lines of homozygous genotypes. In Avena barbata a number of studies have indicated that in some populations a few multilocus genotypes are present in excess over the expected frequency, and appear to be adapted to different microhabitats within the habitat (Allard et al., 1972; Clegg and Allard, 1972)* Multilocus organization in A__ fatua and Festuca microstachys populations has also been studied and a similar situation noted (Allard, 1975). The question that follows logically from the observation of genetic variability in P___ brachystemon is why there might be such relatively strong selection to maintain variability in flowering time and number of nodes at anthesis in the species, to the point that it is nearly as variable as P. congesta. Since P. brachystemon is highly self-pollinated, there is not likely to be a connection with pollinator behaviour as might be the case in the outcrossed species. In P. congesta, variability in flowering time (and number of nodes, which are correlated) could serve to extend the flowering period in the population, thus decreasing the chance of poor fruit set due to lack of synchrony with the activity of particular pollinators. This is particularly likely in a species like P.,congesta which does not appear to rely on one major pollinator, but rather a number of pollinators. One possible explanation for the advantage to be gained from variability in flowering time in P___ brachystemon is that it reflects variability in some physiological or developmental character which is subject to multiniche selection. Alternatively, flowering time may be directly subject to such disruptive selection. For example, it is possible that in microsites which dry early in the season, early flowering is selectively advantageous, while in wetter microsites, delaying flowering may maximize fecundity. - 155 In contrast to the genetic variability present in flowering time are the relatively higher levels of phenotypic variability for height at anthesis observed in J?^ brachystemon, compared to those in P_^ congesta. This plasticity may be advantageous in populations exposed to a mosaic of selective forces, or it may merely be a non-adaptive side effect of the increased homozygosity which has accompanied the evolution of a highly selfed breeding system. If homozygous genotypes are by nature more plastic, there may have been no way for brachystemon to avoid the increased plasticity were it to have proved disadvantageous. Phenotypic plasticity could be adaptive in a number of ways. It could make phenotypically uniform a population of plants which are genetically diverse, maintaining genetic variability in the face of stabilizing selection. Bradshaw (1965) has likened this to the effects of dominance and gives examples in which it may be operating, namely in Plantago maritima (Gregor,1956) and a number of species examined by Turesson (1922, 1925). Alternatively, phenotypic plasticity could allow phenotypic variability in a population of plants which are genotypically relatively uniform. This is the potential which has been attributed to some hypothetical weedy or colonizing species, which would be composed of one or more general-purpose genotypes, at once highly homozygous and highly plastic (Baker, 1974). There are weedy species which do exhibit more phenotypic variability in some characters than their non-weedy relatives. Examples are Sonchus oleraceus (weedy) versus S.  arvensis (Lewin, 1948) and Chenopodium album (weedy) versus C. rubrum (Cumming, 1959). Similarly there are other species pairs in which the more genetically variable species is less phenotypically variable (species of Limnanthes, Lupinus, and Avena mentioned on pages 138 and 139). It is unfortunate that there are no measurements of phenotypic variance yet - 156 available for natural populations of P___ congesta and P. brachystemon, so it is not possible to compare levels of variability under natural and uniform (growth chamber) conditions, nor to estimate the extent of phenotypic plasticity in nature. The relatively large differentiation between populations of P. brachystemon