Plant Speciation 9780231888110

Table of contents :
Contents
Preface
Part One. Populations and Races
1. Plant Reproduction
2. Local Races
3. Geographical Races
Part Two. The Nature of Species
4. The Biological Species
5. Species Units in Uniparental Plants
6. The Species Situation in Plants
7. Types of Species
Part Three. Genetic and Ecological Aspects of Species
8. The Genetic Basis of Species Differences
9. Isolating Mechanisms
10. Ecological Interactions
11. Patterns of Fertility Relationships
Part Four. Evolutionary Divergence
12. Modes of Speciation
13. Primary Speciation
14. Chromosome Repatterning in Speciation
15. Selection for Reproductive Isolation
Part Five. Natural Hybridization and its Products
16. Natural Hybridization
17. Introgression
18. The Syngameon
19. Hybrid Speciation
20. Recombinational Speciation
21. Hyrbrid Complexes
Part Six. Polyploidy
22. Polyploidy: Range and Frequency
23. Types of Polyploids
24. Factors Promoting Polyploidy
25. The Polyploid Complex
26. Natural Autopolyploids
Part Seven. Aneuploidy
27. Aneuploidy
28. Agmatoploidy
Part Eight. Specialized Genetic Systems
29. Permanent Translocation Heterozygosity
30. Permanent Odd Polyploidy
31. Agamospermy
32. The Agamic Complex
33. Vegetative Propagation of Hybrids
34. Natural Triploids
Part Nine. Conclusions
35. Species and Macroevolution
Bibliography
Organism Index
Author Index
Subject Index

Citation preview

PLANT

SPECIATION

OTHER

BOOKS

BY

THE

SAME

AUTHOR

Natural History of the Phlox Family (1959) The Origin of Adaptations (J 963) The Architecture of the Germplasm (¡964 ) Flower Pollination in the Phlox Family. With Karen A. Grant (1965) Hummingbirds and Their Flowers. With Karen A. Grant (1968) Genetics of Flowering Plants (1975) Organismic Evolution

(1977)

Plant Speciation Verne Grant SECOND

EDITION

COLUMBIA UNIVERSITY PRESS NEW YORK 1981

Library of Congress Cataloging in Publication Data Grant, V e m e . Plant speciation. Bibliography: p. 00 Includes index. 1. Plants—Evolution.

2. Species.

I. Title. QH368.5.G7

1981

581.3'8

I S B N 0-231 -05112-3

81-6159

AACR2

COLUMBIA UNIVERSITY PRESS N E W YORK

COPYRIGHT ©

GUILDFORD, SURREY

1971, 1981

COLUMBIA UNIVERSITY PRESS

A L L RIGHTS RESERVED PRINTED IN THE UNITED STATES OF AMERICA

Printed on permanent and durable acid-free paper.

This book is dedicated to George Gaylord

Simpson

Contents

Preface

xi

Part One. 1. Plant Reproduction 2. Local Races 18 3. Geographical Races

Populations and Races 3

Part Two. 4. 5. 6. 7.

30 The Nature of Species

The Biological Species 45 Species Units in Uniparental Plants 64 The Species Situation in Plants 70 Types of Species 77 Part Three.

Genetic and Ecological Aspects of Species

8. The Genetic Basis of Species Differences 9. Isolating Mechanisms 111

97

viii

Contents

10. Ecological Interactions 118 11. Patterns of Fertility Relationships

Part Four. 12. 13. 14. 15.

Evolutionary Divergence

Modes of Speciation 149 Primary Speciation 153 Chromosome Repatterning in Speciation 170 Selection for Reproductive Isolation 178

Part Five. 16. 17. 18. 19. 20. 21.

Natural Hybridization and its Products

193 Natural Hybridization Introgression 205 The Syngameon 234 Hybrid Speciation 242 Recombinational Speciation Hyrbrid Complexes 272

Part Six. 22. 23. 24. 25. 26.

124

256

Polyploidy

Polyploidy: Range and Frequency 283 Types of Polyploids 298 Factors Promoting Polyploidy 307 The Polyploid Complex 324 Natural Autopolyploids 347

Part Seven. 27. Aneuploidy 28. Agmatoploidy

357 369

Aneuploidy

Contents

ix

Part Eight. Specialized Genetic Systems 29. 30. 31. 32. 33. 34.

Permanent Translocation Heterozygosity 379 Permanent Odd Polyploidy 401 Agamospermy 413 The Agamic Complex 434 Vegetative Propagation of Hybrids 462 Natural Triploids 473

Part Nine. 35. Species and Macroevolution

Conclusions 481

Bibliography

Organism Index

489

541

Author Index

553

Subject Index

561

Preface

This book deals with speciation phenomena in higher plants. Speciation is treated as a process consisting of many facets and affected by many factors. We begin with a consideration of populations and races (part 1). This is followed by a discussion of the nature and behavior of species (parts 2 and 3). We then consider the primary divergence of species (part 4), natural hybridization, and hybrid speciation (part S). Polyploidy, agamospermy, and other specialized genetic systems associated with hybridity, and the hybrid complexes based on such genetic systems, are described in parts 6 to 8. The final chapter (part 9) looks briefly at macroevolution in the light of studies at the species level. In short, the book is a treatment of plant evolution at the level of species and species groups with particular reference to higher plants. It is an underlying premise of the book that the natural or evolutionary species is one of the basic units of organization of living material. As such, and in company with other biological units, the evolutionary species possesses certain general properties of its own which can be discovered and elucidated. These properties will not be discovered if the species units are seen only as an endless array of particular entities with particular features, as in purely descriptive taxonomy, although this approach lays the necessary groundwork. They can be discovered by the application of

XU

Preface

the analytical and generalizing methods which have been successful in other branches of theoretical biology. The first edition of this book, published in 1971, has had wide use. But much has happened in plant species biology in the decade since the first edition appeared. Not only has a great deal of new research come out, but also my own understanding of the field has developed. It was time for a thorough revision. A new chapter outline has been adopted for this second edition. A number of new chapters have been added. New evidence has been added throughout the text. Discussions of concepts have been brought up to date. Although the bibliography has been greatly extended for this new edition, it still falls far short of completeness. The literature is now so vast that some selectivity in reviewing it is necessary. In the process, many good papers were inevitably passed over, to my own regret. Karen A. Grant, who critically read the manuscript for the first edition, read all of the new chapters for this edition. Karen Grant also bore most of the burden of proofreading. The typing was done by Douglas Duke and Margaret Boulanger. I extend my appreciation to them. Dr. Vicki P. Raeburn and the other editors at Columbia University Press were helpful and cooperative as always. Department of Botany University of Texas January ¡981

V.G.

PART ONE

Populations and Races 1. Plant Reproduction Departures from Random Cross-Fertilization Vegetative Propagation Agamospertny Autogamy The Recombination System Ecology of Recombination Systems 2. Local Races Dispersal and Gene Flow Panmictic Units and Neighborhoods Gene Flow and Racial Differentiation Interaction Between Gene Flow and Selection Microgeographical Racial Differentiation The Selection-Drift Combination 3. Geographical Races Continuous Geographical Variation Disjunct Geographical Races Ecological Races Types of Selective Factors Involved in Race Formation

3 5 6 10 11 14 15 18 18 22 24 24 26 28 30 30 34 37 39

CHAPTER Plant

ONE

Reproduction

One of the fundamental characteristics of living material is the power of reproduction. This power is vested in organized units of very diverse size and complexity, ranging from DNA macromolecules through chromosomes and cells to individual organisms and breeding populations. In this book we are concerned with one of the more inclusive units of reproduction in the world of life, namely, species. The concept of species as reproductive units is an age-old one, subsumed in the statement that like begets like. In most animals and in many plants, reproduction is inseparably connected with sex. But higher organisms have evolved from unicellular forms in which the sexual process and the reproductive process are distinct and even antagonistic. Reproduction in a protozoan is achieved by fission: one cell divides into two. Sexuality in the same organisms is expressed by conjugation: two cells fuse into one. The whole life cycle consists of alternating phases of cell division, leading to an increase in population size, and sexual conjugation, resulting in a decrease in numbers of individuals. The basic distinctness of sex and reproduction persists in covert form in higher animals and plants. Sexuality in higher organisms means that two individuals are required to carry out a process of reproduction which,

4

Populations

and

Races

in the absence of sexuality, one individual alone could accomplish by some method of budding. The distinctness of the two processes can be seen from the physiological as well as from the populational standpoint in higher plants. Many or most perennial plants possess methods of vegetative propagation along with methods of sexual reproduction. Both methods entail a drain on the food supply of the parental plants; they both require a certain reproductive effort. The potentially conflicting demands of sexual reproduction and of vegetative propagation for a limited supply of parental energy are reconciled in many plants by an inverse correlation between the opposing methods of reproduction (Salisbury 1942, 1976; and later authors, e.g.. Harper 1977). Abundant seed production is often combined with slight vegetative propagation and, conversely, copious vegetative reproduction is often associated with reduced seed output. Thus, in some varieties of the strawberry, removal of the runners brings about a marked increase in the number of fruits, whereas other ever-fruiting varieties produce few or no runners (Salisbury 1942; Holler and Abrahamson 1977). The periwinkle (Vinca major) usually spreads widely by long runners but produces few fruits; if, however, the runners are cut off and kept from forming anew, the plants go on to set seeds (Salisbury 1942). The importance of reproduction per se is obvious enough. It is the condition for the perpetuation of the population or species through time. And it is the only means of multiplication, increase in numbers, and colonization of new territories. The function served by sex which accounts for its establishment in the life cycle, even though it alters the reproductive efficiency of the population or species, is less obvious, and has been and still is a debated question. It is generally recognized that the sexual process, involving an alternation of cross-fertilization and meiosis, is a mechanism for bringing about gene recombination. Gene recombination is in turn the chief source of hereditary variations in a sexually reproducing species. And hereditary variations are the necessary raw materials for response of the species to heterogeneous or changing environmental conditions. It follows that vegetative and other nonsexual means of reproduction are advantageous in a more or less constant environment to which the species is already well adapted, since the formation of poorly adapted recombination types is avoided or minimized by these methods, whereas

Plant

Reproduction

5

the production of seedling progeny by the sexual process is advantageous in relation to a changing or heterogeneous environment (Salisbury 1942, ch. 17; Stebbins 1958b; Grant 1958; Solbrig 1979). Both types of environmental challenges exist and both types of reproductive responses are called for. The overall breeding system of a species which is successful over a long period of time is likely to represent a combination of sexual and nonsexual processes. Sexual and nonsexual methods of reproduction are in fact combined in various ways in plants. Each method serves its function in the life of the species. But each method also has its characteristic effect on the nature and structure of the species.

Departures from Random Cross-Fertilization It is useful for purposes of analysis to postulate an ideal set of sexual reproductive conditions in a hypothetical population, and then to consider the various deviations from this ideal condition which are found in real cases. Let us assume that our hypothetical population is randomly mating or panmictic. Not only does it reproduce exclusively by cross-fertilization, but furthermore, the cross-fertilization takes place under conditions such that the union of gametes in pairs is governed by chance alone. Any female gamete produced by the population is assumed to have an equal chance of being fertilized by any male gamete. Consequently, under panmixia the individuals constituting the descendant generation in the population represent the products of different pairs of gametes drawn at random from the gamete pool produced by the parental generation. Departures from panmixia take four common forms in real plant species. (1) Cross-fertilization takes place preponderantly between neighboring individuals (vicinism), and leads to mating between relatives, or inbreeding. (2) Reproduction is by self-fertilization (autogamy), the closest kind of inbreeding. (3) The new plants arise from buds produced by the mother plant (vegetative propagation). (4) Or the new generation of plants arises from seeds which develop on the mother plant without fertilization (agamospermy). The first two reproductive methods listed above (vicinism and autog-

6

Populations

and Races

amy) involve the sequence of fertilization and meiosis, and are thus methods of sexual reproduction. In the last two methods (vegetative propagation and agamospermy), fertilization and meiosis are circumvented, and the reproduction is therefore asexual or apomictic. All the individuals derived by mitotic divisions from a single parental individual are referred to as members of a clone. Vegetative propagation and agamospermy thus lead to the formation of a clone or array of clones. Another meaningful distinction can be drawn between the first reproductive method listed and all the others. Only in the first case does crossfertilization occur. In the remaining three cases, fertilization is either bypassed entirely or takes place between gametes produced by a single parental plant. Accordingly, reproduction is biparental in the first case, and uniparental in the other three. A population composed of progeny of uniparental reproduction, be they clones or inbred lines, obviously differs in an important respect from a true breeding population, the individuals of which are tied together by mating bonds. Random mating in a large population is an idealized condition useful as a standard of reference. Probably no real, wide-ranging plant species comes very close to this condition. Plant species with biparental reproduction exhibit some degree of vicinism in the known cases. Extreme departures from random mating in the direction of strict uniparental reproduction are found, however, in some real plant species. Vegetative propagation, agamospermy, and perhaps autogamy may replace biparental reproduction completely in some plant groups. It is far more usual, however, to find some combination of uniparental and biparental reproduction in the same plant. Such plants are then intermediate between strictly biparental organisms and strictly uniparental ones. Indeed, as Gustafsson (1946-1947) has pointed out, a series of transitions connects the truly sexual and the various uniparental methods of reproduction in plants.

Vegetative

Propagation

Vegetative reproduction takes place in many ways: by surface stolons and runners, by underground rhizomes and tubers, by offset buds on corms and bulbs, by adventitious buds on cut stems or fallen leaves, and

Plant

Reproduction

1

by vegetative propagules arising within a flower or inflorescence. The latter method has been called vivipary; a more appropriate designation according to van der Pijl (1972) is false vivipary. Vegetative propagation is very widespread in perennial angiosperms, occurring in all major groups. Species of perennials which lack the ability to reproduce vegetatively are exceptional. Gustafsson (1946-1947:272) quotes figures showing that about 80% of all angiosperm species in certain Scandinavian floras have some means of vegetative propagation, and about 50% of the perennial angiosperms in the same floras have these methods developed to such an extent that they can spread rapidly by vegetative reproduction. The redwood tree (Sequoia sempervirens) has the ability to sprout from the root crown, an ability, incidentally, which is exceptional among conifers (Jepson 1923). When an old tree dies, the crown sprouts grow up into saplings and finally into adult trees arranged in a circle around the original parental stump. In time the members of the second-growth tree ring may give way to third-growth rings of their own. Inasmuch as a redwood tree may live to an age of about 1300 years (Jepson 1923; also Harper and White 1974), the life span of the clones has to be reckoned in millennia. Jepson (1923:160) estimates that about 80% of the mature trees in the redwood forest originated from crown sprouts and about 20% from seeds. In our experience the products of asexual and of sexual reproduction have a highly nonrandom spatial distribution in and around the grove. Reproduction in the dense central parts of some groves is virtually exclusively by crown sprouting, whereas seedlings succeed in establishing themselves only in open, peripheral areas. Most species of oaks have creeping rhizomes which produce suckers that can later develop into new plants. In Quercus virginiana in Texas, for example, young shoots can sometimes be shown to be attached to a web of underground rhizomes radiating from a central tree (figure 1.1). Mature trees of this species are often grouped in mottes or clonal colonies (Muller 1951b). The aspens, Populus tremuloides and P. grandidentata, produce vigorous suckers from the root crown. The connections between the suckers and the parent plant later die and separate individuals arise. Aspen trees in nature typically exist in clones of several to many individuals (Barnes 1967a).

Figure 1.1. Cloning in Quercus virginiana. (A) A small motte or clump of young trees. (B) Part of rhizome; note branching. (C) Diagram of distribution of rhizomes in a single clone; black dots indicate positions of aerial shoots. (Rearranged from Muller 1951b)

Plant

Reproduction

9

Some clones of Populus tremuloides in Utah attain large sizes. One such clone occupied 25 acres and included about 15,000 individuals; another occupied 107 acres with about 47,000 individuals; and another occupied about 200 acres. In eastern North America the clones of P. tremuloides are smaller. The largest clone measured in this area was 3.8 acres in extent (Kemperman and Bames 1976). The large aspen clones must be quite old and some may date back to Late Pleistocene time. In the mountains of Utah Populus tremuloides produces much viable seed, in some years at least, but germination of this seed is negligible under the present climatic conditions of low earlysummer rainfall. In this area the aspen apparently reproduces entirely by vegetative means. Since the climatic regime of summer drought commenced about 8000 years ago in Utah, it is possible that some clones of aspen are as much as 8000 years old. In the course of centuries or millennia of vegetative propagation a particular race of the aspen with an early leafing habit has succeeded in colonizing the higher elevations of the Utah mountains which were formerly covered with ice and snow (Cottam 1954). Festuca rubra is a perennial herb which spreads vegetatively by rhizomes and also reproduces sexually, being wind-pollinated and selfincompatible. Harberd (1961) analyzed the composition of populations of this grass in Scotland on the basis of samples taken within large quadrats. Plants possessing different phenotypic characters and exhibiting crosscompatible relationships were considered to represent different genotypes. Cross-incompatible plants with the same phenotypic characters, on the other hand, were held to be clonal divisions of the same genotype. From the phenotypic characters and breeding behavior of his samples, Harberd (1961) concluded that a quadrat 100 yards square contained relatively few genotypes, but many plants belonging to the same clone. One particular genotype was found to be spread over an area more than 240 yards in diameter. As Harberd notes, a clone must be hundreds or perhaps a thousand years old to cover an area of this size. Some other clones were small. The populations sampled in Festuca rubra thus consist mainly of a few genotypes which have spread clonally over large areas, and in addition of some genotypes which have formed only small clones. Cross-pollination by wind in a self-incompatible grass with this clonal structure must usually lead to incompatible unions (Harberd 1961). A related perennial species, Festuca ovina, exhibits a different pattern.

10

Populations

and Races

Festuca ovina spreads much more slowly than F. rubra. In F. ovina, Harberd (1962) found numerous genotypes within a quadrat 10 yards square. Here, as compared with F. rubra, the balance between asexual and sexual reproduction appears to be shifted more toward the latter mode. The coastal grass, Spartina patens, in eastern North America is also outcrossing, rhizomatous, and clone-forming. Individual genotypes can be identified by isozyme markers and their microgeographical distribution mapped. In a study area in North Carolina, 101 genotypes were found on a transect 200 meters long. Some genotypes form fairly extensive clones while other are represented in only a few individual plants (Silander 1979). Multiplication by vegetative means alone is illustrated by the wellknown case of Elodea canadensis (Hydrocharitaceae) in Europe. In eastem North America where it is native, this dioecious pondweed reproduces both sexually by seeds and vegetatively by the breaking off of shoots and the formation of winter buds. Female plants were introduced from North America into Britain about 1840. In the absence of male flowers in the alien territory the Canadian pondweed could only reproduce itself vegetatively. Nevertheless, between 1840 and 1880 it managed to spread through the inland waters of Europe (see Gustafsson 1946-1947).

Agamospermy In agamospermy an individual plant produces viable seeds containing embryos which have arisen without fertilization. The embryological details are complex and vary from case to case (see chapter 31). The new embryo in the seed may develop from an unreduced egg in the embryo sac (parthenogenesis), from some cell or nucleus other than the egg in the embryo sac (apogamety), or from some somatic cell in the ovule (adventitious embryony). Agamospermous seed formation may take place without pollination, or may require pollination (pseudogamy). In the latter case the pollen stimulates the growth of the endosperm which is necessary for the normal development of the seed and its embryo (Gustafsson 1946-1947). The various embryological pathways are alike in bypassing both

Plant Reproduction

11

meiosis and fertilization in the cell lines leading to the new embryo. The result, apart from certain exceptional processes, is the formation of seeds containing embryos which are genetically identical with the maternal parent. Agamospermy is widespread in higher plants. This condition is found in many members of the Gramineae, Compositae, and Rosaceae, and in a host of smaller families. Familiar examples occur in Hieracium, Taraxacum, Crepis, Citrus, and Poa. Agamospermous plants usually have a perennial growth habit. Their sexual relatives, where known, are invariably cross-fertilizing by means of self-incompatibility, dioecism, or some other outcrossing breeding system, suggesting that agamospermous plants have been derived from strongly outcrossing ancestors (Gustafsson 19461947). In some plant groups agamospermy replaces sexual reproduction completely (obligate agamospermy). In other plants some seeds form by agamospermous processes and some by sexual processes (facultative agamospermy). As with vegetative propagation, no sharp line can be drawn between sexual and asexual reproduction, but instead the two modes are bridged by transitional conditions (Gustafsson 1946-1947). The effect of agamospermy on the variation pattern in a facultatively agamospermous population is to break up the variability into a series of groups consisting of identical individuals which differ from one another by minor characteristics. The population is composed of swarms of more or less discrete agamospermous microspecies.

Autogamy Among hermaphroditic angiosperms we find a spectrum of breeding systems ranging from obligate outcrossing at one extreme to virtually complete autogamy at the other. Hermaphroditic flowers may be completely self-incompatible, setting no seeds after self-pollination, as in Gilia capitata capitata (Polemoniaceae); or the self-incompatibility may be incomplete, as in G. capitata tomentosa, where many self-pollinated flowers produce just a few seeds (Grant 1950a). Conversely, self-compatibility may be incomplete and partial. Cheiranthus cheiri (Cruciferae) is fully self-compatible in artificial sellings but, when exposed to equal

12

Populations

and Races

amounts of self- and cross-pollination, most (92%) of the seeds set are products of cross-pollination (Bateman 19S6). Completely self-compatible angiosperms then vary in details of the floral mechanism which promote outcrossing, selfing, or mixtures of both. The rate of outcrossing in natural populations of Clarkia unguiculata (Onagraceae), which has strongly protandrous flowers with exserted sex organs, is estimated from progeny tests to be 96% (Vasek 1965). The floral mechanism of C. exilis, by comparison, permits considerable spontaneous self-pollination, and two populations of this species have 43% and 45% outcrossing (Vasek 1964, 1967). In many self-compatible plants the stamens and stigma of the same flower stand close together and mature simultaneously, so that self-pollination normally occurs automatically. It is generally agreed by students of autogamous plants that the autogamy is usually not so complete as to exclude some outcrossing. Self-pollination may predominate, but some cross-pollination is brought about occasionally by insects or wind, and the breeding system is therefore more properly referred to as predominant autogamy (Allard and Jain 1962; Kannenberg and Allard 1967). Recent estimates of the rate of outcrossing in several species of predominantly autogamous grasses are as follows: Hordeum jubatum, 1% (Babbel and Wain 1977) Hordeum vulgare, 1% and 4% (Allard and Kahler 1971; Jain 1975) Avena barbata, 2% (Jain 1975) Avenafatua, 1-12% and 3% (Iman and Allard 1965; Jain 1975) Bromus mollis, 4% (Jain 1975) We could expect to find some instances in which autogamy is complete or nearly so, and this expectation is confirmed in a few cases. The estimated rate of outcrossing in Festuca microstachys (Gramineae) is 0% (Allard and Kahler 1971). In Polemonium micranthum (Polemoniaceae) the tiny flowers undergo self-pollination in the bud and later open for a few hours after pollination has taken place, and there is no apparent opportunity for insect pollination to occur. Attempts to cross-pollinate the plants artificially were unsuccessful; the only progeny obtained from these artificial crosses were products of selfing. The evidence available suggests that some population of P. micranthum are virtually completely inbreeding (Grant and Grant 1965). Autogamy or, more particularly, predominant autogamy, is a common

Plant

Reproduction

13

and widespread condition in angiosperms, especially among annual herbs (Gustafsson 1946-1947; Stebbins 1950). In the family Polemoniaceae this is the most frequent single method of pollination, being known from breeding tests in some 45 species and predicted from field observations in 30 additional species. Here the autogamous habit is found exclusively, as far as known, among the annual members of the family (Grant and Grant 1965). The breeding system influences the variation pattern of the population. Given two or more homozygous individuals differing with respect to two or more genes, outcrossing generates a much greater amount of individual variability by recombination, and maintains a heterozygous condition for at least one gene in most of these recombination types. Continued selfing has the opposite effect. Given an original population composed of individuals heterozygous for two or more genes, the proportion of heterozygous individuals decreases at a regular rate in each generation of selfing, and the original array of heterozygous and homozygous recombination types becomes sorted out into a smaller number of homozygous types. The strictly autogamous population is expected to consist mainly of a series of tme-breeding pure lines. The expected results of selfing may or may not be realized in actual populations of autogamous plants, depending on other factors. In the first place, autogamy is usually not obligate, as we have already seen, and occasional outcrosses between different biotypes in a predominantly autogamous population will regenerate new variability at periodic intervals (Harlan 1945; Stebbins 1957a; Grant 1958). Furthermore, continued selfing will not lead to a decline in the porportion of heterozygotes at the expected rate if the heterozygous types have a selective advantage over the homozygotes (Hayman and Mather 1953; Jain and Allard 1965). Recent studies by Allard and co-workers provide evidence of persistent genie heterozygosity due to heterozygous advantage in several predominantly autogamous plants. Such evidence has been found in Secale cereale (Jain and Allard 1960; Allard and Jain 1962; Jain and Jain 1962), Phaseolus lunatus (Allard and Workman 1963; Harding, Allard, and Smeltzer 1966), Avena fatua (Iman and Allard 1965; Jain and Marshall 1967), and Avena barbata (Clegg and Allard 1973). The actual composition of populations in many autogamous plant species is thus a good deal more complex than would be expected on the basis of extrapolations from the pure-line concept, as has been empha-

14

Populations

and

Races

sized by many recent students. This is not to say that the simple models are unrealistic in every case, however. The populations of some autogamous species approach a simple composition, consisting of one or a few homozygous biotypes; those of other species contain a greater store of variability and a higher frequency of outcrossing, approaching in these respects the populations of regularly outcrossing species.

The Recombination

System

Plant reproduction involves a combination of variation-producing and genotype-replicating processes, as the foregoing discussion makes clear. The balance between the two types of processes is brought about by the part of the genetic system known as the recombination system. The recombination system brings about a different balance point in different populations and species. It is useful to recognize three modal conditions on the spectrum ranging from maximum recombination to maximum replication. These modes are open, restricted, and closed recombination systems. They can be exemplified by wide outcrossing, predominant autogamy, and agamospermy, respectively. The recombination system is affected by many factors. A classification of these factors is presented in table 1.1. Discussions of these factors are given in recent reviews by Grant (1975, chs. 23, 24) and Solbrig (1979). It is important to note that a given degree of regulation of recombination can be achieved by different combinations of the regulatory factors. Solbrig (1979) notes that some of the recombination-regulating factors come under the direct control of selection and others do not. The components of the recombination system that can be controlled directly by selection are components 1, 2, 4, 5, 6, and 9 (using the numbers in table 1.1.) Components 7 and 10 are controlled by selective forces controlling characteristics other than recombination-regulation per se. And components 3 and 8 are not under direct selection according to Solbrig (1979). Solbrig (1976, 1979) has developed the idea that recombination systems have a cost and that this cost is higher in open recombination systems than in restricted or closed ones. Open recombination entails a genetic or so-called meiotic cost, measured in numbers of ill-adapted recombination types produced, and an energetic cost, measured in flower rewards offered to pollinators and surplus pollen for cross-pollination.

Plant

Reproduction

15

Table 1.1. Components of the recombination system in plants. (Grant 1958, with modifications) I. Factors controlling the amount of recombination per generation; control operative at meiosis 1. Chromosome number 2. Frequency of crossing-over 3. Hybrid sterility II. Factors controlling the amount of recombination per generation; control operative at fertilization 4. Breeding system 5. Pollination system 6. Dispersal potential 7. Seed number and seed germination behavior 8. Population size 9. Crossability barriers and external isolating mechanisms III. Factor controlling the amount of recombination per unit of time 10. Length of generation

Open recombination systems must have benefits that outweigh these costs. The average number of pollen grains per ovule is in fact much higher in outcrossing plants than in inbreeders. The average numbers of pollen grains per ovule for samples of species with different breeding systems are (Cruden 1977): 5859 grains in obligate outcrossers 797 grains in facultative outcrossers 168 grains in predominantly autogamous plants 28 grains in obligately autogamous plants Impatiens capensis produces both cleistogamous and chasmogamous flowers and both selfed and outcrossed seeds. The estimated cost in calories of the two classes of seeds is 135 calories for seeds from chasmogamous flowers versus 65 calories for seeds from cleistogamous flowers (Waller 1979).

Ecology

of Recombination

Systems

The reason for the diverse array of recombination systems in plants is to be sought in the differing demands on reproduction made by different

16

Populations

and Races

types of habitats. Here another spectrum comes into the picture, a spectrum ranging from open habitats to closed communities. Open habitats, whether they are continuously open as in deserts, or intermittently open as in the pioneering stages of ecological succession, or artificially open as in cultivated fields, place a premium on colonization. Colonization in turn is promoted by restricted and closed recombination systems. Such recombination systems emphasize constancy in reproduction. A population adapted to a given open habitat, and possessing a restricted or closed recombination system, can replicate its adaptive genotype(s) and multiply its numbers quickly within the available habitat. Experimentation with new recombination types, which would lower the successful reproductive output, is kept to a minimum. Plant species with restricted and closed recombination systems are in fact found commonly in open and pioneering habitats; for example, desert annuals, mediterranean annuals, weeds, pioneer herbs in ecological successions, and agamospermous plants in areas exposed since the end of Pleistocene glaciation. Plants with open recombination systems, on the other hand, are commonly found in closed plant communities; for example, trees and shrubs of temperate forests, many prairie and plains grasses, and many perennial herbs of forest understory. In closed plant communities the chances of establishment of seedling progeny are small due to competition. The plant population produces its seed crop year after year but new seedlings become established only very rarely when a suitable place opens up in the community. It is estimated for beech trees, Fagus grandifolia, that an average of 1 new seedling has become established per 0.1 acre per 10 years in virgin forest in New Hampshire over the last three centuries (see Cook 1979 for review). Under conditions like these, where the plant population has a high excess fecundity, it can afford an open recombination system, with its corollary production of many poorly adapted recombination-type zygotes. There are two views regarding the selective forces that have led to the development of open recombination systems. My original thought (Grant 1958) was that the high excess fecundity of plants in closed communities permits them to exploit the long-term advantage of an open recombination system, namely, the generation of variability for coping with future environmental changes. To put it the other way around, many plant species with open recombination systems

Plant

Reproduction

17

now found in nature have survived for a long time and reached the present period because of their recombination system. Interdeme or interspecific selection could have brought about this concentration of openrecombination types in many modem communities. A simpler alternative hypothesis has been put forward recently by Levin (1975a) and Solbrig (1976, 1979). Plants in closed communities face an ever-changing biotic environment. New genotypes of pathogens, herbivores, and plant competitors are appearing continually. Open recombination systems generate the new variations that the plant population needs to keep pace with these changing elements in its biotic environment. The open recombination systems could be developed by individual selection in this case. More work is needed to verify this attractive hypothesis. Certain questions of terminology remain to be considered briefly. The terms /--selected type and AT-selected type as used in population ecology correspond to colonizing species and closed-community species, respectively, without, however, having any implication as to the type of recombination system. Nevertheless, the r and K terminology is being extended to cover alternative recombination systems in plants by various authors. In referring to contrasting ecological conditions associated with different types of recombination systems some authors use the terms stableunstable or predictable-unpredictable. These terms are subject to some confusion in usage. Thus an environment that fosters /¡"-selection is supposed to be more predictable than an environment with r-selection (Pianka 1970; Jain 1979). However, as we have just seen, a closed community containing ^-selected types has a more unpredictable biotic environment than an open habitat. This confusion can be avoided by using the straightforward descriptive terms, open habitat and closed community.

CHAPTER Local

TWO

Races

Races are populations or population systems within a species which differ statistically in the composition of their gene pools and in their genetically determined phenotypic characters from other conspecific populations or population systems. This definition covers a wide range in the hierarchy of population units from local breeding populations to major geographical races. It is useful for purposes of evolutionary discussion to recognize two levels of racial variation: local races and geographical races. We will discuss the former in this chapter and the latter in the next. We will approach the subject of local race differentiation from a theoretical standpoint. The controlling factors to be discussed are gene flow and its effects per se, gene flow in interaction with selection, and the selection-drift interaction. Empirical evidence will be introduced in connection with each set of controlling factors.

Dispersal

and Gene

Flow

Sedentary plants are capable of dispersal at two stages in their life cycle. Pollen is carried by wind, insects, birds, or other agents, and fruits or seeds are likewise transported by various physical or biotic agents.

Loca/

Races

19

The botanist who has observed seed dissemination in nature soon comes to recognize the following general pattern. Most of the seeds produced by a maternal plant, which escape being consumed by animals, lodge fairly close to the parental individual, but some seeds are dispersed over longer distances. General observations of this sort have been supported by quantitative evidence in some cases. Salisbury (1961) counted the number of seeds of Verbascum thapsus which were carried by wind to various distances from the parent plant. The main bulk of the wind-borne seeds fell to the ground about 12 feet from the seed parent, but the numbers of seeds fell off rapidly beyond the 12-foot radius. A similar pattern of dispersion is found for the wind-borne fruits of Senecio jacobea (Salisbury 1961). This dispersal pattern turns out to be characteristic in a number of organisms, both plant and animal. The frequency curve of dispersal distances is leptokurtotic and skewed. A high proportion of the dispersal units are distributed close to the parent. The frequency curve than falls off rapidly with distance from this point of origin. This characteristic dispersal pattern is well illustrated by a concrete example. Colwell (1951) released radioactively tagged pine pollen (of Pinus coulteri) and tracked it on two radii from the point of release. Figure 2.1 shows the distribution curve for the wind-borne pine pollen. The. greatest bulk of the pollen fell downwind at a distance of 10 to 30 feet from the source. Thereafter the amount of pollen diminished rapidly. At distances of ISO to 400 feet from the source, some pollen was found, but it was in quite small amounts (Colwell 1931). Pollen dispersal by bees has been studied by sprinkling blue dye in newly opened flowers of cotton (Gossypium hirsutum and G. arboreum) and following its subsequent distribution in the cotton field (Stephens and Finkner 1953). Dye particles were found up to 80 feet away from the source after bee visitations. However, the concentration of dye diminished in a uniform manner away from the source point, suggesting a corresponding concentration of bee activity within a given neighborhood of flowers (Stephens and Finkner 1953). The dispersal pattern of seeds described above, if continued over successive generations, will result in related plants coming to stand as close neighbors. The similar pattern of pollen dispersion will then have the effect that cross-pollination is preponderantly between neighboring related individuals. The net result is a considerable degree of inbreeding in

o

00

"t üü o

o (O i c i o

T3


< -

s8

oc. b. O »-

- s

z 5 a. Z

o

te.

h.

o (O

UI O

z

2 M

5

o

o f

RV o

u

V

1

H-l

V

*

*

>

a> I

BV • »• • •

•

VB T

1

1

•• 1

•• 1

R

1

Colony

1

H-2

RV

BV

••• •

• •• •

VB

• »•

• ••

i

10

5

7

8

Petal size F i g u r e 17.3. Pictorialized scatter diagrams of'Iris fulva, I. he.xagona giganticaerulea. and two hybrid colonies. Same populations as those graphed in figure 17.2. (Drawn by Anderson (1949) from data of Riley (1938); redrawn and rearranged here)

215

Introgression

Explanation of Symbols

TUBE

o> S o

GREEN

YELLOW

GREENISH

0B0VATE ALMOST CLAWLESS

INTERMEDIATE

SPATULATE

EXSERTED

SUBEQUAL

INCLUDED

SMALL

MEDIUM »

LARGE

PETAL SHAPE

STAMENS in

tr

STYLE APPENDAGES

CREST ABSENT

SLIGHT

P e t a l size

WlELL DEVELOPED

—

F i g u r e 1 7 . 4 . S y m b o l s u s e d in f i g u r e 1 7 . 3 . ( R e d r a w n f r o m Introgressive Hybridization, by A n d e r s o n , copyright 1949, John Wiley & S o n s , N e w York, by permission)

to serpentine soils in the foothills, while the introgressive race of the same species occurs as a weed in the central valley along with//, annuus. This ecogeographic distribution suggests strongly that the introduction of H. annuus into California, perhaps by Indians, was followed by hybridization between H. annuus and the original wild form of H. bolanderi (Heiser 1949b). Introgression from Helianthus annuus into H. bolanderi then led to the formation of the valley race of H. bolanderi, which was able to spread throughout the weedy habitats being opened up concurrently. It is likely that//, annuus in central California was affected reciprocally by introgression from H. bolanderi. The introgressive valley race of H. bolanderi may be only a few centuries old (Heiser 1949b). Olivieri and Jain (1977) have reexamined this case on the basis of a larger array of characters, including biochemical ones, and more sophisticated statistical analysis of the data. The resulting histograms differ slightly from that of Heiser (in figure 17.5), as might be expected, but seem to be consistent with Heiser's original interpretation. Helianthus annuus overlaps and hybridizes with other species of annual

216

Natural Hybridization and its Products

8

10

Hybrid index Figure 17.5. Frequency distribution of hybrid index values in five populations of Helianthus annuus and H. bolanderi. The populations are from California unless specified otherwise. (A) H. bolanderi, foothill serpentine race. (B) H. annuus, Missouri race. (C) H. bolanderi, valley weed race. (D) H. annuus, San Joaquin Valley. (E) Hybrid swarm, Sacramento Valley. (Redrawn from Heiser 1949b) sunflowers in other regions of the United States, thus with H. debilis and H. argophyllus in Texas and with H. petiolaris in the Great Plains and Southwest. Helianthus annuus has given rise to introgressive races varying toward H. debilis in Texas and toward H. petiolaris in the central United States. Conversely there is an introgressive race of H. petiolaris (Heiser 1947, 1951a, 1951b, 1954, 1969, 1976).

Introgression

217

Juniperus Juniperus virginiana has a widespread distribution in the eastern United States and J. scopulorum is wide-ranging in the west. Their ranges are mainly allopatric and contiguous with some localized contacts on the borderlines (see figure 17.6). Hybrid swarms were reported by Fassett (1944) on the basis of morphological evidence. Flake, Urbatsch, and Turner (1978) have recently made a population study of these junipers on a long transect from eastern Missouri to western Montana (figure 17.6). They used biochemical characters. These consisted of some 43 volatile oil compounds in the foliage. The data were subjected to a discriminant analysis. The scores resulting from this analysis yielded a hybrid scale which could be used in frequency histograms. The variation pattern in the series of histograms—one for each population on the transect from A to Z— indicates that introgression is occurring from J. scopulorum into J. virginiana in the northwestern part of the area of J. virginiana (Flake, Urbatsch, and Turner 1978). Four histograms in the series are reproduced in figure 17.7.

Pinus An interesting case of introgression in a historical time frame based on paleobotanical evidence has been described by Mason (1949). The case involves two species of closed-cone pines, Pinus remorata and P. muricata, in coastal California over the period from Pleistocene to Recent time. In the Pleistocene the populations of Pinus remorata on Santa Cruz Island were morphologically quite distinct from the related P. muricata, which occurred on the neighboring California mainland at that time. In Late Pleistocene or Post-Pleistocene, P. muricata migrated to Santa Cruz Island. Natural hybridization between the two species ensued. This hybridization has continued into Recent time and has altered the characteristics of some P. remorata populations. On the California mainland the fossil evidence indicates that introgression was occurring in the opposite direction, from P. remorata into P. muricata, in Pleistocene time, and this process is continuing in present times (Mason 1949).