in isozyme patterns (Layton, 1980), the presence of significant genetic variance in some characters, and the fact that neither P. congesta nor P. brachystemon is particularly aggressive or weedy, suggest that the notion of a species composed of general-purpose, highly homozygous, highly plastic genotypes which has been suggested for some selfing species is not an adequate description of the situation in P. brachystemon. Occasional outcrossing and multiniche selection among the segregating lines for the genetic determinants of flowering time are' likely„to have been important,adaptive processes - in P. brachystemon. It is not possible at the moment to say whether the increased phenotypic variability in height at anthesis in P. brachystemon is or is not adaptive, or whether low genetic variability and high phenotypic variability for height in P___ brachystemon and high genetic variability and low phenotypic variability for height in P.  congesta are -different solutions to the same evolutionary problem. Plectritis congesta The presence of genetic variability in flowering time, height at : anthesis, and number of nodes at anthesis in P. congesta is easy to account for as a consequence of the processes of recombination and segregation accompanying the habitual outcrossing in the species. It is impossible to say whether the levels of variability so maintained are more or less than - 157 expected. Since the habitats are similar, presumably the same type of multiniche selective pressures postulated to be operating in P. brachystemon would also affect P. congesta. Populations of the latter, however, would not be able to develop genetically differentiated local subdividions as easily, because of the mixing effect of outcrossing. In addition, unlike P. brachystemon, P. congesta has a pollination biology which could theoretically support selection for greater variability in flowering time in response to pollinator behaviour, to increase the reliability of fruit set; the origin and maintenance of genetic variability in flowering time is easily explained in 7^ congesta. Heterozygote advantage has been shown to operate to maintain at least one of the polymorphisms in P. congesta populations, and inbreeding depression was more marked in the experimental P^_ congesta populations than in the P.  brachystemon populations. Given no evidence to the contrary; one may conclude that the higher levels of genetic variability and the outcrossed breeding system which generates them are both selectively advantageous for P. congesta. Further study in Plectritis The differences between the two Plectritis species in breeding system have definitely affected the genetic structure of their populations, inbreeding in P^ brachystemon having decreased the genetic diversity of isozyme and fruit morph characters and the genetic variance in height. Yet it is obvious from the genetic variance in flowering time and the phenotypic variance in height and number of nodes that brachystemon has not sacrificed all of its sources of variability to its breeding system. - 158 More study of both species is warranted, to clarify the differences and to answer what may be the fundamental evolutionary and systematic question, namely, how and why P. brachystemon evolved a selfing breeding system at some point in the past, and how and why both species are equally successful in sympatry at present. Experiments with P. brachystemon could be designed to determine whether this taxon does in fact have a population genetic structure affected by a number of different microhabitats and multiniche selection. First, multilocus isozyme genotypes could easily be determined for a population. Since 13 isozyme loci have already been studied, and expansion of the number of loci available for characterization is quite feasible (Layton, personal communication), one should be able to differentiate a number of multilocus genotypes, if present, even within a relatively invariable P.  brachystemon population. If microhabitat patterning is a factor in a population, it might be possible to detect at least some of its