218

Natural

Hybridization

and its

Products

Character Coherence as a Criterion of Hybridity Anderson (1939, 1949, 1953) regarded character coherence—the loose association between different characters in a variable population—as a diagnostic feature of hybridity, as noted earlier. This conclusion was based on both theoretical and empirical grounds. Character coherence is an expected result of multifactorial linkage in a segregating hybrid population. The expectation is confirmed experimentally in artificial hybrid populations. And character coherence is observed generally in natural hybrid populations. Anderson's conclusion has been widely accepted for many years. Recently, however, it has been shown to be an overgeneralization (Grant 1979). Consider first the empirical evidence in natural hybrid populations. This evidence pertains almost exclusively to perennial plants, usually perennial herbs. In such plants individuals of one or both parental species are likely to persist and be intermixed with the F, hybrids and later-generation products. If they are included in the population sample, as they routinely were in the earlier studies, they will bias the results in favor of character coherence. An examination was made of the original data in several older examples in Iris, Aquilegia, Oxytropis, and other plant groups in which character coherence had been reported. In each case parental-type individuals were found in the population samples. When these were removed from the sample, and the data were reanalyzed for the hybrid subpopulation alone, the amount of character coherence showed a substantial decrease (Grant 1979). The amount of character coherence can be measured by the correlation coefficients for various character pairs. And the magnitude of the decrease in character coherence, in comparing the hybrid subpopulation with the whole population, can be measured by the difference between correlation coefficients for the same character pair in the two types of samples. Proceeding on this basis, it was found that the magnitude of the decrease varies greatly from one example to another. A hybrid population in Opuntia, for instance, showed statistically significant correlation between characters in nearly all paired combinations, when the whole pop-

Introgression

219

ulation was analyzed; but it showed only one significant correlation coefficient in one character pair in the hybrid subpopulation. The evidence for character coherence in this hybrid population is an artifact of the method of sampling. In a hybrid population in Iris, on the other hand, the amount of statistically demonstrable character coherence, while less than previously thought, is still quite high (Grant 1979). Character coherence, then, is a common feature of natural hybrid populations in plants, but it is not a universal one. It is exhibited, for instance, by particular hybrid populations in Iris and Aquilegia, when these are sampled properly, but not by certain hybrid populations in Opuntia, Oxytropis, and Gilia (Grant 1979). The older generalization cast its net too far. In order to understand this we need to reexamine the theoretical assumptions on which the older generalization was based. Some of the theoretical assumptions pertain to the external factors determining character coherence, and others to the internal factors. The most important external factor is natural selection. It had been assumed that selection in natural hybrid populations would always favor genotypes that approach the parental types. No doubt this assumption is valid in many or even most cases. But there is no reason to believe that it is universally true. Selection will also favor new recombination types where new and different environments are available for colonization. As regards the internal factors, multifactorial linkage is a real force, and it promotes character coherence; but gene recombination is also a real force and it works in the opposing direction to produce character recombinations. The result in any given hybrid population represents a balance between these opposing tendencies. The balance point in each particular case is determined to a large extent by the recombination system of the plants involved (see chapter 1). In general, hybrid populations can be expected to show less character coherence in plants with high basic chromosome numbers than in those with low numbers; and, likewise, less coherence in annuals than in perennial plants.

Ecological

Segregation

of Morphological

Types

Physiological characters determining ecological preferences segregate in hybrid populations just as do morphological and biochemical charac-

3

a.

o o.

T3

e

u

a >> -O

© JS .SP

>. a o o

g o 05 ü

•S-

s a

3

3

C

=3

a ta

« •fi J SJ C/3 C X>

« -e .j a = 5„

& «

2 M so , 3 ° Un O g

. o Ph

>e

Sol -

E

221

Introgression B, C , D

Jl

_T JZ-

-90

-70 - 3 0 -SO -10

10

SO SO 70 9 0 MVBRlO SCALE

MO ISO |

ISO 170

1*0

W

-90 -70 - 5 0 -3 0 - 1 0

10 !

3 0 30 70 9 0 HYBRID SCALE

IIO 13 0 |

13 0 170

19 0

- 9 0 - 7 0 - 3 0 -3 0 - 1 0

10 3 0 3 0 TO 9 0 i HYBRID SCALE

IIO 13 0 ISO 17 0 ISO 1

Figure 17.7. Histograms of hybrid scales for Juniperus populations on an eastwest transect from Missouri to Montana. The hybrid scale values are derived from a discriminant analysis of 43 volatile oils in the leaves and stems. (B,C,D) Good J. virginiana from Missouri and southeastern Nebraska. (X) Good J. scopulorum from Montana. (H,W) Two introgressive populations from the border zone between the two species. The map localities of these and other populations on the transect are shown in figure 17.6. (Rearranged from Flake, Urbatsch, and Turner, in Systematic Botany, copyright © 1978, by permission) ters. The former are difficult to score accurately, however, unlike the latter, and are rarely reported. Nevertheless, field observations indicate that segregation of ecological preferences does take place in hybrid populations. An introgressive population or a hybrid swarm often grows in a heterogeneous or somewhat hybridized habitat lying between the habitats of the parental species populations. In such a situation the microgeographi-

222

Saturai

Hybridization

and its

Products

cal distribution of different backcross types is not at random. On the contrary, as might be expected, the individuals which approach one parental species in morphological characters tend to be clustered in the fad e s of the hybridized habitat which is most like the normal environment of that same species. Conversely, the opposite class of hybrid segregates and backcross types is found growing preponderantly in environmental niches approaching that of the opposite parental species. I have observed this ecological segregation of different morphological forms in hybrid populations which I have examined in Aquilegia, Mimulus section Diplacus, Ipomopsis, and Gilia. No doubt other botanists have made similar observations. The phenomenon has been documented by Benson, Phillips, and Wilder (1967) for a hybrid population of Quercus. The parental semispecies in this case are Quercus douglasii and Q. turbinella californica. Quercus douglasii is a deciduous tree with glaucous blue leaves in the interior foothills of central and northern California. Quercus turbinella californica is an evergreen scrub oak of desert and semidesert slopes in central and southern California. The two semispecies come into contact and hybridize at various localities in the foothill zone of south-central California. The population analyzed in detail by Benson, Phillips, and Wilder (1967) is in Tejon Pass in this general zone of contact. The Tejon population is spread over an ecologically diversified area with hill slopes exposed to different points of the compass. The hot dry slopes facing the south and southeast are most like the normal habitat of the desert scrub oak, Quercus turbinella californica. The slopes facing north and northeast, which are also dry but receive less insolation, approach the usual conditions of the more northern tree oak, Q. douglasii. Benson, Phillips, and Wilder (1967) measured a sample of specimens in the Tejon population for 11 vegetative characters and scored them by the hybrid index method. This revealed the presence of numerous backcross types approaching Quercus douglasii and others approaching Q. turbinella californica in about equal proportions in the hybrid population as a whole. But, when the data were reanalyzed, taking into consideration the microgeographical position of the individual plants, a different picture emerged. The plants on the slopes facing north and northeast tend to resemble Q. douglasii in morphology. The slopes facing south and southwest, on the other hand, harbor mainly plants with hybrid index values

Introgression approaching Q. turbinella californica 1967).

223

(Benson, Phillips, and Wilder

M-V Linkage in Relation to

Introgression

Genes determining various morphological characters in higher plants, including those which are first expressed in the flowers, fruits, or late stages of vegetative development, are commonly linked with genes which affect growth and vigor in early developmental stages. And related plant species often differ allelically with respect to the linked morphological and viability genes. This is the M-V linkage which we discussed briefly in chapter 8. There is some evidence, furthermore, for the existence of M-V linkage systems on the rearranged segments differentiating chromosomally intersterile plant species as well as on the homologous segments of interfertile races or species (see chapter 8). The M-V linkage is certainly a widespread and probably a general condition in plant species. Here we will discuss the bearing of this widespread type of linkage on the analysis of introgression. Suppose that we find a plant population exhibiting the following features. (1) The population is variable. (2) The plants closely approach a certain species, A, in a combination of morphological characters, but vary in the direction of another species, B. (3) They also approach species A in their ecological preferences. (4) They are interfertile with species A but form chromosomally sterile F, hybrids with species B. By the standard methods of introgression analysis we would conclude that our hypothetical population is a product of introgression of particular genes from species B into species A. We would postulate a former event of hybridization of A x B, followed by repeated backcrossing to A, with selection for favorable backcross types in the environment of A. Undoubtedly, introgressive hybridization will produce the effects described above. But it must be pointed out that the same end results can also be attained by another and different mode of hybridization. If morphological genes are commonly linked with viability genes, and if plant species differ allelically in respect to these M-V linkage systems, then interspecific hybridization followed by straight inbreeding and selection of the inbred products will bring about the combination of effects listed

224

Natural

Hybridization

and its

Products

in a preceding paragraph. That combination of features is therefore not absolutely diagnostic of introgression (Grant 1967). We have an experimental case of introgression-like effects without introgression in Gilia. The parental species involved in this case are Gilia modocensis and G. malior. They are autogamous and tetraploid species belonging to the G. inconspicua complex, which we introduced in chapter 11. Our interest here centers on certain inbred lines derived from the Fi hybrid of G. malior x modocensis, particularly the lines designated Branch II (Grant 1966b, 1967). See figure 17.8. The Fi hybrid of Gilia malior x modocensis is highly but not completely sterile and has a low degree of chromosome pairing. There is cytogenetic evidence apart from the observed pairing relationships in the hybrid to indicate that the parental species differ by numerous independent segmental rearrangements (Grant 1964c, 1966c). The Fi hybrid produced some progeny without change in ploidy. These

Gilia

malior

G. modocensis

Branch II,

F i g u r e 1 7 . 8 . T w o intersterile species of Gilia and one of their hybrid derivatives. The latter was derived by inbreeding and selection for viability. It is close to one parental species in morphology and fertility relationships. (Grant 1967)

F|

Introgression

225

were propagated as a series of inbred lines. They were selected for fertility and for vigor in an environment favorable to the modocensis-like types. They were not selected on the basis of morphological characters as such. The later-generation selection products were vigorous and fertile (Grant 1966b, 1966c). One set of related lines derived from the same parental individual in F2 and referred to collectively as Branch II turned out to resemble closely the Gilia modocensis parent in morphology (figure 17.8). Branch II varied, however, in the direction of G. malior. Fertile plants of Branch II were backcrossed to both parental species. The backcross hybrid between Branch II and Gilia modocensis was fertile with complete bivalent pairing, while the reciprocal backcross hybrid with G. malior was highly sterile with reduced pairing (Grant 1966b). Let us suppose now that Branch II plants were found growing in nature and were analyzed morphologically, ecologically, and cytogenetically. It would be routinely identified as a deviant race of Gilia modocensis containing particular genes and their morphological expressions introduced by introgression from G. malior. But we know the whole pedigree in this case and can positively rule out backcrossing in the ancestry of Branch II (Grant 1966c, 1967). It is apparent, therefore, from experimental studies as well as from theoretical considerations, that effects indistinguishable from those brought about by introgression can arise along an alternative pathway. The existing methods of introgression analysis do not discriminate between (a) backcrossing and selection and (b) selection in interaction with M-V linkage (Grant 1967). It becomes necessary under the circumstances to reexamine the criteria for identification of introgression and to reconsider some of the cases identified by these criteria. Perhaps the best criterion to use in the present stage of our understanding is the breeding system of the plants involved. In predominantly autogamous plants the probability of a hybrid reproducing by backcrossing is exceedingly low. To infer introgression from the observation of introgression-like variations in such plants is unwarranted. Introgression has in fact been inferred from an introgression-like variation pattern in certain hybridizing autogamous plants, particularly the Elymus glaucus group and the Aegilops variabilis group (Snyder 1951; Stebbins 1956; Zohary and Feldman 1962). But the interpretations in these cases should perhaps be reconsidered in the light of M-V linkage.

226

Natural

Hybridization

and its

Products

In outcrossing plant groups, on the other hand, backcrossing by hybrids is their most likely method of reproduction. This is a consequence, in the first place, of the breeding system in itself. It follows, moreover, from the relative numbers of viable gametes contributed to the gamete pool by the numerous fertile species plants and by the rare sterile hybrid plants, respectively, as discussed earlier in this chapter. The inference of introgression from an introgression-like variation pattern in outcrossing plants has a high probability of being correct, in line with the views of Anderson (1949, 1953), Stebbins (1950), and others. The examples of introgression selected for purposes of illustration in this chapter—iris, Helianthus, Juniperus, etc.—all involve outcrossing plants. In plant groups possessing a breeding system intermediate between outcrossing and inbreeding, hybrid reproduction can be expected to follow a mixture of pathways. This may be the case in parts of the Aegilops variabilis group, to judge from work of Pazy and Zohary (1965).

Transgression

of Chromosomal

Sterility

Barriers

Introgression between plant species possessing different structural karyotypes must involve the passage of genes through a chromosomal sterility barrier. The problem before us in this section is to consider the mechanism by which this transgression of chromosomal sterility barriers probably or necessarily takes place. The account given here is a revision and extension of earlier treatments (Grant 1958; 1963:487 ff.). The preconditions are the following. We have a hybrid population consisting of individuals of two species and their hybrids. The plants are outcrossing. The chromosomes of the parental species differ by segmental rearrangements, which lead to reduced pairing or unbalanced gametes or both in the hybrids, and the hybrids consequently exhibit chromosomal sterility or semisterility. Certain genes borne on these chromosomes in one parental species, when introduced into the genotype of the second parental species, form a recombination product which is adaptively superior in the environment of the second species. The foregoing conditions are probably often realized in nature. The process of sterility transgression which is expected to take place

Introgression

227

under the preceding conditions is complicated. It involves the simultaneous operation of several separate subprocesses. In the first place, the plants constituting the hybrid population reproduce by outcrossing. Second, the homeologous chromosomes form bivalents and undergo crossing-over in at least some of the pollen and egg mother cells in the structural heterozygotes. Gamete selection is occurring so as to eliminate the deficiency-duplication products of meiosis and favor the genically balanced gametophytes. At the same time Darwinian selection between different individuals in the population is also going on. The latter selective process is favoring certain interspecific recombination genotypes in the environment of one parental species. The process of sterility transgression will be considered first in relation to a particular model. We have a recipient species (R) and a donor species (D) which contributes some of its genes to the former by introgression. Recombination genotypes containing certain D genes in the genetic background of R have a high adaptive value in the environment of species R. These interspecific recombination types have a higher selective value in this environment than the corresponding allele combination found in unintrogressed species R . The genes involved in the adaptively significant gene combination are borne on chromosomes which differ between the two species in structural rearrangements. In order to simplify the discussion, we focus our attention on one particular chromosome (/) present in two structural forms in the parental species (/ r and I d ). Chromosomes l r and l d differ with respect to a transposition, as shown in figure 17.9A. A segment in the left arm has exchanged places with a segment in the right arm during the phylogenetic history of chromosome I d , but this rearrangement has not occurred in chromosome / r , and the two chromosome types therefore differ structurally in the manner indicated in Figure 17.9A. The central or interstitial segments of chromosomes I r and l A are homologous (see figure 17.9A again). We postulate that the genes of species D which form an adaptively valuable recombination with genes of R are borne on these interstitial segments of chromosome / d . The R genes involved in the same valuable gene combination are on the corresponding homologous segments of / r , or here and on other chromosomes of the complement too. The R x D hybrids may have variable reduced pairing of the / chro-

228

Natural

Hybridization

and its Products

mosomes in consequence of the lack of complete homology. In the lrld bivalents, when such bivalents do form, crossing-over can be expected to occur fairly regularly in the homologous interstitial segments (figure 17.9B). Half of the meiotic products of such crossing-over carry deficiencies and duplications for the transposed segments (figure 17.9B), and yield inviable gametophytes, making the hybrid semisterile. The transposed segment, in other words, acts as a sterility factor when crossed over into the homologous chromosome in such a way as to form a deficiency-duplication product. The deficiency-duplication chromosomes will drop out of the gamete pool in each generation. And so will the D genes borne on their interstitial segments. The only functioning gametes are those carrying the noncrossover / r or l d chromosomes (figure 17.9B). Therefore the Fi hybrid, after sib crossing, yields F 2 progeny segregating into the two parental and the hybrid classes ( / r / r , ldId , lrId); and, on backcrossing to R give progeny segregating into one parental and one hybrid class ( / r / r , / r / d ) . The lTld types are semisterile in F 2 , B\ , and B2, as they are in F ^ Conversely, the only fertile progeny in later generations are the parental types, lrlr and l d I d .

.

With the usual mode of crossing-over in chromosome /, as diagrammed in figure 17.9B, therefore, the hybrid population segregates into the parental types and the Fi type for morphological and physiological traits determined by genes on this chromosome. Introgression of genes on chromosome I from species D into species R is blocked by the linkage of these genes with the sterility factors on the same chromosome. But the genetic results of hybridization are very different if crossingover takes place in the structural heterozygote at crossover points that separate the transposed segments from the homologous interstitial segments. Double crossing-over that will accomplish this separation is diagrammed in figure 17.9C. This mode of crossing-over may or may not happen to occur in the Fi generation; if not, it can take place in any structural heterozygote in any subsequent backcross generation. Double crossing-over of this sort yields crossover chromosomes of the type labeled lTd in figure 17.9C. These lrd chromosomes are genically balanced, and hence form viable gametes. They are structurally homologous with / r chromosomes, so that they can form fertile heterozygotes of

Introgression

229

the constitution l r ! r d , and thus they can be incorporated into the chromosome pool of species R. These balanced lrd chromosomes also carry large blocks of D genes on their central interstitial segments. The I r d chromosomes which are at once balanced internally and homologous with / r can therefore serve as a bridge for the passage of D genes into species R. Darwinian selection can now bring about an increase in frequency of the new I r d chromosomes, or parts of them, in the environment of species R. We have discussed the process of sterility transgression with reference to a simplified model involving a transposition. It can be shown that the process will work in a similar fashion with other types of rearrangements, particularly translocations and reinversions, under appropriate conditions of crossing-over. Furthermore, and significantly, it will work on larger numbers of rearrangements on separate chromosomes or chromosome arms, so as to circumvent strong chromosomal sterility barriers as well as weak ones. Environmental selection for a new interspecific gene combination pulls the significant genes of the donor species through a sterility barrier composed of various types and numbers of chromosomal rearrangements. Such a selective process can build up a new race possessing various morphological and physiological characters like the donor parental species, but belonging to the fertility group of the recipient species. It remains to consider the role of the genes on the rearranged segments in the process of introgression through a chromosomal sterility barrier. The convergence referred to above is limited by these genes. The morphology-determining genes on the rearranged segments of species D will be weeded out along with the segments themselves by gamete selection in the hybrid population. If the morphological genes of D are linked with viability genes of D on the same segments, moreover, they will be discriminated against by Darwinian selection also in the environment of species R. The two types of linkage—that between the morphological genes and the chromosomal sterility factors and the M-V gene block on the rearranged segments—prevent the introgressive race from having too close a resemblance to the donor species in its morphological characters. The rearranged segments, in other words, restrict the introgression of alien genes, and act as a brake on the process of convergence in morphological and physiological traits.

230

\atural

Hybridization

Introgression

and its

and Race

Products

Formation

Particular races have been formed by introgression in many species groups. The example of the Helianthus annuus group described earlier is representative. Races of introgressive origin have been described in the following genera among others: Dracophyllum (DuRietz 1930), Coprosma (DuRietz 1930), Abies (Mattfeld 1930; review in Stebbins 1950), Tradescantia (Anderson and Hubricht 1930), Cistus (Dansereau 1941), Helianthus (Heiser 1949b), Gilia (Grant 1950a, Grant and Grant 1960), Pi nus (Haller 1962), and Purshia (Stebbins 1959). In many cases the races of introgressive origin are sympatric, at least partially, with the donor species or semispecies. Introgressive race formation is fairly common in the diploid, largeflowered, outcrossing Cobwebby Gilias (Gilia section Arachnion) of southern California. The species and semispecies are sympatric in numerous combinations and occupy recent arid and semiarid habitats. Races of introgressive origin appear to be favored adaptively in some of these habitats. In this group the following geographical races, named as subspecies, have the parentages noted in parentheses, on morphological, ecological, and cytogenetic evidence (Grant and Grant 1960): G. latiflora davyi

(G. latiflora introgressed by G.

G. leptantha transversa

(G. leptantha leptantha introgressed by G. latiflora ) (G. carta introgressed by G. leptantha purpus ii) (G. ochroleuca bizortata introgressed by G. leptantha leptantha)

G. caria speciosa G. ochroleuca

vivida

tenuiflora)

Introgressive hybridization is a force promoting convergence in morphological characters and ecological preferences in a zone of sympatric contact between two species or semispecies. Other forces work in the opposite direction. Sympatric contacts under the appropriate conditions can also lead to selection for reproductive isolation and/or character displacement, as we noted in chapter 15. The results in these cases are enhanced divergence

¡ntrogression

231

between the species in their sympatric zone. Sympatry is a challenge which evokes a variety of responses. Character displacement is well known in animals, but has been reported only occasionally in plants. Some botanists believe that character displacement has not been looked for sufficiently in plants. This is probably true and may be part of the story. The aspect of the story that I would emphasize, however, is the role of introgression. Convergence in morphological and physiological traits due to introgression is common in plants and probably predominates over character displacement in many or most cases.

Conclusions Natural hybridization occurs fairly commonly between sympatric species or semispecies of plants where the ecological conditions are permissive. A large amount of evidence which is descriptive but not quantitative indicates that natural hybridization is more common in plants than in higher animals with internal fertilization. The occurrence of hybridization in plants is not well correlated with the strength of internal reproductive barriers between the parental species, up to the limiting point of absolute cross-incompatibility. The occurrence of hybridization is, on the other hand, closely correlated with the availability of open habitats for the establishment of the hybrids and hybrid progeny. Such open habitats are formed on a wide scale by human activities in the modem world, but they have also been created by various natural processes throughout geological history. The reproduction of natural hybrids in plant groups with an outcrossing breeding system commonly follows the pathway of back-crossing and introgression. The infiltration of genes from the donor species to the recipient species is much more strongly restricted where these parental species are differentiated in respect to chromosomal rearrangements than where they possess completely homologous chromosomes. In autogamous plant groups, hybridization may lead to end results which are quite similar to those usually taken as diagnostic of introgression. The alternate pathway involves inbreeding (rather than backcrossing), segregation, and selection.

(A) Parental Types Species R

t/zz/jzi/z

M r

Species D

Id!d

F i zygote

M d

(B) Products of Meiosis in Fi or Backcross Hybrid, with crossing over between arms Pairing

Gametes

'r [

rd

^dr Id (C) Products of Meiosis in Fi or Backcross Hybrid, with double crossing-over between rearranged segments Pairing

r-r-rrr-m

F i g u r e 1 7 . 9 . Cytogenetic behavior of a structurally differentiated pair of chromosomes in a hybrid population. The parental species R and D carry and introduce into the population two partially h o m o l o g o u s forms of this chromosome, designated l r and l d respectively, which differ by a transposition. Further explanation in text.

Introgressioti

233

Introgressive hybridization plays a key role in race formation in many predominantly outcrossing plant groups. Races of introgressive origin are found in species groups in Pinus, Abies, Quercus, Purshia, Cistus, Coprosma, Dracophyllum, Helianthus, Gilia, Tradescantia, and other genera.

CHAPTER

The

EIGHTEEN

Syngameon

Localized natural hybridization and limited gene exchange between semispecies which are ordinarily reproductively isolated serves to link these semispecies together into a more inclusive unit of interbreeding. This larger unit is the syngameon. A situation of this sort was described by Lotsy (1925, 1931) in a segment of the genus Betula. On the European continent there are many forms of Betula, of various and disputed taxonomic rank, which hybridize on a wide scale. Lotsy (1925) concluded that " w e have in Betula one very large pairing community, one syngameon." DuRietz (1930:367 ff.) reviewed many similar cases in Nothofagus in New Zealand, Geum in Europe, Melandrium in Sweden, Salix in northern Europe, etc. All these groups represent "large and highly polymorphic 'hybrid'-syngameons in which the species have got more or less lost . . . " (p. 384). Several later workers have developed the concept of the syngameon (Cuenot 1951; Grant 1957; Beaudry 1960). The syngameon can be defined in modern terms as "the sum total of species or semispecies linked by frequent or occasional hybridization in nature; [hence] a hybridizing group of species. . . . " (Grant 1957:67). The syngameon is the most inclusive unit of interbreeding in a hybrid-

The Syngameon

235

izing species group. It may have a relatively simple internal structure, consisting of two or three semispecies, but it may also be highly complex. In some plant groups the syngameons that have developed are vast networks of semispecies, composed of numerous taxonomic species and extending across geographical areas of continental extent. In its external face, and in its reproductive interactions with other population systems, a syngameon behaves like a biological species, i.e., it is a reproductively isolated entity. It differs from a biological species in its more complex internal structure. It should be noted that the principal components of syngameons are generally treated, and properly so, as species in formal systematics; the components are usually good taxonomic species. Syngameons are developed in many plant genera. Some older examples in Europe and New Zealand were listed above. Many examples are known in western North America in Pinus, Juniperus, Quercus, Mimulus section Diplacus, Iris, Aquilegia, Gilia section Arachnion, and other genera. Selected aspects of these syngameons have been described in previous chapters. Here we will outline the structure of the syngameon as a whole in the case of two plant groups.

Pacific Irises The most complex syngameon which has been analyzed thoroughly by a single investigator is probably the one in the Pacific coast irises of North America (Lenz 1959a). The group is Iris section Apogon series Californicae. It is a natural group consisting of 11 taxonomic species ranging from southern California to Washington. Their fertility relationships were discussed in chapter 11. Strong internal isolation does not exist. The species or semispecies occupy a variety of ecological zones within their overall distribution area. Thus Iris douglasiana occurs along the coast, /. macrosiphon in foothills back from the coast, and so on. The habitats of these and other semispecies are often contiguous, permitting marginal or neighboringly sympatric contacts. Furthermore, the ecological barriers have been broken down in many places by logging, road building, and other human disturbances of the environment. The semi-

236

Saturai

Hybridization

and

its

Products

species then engage in natural hybridization in numerous combinations (Lenz 1958, 1959a). All of the taxonomic species except /. munzii are linked by natural hybridization. In other words, the eleven taxonomic species consist of one isolated species, I. munzii, and ten semispecies linked together in a syngameon, as shown in figure 18.1 (Lenz 1959a). Some semispecies like I. douglasiana are central in the syngameon, and have numerous linkages, while others like /. hartwegii are peripheral with only one linkage. Some pairs of semispecies hybridize extensively; some hybridize to a moderate extent; and some only slightly, as indicated by the lines in figure 18.1.

White Oaks Syngameons have developed independently in both subgenera of Quercus, the black oaks and the white oaks. The two series of syngameons

Figure 18.1. Structure of the syngameon in the Pacific coast irises. The breadth of the connecting lines indicates the amount of natural hybridization. Bars = extensive hybridization; solid lines = moderate hybridization: broken lines = slight hybridization. (Lenz 1959a) Abbreviations of iris species and semispecies are as follows. BRA = / . bructeatu. CHR = / . chrysophylla. DOU = /. douglasiana. FER = 1. fernaldii. HAR = /. hartwegii. INN = /. innominata. MAC = /. macrosiphon. MUN = /. munzii. PUR = /. purdyi. TEN = /. tenax. TMA = /. tenuissima. x TH = I. thompsonii = I. douglasiana x I. innominata.

The

Syngameon

237

are strongly isolated reproductively from one another. The syngameon in the white oaks is the most extensive geographically and taxonomically. The white oaks in the United States have been studied by many different botanists. T h e literature on natural hybridization in these oaks is voluminous and scattered. Palmer (1948) summarized the known natural hybrid combinations in North American oaks, but many new papers have appeared since 1948, and there has been no recent general review. 1 will attempt to sketch the white oak syngameon in the United States in its main outlines here. Key references will be cited; but a complete literature survey cannot be given here. It is convenient for purposes of presentation to break the white oak syngameon down into regional segments. W e will start with the California segment and proceed across the country to the eastern United States. Seven species of white oaks in California have ranges that overlap in various combinations. In some parts of their overlapping areas, where ecological isolation breaks d o w n , they hybridize. In figure 18.2 these semispecies are arranged in their approximate geographical order from

238

Saturili

Hybridization

and its

Products

States.

north to south, and the known hybrid connections between them are indicated. The California segment of the syngameon is interesting in terms of the morphological and ecological extremes that are linked together. Quercus garryana is a forest tree with large deeply cleft leaves in the Pacific northwest. Quercus dumosa and Q. turbinella are low shrubs with small spiny leaves in arid chaparral and desert associations respectively. Yet these extreme forms are connected directly by hybridization or indirectly through intermediary populations. At particular sites in northern California the forest tree, Q. garryana, hybridizes with the dry foothill tree, Q. douglasii (Stebbins 1950), and with the scrub oaks, Q. dumosa and Q. durata (Tucker 1953). In southern California Q. douglasii hybridizes with the chaparral Q. dumosa, and Q. dumosa with the desert Q. turbinella (Tucker 1952; Benson 1962). These and other combinations are shown in figure 18.2. A single hybrid oak tree in desert mountains in southern California is believed, on morphological characters, to have Q. lobata as one parent (Tucker 1968). Quercus lobata, a cismontane valley oak, does not occur anywhere near the hybrid site today, but could have occurred there during

239

The Syngameon

alba

hybrids

bicolor

? Wi •• alba

Figure 18.4. Leaves of Quercus

hybrids

prinus

alba and two of its hybrids. (Hardin 1975)

a pluvial stage in the Pleistocene. The other parent of this hybrid oak has been thought to be the locally abundant Q. turbinella (Tucker 1968); but recent reinvestigations suggest that the other parent may instead be a new southwestern species currently being described as Q. cornelius-mulleri (Kevin Nixon, personal communication, and in Madrono). The white oak syngameon continues in the southwest from Arizona and Utah to west Texas with the constitution shown in figure 18.3. The semispecies are again arranged in their approximate geographical positions in this graph. Hybridization between the desert scrub oak, Q. turbinella, and the Rocky Mt. tree, Q. gambelii, occurs in northern Arizona today, and occurred in Utah in the past as testified to by old isolated hybrid clones (Cottam, Tucker, and Drobnick 1959). The Rocky Mt. Q. gambelii and the bur oak of the eastern United States, though geographically isolated today, were connected by hybrid-

240

Natural Hybridization

and its Products

Figure 18.5. Structure of the white oak syngameon in the eastern United States. (Hardin 1975) ization in the past, probably in the Pleistocene, on the basis of intermediate characters in s o m e aberrant populations ( M a z e 1968). In the southwest Q. turbinella

also hybridizes with Q. grísea,

in turn hybridizes with Q. mohriana, havardi

which

which in turn hybridizes with Q.

(Muller 1951a) (figure 18.3).

In T e x a s the post o a k , Q. stellata, white oak s y n g a m e o n . Quercus

stellata

o c c u p i e s a central position in the hybridizes in T e x a s with:

The

Syngameon

241

Q. havardi, Q. mohriana, Q. virginiana, Q. minima, Q. oleoides, Q. margaretta, Q. alba, Q. Ivrata, Q. macrocarpa, Q. prinus, and Q. sinuata. Some of these semispecies also hybridize with one another, e.g., Q. virginiana x Q. minima (Muller 1951a, 1970). Furthermore, in the eastern United States, Q. stellata hybridizes with some of these same semispecies as well as with other semispecies, e.g., Q. bicolor and Q. prinoides (Palmer 1948; Hardin 1975). In eastern North America the white oak, Q. alba, is a central figure in the syngameon, hybridizing with eleven other semispecies. Quercus alba hybridizes in this region with: Q. michauxii, Q. muehlenbergii, Q. prinoides, Q. lyrata, Q. macrocarpa, Q. stellata, Q. bicolor, Q. austrina, Q. prinus, Q. margaretta, and Q. robur (Hardin 1975). Leaf characters and leaf variation in two of these hybrid combinations are shown in figure 18.4 Hardin (1975) has presented an excellent treatment of the white oak syngameon in eastern North America, with particular reference to Quercus alba. Many of the species in this area are wide-ranging and have broad distributional overlaps with other species. Natural hybridization occurs sporadically within the synpatric areas. The composition of the resulting syngameon is shown graphically in figure 18.5. It will be seen that 14 of the 16 species recognized are members of the syngameon (Hardin 1975). The English oak, Q. robur, is cultivated in eastern North America and hybridizes spontaneously with Q. alba in a few locations (Hardin 1975). In the British Isles where it is native, Q. robur hybridizes with Q. petraea (Cousens 1963, 1965; Stace 1975). The white oak syngameon thus bridges the Atlantic and continues in Europe.

CHAPTER

Hybrid

NINETEEN

Speciation

One of the historical problems of plant evolution, a problem which was discussed by successive authors through the eighteenth and nineteenth centuries, is the role of hybridization in species formation. Is natural hybridization a mechanism for the production of new plant species? This question was answered in the affirmative by several early authors. But, as we shall see, these same authors failed to consider some essential parts of the problem, especially segregation in the hybrid progeny, and their affirmative answers were therefore premature and prescientific. The early discussions did, however, serve to keep the problem alive until it could be approached in the light of genetics by modern workers. We saw in an earlier chapter that new races arise sometimes as products of natural hybridization. This is hybrid race formation but it is not necessarily hybrid speciation. The new race, once formed by processes involving hybridization, could go on to diverge to the species level in geographical isolation from the original ancestral race. This would be a special case of geographical speciation in which hybridization produced the necessary variations at an intermediate stage. We do not need to consider this case further here. By hybrid speciation we mean the origin of a new species directly from a natural hybrid. This definition brings the problem into focus. It is clear

Hybrid Speciation

243

that the same sexual process which produced the natural hybrid will also bring about the breakup of its gene combination by segregation in later generations. Therefore an essential part of the mechanism of hybrid speciation is the stabilization of the breeding behavior of the hybrids.

Stabilization

of Hybrid

Reproduction

Segregation in the Fi or any subsequent generation derived from an interspecific cross can be completely stopped or greatly restricted by various mechanisms. The known methods of stabilization in hybrid reproduction are: (1) vegetative propagation; (2) agamospermy; (3) permanent translocation heterozygosity; (4) permanent odd polyploidy; (5) amphiploidy; (6) recombinational speciation; and (7) the segregation of a new type isolated by external barriers. Methods 1 and 2, known collectively as apomixis, are introduced briefly in a later section of this chapter and discussed more fully in part 8. Methods 3 and 4 are exemplified by the genetic systems in the Oenothera biennis group and the Rosa canina group, respectively, which are also discussed in part 8. Method 5, amphiploidy, or allopolyploidy, is introduced in this chapter and covered in more detail in part 6. Methods 6 and 7, in contrast to method 5, both involve recombination on the homoploid level. Method 6, designated recombinational speciation, is the formation of a new homozygous recombination type for chromosomal sterility factors. It is introduced in this chapter and discussed more extensively in chapter 20. Method 7 is the formation of a new recombination type isolated by external barriers. We call this process hybrid speciation with external barriers, and discuss it briefly later in this chapter. The various methods listed above permit the multiplication of any given genotype of hybrid origin. The genotype in question may be that of the F, hybrid itself or of a later-generation derivative, and it may be heterozygous or homozygous. Under methods 6 and 7 the hybrid derivative attains a true-breeding condition by becoming homozygous, at least temporarily and at least partially. The hybrid derivative breeds true for a highly heterozygous condition under methods 1 to 4. Method 5 is a compromise solution combining features of both homozygosity and heterozygosity. The hybrid

244

Natural

Hybridization

and

its

Products

derivative under method 5 is homozygous as far as meiosis is concerned, and is therefore tme-breeding, but retains the opposite gene systems of two parental species. In any case the particular gene combination of hybrid origin, if adaptively valuable in some available environment, is able to increase in numbers and form a population of its own in the area covered by the favorable environment. The derivative population may be a new species or microspecies of hybrid origin. Now the method of stabilization of hybrid reproduction determines the type of species resulting from the hybridization. Vegetative propagation and agamospermy are asexual methods of reproduction. These means of uniparental multiplication can produce clonal or agamospermous microspecies. Permanent structural or numerical hybridity (methods 3 and 4) also produce uniparental microspecies. Amphiploidy, recombinational speciation, and hybrid speciation with external barriers, on the other hand, are methods of stabilization which take place in a context of sexual reproduction. They give rise, therefore, to new biological species. The new species descended from hybrids or hybrid derivatives are evolutionary species in every case, according to the concepts discussed in chapter 7. But only where sexuality and interbreeding persist throughout the critical formative stages can we speak of the hybrid origin of new biological species. The difficulties of hybrid speciation are greatest in the latter case where sexuality remains in force. The hybrid formation of a new biological species is theoretically the more difficult but also the more interesting process.

Historical

Background

In the pre-Mendelian period the hybrid origin of new plant species was proposed by Linnaeus in 1744 and 1760, William Herbert in 1820, Charles Naudin in 1863, and Anton Kemer in 1891. The line of thought was continued in the early post-Mendelian period by Lotsy, Hayata, and others. The references as cited in our bibliography are: Linnaeus (1760, 1790), Herbert (1820), Naudin (1863), Kerner (1894-1895 2:576 ff.), Lotsy (1916), and Hayata (1921): see also DuRietz (1930:400 ff.) for a review.

Hybrid

Speciation

245

Linnaeus and his student, Rudberg, first suggested the hybrid origin of new plant species in 1744 in connection with Peloria, a form of Linaria with radially symmetrical flowers. Hybrid speciation is suggested again in the Disquisitio de Sexu Plantarum (1760). Here Linnaeus states that: "It is impossible to doubt that there are new species produced by hybrid generation." Kemer (1894-1895) started with the known facts that crossing generates new forms and that natural interspecific hybrids occur in many plant groups, and concluded from these premises that "hybrids could originate new species." To be sure, many hybrids are formed in nature, but only a few of them succeed in becoming new species. The important limiting factor, according to Kerner, is the environment. Only where there is a favorable and open habitat not occupied by the parental species can the hybrid plants establish themselves and increase in numbers. But such open habitats do occur here and there in nature. And so do species of hybrid origin. Kerner cites examples in Rhododendron, Salvia, and Nuphar in Europe. Rhododendron intermedium, for example, is a product of hybridization between R. ferrugineum and R. hirsutum. In some places in the Alps, R. intermedium occurs as a rare associate of the parental species, but in other places it forms large populations of its own and outnumbers the parental types. Here R. intermedium has attained the status of a true species (Kerner, 1894-1895 2:588-589). Linnaeus and Kerner were the outstanding early exponents of the idea that natural hybrids can be the starting points of new species. Analysis of their discussions reveals that they both ignored two important aspects of the problem which any satisfactory hypothesis of hybrid speciation must take into consideration, namely, sterility and segregation. Linnaeus and Kerner did not recognize the phenomenon of segregation. Indeed, Kerner postulates that: "hybrids transmit their form unchanged to their posterity . . . provided the pollen from other species is excluded." Linnaeus had done little experimental work on plant hybridization, and Kerner's experimental work dealt with other aspects of plant reproduction. Consequently neither author was prepared to realize that there is any problem of hereditary transmission in passing from natural hybrid to daughter species in fertile species crosses. The list of early plant hybridizers who did know about segregation and

246

Saturai

Hybridization

and

its

Products

did not discuss species formation through hybridization is a long one, ranging from Kolreuter ( 1 7 6 1 - 1 7 6 6 ) to Mendel, Darwin and

Focke

(1881). Naudin's (1863) paper is very interesting because it contains discussions of both segregation and hybrid speciation. Naudin treated the former as a fact and the latter as a possibility. He recognized that the fact of segregation created difficulties for the theory of hybrid speciation. However, he allowed that exceptions might occur in which hybrid characters become fixed in later generations, and such exceptional cases could lead to the formation of new species (Naudin 1863). The difficulty of sterility barriers was also set to one side in the early discussions of hybrid speciation. It was known that some species pairs are intersterile and others are interfertile. T h e possibility of hybrid speciation was restricted to cases in which the hybrid is fertile (Kerner, 1 8 9 4 - 1 8 9 5 2:587). This focused the attention arbitrarily on one special aspect of the whole question. It made the general problem of hybrid speciation appear to be simpler than it actually is. With the rise of modem genetics in the early decades of this century, the old problem of hybrid speciation could at last be stated correctly. The problem could be seen to involve sterility barriers and segregation as well as morphological and physiological traits. In order to solve this problem it was necessary to find a mechanism by which a new, internally isolated, constant type could arise from species hybrid without loss of sexuality. The first major breakthrough was W i n g e ' s (1917, 1924) hypothesis of amphiploidy, to use a much later term, which postulates that a new constant species can arise from a hybrid between two preexisting species following chromosome number doubling. W i n g e ' s hypothesis was soon confirmed experimentally by other workers in Nicotiana, sica, Galeopsis,

Raphanus-Bras-

and other plant groups (Clausen and Goodspeed 1925:

Kaipechenko 1927: Miintzing 1930b, 1932). Amphiploidy is a mode of speciation which involves unidirectional increases in number of chromosome sets, whereas many plant species have obviously evolved on the diploid level, and the question therefore remained whether hybrid speciation could take place without change in ploidy. Winge himself returned to this question in his later work on f>on a n d Erophila,

a n d L a m p r e c h t t a c k l e d it in Phaseolus

Tragopo-

(Winge

1938,

1940; Lamprecht 1941). The stated objective of the experiments in Tru-

Hybrid Specialion

247

gopogon, Erophila, and Phaseolus was to determine whether new fertile types or microspecies could be produced by hybridization without the accompaniment of amphiploid doubling. These experiments paved the way for studies by other workers on a process of hybrid speciation, alternative to amphiploidy, which is now known as recombinational speciation.