dimensions with suitable measurements of physical factors (soil, moisture, micro-topography) and biotic factors (correlated plant species), and relate the multilocus isozyme patterns to particular microhabitats, as has been done with other species (Allard et al., 1972; Allard, 1975). Since comparisons with sympatric P. congesta populations are possible, one could do the same thing with that species, to verify whether the effect of oucrossing has been to produce the expected larger number of isozyme genotypes, but without any subdivided pattern based on microhabitat differences. One of the most interesting experiments would involve measurements of the phenotypic variance in natural populations of both species for the characters used in this experiment, in order to estimate the naturally occurring levels of phenotypic plasticity. For example, under experimental - 159 conditions V\_ congesta was genetically more variable for height than was P. brachystemon, but phenotypically less variable. One could measure the phenotypic variance for height in natural populations to see if the different levels of genetic variability translate/into similar or dissimilar levels of phenotypic variability under natural conditions. As a final example, one could extend the studies to other species in the genus, which because of their floral morphology are presumed to be more highly selfed than P^ congesta, and to other sections of the ranges of P. congesta and P_j_ brachystemon. There are reports of populations of the two species in;the more southerly parts of their ranges which are more similar to each other than are any that have been studied in British Columbia (Morey, 1962). This appears to be the case in some populations observed recently in California, and it would be interesting to see how extensive the apparent intermediacy is, in terms of breeding system and population genetics and biology. Jain ,et al. (1970) have pointed out the danger in extending arguments from studies of a population in one area to other species or other populations in the same species; populations of P. congesta in the southern part of the range appear to be far less polymorphic within populations for some fruit characters than are those in British Columbia. Implications for other studies The estimation and comparison of levels and organization of genetic variability in relation to breeding systems in various organisms is an active area of study among population geneticists. In this field it is important to distinguish between potential or hidden genetic variability - 160 (genetic diversity) and realised or free genetic variability (genetic variance). In the two Plectritis species studied here, a reasonably good relationship was observed between the breeding system, the amount of genetic diversity, and the level of genetic variance present in a population. In comparison to the outcrossed P. congesta, the selfed P. brachystemon has less genetic diversity and less genetic variance.. There is, however, still considerable genetic variance in the Pv brachystemon for at least one character, flowering time, perhaps more than one would expect from such a highly selfed organism, or on the basis of the lack of diversity in the isozymes. Moreover, the levels of phenotypic variability measured in some of the characters indicate more variability in P. brachystemon than in P.  congesta; this may be misleading, however, as in at least one case (height at anthesis) the difference is caused by high phenotypic plasticity rather than genetic variability. In studies where isozymes are being examined, it is important that a lack of genetic diversity in isozymes not be interpreted to indicate a general lack of genetic variance in all characters, without other supporting evidence. Isozymes have so far proved to be good indicators of relative levels of variability, for instance between outcrossers and selfers in general; nevertheless, amounts of genetic variance which are significant from an evolutionary point of view can be maintained along with isozyme monomorphism. Isozymes will and should continue to be used to estimate population genetic parameters because they are more nearly estimates of the genotype than are characters which have a large environmental component to their variation. Allele frequencies at many individual loci can be examined, and the sampling is relatively fast, easy, and precise. Their contribution to the phenotype, other than their electrophoretic mobility, - 161 is often unknown, and their selective values are correspondingly difficult to interpret. When studies are directed to variability of quantitative characters in populations, the roles of genetic variance and phenotypic plasticity or environmental variance should be assessed carefully. Lacking evidence as to the genetic components of variance, it may be tempting to argue that an outcrossed population which is phenotypically more variable than a selfed population is expressing higher levels of genetic variance. By the same reasoning, a selfed population that is more variable than an outcrossed one might be regarded as exhibiting characteristically higher levels of phenotypic plasticity. Some of the earlier descriptions of increased phenotypic plasticity in weedy or selfing plant species have been based on observations of phenotypic variance in comparison with related non-weedy or outcrossing species, and are not accompanied by an estimate of the genetic component of the variance (for example Cumming, 1959 and Lewin, 1948). Later studies, such as those with Lupinus (Harding et al., 1974), Limnanthes (Brown and Jain, 1979),and Avena (Jain and Marshall, 1970), have estimated the genetic components; difficult as they may be to obtain, such estimates do a great deal to increase our confidence in speculations about the contribution of genetic variance vs. phenotypic plasticity in strategies of populations of various species. Selection experiments may: not be the fastest way to study the genetic components of variation in quantitatively inherited characters. It may be faster to do large scale breeding experiments of the type which are common in crop science and agricultural research (various controlled crossing methods and progeny testing methods). Selection experiments have the advantage of smaller space requirements, and consequently better control - 162 of the environmental heterogeneity may be possible. Estimates of genetic variance in quantitative characters are an invaluable addition to electrophoretic studies, if the population genetics or biology of a plant are to be fully understood. They reflect a different and in some ways more biologically meaningful portion of the phenotype than do isozymes. - 163 Literature Cited Allard, R. W. 1975. The mating system and microevolution. Genetics 79: 115 - 126. Allard, R. W., G. R. Babbel, M. T. Clegg, and A. L. Kahler. 1972. Evidence for coadaptation in Avena barbata. Proc. Natl. Acad. Sci. U. S. 69(10): 3043 - 3048. Allard, R. W. and A. D. Bradshaw. 1974. The implications of genotype-environment interactions in applied plant breeding. Crop Sci. 4: 503 - 508. Allard, R. W., S. K. Jain, and P. L. Workman. 1968. The genetics of inbreeding populations. Adv. in Genetics 14: 55 - 131. Arboleda-Rivera, F. and W. A. Compton. 1974. Differential response of maize (Zea mays L.) to mass selection in diverse selection environments. Theor. Appl. Genet. 44: 77-81. Baker, H. G. 1953. Race formation and reproductive method in flowering plants. Symposia of the Soc. for Exp. Biology 7: 114 - 145. 1974. The evolution of weeds. Ann. Rev. Ecol. Syst. 5: 1-24. Barnes, D. K., C. H. Hanson, F. I. Frosheiser, and L. J. Elling. 1971. Recurrent selection for bacterial wilt resistance in alfalfa. Crop Sci. 11: 545 - 546. Bradshaw, A. D. 1965. Evolutionary significance of phenotypic plasticity in plants. Adv. in Genetics 13: 115 - 155. Brown, C. R. and S. K. Jain. 1979. Reproductive system and pattern of genetic variation in two Limnanthes species. Theor. Appl. Genet. 54: 181 - 190. Busbice, T. H., R. Y. Gurgis, and H. B. Collins. 1975. Effect of selection for self-fertility and self-sterility in alfalfa and related characters. Crop Sci. 15: 471 - 475. Carey, K. and F. R. Ganders. 