Introduction

to

Amphiploidy

Chromosomal sterility stems from segmental rearrangements between the parental genomes which upset the course of meiosis, visibly or invisibly, in the hybrid. Either the homeologous chromosomes do not pair normally in bivalents and do not separate properly to the poles at anaphase, or they pair but segregate to yield daughter nuclei carrying deficiencies and duplications. In either case a proportion of the meiotic products are unbalanced and do not develop into functional spores and gametes. This proportion of inviable spores and, hence, the degree of sterility rises rapidly with increase in number of heterozygous rearrangements (see chapters 8 and 9). Let us suppose that the chromosomally sterile hybrid undergoes doubling of chromosome number. Then, disregarding other possible complicating factors for the present, it will become meiotically normal and gametically fertile. This recovery of fertility on the new polyploid level is most clear-cut in the first case mentioned above where the chromosomes do not form bivalents regularly in the diploid hybrid. The structurally well-differentiated genomes of the parental species can be symbolized as A and B respectively. The genomic constitution of the two diploid species is AA and BB, and that of their hybrid is AB. The chromosomes belonging to the A set have no homologous partners to pair with at meiosis in the hybrid, and neither do the B chromosomes. But the situation is entirely different in the allotetraploid (or amphidiploid) derivative of this hybrid with the genomic constitution to AABB. Now there exists a homologous pair of each chromosome type in each genome. Consequently meiosis and gamete formation can proceed normally. In the second case mentioned earlier the homeologous chromosomes in the hybrid do form bivalents but segregate to produce genically unbal-

248

Natural

Hybridization

and its

Products

anced meiotic products. The diploid parents possess different subgenomes. A, and A,. Their F, hybrid has the genomic constitution A,A, and has chromosomal sterility associated with visibly normal meiosis. In this case also the tetraploid derivative, ASA,A,A,. is likely to recover fertility, or at least semifertility, as a result of what Darlington called differential affinity (Darlington 1932). Each A, chromosome and each A, chromosome has a completely homologous partner in the tetraploid. These homologous chromosomes pair preferentially in the tetraploid and pass to opposite poles to yield segmental ly and genically balanced products of meiosis. The tetraploid derivative of the chromosomally sterile hybrid, whether it has the genomic constitution AABB or AsA,AtA,, is not only fertile itself, but is also reproductively isolated from its diploid parents. Crosses between tetraploid and diploid plants frequently run into incompatibility barriers, and the hybrids, if any arise, are diploid and hence usually sterile. The classical cases of Primula kewensis, Raphanobrassica, and Galeopsis tetrahit illustrate the various types of genomic constitution in experimental amphiploids. The F, hybrid of P. floribunda (2n = 18) x P . verticillata (In = 18), widely known by the name P. kewensis, is a diploid perennial herb like its parents. It showed normal bivalent pairing but was highly sterile. In three different years the otherwise sterile hybrid plant spontaneously gave rise to fertile branches, which, when studied cytologically, turned out to be tetraploid (2n = 36) and to have predominantly bivalent pairing. Evidently, preferential pairing of completely homologous chromosomes occurred in the tetraploid branches. These fertile shoots arose by somatic doubling in a bud or sector of a bud. They produced seeds which developed into fertile and fairly uniform F 2 progeny (Newton and Pellew 1929; Upcott 1939). the amphiploid A contrasting condition was found in Raphanobrassica, derivative of Raphanus sativus x Brassica oleracea. The parental species are both diploid with 2/i = 18 chromosomes. The Fi hybrid exhibits complete failure of chromosome pairing and is highly sterile. It produces some unreduced diploid gametes. Union of these gave rise to tetraploid (In = 36) plants in F 2 , which had normal meiosis with regular bivalent formation and were mostly quite fertile. They yielded a morphologically

249

Hybrid Speciation

uniform F3 generation. The new, fertile, true-breeding line is isolated by sterility barriers from the parental diploid species (Karpechenko 1927). Galeopsis tetrahit (2/i = 32), unlike Primula kewensis and Raphanobrassica, is a naturally occurring tetraploid species in northern Europe and Asia. Miintzing proved that this annual herb is an amphiploid derived from two related diploid species in Europe, G. pubescens (2/i = 16) and G. speciosa (2n = 16) (Miintzing 1930b, 1932). The artificial F! hybrid of Galeopsis pubescens x speciosa is meiotically irregular and fairly sterile with 8% good pollen and 5 to 8 bivalents. The Fi hybrids, despite their sterility, produced an F 2 generation consisting of some sterile diploids and one sterile triploid plant. The latter arose from the union of one unreduced gamete and a reduced gamete in F,. The triploid when backcrossed to G. pubescens yielded a single seed which gave a tetraploid plant in F 3 . The tetraploid probably arose from the fertilization of an unreduced 3n egg by a normal In sperm. The tetraploid F 3 plant had 16 bivalents at metaphase of meiosis and produced 70% good pollen. It yielded fertile tetraploid F4 progeny (Miintzing 1930b, 1932). The artificial allotetraploid resembled natural Galeopsis tetrahit in morphology as well as in chromosome number. Furthermore, it is isolated by an incompatibility barrier from G. pubescens and G. speciosa, as wild G. tetrahit is. It could be considered to be a synthetic form of G. tetrahit. To test this assumption, Miintzing crossed the artificial allotetraploid with natural G. tetrahit. The artificial and natural tetraploids crossed easily to produce FiS which were fertile with good chromosome pairing. This was the first experimental resynthesis of a naturally occurring amphiploid species (Miintzing 1930b, 1932).

Introduction

to Recombinational

Speciation

Recombinational speciation is the term adopted to denote one of the lesser known forms of hybrid speciation (Grant 1966e). The process takes place within a breeding system of sexual reproduction. It leads to the formation of a daughter species which is isolated from the parental species by a chromosomal sterility barrier, as in amphiploidy. But, in con-

250

.Watural Hybridization

and its

Products

trast to amphiploidy, the daughter species remains on the same ploidy level as its parents. This result comes about by the formation and establishment in the later hybrid generations of a new homozygous recombination type for the various independent chromosomal rearrangements separating the parental species. The new homozygous recombination type is fertile within the line but intersterile with either parent. The process of recombinational speciation can produce daughter species which are surrounded by sterility barriers of varying strength. The degree of isolation of the daughter species is directly proportional to the degree of intersterility between the parental species. The leading idea for the hypothesis of recombinational speciation was provided by Miintzing (1930a, 1934, 1938). He proposed but did not develop the concept that separate pairs of chromosomal sterility factors can be recombined to give a new fertile type. Gerassimova (1939) confirmed this suggestion experimentally for translocations in Crepis. Stebbins then discussed recombination of sterility factors in relation to hybrid speciation in a general way (Stebbins 1942, 1945, 1950:249, 286-289). Later Stebbins (1957a) and Grant (1958) presented a definite genetical model of recombinational speciation. Still later, in 1963, I reviewed the hypothesis and summarized the evidence for it existing at that time (Grant 1963:469 ff.). The subject of recombinational speciation is introduced here so it can be compared with other modes of hybrid speciation. It is explained and discussed in more detail in the next chapter. It may be noted in passing that the recombination process is just as capable of generating new genie sterility barriers out of preexisting ones as it is of compounding chromosomal sterility barriers. There is some evidence suggesting a recombinational formation of new genie sterility barriers in Phaseoius and new gene-controlled incompatibility barriers in Gilia (Lamprecht 1941; Grant 1954a). Too little is known about this aspect of the problem, however, to warrant a separate discussion in this book. In general the process would be essentially similar to the one, to be described later, involving chromosomal sterility factors.

251

Hybrid Spedation

Hybrid Speciation

with External

Barriers

Hybridization between two interfertile plant species may lead to the formation of diverse recombination types for morphological and physiological character differences between the parental forms. Some of the new character combinations may bring about external isolation of one sort or another between the hybrid derivatives and the parental species. Such externally isolated recombinants, if well adapted to some available environment, can then increase in numbers within the territory of the parent species. Earlier in this chapter we mentioned a historical example of possible hybrid speciation in Rhododendron studied by Kemer (1894-1895). Kerner postulated that R. intermedium arose from a hybrid between R. ferrugineum and/?, hirsutum and became successful in certain places in the Alps. The daughter species of hybrid origin differs from its parents in flower color and, to some extent, in soil preferences. Kemer's account of the case suggests that/?, intermedium is partially isolated f r o m R . f e r r u gineum and R. hirsutum by the pollinating behavior of bees and perhaps by edaphic factors. For a more recent description of the hybrid/?, intermedium and its parental species see Hegi (1909-1931 5:1627-1644). Delphinium recurvatum, D hesperium, and D. gypsophilum form a group of interfertile diploid species in the foothills and valleys of California. On morphological and ecological grounds, D. gypsophilum is believed to have originated as a segregate in a hybrid swarm of D. recurvatum x hesperium. The former is intermediate between the putative parents in its ecological preferences and in its morphological characters. Some individuals in an artificial F 2 of D. recurvatum x hesperium resemble D. gypsophilum in morphology. In nature, D. gypsophilum is kept apart from the other two species mainly by ecological isolation (Lewis and Epling 1959). Three species of Penstemon in California, P. centranthifolius, P. grinnellii, and P. spectabilis, are interrelated, diploid (2/i = 16), and somewhat interfertile. They differ in their floral mechanisms, mode of pollination, and vegetative ecology. Morphological comparisons between the three species, ecological field studies, and the characteristics of natural hybrids combine to suggest that Penstemon spectabilis arose as a product of hybridization between P. centranthifolius and P. grinnellii. However, the possibility that P. spectabilis is a product of primary speciation has

252

Natural Hybridization and its Products

not been ruled out. The three species are isolated in nature by mechanical, ethological, and ecological isolating mechanisms (Straw 1955, 1956). Stebbins and Ferlan (1956) postulated a hybrid origin of Ophrys murbeckii and O.fusca x lutea in the Mediterranean region. Ethological and ecological isolation are apparently the chief factors keeping the putative daughter and the parental species distinct in this case (Stebbins and Ferlan 1956). Tucker and Sauer (1958) have described a series of diploid populations of Amaranthus on the delta of the Sacramento River in California where diploid hybrid speciation may be in statu nascendi. Several well-established species are found in the region—A. powellii, A. caudatus, A. retroftexus, and along with them are aberrant local populations exhibiting different combinations of the characters of the aforementioned species. The aberrant populations, or some of them, may represent different diploid hybrid derivatives of the preexisting Amaranthus species (Tucker and Sauer 1958). In Alaska, Carex rostrata typically occurs in sweet-water lakes and C. rotundata in peat bogs. Hybrid swarms between these species are found in the transition zone between their respective habitats. In some areas a new, morphologically distinct, fertile population is segregating out of the hybrid swarms and is increasing in numbers relative to the parental species. This new form is regarded as a daughter species of hybrid origin and is named C. paludivagans (Drury 1956). New Zealand is a center of origin for the genus Epilobium with 37 native species occurring in a wide range of habitats. The plants are perennial herbs with vegetative reproduction, small flowers, and a predominantly autogamous breeding system. They are homoploid with 2n — 36. The species are interfertile among themselves in most combinations. Sympatric occurrences between them are common. In sympatric situations, reproductive isolation results from the autogamy as well as from ecological differentiation (Raven and Raven 1976). However, the reproductive isolation breaks down occasionally, and hybrids and hybrid swarms are well known. In some cases uniform populations of hybrid origin are found. Some of these are at the race level. Others are at the species level. Thus Epilobium purpuratum on South Island is believed to be derived from E. crassum x porphyrium. It is difficult to be certain about this and other suspected cases, but in general

Hybrid

253

Speciation

it seems probable that the species diversity of Epilobium in New Zealand has been enriched by hybrid speciation (Raven and Raven 1976). An interesting case has been reported recently of hybridization in tree fems (Cyatheaceae) in Puerto Rico (Conant and Cooper-Driver 1980). The species Alsophila bryophila, A. dryopteroides, and Nephelea portoricensis (all with 2n = ca. 69 II) hybridize in all possible combinations. The hybrids are fertile and produce F 2 segregates on the homoploid level, some of which are established in nature. One homoploid hybrid product of A. dryopteroides x N. portoricensis forms a uniform population which is larger in size than those of the parental species in the same locality. The hybrid population is reproducing sexually. No evidence was found of vegetative reproduction (Conant and Cooper-Driver 1980). The true-breeding condition in this hybrid derivative appears to be determined by the following breeding system. The Fi fern hybrid produces recombinant spores which develop into hermaphroditic gametophytes. The female and male gametes on a gametophyte result from mitotic divisions and are thus genetically identical. Intra-gametophyte selfing takes place. This produces a homozygous sporophyte carrying the gene combination of the original recombinant spore. This sporophyte gives rise to spores, new gametophytes, and eventually a colony of identical individuals of the new type (Conant and Cooper-Driver 1980). The hybrid derivative population is isolated by external barriers of ecology and selfing (Conant and Cooper-Driver 1980).

Introduction

to Apomictic

Speciation

An interspecific hybrid in the Fi or some subsequent early generation may prove to be adaptively superior. Its highly heterozygous genotype may possess properties of heterosis or homeostasis or both. Such a hybrid is potentially worth perpetuating and multiplying. But the reproduction of the adaptively superior species hybrid is beset with two obvious difficulties. The first is hybrid sterility: the second is inability to breed true to type. However, there are also some ways around these difficulties, as Darlington pointed out long ago (1932, ch. 16). A heterozygote derived from a species cross can surmount its inherent reproductive difficulties of sterility and segregation, and can multiply its

254

Saturai

Hybridization

and iti

Products

particular heterozygous genotype, by several of the methods of stabilization listed at the beginning of this chapter (methods 1 to 5). In this section we mention the first two methods, known collectively as apomixis, in relation to hybrid speciation. Apomictic speciation is introduced here and is discussed more fully later in part 8. Opuntia fulgida

and O. spinosior

are two distinct species of cholla

cactus in the southwestern American desert. They hybridize occasionally to produce fairly seed-sterile hybrids. Hybrid plants, despite their seed sterility, have become abundant in a local area along the Gila River in Arizona, owing to their capacity for vegetative propagation by the fallen stem joints (Grant and Grant 1971a). The European blackberries belonging to the genus Rubus section Moriferus are a large complex of hybridizing sexual species and their apomictic derivatives. The hybrids possess both agamospermous and vegetative means of reproducing. Each hybrid type can multiply and spread by asexually produced seeds and by rooting of the stem tips, and many microspecies have arisen in this way (Gustafsson 1943). A hybrid with distinctive morphological characters and ecological preferences can thus spread apomictically and form a population of its own. The population composed of similar hybrid individuals of apomictic derivation is a clonal or agamospermous microspecies when it is poorly distinguished morphologically or narrowly endemic or both. If the hybrid derivatives possess a character combination which is recognizable in ordinary taxonomic practice, and if this character combination has been able to spread by apomictic means throughout a definite geographical area, the resulting population has attained the status of a taxonomic species.

Conclusions It may be useful to summarize the known modes of hybrid speciation in plants. I. Hybrid speciation with sexual reproduction; the origin of a new biological species directly from an interspecific hybrid. 1. Hybrid speciation with external barriers; the formation of a new recombination type in the progeny of a hybrid which is separated from the parental species by external isolating mechanisms.

Hybrid Speaation

255

2. Recombinational speciation; a chromosomally sterile or semi-sterile hybrid gives rise to homoploid or undoubted derivatives which possess a new homozygous recombination of the various independent chromosomal rearrangements separating the parental species; the new fertile line has its own character combination and is chromosomally intersterile with the paren'al species. 3. Amphiploidy; a chromosomally sterile hybrid gives rise by doubling of the number of chromosome sets to a fertile polyploid derivative; the amphiploid derivative is separated from the parental species by a chromosomal sterility barrier. II. Hybrid speciation with asexual or subsexual reproduction; the multiplication of a hybrid by mechanisms circumventing the normal sexual and chromosomal cycle, and the spread of this hybrid type throughout a definite geographical area, so that it attains the population size and composition of a uniparental microspecies or species. 4. Apomictic speciation; a new clonal or agamospermous microspecies develops from an interspecific hybrid by means of vegetative propagation or agamospermous seed formation, respectively (part 8). 5. Permanent structural or numerical hybridity; a translocation heterozygote multiplies by the Oenothera biennis genetic system, or an odd polyploid increases in numbers by the Rosa canina genetic system, to the population size of a heterogamic microspecies (part 8).

CHAPTER

Recombinational

TWENTY

Speciation

Recombinational speciation is the formation and establishment, in the progeny of a chromosomally sterile or semisterile hybrid, of a new homozygous recombination type for the two or more independent rearrangements differentiating the parental species. The hybrid derivative is fertile itself, and on the same ploidy level as the parental species, but is isolated from the latter by a chromosomal sterility barrier. The process of recombinational speciation can produce daughter species which are surrounded by weak or by strong sterility barriers. The hybrid origin of a new homoploid line which is fertile itself but highly intersterile with its congeners is clearly the more difficult case to explain. But, if we can find a satisfactory explanation of this case, we will have all the elements needed for understanding the theoretically easier cases involving weak sterility barriers. In this chapter we describe the hypothetical mechanism of recombinational speciation, review the relevant experimental evidence, and inquire into the possible role of the process in nature.

Recombinational Spedatiort

The

257

Hypothesis

A definite genetical model applicable to the theoretically interesting case of recombinational speciation with strong sterility barriers was presented by Stebbins (1957a) and Grant (1958). This model employs a chromosomal sterility barrier between two hybridizing parental species which is composed of separable segmental rearrangements. The model was based on earlier work by Miintzing and others as noted in our introduction to the subject in chapter 19. Suppose that two parental species are isolated by a chromosomal sterility barrier composed of two or more separable segmental rearrangements. Then their interspecific hybrid, which is chromosomally sterile but not completely so, can give rise to one or more new homozygous recombination types for the segmental rearrangements. These new types are fertile themselves but chromosomally intersterile with one another and with the parents (Stebbins 1950:287-288; 1957a; Grant 1958; 1963:469 ff.). This process of recombination can be illustrated by a simple example. Let the two parental types differ with respect to two translocations on four chromosomes. Their chromosomal constitutions are AA BB CC DD and AbAbBaBa CdCd DCDC respectively (see figure 20.1). Their F! hybrid, with the constitution AAb BBa CCd DDC, is heterozygous for two independent translocations and therefore has an expected gametic fertility of 25% (see figure 20.1 again). The functioning gametes include the two parental types, which do not interest us further here, and also two recombination types (ABCdDc and At,B 9 II + < 1 8 I in triploid hybrids derived from crosses between tetraploid and diploid species belonging to an ancestor-descendant lineage. Just this type of pairing was in fact observed in triploid hybrids of Gilia sinuata (4x) x G. latiflora (2t) and G. modocensis (4x) x G. latiflora (2r) (Grant, 1964c). Therefore the latiflora or T genome is present in both G. sinuata and G. modocensis. On morphological and ecological evidence it can be concluded that G. latiflora itself, and not some other member of the T genome group, was a diploid parent of both G. sinuata and G. modocensis. These two tetraploid species differ genomically and morphologically as regards their other diploid parent. Day (1965) made a detailed taxogenetic study of two tetraploid species and their putative diploid parents in the Gilia transmontana subgroup. The morphological, ecological, cytological, and cytogenetic evidence pointed to a reticulate derivation of the two tetraploids, G. transmontana and G. malior, from three diploid species, as shown in figure 25.4. Day then confirmed this phylogeny by artificially resynthesizing G. transmontana and G. malior from their diploid parents and showing that each artificial amphiploid is interfertile with the corresponding natural tetraploid species (Day 1965).