1980. Heterozygote advantage at the fruit wing locus in Plectritis congesta (Valerianaceae). Evolution 34(3): 601 - 607. Chandhanamutta, P. and K. J. Frey. 1973. Indirect mass selection for grain yield in oat populations. Crop Sci. 13: 470 - 473. Clegg, M. T. and R. W. Allard. 1972. Patterns of genetic differentiation in the slender wild oats species, Avena barbata. Proc. Natl. Acad. Sci. U. S. 69: 1820 - 1824. - 164 Cortez-Mendoza, H. and A. R. Hallauer. 1979. Divergent mass selection for ear length in maize. Crop Sci. 19(2): 175 - 178. Cumming, B. G. 1959. Extreme sensitivity of germination and photoperiodic reaction in the genus Chenopodium (Tou*r\) L. Nature (London) 184: 1044. Devine, T. E., C. H. Hanson, S. A. Ostazelski, and T. A. Campbell. 1971. Selection for resistance to anthracnose (Colletotrichum trifolii) in four alfalfa populations. Crop Sci. 11: 854 - 855. Dobzhansky, Th. and B. Wallace. 1953. The genetics of homeostasis in Drosophila. Proc. Natl. Acad. Sci. U. S. 39: 162 - 171. Dudley, J. W., R. J. Lambert, and D. E. Alexander. 1974. Seventy generations of selection for oil and protein concentration in the maize kernel. In "Seventy generations of selection for oil and protein in maize." Crop Sci. Soc. of Amer., Inc., Madison, WI. Ellstrand, N. C. and D. A. Levin. 1980. Recombination system and population structure in Oenothera. Evolution 34(5): 923 - 933. Falconer, D. S. 1960. "Introduction to Quantitative Genetics". The Ronald Press Co. New York. Fehr, W. R. and C. R. Weber. 1968. Mass selection by seed size and specific gravity in soybean populations. Crop Sci. 8: 551 - 554. Foster, K. W., S. K. Jain, and D. G. Smeltzer. 1980. Responses to 10 cycles of mass selection in an inbred population of grain sorghum. Crop Sci. 20:- 1-4. Fox, D. J. and K. E. Guire. 1976. "Documentation for MIDAS". Statistical Research Laboratory, the University of Michigan. Ganders, F. R., K. Carey, and A. J.. F. Griffiths. 1977. Natural selection for a fruit dimorphism in Plectritis congesta (Valerianaceae). Evolution 31: 873 - 881. 1977. Outcrossing rates in natural populations of Plectritis brachystemon (Valerianaceae). Can. J. Bot. 55: 2070 - 2074. Gardner, C. 0. 1961. An evaluation of effects of mass selection and seed irradiation with thermal neutrons on yield of corn. Crop Sci. 1: 241 - 245. Geadelmann, J. L., and K. J. Frey. 1975. Direct and indirect mass selection for grain yield in bulk oat populations. Crop Sci; 15: 490 - 494. Gottlieb, L. D. 1973. Genetic confirmation of the origin of Clarkia  lingulata. Evolution 27: 205 - 214. - 165 Gottlieb, L. D. 1977. Phenotypic variation in Stephanomeria exigua ssp coronaria (Compositae) and its recent derivative species "Malheurensis". Amer. J. Bot. 64(7): 873 - 880. Gregor, J. W. 1956. Genotypic-environmental interaction and its bearing on a practical problem of international interest. Proc. Inter. Grassland Congr., 7th, Palmerston, New Zealand. Pp. 202 - 211. New Zealand Senior Trade Commissioner in the U. S. A. Washington, D. C. Hamrick, J. L.,' Y. B. Linhart, and J. B. Mitton. 1979. Relationships between life history characteristics and electrophoretically detectable genetic variation in plants. Ann. Rev. Syst. Ecol. 10: 173 - 200. Harding, J., C. B. Mankinen, and M. H. Elliott. 1974. Genetics of Lupinus. VII. Outcrossing, autofertility, and variability in natural populations of the nanus group. Taxon 23: 729 - 738. Hill, W. G. 1972. Estimation of realised heritabilities from selection experiments. I. Divergent selection. Biometrics 28: 747 - 765. Hillel, J., M. W. Feldman, and G. Simchen. 1973. Mating system and population structure in two closely related species of the wheat group. I. Variation between and within populations. Heredity 30(2): 141 - 167. Hilu, K. W., and J. M. J. deWet. 1980. Effect of artificial selection on grain dormancy in Eleusine (Gramineae). Syst. Bot. 5(1): 54 - 60. Hitchcock, C , L., and A. Cronquist. 1973. "Flora of the Pacific Northwest". University of Washington Press, Seattle. Imam, A. G., and R. W. Allard. 1965. Population studies in predominantly self-pollinated species. VI. Genetic variability between and within natural populations of wild oats from differing habitats in California. Genetics 51: 49-62. Jain, S. K. 1979. Response to mass selection for flowering time in meadowfoam. Crop Sci. 19: 337 - 339. Jain, S. K., and R. W. Allard. 1960. Population studies in predominantly self-pollinated species. I. Evidence for heterozygote advantage in a closed population of barley. Proc. Natl. Acad. Sci. U. S. 46: 1371 - 1377. Jain, S. K., and D. R. Marshall. 1967. self-pollinating species. X. of Avena fatua and A. barbata. Population studies in predominantly Variation in natural populations Amer. Nat. 101: 19 - 33. 1970. Within-family selection in Avena fatua and A. barbata. Theor. Appl. Genet. 40: 73 - 75-. - 166 Jain, S. K., D. R. Marshall, and K. Wu. 1970. Genetic variability in natural populations of softchess (Bromus mollis L.). Evolution 24: 649 - 659. Jones, A. 1972. Mass selection for low oxidation in sweetpotato. J. Amer. Soc. Hort. Sci. 97: 714 - 718. Jones, A., P. D. Dukes, and F. P. Cuthbert, Jr.. 1976. Mass selection in sweet potato: breeding for resistance to insects and diseases and for horticultural characteristics. J. Amer. Soc. Hort. Sci. 101:. 701 - 704. Josefsson, A. 1963. Effects of selection in fodder beets. In "Recent Plant Breeding Research". E. Akerberg, A. Hagberg, G. Olsson, and 0. Tedin, eds. John Wiley & Sons, New York. Kannenberg, L. W., and R. W. Allard. 1967. Population studies in predominantly self-pollinated species. VIII. Genetic variability in the Festuca microstachys complex. Evolution 21: 227 - 240. Lande, R. 1977. The influence of the mating system on the maintenance of genetic variability in polygenic characters. Genetics 86: 485 - 498. Layton, C. L. 1980. The genetic consequences of contrasting breeding systems in Plectritis. M. Sc. Thesis, University of British Columbia, Vancouver, B.C., Canada. Lerner, I. M. 1954. "Genetic Homeostasis". Oliver and Boyd, Edinburgh and London. Levin, D. A. 1975. Genetic consequences of translocation heterozygosity in plants. Bioscience 25: 724 - 728. 1978. Genie heterozygosity and protein polymorphism among local populations of Oenothera biennis. Evolution 32(2): 245 - 263. Lewin, R. A. 1948. Biological flora of the British Isles. Sonchus  oleraceus L. J. Ecol. 36: 204 - 216. Lewontin, R. C. 1957. The adaptation of populations to varying environments. Cold Spring Harbor Symp. Quant. Biol. 22: 395 - 408. 1966. On the statistical measurement of relative variability. Syst. Zool. 15: 140 - 141. Marshall, D. R., and S. K. Jain. 1969. Genetic polymorphism in natural populations of Avena f atua and _A;_ barbata. Nature (London) 221: 276 - 278. Matzinger, D. F., and E. A. Wernsman. 1968. Four cycles of mass selection in a synthetic variety of an autogamous species Nicotiana tabacum L. Crop Sci. 8: 239 - 243. - 167 McConnell, R. L., and C. 0. Gardner. 1979. Selection for cold germination in two corn populations. Crop Sci. 19: 765 - 768. Morey, D. H. 1962. The biosystematics of the genus Plectritis. Ph. D. Thesis, Stanford University, Stanford, CA. Murty, B. R., V. Arunachalam, P. C. Doloi, and J. Ram. 1972. Effects of disruptive selection for flowering time in Brassica campestris var brown sarson. Heredity 29.: 287 - 295. Olsson, G., and G. Andersson. 1963. Selection for oil content in cruciferous plants. In "Recent Plant Breeding Research". E. Akerberg, A. Hagberg, G. Olsson, and 0. Tedin, eds. John Wiley & Sons, New York. Pedersen, M. W., B. Berrang, M. E. Wall, and K. H. Davis, Jr. 1973. Modification of saponin characteristics of alfalfa by selection. Crop Sci. 13: 731 - 735. Rick, C. M., and J. F. Fobes. 1975. Allozymes of Galapagos tomatoes: polymorphism, geographic distribution, and affinities. Evolution 29: 443 - 457. Rick, C. M., J. F. Fobes, and M. Holle. 1977. Genetic variation in Lycopersicon pimpinelljfolium: evidence of evolutionary change in mating systems. Plant Syst. Evol. 127: 139 - 170. Rogers, S. 1971. Studies on British poppies. 