The Polyploid Complex

335

Figure 25.4. Phytogeny of two segments of the polyploid complex in the Gitia inconspicua group. (Drawn from data of Grant and Grant I960, Grant 1964a; Day 1965) Similar evidence was obtained for the ancestry of the tetraploid species known as Gilia ophthalmoides III. This species was believed to be derived from G. clokeyi (2c) x G. mexicana (2t). Spontaneous tetraploid progeny obtained from this diploid interspecific hybrid turned out to be good G. ophthalmoides III, thus confirming the phylogenetic hypothesis (Grant and Grant 1960, and unpublished). Additional evidence points to a partly common, partly different ancestry of a related tetraploid species, G. ophthalmoides I. The reticulate pattern of relationships can be extended to include G. sinuata, G. modocensis, and other tetraploid species. The reticulate phylogeny of two segments of this polyploid complex is shown in figure 25.4.

336

Polyploidy

Triticum and

Aegilops

The well known polyploid series in wheat is part of a larger polyploid complex involving Aegilops. Triticum and Aegilops are interconnected at the polyploid level. The distinctions between the old generic units, wheat and goatgrass, based on spike and spikelet characters, thus break down. Apparently Triticum and Aegilops comprise a single genus, and the trend in wheat genetics is to transfer species from Aegilops to Triticum. The task of taxonomic and nomenclatural reorganization is still unfinished, however, and we have two overlapping sets of genus-species names at the present stage. Triticum-Aegilops is a group of annual grasses belonging to the subtribe Triticinae of the tribe Hordeae. The basic chromosome number is x = 7 in Triticum-Aegilops as in the tribe as a whole. Within Triticum in the old narrow sense the taxonomic trend in recent years has been to reduce many of the formerly recognized species to subspecies or cultivars, thus simplifying the taxonomy of the wheats. An outline of the classification of the wheats, based on Feldman (1976), is given in table 25.1.

T a b l e 2 5 . 1 . Classification and g e n o m e constitution of wheats (Triticum). (Based on Feldman 1976; distribution notes from Harlan and Zohary 1966, and Z o h a r y et al. 1969) 1. T. monococcum. Einkom 2x. AA. (a) Wild race: T. m. boeoticum. Western Asia from Turkey to Iran, and southern Balkans. (b) Cultivated race: T. m. monococcum. 2. T. turgidum. Emmer and durum. 4jr. AABB. (a) Wild race: T. t. dicoccoides. Western Asia from Turkey and Israel to Iran. (b) Cultivated race with hulled grains: T. t. dicoccum. (c) Cultivated races with free-threshing grains: T. t. durum, turgidum, polonicum, caithlicum. 3. T. timopheevii. 4jt. AAGC. (a) Wild race: T. t. araraticum. Southern Soviet Union in Georgia and Armenia. (b) Cultivated race: T. t. timopheevii. 4. T. aestivum (= T. vulgare). Bread wheat. 6x. AABBDD. (a) Cultivated races with hulled grains: T. ae. spelta, macha, vavilovii. (b) Cultivated races with free-threshing grains: T. ae. aestivum, compactum, sphaerococcum.

The Polyploid

Complex

337

Genome analysis of wheats has been carried out by Kihara (1924, 1954), Sears (1948, 1959), and many others. The/4 genome in polyploid wheat is derived from the diploid species T. monococcum. The source of the B genome in tetraploid T. turgidum is not known; we will return to this question in a moment. The D genome in hexaploid T. aestivum comes from the diploid species Aegilops squarrosa (= T. tauschii) which has a wide range in western and south-central Asia. The B genome occurs in both T. turgidum and T. aestivum. A diploid species with the constitution BB has been looked for but never found. At one time Aegilops speltoides (= Triticum speltoides) was thought to be the carrier of the B genome, but cytogenetic and electrophoretic evidence is against this hypothesis, which is no longer accepted in its original form. The genome constitution currently assigned toAe. speltoides is 55. As regards the missing BB diploid, one possibility is that it is now extinct. Another possibility is that there never was a BB diploid. The B genome in T. turgidum may have diverged from some other related genome, perhaps 5, within tetraploid T. turgidum (Feldman 1976). The phylogeny of the polyploid series leading to bread wheat can be summarized as follows. (1) Hybridization and amphiploidy between wild einkorn (AA) and an unidentified diploid canying B or a precursor of B to yield wild emmer (AABB). (2) Changes brought about by selection under domestication from wild emmer to cultivated emmer with hulled grains and then to cultivated durum wheat with free-threshing grains. (3) Hybridization and amphiploidy between cultivated emmer and Aegilops squarrosa (DD) to produce bread wheat (AABBDD). (4) Selective changes in cultivation from hulled bread wheat to free-threshing bread wheat. These events took place in western and southwestern Asia. This area is an important center of early agriculture as is well known. Carbonized grains of T. monococcum, T. turgidum, and T. aestivum are found in various archeological sites in this area. The ages of the oldest known grains of the three species are about 9000 B.P. (Feldman 1976). Wild emmer occurs today in western Asia from Israel and Lebanon to the Tigris river and Caspian sea (Harlan and Zohary 1966). Durum wheats have mostly been grown in the Mediterranean region with its relatively mild climate (Zohary et al. 1969). Aegilops squarrosa ranges from the Caspian region into the central Asian steppes as far east as Tashkent and Alma Ata (Zohary et al. 1969). It evidently includes a

338

Polyploidy

variety of ecotypes, including those adapted to highly continental climates. The ecological adaptations of Ae. squarrosa represented that species' most important contribution to bread wheat. T h e D genome in hexaploid bread wheat accounts for its wide ecological amplitude and its ability to be grown in cold dry steppe climates (Zohary et al. 1969; Harlan, personal communication). It is interesting to note that this process is continued in the artificially synthesized allooctoploid, Triticale, an amphiploid derivative of T. aestivum (6x) x Secale cereale (2x), which can be grown in northern latitudes and montane altitudes. Triticum timopheevii has the genomic constitution AAGG (table 25.1) and thus represents another tetraploid product of T. monococcum. The other diploid parent could be Ae. speltoides (SS) or some related species such as Ae bicornis (S"Sb). The S genome is close to the G genome. It is assumed on this phylogenetic hypothesis that the S genome changed to G within tetraploid T. timopheevii (Feldman 1976). The larger number of species in the old genus Aegilops includes diploids, tetraploids, and hexaploids. Five genomes and a larger number of subgenomes are recognized in Aegilops. At the diploid level the five genome groups mostly coincide with taxonomic sections of the genus. The main genomes, subgenomes, and their diploid carriers are listed in table 25.2. Some subgenomes ( M " , D2, Sv, etc.) are known only in polyploids; these are not listed in table 25.2. The genomic constitution of the polyploid species of Aegilops is given in table 25.3. The crisscross phylogenetic derivation of the polyploid species can be visualized by reference to this table. Thus two diploid species Table 2 5 . 2 . Genomes and their diploid carriers in Aegilops. Zohary and Feldman 1962) Genome group

Subgenome

Mt S

S" S>

s D

C

M

C C" M M"

(Kihara 1954;

Diploid species Ae. Ae. Ae. Ae. Ae. Ae. Ae. Ae. Ae.

mutica bicornis sharonensis, longissima speltoides, ligustica squarrosa caudata umbellata comosa uniaristata

The Polyploid Complex Table 2 5 . 3 . G e n o m e constitution of polyploid species of Aegilops. Feldman 1962) Polyploid species Ae. crassa (4JT) Ae. crassa (6x1 Ae. juvenalis Ae. ventricosa Ae. cylindrica Ae. variabilis and Ae. kotschyi Ae. triuncialis Ae. columnaris Ae. biuncialis Ae. triaristata (4JT) Ae. triaristata (6jr) Ae. ovata

339 (Zohary and

Genome constitution

DDM^M" DDD*D*McrMer DDCC'M'M' DDM'Mr DDCC cx:"s°sr C"C"CC C"C"M'Mr C"CuMbMb C"C"M'M' C"C"M'M'MaMn C"C"M"M°

carrying the DD and MM genomes have produced tetraploids with the constitution DDMM One of these diploids (DD) and another (CC) have then produced another tetraploid DDCC, and so on. The different tetraploid species often have one common and one different diploid ancestor. The process of recombination of genomes is continued at the hexaploid level. The reticulate pattern of relationships in Triticum-Aegilops is presented here in terms of the addition of different genomes by amphiploidy. This is not the whole story, however. There is evidence, both direct and indirect, for natural hybridization and introgression between some of the species involved. This hybridization has altered the morphological characters and probably also the genomes of these species (Zohary and Feldman 1962; Pazy and Zohary 1965; Zohary et al. 1969; Feldman 1976).

Old Polyploid

Complexes

In the course of time the diploid species of a polyploid complex may die out. Or the lower polyploids may become extinct too, leaving the higher polyploids behind as the only living representatives of the complex. The Bromus carinatus group (Gramineae) illustrates a polyploid com-

340

Polyploidy

plex which has reached an advanced stage of development. This group reaches high ploidy levels, B. arizonicus being 12*, but has suffered much extinction at the diploid and tetraploid levels. The ancestry of the I2x and &t species has been worked out by Stebbins and collaborators (Stebbins and Tobgy 1944; Stebbins, Tobgy, and Harlan 1944; Stebbins 1947b, 1956). The Bromus carinatus complex belongs to the section Ceratochloa of the genus but contains genomes from two related sections, Bromopsis and Neobromus. The plants are perennial grasses, often bunch grasses, of North and South America. The basic number is x = 7. There are no known diploid or tetraploid species in Bromus section Ceratochloa. Bromus catharticus and its relatives in South America are allohexaploids carrying genomes derived from diploid species which are now extinct (figure 25.5). Bromus carinatus and B. marginatus in North America are allooctoploids. Three of their four genomes are homologous with those in the South American hexaploids. The fourth L genome is believed on cytological and morphological evidence to be derived from some North American diploid species of the section Bromopsis (figure 25.5). The amphiploid origin of the octoploids probably took place in western North America in the late Tertiary, perhaps the Pliocene. It is necessary to suppose that some hexaploid species homologous with the present-day South American hexaploids occurred in North America in former times (Stebbins and Tobgy 1944; Stebbins 1947b). Bromus arizonicus is an amphiploid at the 12x level. It, too, is homologous with the South American hexaploids in three of its six genomes (figure 25.5). The other three genomes in B. arizonicus are probably derived from an allohexaploid in the related section Neobromus (figure 25.5) (Stebbins, Tobgy, and Harlan 1944). The 12-ploid species, Bromus arizonicus, is a common and aggressive weed throughout much of its range in Arizona and California. Its diploid and tetraploid ancestors are extinct. Its hexaploid ancestor in the section Ceratochloa became extinct in North America, perhaps during adverse climatic changes in Pliocene-Pliestocene time, and survives today under more moderate climatic conditions in South America. Yet the 12-ploid derivative of these extinct diploids and lower polyploids is a successful and aggressive species today, showing how continued amphiploid doubling can give an old polyploid complex a new lease on life (Stebbins 1947b, 1956).

The Polyploid

341

Complex

A1A1A2A2B16182B283636484 8. arizonicus

A,A,8,8,8282^. 8. carinatus B. marginatui 8. pitensn

A, A , 8 , 8] 8282 B. catharHcus B. haenkeanus 8. stomineus

A2A2B383B4B4 B. trinii

A,A,6,B,

63638484

LL B. ciliahjt 8. anomalut 8. orcuttionus BROMOPSIS

B B

A A

ii

A,A,

77

8383

8382 CERATOCHLOA

8484

NEOBROMUS

G e n o m i c Relationships Figure 25.5. Phylogenetic relationships of high polyploid species of Bromus. Some of the basic genomes are unknown at the diploid or tetraploid level. (Redrawn from Stebbins 1956)

Vestiges of Ancient Polyploid Stebbins (1971,

Complexes

1980) h a s c a l l e d a t t e n t i o n to a n i n v e r s e

correlation

b e t w e e n d e g r e e of a d v a n c e m e n t in m o r p h o l o g y a n d t h a t in p l o i d y l e v e l in v a r i o u s a n g i o s p e r m f a m i l i e s . W i t h i n t h e p l a n t f a m i l y , t h e m o r p h o l o g ically p r i m i t i v e g e n e r a h a v e b a s i c c h r o m o s o m e n u m b e r s t h a t a r e p o l y -

342

Polyploidy

ploid. Conversely, the genera with more specialized morphological characters have diploid basic numbers. Some examples are the following (Stebbins 1980):

Liliaceae Winteraceae Polemoniaceae Hydrophyllaceae Compositae

Morphologically primitive, old-polyploid basic numbers Tofieldia Drimys Cantua, Cobaea Wigandia Heliantheae

Morphologically specialized, diploid basic numbers Trillium, Scilla Tasmannia Gilia, Phlox Nemophila Cichorieae

The evolutionarily advanced genera in these families have retained the ancestral diploid condition. The retention of diploidy was probably an important factor in their ability to attain high levels of evolutionary advancement. The high-number genera in the same families, on the other hand, are old polyploids, and often remnants of old polyploid complexes, which have remained conservative (Stebbins 1971, 1980). Some whole families of woody dicotyledons have high basic chromosome numbers which have been plausibly considered by various authors to represent an old polyploid condition (Stebbins 1950, 1971; Tischler 1954; Darlington 1954b); Magnoliaceae Trochodendraceae Cercidiphyllaceae Salicaceae Hippocastanaceae Platanaceae

x x x x X X

19 19 19 19 = 20 21

= = =

Most of the main groups of pteridophytes have basic numbers in the polyploid range, and in some groups the basic numbers are in the high polyploid range (Manton 1950; Love, Love, and Pichi Sermolli 1977). For example: Polypodium, etc. (Polypodiaceae) Psilotum (Psilotaceae) Tmesipterus (Psilotaceae)

x = 37 x = 50, 52 x= 104

The Polyploid Complex Equisetum (Equisetaceae) Ophioglossum (Ophioglossaceae)

343 x

— 108 120

The most likely conclusion is that many groups of recent pteridophytes are the surviving remnants of geologically ancient polyploid complexes (Manton 1950).

Irreversibility The evolutionary trend from diploid to tetraploid to higher polyploid is in the main an irreversible trend. This trend can be reversed partially by polyploid drop (see chapters 22 and 27). Thus in Hesperis, where x = 7, tetraploid species have gametic numbers of n = 14, 13, and 12. Some workers have suggested the possibility of abrupt and complete reversals from the tetraploid to the diploid condition by means of polyhaploids (Raven and Thompson 1964; De Wet 1968). The haploid or polyhaploid progeny of a tetraploid parent is, of course, diploid. Such polyhaploids do arise spontaneously in polyploid plant populations. The polyhaploid derivative of an allotetraploid is, however, by definition and by observation, a chromosomally sterile diploid hybrid whose main chance of reproducing sexually is by doubling once again. Polyhaploids offer little hope for reversals in polyploid trends.

Suppression

of

Variability

The polyploid condition places certain restrictions on the generation of mutational and recombinational variability, as compared with the diploid condition, and these restrictions become tighter at the higher polyploid levels. The expression of new mutations is suppressed by gene duplication and polysomic inheritance. The formation of new recombination types is restricted by these factors and by preferential pairing. Tischler points out that seedling mutants are common in Triticum monococcum (2r), are much rarer in T. durum (4t), and are quite rare in T. aestivum (dr) (Tischler 1953). The generation of variability is restricted but not blocked completely

344

Polyploidy

by the polyploid condition. Segmental allopolyploids can segregate to a limited extent for the interspecific gene differences. Polyploids can segregate mutant types rarely in polysomic ratios. Nevertheless, the variation-producing mechanism of the sexual process is inhibited in polyploids as compared with diploids. The ability of polyploid species to respond to changing environmental conditions is therefore expected to be less efficient than that of diploid species. Certainly new polyploids are often very successful, more so than their diploid relatives, in various types of habitats, as we have seen in chapter 24. They are successful by virtue of a genetic system which permits a particular hybrid genotype to breed true. But this genetic system also imposes restrictions on the release of new variations and thereby sets limitations on the evolutionary potential of the polyploid lineage. As the mainly irreversible trend to higher ploidy levels continues in the phyletic lineage, these limitations of evolutionary potential must become greater. The polyploid system, as Stebbins states (1950:366), is capable of producing numerous related species or even genera with generally similar characteristics, but has apparently played no role in the origin of new families or orders.

Diploidization Old polyploids tend to be more diploid-like than newly formed polyploids. This change is known as diploidization. Diploidization affects both the cytological behavior and the genie constitution of polyploids. A new or raw polyploid is expected to have a high frequency of multivalents, if it is an autopolyploid, and some multivalent formation, if a segmental allopolyploid. The multivalents reduce the fertility of the polyploid. Natural or artificial selection for fertility will favor genes, where these are available in the gene pool, that promote bivalent formation at the expense of multivalent formation. Gilles and Randolph (1951) compared the multivalent frequency in an autotetraploid line of maize (Zea mays, x = 10) at the beginning and the end of a 10-year period of selection for fertility. The newly induced autotetraploid plants had 8 - 1 0 IV in 89% of the pollen mother cells, whereas their tenth -generation descendants had 8—10 IV in only 52% of the cells. The frequency of bivalents increased as the quadrivalent fre-

The Polyploid

Complex

345

quency decreased. The artificial selection for fertility which brought about this change probably involved an unconscious selection for genie factors influencing the mode of chromosome pairing at meiosis. Selection for meiotic genes must be important in the history of many natural amphiploids. There is much homology between the different genomes in hexaploid wheat, Triticum aestivum. Yet the homeologous chromosomes do not normally pair; instead they form bivalents consisting of homologous chromosomes. This meiotic behavior is due to a gene on the long arm of chromosome 5 of wheat. When this gene is absent, multivalents are formed. The 5B gene, when present, inhibits multivalent formation and promotes homologous bivalent pairing (Riley 1960). Duplicate factors are common in new amphiploids and present in old ones. The duplicate loci are a consequence of the summation of related genomes. The duplicate factors may not be necessary for the functioning of the organism. One duplicate locus is then free to diverge from its original function. If the divergent gene finds another related function which is beneficial for the organism it will be preserved by selection. An interesting case of divergence between originally duplicate factors is found in the allotetraploid cottons, Gossypium hirsutum and G. barbadense. Their genome constitution is AADD. The AA diploids are Old World cultivated cottons, and the DD diploids are wild American species. A linkage group including the gene R for anthocyanin pigment and the gene CI for inflorescence congestion occurs in both the A and D genomes and again in the AADD tetraploids. The R gene is present as a single locus in IheAA diploids and theDD diploids, but as two loci (Rt and/? 2 ) in the AADD tetraploids. Similarly the gene CI is present as a single locus in theAA diploids, and probably also in th eDD diploids, but occurs in two loci (CI i and Cl2) in the AADD tetraploids. The genomic location of these genes in the tetraploids is R2—C/2 in the A genome, and /?!—C7, in theZ) genome (Stephens 1951a, 1951b). The genes Rt and R2, and CI j and C7 2 . were undoubtedly duplicate factors when tetraploid cotton first originated. But they are not duplicate factors in modern tetraploid cotton. Instead R x and R2 interact as a pair of complementary factors, and so do C7, and Cl2 (Stephens 1951a, 1951b). The evolutionary trend to higher levels of polyploidy is accompanied by greater difficulties of meiosis and greater restrictions on segregation and recombination. The process of diploidization works against these

346

Polyploidy

tendencies. This process enables medium or high polyploids to behave genetically like diploids to a greater or lesser extent. We can suspect that diploidization is a factor making it possible for some polyploid complexes to evolve to high ploidy levels. Likewise it is probably a factor in the ability of some genera with high basic numbers (e.g., Epilobium, x = 18; Equisetum, x= 108) to speciate extensively on their old polyploid levels. Diploidization, in short, can give old polyploids a new lease on life.

CHAPTER Natural

TWENTY-SIX Allopolyploids

The principal criteria for distinguishing autopolyploids from amphiploids in natural occurring plant groups, as noted in chapter 23, are multivalent formation, sterility, tetrasomic inheritance, and morphological resemblance to related diploids. These criteria are relevant but are not diagnostic and can easily lead to erroneous conclusions in particular cases. At one time natural autopolyploidy was in fact thought to be common in plants on the basis of these criteria. The critical réévaluation of the evidence by Clausen, Keck, and Hiesey (1945) changed this conclusion. Natural autopolyploidy is uncommon as compared with amphiploidy in higher plants, but does occur in a number of plant groups. Nevertheless, reports of autopolyploidy continue to appear in the literature. Such reports are all too often based on inadequate evidence. For example, autopolyploidy is claimed for tetraploid members of the Epilobium angustifolium group on cytogenetic evidence which is extensive and suggestive but inconclusive (Mosquin 1967; Mosquin and Small 1971). The best documented cases of natural autopolyploidy in plants are those in which the parentage of the polyploids is known with reasonable certainty, and the fertility and populational relationships between the parental diploids are also known. These lines of evidence, of course, require combined field and experimental methods, which go well beyond the usual cytotaxonomic methods in their analytical resolving power.

348

Polyploidy

In this chapter we will review three authentic cases of natural autopolyploidy and list briefly two additional examples. Lack of space prevents us from discussing various recent reports in which autopolyploidy is claimed on insufficient evidence.

The Biscutella Laevigata

Group

Biscutella laevigata and its relatives (Cruciferae) comprise a group of perennial herbs of rocky places in Europe. The classical studies of Manton (1934, 1937) establish the existence of lx, 4x, and 6r forms (x = 9). The hexaploid is known only from a locality in Spain. The main part of the story as it is understood to date concerns the diploids and tetraploids. The tetraploid form, B. laevigata sens, str., has a continuous distribution throughout the Alps. The diploids, known collectively as B. coronopifolia, form a series of disjunct populations in the lowlands north and east of the Alps. In addition, one exceptional diploid occurs in the mountains near Vienna. The diploids from separate areas differ morphologically and had been described previously as different species. But Manton (1937) crossed them inter se and found that the F, hybrids were fertile with regular chromosome pairing. They thus appear to be races of one species. This indicates that their tetraploid derivative, B. laevigata, is an interracial autotetraploid. The occurrence of some quadrivalents in B. laevigata is suggestive but not conclusive of autotetraploidy. More convincing is the finding of trivalents in the artificial triploid hybrids produced by crossing B. laevigata with different diploid races (Manton 1937). The geographical and ecological distribution of the diploids and tetraploids, when combined with the known geological history of the area, suggests that the lowland diploid populations are relicts of an interglacial flora. The alpine tetraploid B. laevigata, on the other hand, occurs in an area that was covered by ice during the last glacial period, and therefore appears to be a post-glacial immigrant into its present area (Manton 1934, 1937). The interracial gene combinations were probably adaptively valuable for the autotetraploid. In this connection, the diploid race from the mountains near Vienna may have been one parent in the interracial cross or

Natural

349

Aulopolyploids

crosses that produced the successful alpine autotetraploid (Manton 1934, 1937; Clausen, Keck, and Hiesey 1945).

The Solarium Tuberosum

Group

The complex Solarium tuberosum group has been thoroughly studied by a series of workers and is the best analyzed case of natural autopolyploidy. Let us begin by putting this group into taxonomic perspective. The section Tuberarium of the genus Solatium is a large section consisting of 32 series in North and South America (Correll 1962). The series Tuberosa contains six species including 5. tuberosum sens. lat. in Peru, Chile, and Argentina. Five of these species are wild, and the sixth, S. tuberosum, contains wild races in Patagonia and Chile, as well as numerous cultivated races in the Andean region (acc. Correll 1962; for a different taxonomic treatment of series Tuberosa see Hawkes and Hjerting 1969). The plants are outcrossing, tuber-bearing perennial herbs. The diploid forms are self-incompatible and are cross-pollinated by Bombus and other bees; seed reproduction is supplemented by vegetative propagation by tubers in them (Dodds and Paxman 1962). The cultivated polyploids are mostly sterile as to pollen and seeds and are propagated vegetatively (Swaminathan and Howard 1953; Swaminathan 1954; Hawkes and Hjerting 1969). Solanum tuberosum sens. lat. is a composite taxonomic species. It contains not only an array of wild and cultivated forms, as mentioned above, but also a polyploid series of 2x, 3a, Ax, and 5* forms on the base of x = 12 (Swaminathan and Howard 1953). Dodd's (1962) classification of the cultivated potatoes in 5. tuberosum sens. lat. recognizes seven main forms on morphology, distribution, and ploidy. Dodds uses the informal category of " g r o u p " for these taxa; his groups are equivalent to cultivated races in other classifications. The seven groups are listed below with notes on ploidy and other features. 1. 2. 3. 4.

Stenotomum. 2x. High Andean altiplano. Phureja. 2x. Andean valleys. Andigena. 4r. Andean cultivated potato. Tuberosum ( = S. tuberosum sens, str.) Ax. Common cultivated potato, Europe and North America.

350

Polyploidy

5. Chaucha. 1c. Seed sterile. 6. Juzepczukii. 3k. Seed sterile. 7. Curtilobum. 5x. Seed sterile. The phylogenetic relationships between these taxa are shown graphically in figure 26.1 (following Dodds 1962, and Dodds and Paxman 1962). Involved in the phylogeny is a species belonging to a related series, 5. acaule of the series Acaulia. The Andean potato, andigena, is held to be a tetraploid derivative of interracial crosses between stenotomum (2x) and phureja (2x). The tetraploid must have arisen spontaneously under primitive agricultural conditions. The common cultivated potato, also tetraploid, is then a selection product of this more primitive Andean potato.

Figure 26.1. Phylogeny of the main cultivated races of the potato, Solanum erosum. Further explanation in text. (Drawn from data of Dodds 1962)

tub-

Natural

351

Allopolyploids

The common potato has long been thought to be an autotetraploid on highly suggestive cytogenetic evidence. The common potato has a high frequency of quadrivalents at meiosis. The maximum number is 9 IV per cell in the subvariety Chippewa (Swaminathan 1954). The average frequency of quadrivalents is 4.4 IV in some subvarieties (Howard 1961). In line with this cytological behavior is the occurrence of tetrasomic ratios for some genes in cultivated potatoes (Cadman 1942). Other lines of evidence clinch the case for autotetraploidy. The artificial triploid hybrid between tetraploid tuberosum and diploid stenotomum has an average of 8.2 III per cell. This high frequency of trivalents indicates that the cultivated potato is an autotetraploid with stenotomum chromosomes in it (Howard 1961). Further evidence comes from the reduced or polyhaploid form of S. tuberosum sens. str. with 2n = 24 instead of the normal 2n = 48. Some polyhaploids show 12 II, in other words, full bivalent pairing, in a majority of the cells, and have high pollen fertility. These polyhaploids also form fertile hybrids with some of the diploid cultivars, (Swaminathan 1954; Howard 1961). Finally, the diploid races are closely related genetically. There are no crossability barriers between them. The interracial Fi hybrids have regular bivalent pairing. These hybrids are fertile in some combinations but are pollen sterile in others (Swaminathan and Magoon 1961; Dodds and Paxman 1962). Attention has been focused here on autotetraploidy in the Solanum tuberosum group. It should be noted to round out the picture that amphiploidy is common among the wild species of section Tuberarium. Thus S. acaule (in series Acaulia) is an allotetraploid, and 5. demissum (series Demissa) is an allohexaploid (Swaminathan and Howard 1953; Howard 1961).

The Dactylis

Glomerata

Group

Dactylis glomerata (Gramineae), 2n = 14 and 28. Orchard grass is selfincompatible, outcrossing, and vegetatively propagated. It has a wide range in Eurasia, and encompasses a wide range of morphological and ecological variation at two ploidy levels (Stebbins and Zohary 1959). At the diploid level, Stebbins and Zohary (1959) recognize 11 subspe-

352

Polyploidy

cies. Four of these represent morphological extremes. They occur in old types o f vegetation and have disjunct distributions ranging from the Himalaya Mts. to the Canary and Cape Verde Islands. These four geographical races apparently represent products of an old primary divergence at the diploid level. The other seven subspecies are intermediate between the four extremes and are associated with more modern types of vegetation, including grazing subclimaxes and weedy places. They are probably products of interbreeding and intergradation between the primary races (Stebbins and Zohary 1959). The diploid races are interfertile. The F , hybrids are semifertile to fully fertile as to pollen and are seed fertile (Stebbins and Zohary 1959). The tetraploids are common and widespread, more so than the diploids, and are very variable. Morphological and ecological evidence suggest that the tetraploids are derived from different interracial hybrid combinations. They are thus interracial amphiploids (Stebbins and Zohary 1959). Crosses between tetraploid and diploid orchard grass are hard to make, and the F , hybrids are highly sterile (Stebbins and Zohary 1959). Dactylis as presently constituted is thus a taxonomic species consisting

glomerata

of two biological species. The sterility barrier between the diploids and tetraploids is strong, but not absolute. Triploid hybrids set some seed when pollinated by tetraploid plants. The backcross progeny that are the most vigorous are tetraploid. Therefore introgression is possible from diploids into tetraploid populations via intermediate triploids. Such introgression is another factor contributing to the variability of the tetraploid population system (Stebbins and Zohary 1959).

Other Examples Two other examples of natural autopolyploids will be listed here with references to literature where further details can be found. Medicago coerulea)

sativa, 2n = 16 and 32. The diploid form of M. sativa ( = M.

is a wild plant of western Asia from Turkey to Iran and the

Caspian Sea. The tetraploid form (M. sativa sens, str.) is a cultivated

Natural

Autopolyploids

353

plant. It apparently arose spontaneously in the area of the diploid, and later spread under cultivation into Europe (Lesins 1976). Galax aphylla (Diapensiaceae), 2/i = 12 and 24. The genus is monotypic, containing only this one taxonomic species in the southeastern United States. Diploid and tetraploid individuals were found growing together at numerous sites through the range of the taxonomic species (Baldwin 1941). The setup strongly suggests autotetraploidy (Clausen, Keck and Hiesey 1945; Stebbins 1947, 1980).

PART

SEVEN

Aneuploidy 27. Aneuploidy Descending Basic Aneuploidy Ascending Basic Aneuploidy Polyploid Drop Aneuploid Increase in Polyploid Hybrids Carcx Gaytonia virginica High Basic Numbers in the Angiosperms

357 358 360 363 363 364 365 367

28. Agmatoploidy Polycentric Chromosomes Carex Luzula Cyperaceae and Juncaceae

369 370 372 373 374

CHAPTER

TWENTY-SEVEN

Aneuploidy

Chromosome number differences between individuals or populations (or cells) take three main forms: polyploidy, aneuploidy, and agmatoploidy. Whereas polyploidy represents numerical differences with respect to whole sets of chromosomes, aneuploidy refers to differences in number of individual chromosomes, and agmatoploidy is differences in number of chromosome fragments. Aneuploid variation encompasses a very heterogeneous array of phenomena. The deviations from a standard chromosome number may be up or down. The numerical deviations may involve a single chromosome (polysomy, monosomy), a single chromosome pair, or more than one pair. The extra chromosomes may be homologs of members of the regular complement, or not a part of the regular complement (B chromosomes or accessory chromosomes). And the aneuploidy can occur at the diploid or polyploid level. Aneuploidy at the polyploid level includes several distinct phenomena in itself. Different paired combinations of different basic diploid numbers yield aneuploid variation among the allotetraploid derivatives. Thus two 8-paired species produce a 16-paired allotetraploid, while an 8-paired x 9-paired cross gives a 17-paired allotetraploid. Extra chromosome pairs arise from meiotic irregularities in many polyploids. And there is the

358

Aneuploidy

possibility of a polyploid drop, that is, descending aneuploidy in polyploids. Various types of aneuploid variation may occur in a mixed condition in the same plant individual or population. A species group may display aneuploidy at both the diploid and polyploid levels. Aneuploidy may also be mixed with agmatoploidy in the same plant group. In this chapter we will consider a few examples of ordinary aneuploidy at the polyploid and diploid levels. Polysomy and B chromosomes will not be discussed here (see Khush 1973).

Descending

Basic

Aneuploidy

Aneuploid reduction series at the diploid level are known in quite a few plant groups. Stebbins (1950:455-456) lists 25 such groups. We presented some examples with other relevant information in the chapter on chromosome {«patterning. Several examples are listed below; the trends are expressed in n numbers and missing intermediate numbers are indicated by (). Crepis (Compositae), 6-5-4-3 (Babcock 1947) Youngia (Compositae, 8-()-5 (Babcock and Stebbins 1937) Eriophyllum and Pseudobahia (Compositae), 8-7-()-4-3 (Carlquist, 1956) Nicotiana (Solanaceae), 12-0-10-9 (Goodspeed 1954) Phacelia (Hydrophyllaceae), 11-10-9-8-7- (Cave and Constance 1947, 1963) Clarkia (Onagraceae), 7-6-5 (Lewis 1953b) Knautia (Dipsacaceae), 10-()-8 (Ehrendorfer 1964) Cephalaria (Dipsacaceae), 9-()-5 (Ehrendorfer 1964) Lesquerella (Cruciferae), 8-7-6-5 (Rollins and Shaw 1973) Persoonia and related genera (Proteaceae), 7-0-5 (Johnson and Briggs 1963, 1975) Fritillaria (Liliaceae), 12-()-9 (Darlington 1956b, 1973) Descending aneuploid trends are sometimes apparent between genera in a family or tribe. Two examples will suffice. In the tribe Polemonieae of the Polemoniaceae we have the trend 9-8-7-6; and in the Loranthaceae the trend 12-11-10-9-8 (Grant 1959; Barlow and Wiens 1971).

Aneuploidy

359

Some 783 species of angiosperms have low diploid numbers in the range n = 2 to n = 6. The numbers of species with a given low number are as follows (from Grant 1963): 408 species 225 species 134 species 15 species 1 species

with n = 6 with n = 5 with n = 4 with n = 3 with n = 2.

All of these species are products of phylogenetic reduction in basic number. The mechanism of aneuploid reduction at the diploid level involves unequal reciprocal translocations. (Navashin 1932; Tobgy 1943; Babcock 1947). Chromosomes A and B in the ancestral complement exchange segments of very unequal length, so that all of the genetically active material on chromosome A becomes translocated onto chromosome B, while the centromere of A receives in exchange a small inert segment. The now inert chromosome^ is next lost in some cell division. The derived chromosome complement is thus reduced by one pair. The classical demonstration of this mechanism is Tobgy's (1943) cytological analysis of Crepis neglecta (2n = 8) and C. fuliginosa (2n = 6). The latter species with the reduced chromosome number is derived from an ancestor close to C. neglecta. The individual chromosomes in the two species can be identified by size and morphology in somatic and meiotic cells. The four chromosomes of the set in C. neglecta are designated A, B, C, D; and the three in C. fuliginosa, A, B, D (figure 27.1). The pairing configurations in the interspecific hybrid show that the two species differ by three inversions, by a translocation in the A and D chromosomes, and by a second translocation in the B and C chromosomes. This latter translocation is particularly significant. In the translocation chain one part of the B fuliginosa chromosome pairs with B neglecta and another part with C neglecta (figure 27.1). This shows that the B chromosome of fuliginosa contains segments homologous with the C of neglecta. There is no C chromosome as such in the C. fuliginosa complement, and the basic chromosome number is reduced to this extent, but the genetic material on the ancestral C chromosome is carried on the B chromosome of C. fuliginosa. Phylogenetically this came about by an unequal

360

Aneuploidy

Figure 27.1. Chromosomes of two species of Crepis and their hybrid. (A) Crépis neglecta, haploid set of somatic chromosomes. (B) Crepis fuliginosa, haploid set of somatic chromosomes. (C) F! hybrid, diploid set of somatic chromosomes. (D) Fi hybrid, meiotic chromosomes paired at metaphase I. (Togby 1943) reciprocal

translocation between the ancestral B and C chromosomes, re-

sulting in a new composite B chromosome in C. fuliginosa

Ascending

Basic

(Tobgy 1943).

Aneuploidy

Aneuploid increase at the diploid level is also known in a number of plant groups. Several examples are listed below; the n numbers are given and the missing numbers are indicated by () as in the preceding list.

Aneuploidy

361

Fritillaria (Liliaceae), 12-13 (Darlington 1956b, 1973) Narcissus (Amaryllidaceae), 7-0-10-11 (Fernandes 1951) Crocus (Iridaceae), 12-13-14-15 (Brighton 1978) Phacelia (Hydrophyllaceae), 11-12-13 (Cave and Constance 1947, 1950; Constance 1963) Clarkia (Onagraceae), 7-8-9 (Lewis 1953b) Cephalaria and Succisia (Dipsacaceae), 9-10 (Ehrendorfer 1964) Not surprisingly, there are some aneuploid series that are considered to be ascending by some students and descending by others. Podocarpus (Podocarpaceae) has the haploid numbers 10-11-12-13-0-17-18-19. The phylogenetic trend is ascending according to Khoshoo (1959), but descending according to Hair and Beuzenberg (1958). The tribe Astereae (Compositae) has the haploid numbers 4-5-6-0-8-9. One school of workers favors an ancestral basic number of x = 5 (or 4) (Turner and Home 1964; Turner 1978); another school an original basic number of x = 9 (Solbrig et al. 1964; Semple and Brouillet 1980). One of the ways in which aneuploid increase can occur is by misdivision of the centromere. The centromere, a compbund structure, rarely divides transversely, producing two halves each of which retains the ability to coordinate with the spindle. Then one metacentric (or submetacentric) chromosome gives rise to two telocentric (or subtelocentric) chromosomes (Darlington 1937a, 1973). The basic number in Fritillaria is x= 12 and the complement usually consists of 10 telocentric and 2 metacentric chromosomes. Fritillaria pudica with « = 1 3 has 12 telocentrics and 1 metacentric. Hence the aneuploid increase from n = 12 to n = 13 probably involved the conversion of one metacentric to two telocentric chromosomes (Darlington 1956b, 1973). A similar increase in the number of telocentric chromosomes is associated with aneuploid increase in Narcissus and Crocus (Fernandez 1951; Brighton 1978). Another mechanism of aneuploid increase involves the incorporation of an extra pair of chromosomes of the regular complement as a result of lagging during cell division. This results in a tetrasomic type (2n + 2). The extra or tetrasomic chromosomes can then diverge from their homologs through interchanges with other non-homologous chromosomes and by divergences in gene function. The original basic number in Clarkia (Onagraceae) is x = 7 and the 8-

Aneuploidy

362

and 9- paired species are derived (Lewis 1953b). The aneuploid increase in Clarkia, in some cases at least, has taken place by the incorporation of tetrasomic chromosomes which have undergone translocations. Lewis and Roberts (1956) made a detailed cytogenetic study of the closely related species pair, Clarkia biloba (n = 8) and C. lingulata (n = 9). The latter species is derived from the former. Chromosome pairing in the interspecific hybrid indicates that the biloba and lingulata genomes differ by two translocations and one inversion but are otherwise very homologous. One of the translocations is on the ninth chromosome of C. lingulata. The ninth chromosome of C. lingulata forms a chain of 5 with two other lingulata chromosomes and two biloba chromosomes. The configuration and interpretation are both shown in figure 27.2. Evidently in the origin of the ninth chromosome, two biloba chromosomes with the end arrangements 1-2 and 3-4 exchanged arms to produce a new chromosome 1-4. The two original and one new chromosome then became established in the ancestor of C. lingulata. The latter consequently possesses one more chromosome pair than the ancestral C. biloba (Lewis and Roberts 1956).

C. biloba

I 2 m o b b

3 i

u

4 i

C. lingulata

I 2 h k ^ h

3 i

u

4 i

hybrid

2^

^ '

l V 4

3

3

4

Figure 27.2. Pairing configuration and homologies of certain chromosomes in C. biloba and C. Ungulata. (Rearranged from Lewis and Roberts 1956)

363

Aneuploidy

Polyploid

Drop

Aneuploid reduction at the polyploid level has been aptly termed polyploid drop (Darlington 1956b, 1963, 1973). The loss of one or more chromosome pairs in a polyploid can often be tolerated because of the presence of the duplicate factors in the homologous or homeologous chromosomes. Darlington (1973:105) cites examples of polyploid drop in 20 plant groups. Several of these are repeated below. The n numbers are given for the diploids, polyploids, and reduced aneuploids, in that order. Hesperis (Cruciferae), 7-14-13-12 Linum (Linaceae), (8, postulated)-16-15 Veronica (Scrophulariaceae), 7-21, 20 Maydeae (Gramineae), 5-10-9. In the Proteaceae the ancestral basic number is x = 7 and polyploid doubling to x = 14 occurred early in the history of the family (in the primitive subfamily Persoonioideae). This old tetraploid number also occurs in the derived subfamilies Proteoideae and Grevilleoideae. In the Proteoideae there is in addition a descending series x = 14-13-12-11-10; and in the Grevilleoideae a series x = 14-13-11-10 (Johnson and Briggs 1975).