4. Some aspects of variability in the British species of Papaver and their relation to breeding mechanisms and ecology. Watsonia 8: 263 - 276. Romero, G. E., and K. J. Frey. 1966. Mass selection for plant height in oat populations. Crop Sci. 6: 283 - 287. Schaaf, H. M. 1968. Phenotypic selection in crested wheatgrass. Crop Sci. 8(6): 643 - 647. Solbrig, 0. T. 1972. Breeding system and genetic variation in Leavenworthia. Evolution 26: 155 - 160. Strid, A. 1970. Studies in the Aegean flora. XVI. Biosystematics of the Nigella aryensis complex with special reference to the problem of non-adaptive radiation. Opera Botanica 28: 5 - 169. Taylor, R. L., and R. P. Brockman. 1966. Chromosome numbers of some western Canadian plants. Can. J. Bot. 44(8): 1093 - 1103. Turesson, G. 1922. The genotypical response of the plant species to the habitat. Hereditas 3: 211 - 350. Turesson, G. • 1925. The plant species in relation to habitat and climate. Hereditas 6: 147 - 236. - 168 Villegas, C. T., C. P. Wilsie, and K. J. Frey. 1971. Recurrent selection for high self-fertility in vernal alfalfa (Medicagd sativa L.). Crop Sci. 11: 881 - 883. Zuber, M. S., M. L. Fairchild, A. J. Keaster, V. L. Fergason, G. F. Krause, E. Hildebrand, and P. J. Loesch, Jr.. 1971. Evaluation of 10 generations of mass selection for corn earworm resistance. Crop Sci. 11: 16 - 18. Between family / within family variance ratios in experimental populations. a. PCO populations. Treatment Generation Control Early anthesis Late anthesis Short height Tall height rs to .Height at Nodes at Primary Days to Fruit irgence anthesis anthesis branches anthesis production 1.02* 1.82 2.30 1.31* 2.15 1.40* 3.13 10.82 5.13 4.16 8.52 2.93 3.82 5.71 3.21 1.28* 6.63 1.53* 3.11 9.44 3.91 3.82 6.36 2.39 6.34 3.85 1.93 1.89 2.54 1.54* 2.53 2.99 1.87 1.40* 1.01* 2.43 7.05 3.74 7.44 2.74 2.34 5.82 2.60 4.68 1.63* 1.86 2.22 2.89 6.61 3.71 3.52 3.11 1.79 2.99 6.63 4.49 3.15 2.38 1.40* 4.58 2.23 1.75 2.07 1.24* 1.91 3.43 4.01 1.37* 3.11 1.74 2.35 8.14 2.56 2.73 2.51 1.12* 4.00 3.36 4.76 2.14 3.95 3.45 3.22 5.30 3.75 3.18 3.35 2.66 3.49 2.90 2.56 6.44 7.30 1.98 2.66 3.11 2.31 4.72 2.25 1.03* 2.87 5.31 3.20 4.54 2.37 1.79 4.09 5.22 1.58* 4.41 1.54* 3.61 5.43 4.83 7.84 4.47 4.25 1.06* 2.26 2.75 1.59* 1.04* 3.40 4.72 1.87 2.28 2.97 2.38 6.04 3.33 3.58 0.75* 2.69 1.40* 2.91 4.17 6.14 2.73 5.15 1.57* 2.86 1.80 5.86 3.60 2.95 b. PCS populations. Treatment Generation Days to Height at Nodes at Primary Days to Fruit Control G1 G2 G3 G4  G5 Early G1 anthesis G G3 G5 Late Gj^ anthesis G„ G3 G4 Short G height G 4 G5 Tall G height G G3 G4  G5 rgence anthesis anthesis branches anthesis production 1.68 * 4.99 2.93 1.73 3.36 0.86 * 2.92 6.03 9.97 3.09 9.76 2.40 6.16 3.07 6.61 1.70 6.96 1.85 3.75 5.99 11.95 6.66 11.53 1.57 * 3.17 5.37 6.92 2.73 6.09 1.77 2.28 2.59 1.95 2.84 1.44 * 2.20 5.07 4.00 6.78 5.34 3.91 9.78 3.71 3.35 3.38 4.64 2.15 1.65 * 1.34 * 3.94 2.73 4.05 1.91 3.02 7.01 6.58 1.94 6.73 1.51 * 4.41 3.39 1.17 * 2.59 1.17* 3.62 5.40 4.50 1.65 * 1.87 2.48 5.27 2.91 5.24 6.99 2.13 1.24 * 2.60 6.96 4.45 2.74 2.51 3.48 2.00 5.14 6.43 4.22 3.66 2.52 1.57 7.50 4.63 6.20 1.23 * 2.04 2.67 6.06 2.74 4.97 4.59 3.96 5.39 8.40 6.44 8.88 2.26 1.60 * 5.83 15.09 4.24 6.59 1.17 * 3.41 3.74 2.48 6.88 3.72 1.17 * 3.06 5.53 2.54 5.40 3.47 2.42 3.83 10.71 2.44 4.80 1.13 * 1.87 7.53 6.24 4.63 4.35 2.06 1.63 * 4.22 4.00 2.12 4.40 c. PBS populations. Treatment Generation Days to Height at Nodes at Primary Control Days to Fruit Early anthesis Late anthesis Short height Tall height rgence anthesis anthesis branches anthesis production 2.15 4.66 3.23 4.19 4.47 1.36* 2.34 21.49 5.18 10.69 3.57 3.29 3.09 5.13 3.64 4.83 6.53 2.19 2.27 2.81 3.18 4.76 5.60 3.03 1.33* 2.42 1.88 2.59 1.39* 0.71* 1.19 2.81 1.36 * 2.58 0.70* 4.03 5.93 2.98 2.84 6.07 2.85 1.67 2.55 2.15 1.92 2.04 1.94 3.95 13.17 3.08 3.69 3.67 4.86 1.59* 1.95* 1.90* 3.43 1.44* 1.32* 7.91 1.84 8.63 3.19 3.39 1.90 4.41 2.17 1.34* 1.31* 2.06 4.00 3.92 2.12 6.01 3.71 2.08 1.68 2.88 2.60 1.95 3.27 4.46 1.59* 2.09 1.36* 1.71* 4.82 2.60 1.86 5.16 4.63 5.08 2.53 5.33 16.20 4.37 10.36 5.72 4.96 4.57 1.57* 5.06 2.30 3.66 1.12* 2.79 1.39* 1.68 1.02* 1.91 2.47 4.59 4.31 4.72 5.57 3.97 1.78 8.66 2.78 5.04 1.20* 1.92 1.54* 1.47* 4.30 2.16 2.80 1.38* 2.86 5.85 8.57 3.58 10.22 3.02 2.28 3.49 1.63* 2.88 10.57 Coefficients of variation, unselected characters. -• . «d PCO PCS PBS § a. Early Late Short' Tall Early Late Short Tall Early Late Short Tall ' to' Days to emergence GQ (Control) 22.7 22.7 27.17 G ' 18.09 20.29 17.63 22.18 26.05 21.3 18.58 17.46 37.95 32.32 24.04 19.42 G2 32.47 42.03 35.36 31.79 , 30.89 30.11 35.78 32.73 17.64. 20.58 37.59 39.42 G3 18.85 26.09 35.10 25.66 21.62 28.01 27.43 24.90 29.82 28.07 26.31 28.21; G. 24.38 27.48 24.86 23.32 24.4 16.98 39.76 22.38 35.73 37.95 39.76 32.36 4 - ... G5 ' Height at anthesis GQ (Control) 20.16 20.16 23.97 -Gl 17.31. 18.13 17.39 18.46 138.76 28.15 G2 16.53 17.62 .22.23 25.13 21.81 29.62 G3 14.05 20.54 15.41 20.95 19.5 22.2 G. 17.29 16.53 17.52 21.57 "30.11 19.9 4 • G5 to Appendix 2, continued. - 173 pq co cj O a cfl H M o xi to QJ 4J cfl rH W Cfl W cd H H O XI CO CO r4 H W CO H 4J M O XI cu co to w ro O ro CO cn rH •H cn cu xi • 4-1 ' C ^ cd O H 4J 4-1 cd C o • CO . C_J OJ ' o c .53 CJ VO CM in 00 rH • • • a VO rH CM VO r-1 rH rH rH 00 rH VO rH rH o- CM • • a a LO rH <r rH rH rH rH CN CO <r LO CO CM CO >• • • • CM oo r-l rH rH CO oo CO CN LO CO VO • • • • rH o CM CN rH rH rH CO CM CO rH O 00 rH a • • a a-* CO OY rH rH 00 rH r~- rH CN O CT. CM • • a a CN o •CM r-» rH rH i—1 rH VO 00 CM rH m CM CM VO • • • • a o rH rH 00 »—1 rH rH rH r-» VO 00 VO • * a • rH rH CTN 00 rH <r <o <r O LO LO • • a • CTi CM o rH rH rH r-» CM rH rH LO VO • • a a CN CTi rH rH rH •H rH CM CT. VO • • a * a O o CM rH rH rH rH LO VO CO CO CO O rH LO • • a a m o CN CO rH rH rH rH rH CN CO < O o CM rH 00 00 1-^ ,—1 h~ 00 o CO CO .00 CM rH vO. <r o-> 00 VO o LO o o> VO 'CM rH VO f» rH ro o CO rH o rH a a a a o- CM o 00 LO O o •—< rH co CO 00 CTi CM rH LO OY <r a a a CTi rH rH ' rH #* rH o 1^ CO rH rH. LO CO CO LO LO ON O VO CO CM O O CN LO rH rH rH rH OO LO o> O ro CO VO CO a a a CM 00 VO o •* co CM VO O o- rH CM rH VO rH CN rH LO CM LO' CO 00 rH a a a a VO CM o r-. ro rH LO rH rH rH . CM rH CM LO rH rH LO a a a • CO H o> r—i CO rH CO 00 CM ' rH • rH 00 CO LO ro CO cj cn CM CM •rH rH rH CM rH ' cn a) LO VO XI rH ; c^ o 4J a a a a s VO CO CM 00 cO VO CM o o o LO rH CM 4J a CO VO <t- ro r~ •—t CM LO m rH a a a 0) LO LO LO Xi CN CO ro CJ rH rH CM a /•—N cd rH LO ON M •o LO r~- O Xi H a a a 4-1 rH CO CM rO a LO rH u . o CM rH H CO cj S -•H U o rH CN CO-Pn CJ O o •o o CJ - 174 Appendix 2, continued. CO CO CO rH rH CO CM rH in CO vo . 1—1 o CO CO" rH • • • * • • • cd VO in rH m ON H . m CO 4J CO. in ON CN r—1 CN VO 1H ON o in m rH CO ON O • • • • • • • xi • vo CO vo m rH CO r-. r-H co CO m in vo o-ON CO vo CO CO m CO VO CN CO •M • • • • • . cd VO vo o VO r-1 <!-VO CO CO pq U cd W m m ON CO . m ON • • • CN . CO in m vo CO CN CO -o- ON CO CN m rH rH CO m CN ON CN rH • • .. • • • • • CO vo in rH o CO H CO ON 00 4J vo rH rH CN vo vO rH ON in r-H in ON m CO o • • • • • • • • XI VO in CO VO rH as CO VO oS o ON as CO CN m in OS a) r-. ON ON ON •. • . • cd CO CO as CN ON CO as 00 CN >N rH as CO rH ON 00' <r u • • CO O r-- CO, o w VO VO m o a PH as ON m rH . o in CO as CM co rH • • • ' • • • • • CJ vo ; m VO in vO CO CO VO H CO ^> VO vo 4-1 <!• o CN vo rH CO CO U rH rH VO VO ; as r- CO o . . . • • Xi in VO in CN 00 VO CO VO OS CO m r-» as . -CM CM r-. r-. a) CM VO r--4J CO Cl • • • CO •rl o rH VO ' -vi 00 cn •H CO 00 vo VO •  a) 4J X /—s a r»N 4J rH 3 rH ON m m rH a o -d O 00 rH U cd U o u • • • cd ' 4J u 4J . ON CO CN w o Cl a, Cl 00 m in 4J o o 4J O CO v_^ •rl 3 cd o rH CN CO m U - o rH CM CO < o o O . O o o O O CJ 

Cite

Citation Scheme:

    

Usage Statistics

Country Views Downloads
United States 17 2
China 14 20
Russia 4 0
United Kingdom 3 0
Japan 3 0
India 2 0
Serbia 2 0
Indonesia 2 1
Unknown 1 0
Hashemite Kingdom of Jordan 1 0
Czech Republic 1 0
Philippines 1 0
Germany 1 19
City Views Downloads
Unknown 14 20
Ashburn 11 0
Beijing 9 0
Shenzhen 5 20
Saint Petersburg 4 0
Tokyo 3 0
Delhi 2 0
Fort Worth 2 0
Mountain View 2 1
Sunnyvale 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0095380/manifest

Comment

Related Items