Aneuploid

Increase in Polyploid

Hybrids

Species hybrids with unpaired univalent chromosomes at meiosis are a source of aneuploid variation in experimental plants and to a lesser extent in natural populations. The aneuploidy is usually ascending, since the gain of extra chromosomes can be tolerated better than the loss of members of the basic complement. This form of aneuploidy is also more likely to be found at the polyploid level than at the diploid level, because of the better buffering properties of polyploid genotypes. One good example is furnished by a hybrid cross and its progeny in the Erophila verna group (Cruciferae) (Winge 1940). The case was reviewed earlier in the chapter on recombinational speciation, and will only be mentioned briefly here in connection with aneuploidy.

364

Aneuploidy

The hybrid cross was between a 15-paired type and a 32-paired type. The Fi hybrid had numerous unpaired chromosomes. In F2 segregation occurred for chromosome number and other features. In later inbred generations the segregates sorted out into different stable lines. Eight stable derivative lines were analyzed cytologically and found to have the following aneuploid numbers: n = 22, 23, 25, 29, 31, and 34. Aneuploidy also occurs in the wild populations of the Erophila verna group in Europe and may well have arisen in a similar way (Winge 1940). Two tetraploid species of Gilia (G. malior and G. modocensis) with the same chromosome number (In = 36) produced an Fi hybrid with numerous unpaired chromosomes at meiosis. This hybrid produced in turn, some homoploid derivates and also a stable aneuploid type with In = 50 (Grant 1966c). Aneuploid numbers of hybrid origin in polyploids are sometimes frozen by agamospermy. For example, the combination of polyploidy, hybridization, and agamospermy is associated with aneuploid variation in theBouteloua curtipendula group (Gramineae) (Gould and Kapadia 1962, 1964). This and other similar examples will be discussed in later chapters on agamospermy.

Carex The genus Carex is remarkable for its long and nearly continuous aneuploid series. The series ranges from n = 6 to n = 56 and includes every gametic number from n = 12 to n = 43 (Heilbom 1932; Stebbins 1950:453; Davies 1956). Short, consecutive aneuploid series (sensu lato) are found within several of the sections of Carex. Examples are Distantes (n = 28-37), Extensae (n = 30-35), Acutae (n = 34-42), and the Carex caryophyllea group (n = 31-34) (Davies, 1956). The basic numbers Jt = 6, 7, 8, and 9 are known in Carex and an additional and perhaps original basic number, x = 5, is inferred (Heilbom 1939; Wahl 1940; Love, Love, and Raymond 1957). Polyploidy occurs in Carex on the above and other secondary basic numbers. Thus the C. siderosticta group has forms with n = 6 and n = 12; the section Capillares (jc = 9) contains diploid, tetraploid, and hexaploid species; and the C. stenantha group has forms with n = 17 and n

Aneuploidy

365

= 34. There are polyploid series in Carex based on x = 5, 6, 7, 8, and 9 (Wahl, 1940; Love, Love, and Raymond, 1957). Some of the observed aneuploid variation in Carex can therefore be attributed to various side effects of polyploidy (Wahl 1940; Love, Love, and Raymond 1957). Monobasic polyploidy on different base numbers will give similar but different derivative numbers, and dibasic polyploidy will fill in the gaps. Thus, different hybrid combinations of 7-paired and 8-paired diploids will produce a series of tetraploids with n = 14, 15, and 16. Polyploid drops and gains may extend the series further in both directions. Natural hybridization is also known to occur in the genus Carex. Drury (1956) reports that nearly all of the species of section Vesicariae in Alaska are involved in hybridization. Natural hybrids occur between Carex rostrata on the one hand and C. rotundata, membranacea, and physocarpa on the other. Hybrid swarms between C. rostrata and C. rotundata in one part of Alaska are segregating out a new constant and fertile form which is increasing in numbers and developing into a new species of hybrid origin named C. paludivagans (Drury 1956). Hybridization and polyploidy account for much of the aneuploid variation in Carex, but this is not the whole story. Agmatoploidy also appears to occur in Carex, as will be noted in the next chapter.

Claytonia

Virginica

Claytonia virginica (Portulacaceae) is a small corm-bearing perennial herb with a wide distribution in central and eastern North America. An extensive aneuploid series occurs in this group. The following In numbers have been counted by Roth well (1959) and Lewis (1970): 2n= 12, 14, 16, 17, 18, 19, 20, 22, 24, 26, 28, 30, 31, 32, 34, 36, 41, 48, 72. Some of these numbers are illustrated in figure 27.3. The original basic number in the group is probably x = 8 (Rothwell 1959). The chromosome number variation can be resolved into several components of diploid aneuploidy and polyploid aneuploidy. First, there is descending basic aneuploidy from x = 8 to 7 to 6. Then a series of tetraploids is built up from different pairs of these diploids to give n numbers of 16, 15, 14, and perhaps 12. The n number 12 is found also in triploids C* = 8, 2/i = 3r = 24) and the n number 11 is also triploid.

366

Aneuploidy

mtft

2n = 12

2n=l7

2n=20

fr

2n =14

2n=l6

2n = l8

2n = 2 6

2n = l9A • • •

2n = 2 8

» 2n = 3 0

2n = 3 2

2n = 3 4

Figure 27.3. Aneuploid variation in Claytonia virginica. Meiotic chromosomes, mostly at metaphase I. Unpaired chromosomes shown white, multivalents stippled. (Rearranged from Rothwell 1959)

Higher polyploids account for larger numbers; the 48-chromosome type is 6r, and the 72-chromosome is probably 9r. Finally, there is some aneuploid variation around the even polyploid levels, as exemplified by the numbers n = 17 and 18 (Rothwell 1959). Aneuploid variation is sometimes found within populations. A population at Carthage, Texas, showed a range of chromosome numbers from

Aneuploidy

367

2/i = 24 to 41, the common numbers being 2n = 24-26 and 28-32. The modal numbers were found to differ from year to year in the Carthage population and from early spring to summer in a St. Louis population (Lewis 1970, 1976). Differences in chromosome number between different parts of the same plant also occur. For example, one individual in the Carthage population had 2n = 30 in the stems and pollen mother cells, but 2n = 28 in the root (Lewis, Oliver, and Luikart 1971). The diploids (jc = 8), tetraploids (x = 14, 15, 16), and diploids (x = 11, 12) are all widespread and broadly sympatric in eastern North America. The diploids with x = 7 and 6 have more restricted geographical ranges (Lewis and Semple 1977). These types are treated as "cytotypes" and "races" of a single species (Lewis and Semple 1977). From an evolutionary standpoint, however, Claytonia virginica has to be regarded as an unanalyzed polyploid species group with superimposed aneuploidy.

High Basic Numbers in the Angiosperms Many genera and some families of woody angiosperms have high basic numbers in the range x = 12 to x = 21. The question of the origin of these basic numbers has been discussed extensively. Old polyploidy is a possibility which is often suggested. It will facilitate our analysis of the problem to recognize two subgroups of high numbers. Quite high numbers in the range x = 19-21 occur in the Magnoliaceae, Trochodendraceae, Hippocastanaceae, Platanaceae, etc. Moderately high basic numbers in the range x = 12-16 are found in some other woody families, e.g., Fagaceae (x = 12), Aceraceae (,r= 13), Betulaceae U = 14), Ulmaceae (JC = 14), Juglandaceae (x = 16), etc. The quite high basic numbers in the first subgroup (JC = 19-21) have been considered by various authors to be of ancient polyploid derivation (Stebbins 1950; Tischler 1954; Darlington, 1956b). I agree that this is the most probable conclusion in this case (see chapters 22 and 25). Consider next the subgroup of moderately high basic numbers (.r = 1216). Stebbins suggests that these are also old polyploid numbers (Stebbins 1938, 1947a, 1950, 1971). An alternative possibility, which has been ignored, is that these numbers are products of ascending aneuploidy (Grant 1971).

368

Aneuploidy

This brings us to the question of the original basic number of the angiosperms. One suggestion is that the original number lies in the range x = 7 - 9 (Grant 1963, 1971; Ehrendorfer 1964). The gametic numbers n = 7, 8, and 9 occur together in many angiosperm families, including some primitive ones such as the Annonaceae (Walker 1972), and apparently represent a basic condition in the class. Many students have gone on to pick x = 7 as the original basic number (Darlington and Mather 1949; Darlington 1956b; Hair 1966; Ehrendorfer et al. 1968; Raven and Kyhos 1965; Raven et al. 1971; Raven 1975; Walker 1972). However, I cannot see any compelling reason for picking x = 7 specifically. The suggestion of a primitive rangex = 7 - 9 still seems to be the safest conclusion that can be drawn with any degree of reliability from the available evidence. Taking x = 7 - 9 as an original starting point, we can readily see that moderately high basic numbers like x = 12, 13, and 14 could be of old tetraploid origin, with or without polyploid drop. But they could equally well be products of ascending aneuploidy. We would expect ascending aneuploidy to occur in many groups of forest trees and shrubs for the following reason. A high diploid chromosome number is one of the means of maximizing the generation of recombinational variability in a plant population. And the ecological conditions in stable forest communities are such as to favor open recombination systems (see chapter 1). Consequently, the dominant and subdominant species of forests are often exposed to selection for aneuploid increases. It is to be expected, therefore, that many high-number woody plants are true diploids.

CHAPTER

TWENTY-EIGHT

Agmatoploidy

Aneuploidy (sensu stricto) is numerical variation in whole chromosomes. The term agmatoploidy, in contrast, refers to differences in the number of independently assorting pairs of chromosome fragments in a group (Malheiros-Gardé and Gardé 1951; Love, Love, and Raymond 1957). The two types of chromosome number variation are correlated with the type of centromere. In most plants and animals the chromosome arms are attached to localized centromeres which perpetuate themselves and their chromosomes through a series of cell generations. The gain or loss of one or more chromosome pairs in the complement depends on the gain or loss of the corresponding centromeres. Conversely, chromosome arms or fragments cannot persist in the cell line or population unless they are attached to a centromere. Changes in chromosome number are aneuploid (sensu stricto) or polyploid. A few groups of plants and animals have diffuse centromeres. The centromeric activity is not localized in a single region but is spread out through the length of the chromosome. Chromosomes containing diffuse centromeres have been termed polycentric (Lima-de-Faria 1949; Bemardini and Lima-de-Faria 1967). Fragments arising by transverse divisions of an original polycentric chromosome can perpetuate themselves through cell divisions, if their broken ends heal, and can thus become permanent

370

Aneuploidy

components of the population. Diffuse centromeres open up the possibility of agmatoploidy (Lima-de-Faria 1949; Love, Love, and Raymond 1957). Agmatoploid series cannot be distinguished from aneuploid series proper in a simple list of chromosome numbers. One may encounter 2n = 10, 12, 14 . . . i n either case. Therefore the term aneuploidy in its general sense is still useful for descriptive purposes in dealing with nonpolyploid chromosome number variation. The phenomenon now known as agmatoploidy was formerly subsumed under aneuploidy, but was segregated out as its nature and special features became better understood (Malheiros-Gardé and Gardé 19S1). As a corollary of this terminological segregation, aneuploidy acquires a more narrowly defined usage as well as the original broad usage. It will be necessary in this chapter to use the term aneuploidy in both the strict and the general sense; the usage will be duly identified. It will be evident from the discussion in this chapter that our understanding of agmatoploidy is still in its early exploratory stages. The cytogenetic facts are not settled beyond dispute in many cases. The relation between agmatoploidy and plant speciation is unclear. We do not know how different agmatoploid populations are isolated reproductively. And we do not understand the adaptive value of agmatoploid changes.

Polycentric

Chromosomes

In the animal kingdom, chromosomes with diffuse centromeres are found in some Hemiptera or true bugs, i.e., in the Coccidae, Aphidae, and Heteroptera, and in the Acarina or mites (Lima-de-Faria 1949). Among higher plants, diffuse centromeres occur in the Juncaceae (Luzula, Juncus) and Cyperaceae (Eleocharis, Scirpus, Carex) and have been suggested for the Zingiberaceae and Musaceae (Love, Love, and Raymond 1957, for review with references). Polycentric chromosomes containing diffuse centromeres differ in morphology and behavior from the ordinary type of chromosome with a localized centromere. It will be recalled that ordinary chromosomes exhibit a definite centromeric constriction at metaphase of mitosis, and that these centromeres become aligned on the metaphase plate, with the chromosome arms dangling in various positions. Polycentric chromosomes lack

Agmatoploidy

371

the constriction and, moreover, the whole chromosome becomes aligned on the metaphase plate. At anaphase ordinary chromosomes move toward the poles with the centromeres in the forward position and drawing the arms behind. The polycentric chromosome, by contrast, drifts toward the pole in a nearly single plane or even with its ends ahead of the midsection (Love, Love, and Raymond 1957). The above-mentioned features of polycentric chromosomes are exhibited by mitotic cells of Luzula campestris (Juncaceae) and are shown in figure 28.1 (Brown 1954). There has existed a question as to whether the centromeric activity is continuously or discontinuously distributed throughout the length of a polycentric chromosome. In other words, is the diffuse centromere, fully or incompletely diffuse? The evidence in Luzula purpurea supports the latter possibility. The chromosomes of this species apparently contain numerous centromeric segments (Bernardini and Lima-de-Faria 1967). It is well known that acentric fragments of ordinary chromosomes tend to lag during cell division and sooner or later become lost from the viable cell lines. In organisms with diffuse centromeres, however, chromosome fragments could be expected to complete the normal metaphase and anaphase movements. H&kansson induced chromosome breakage by X-ray treatment in Eleocharis palustris (Cyperaceae). He found, as expected,

Figure 28.1. Mitosis in Luzula campestris (2« = 1 2 ) . Lateral views showing chromosome orientation at metaphase and early anaphase. (Brown 1954)

372

Aneuploidy

that the chromosome fragments, even small ones, arranged themselves on the equatorial plate at metaphase and moved to the poles at anaphase (Hikansson 1954, 1958). Spontaneous chromosome breakage in organisms with polycentric chromosomes may therefore lead to the formation of ascending agmatoploid series. A species group may show much variation in the number of chromosome pairs. And the chromosome sets containing the larger numbers will include at least some smaller-sized members (Hikansson 1954; Love, Love, and Raymond 1957). This is not to say that agmatoploidy is an inevitable result of fragmentation of polycentric chromosomes. Other factors, cytological as well as selective, are undoubtedly involved. White (1957a, 1957b) has called attention to one probably important cytological factor, namely, the role of the telomere or chromosome end. If the fragments do not acquire normal healed ends, they may not survive long (White 1957a, 1957b). In fact, aneuploidy (sensu lato) does not occur universally in all groups of organisms which possess polycentric chromosomes. Among Heteroptera with diffuse centromeres, some families such as the Pentatomidae and Corixidae show great constancy in chromosome number, while other families have a wider range of chromosome numbers (White 1957a). Various species of Luzula, Eleocharis, and Carex have retained low basic numbers. There is relatively little aneuploidy (sensu lato) in Eleocharis.

Carex Aneuploidy (sensu stricto) is well developed in the genus Carex, as noted in the preceding chapter, but there is evidence for the occurrence of agmatoploidy also. Carex chromosomes do not have a visible localized centromere. On chromosome morphology they appear to have a diffuse centromere. Moreover they often move end first to the poles at anaphase. These facts indicate that the chromosomes are polycentric, and suggest that fragmentation could contribute to the observed variation in chromosome number (Davies 1956; Love, Love, and Raymond 1957). Heilborn (1924, 1932) compared Carex pilulifera and C. ericetorum with respect to total length as well as number of the somatic chromosomes. He found that the species with the higher number, C. ericetorum

Agmatoploidy

373

(rt = 15), has shorter chromosomes than the lower-number species, C. pilulifera (n = 9). The number of long chromosomes in the complement decreases as the total number of chromosomes increases (Heilbom, 1924, 1932). Later workers have confirmed this observation in other species groups of Carex. The species with higher numbers usually have more short chromosomes than their low-number relatives (Davies 1956; Love, Love, and Raymond 1957). The inverse relation between chromosome number and chromosome size could be due to several factors other than fragmentation. Consequently this line of evidence does not prove that fragmentation and agmatoploidy have contributed to the observed chromosome number variation in Carex. The size-number relation, however, when taken in conjunction with the occurrence of polycentric chromosomes and the unusually extensive development of aneuploidy (sensu lato), does suggest strongly that agmatoploidy has played a role in speciation in the genus Carex.

Luzula The chromosomes of Luzula (Juncaceae) have been shown to possess diffuse centromeres (Malheiros, De Castro, and Camara 1947; Nordenskiold 1951, 1962; Brown 1954). These and other authors have considered the possibility that chromosome fragmentation has contributed to chromosome number differences within this genus. Chromosome numbers in Luzula range from 2n = 6 in L. purpurea to In = 66 and 72 in L. pilosa. The number In = 12 is found in many species and is a basic number in the genus. There is an ordinary polyploid series on this base of x = 6 which reaches the octoploid level (Nordenskidld 1951). Aneuploidy (sensu lato) is also present. Thus Luzula spicata has the contiguous numbers n = 6 and 7, and L. orestra has n = 10 and 11. In each case the complement with the higher number includes two small chromosomes corresponding to one large chromosome in the lowernumber complement. The aneuploid increase is therefore probably due to fragmentation and agmatoploidy (Nordenskiold 1951, 1956). In addition, Nordenskiold (1951) finds polyploid series of a special kind in which the total length of all chromosomes combined remains

374

Aneuploidy

about constant. The individual chromosomes of a tetraploid are about half the size of those in the diploids; and the chromosomes of an octoploid are about half as large as those in the tetraploids. This situation is found in the Luzula campestris and the L. spicata groups. Nordenskiold attributes the simultaneous changes in number and size to fragmentation of all the chromosomes in the ancestral complement. The numerical relations between species are polyploid, but the mode of doubling may be fragmentation rather than multiplication, and the series then constitutes a special type of agmatoploidy (Nordenskiold 1949, 1951, 1956, 1961). Confirmation of this hypothesis of agmatoploidy is provided by the pairing behavior of large chromosomes with small ones in hybrids between species of different ploidy levels. Luzula campestris and L. pallescens have 6 pairs of large chromosomes; L. sudetica has 24 pairs of small chromosomes and is technically octoploid as to number. At meiosis in the interspecific hybrids, each of the six large chromosomes belonging to the diploid parental species pairs with several small chromosomes contributed by the high-number species, L. sudetica (Nordenskiold 1951, 1956, 1961).

Cyperaceae

and Juncaceae

The family Cyperaceae possesses several unusual and distinctive cytological and embryological features. These features have been pointed out by Heilborn (1924, 1932) and many later students. In the first place, polycentric chromosomes occur in some genera of Cyperaceae. Diffuse centromeres occur in Eleocharis and Carex, as we have already seen. Fimbristylis and some species of Cyperus, on the other hand, have localized centromeres. Both types of centromere apparently occur in Bulbostylis (Love, Love, and Raymond 1957). Agmatoploidy also occurs in the Cyperaceae as a correlated condition. Remarkably extensive and continuous aneuploid series (sensu lato) occur in Carex. Some aneuploidy is found in Eleocharis. But the species of Fimbristylis fall into an even or nearly even polyploid series. In various members of the Cyperaceae, moreover, the two meiotic divisions occur in an order which is the inverse of that in other plants. The paired chromosomes do not disjoin in the first meiotic division, but do separate in the second division. In contrast to the normal sequence, di-

Agmatoploidy

375

vision I is equational and division II reductional (Wahl 1940; E>avies 1956). Davies (1956) has pointed out that postreductional meiosis is closely correlated with diffuse centromeres in both the angiosperms and the insects. The explanation of this correlation is not clear. A fourth peculiar characteristic of the Cyperaceae is the mode of pollen formation. In each tetrad of microspores, three degenerate and only one develops into a functional grain (Heilborn 1932; Wahl 1940; Davies 1956). It is very interesting that most of these unusual cytological and embryological features appear again in the family Juncaceae. Diffuse centromeres are found in Luzula, as mentioned earlier in this chapter, but not apparently in Juncus. Correspondingly, Luzula exhibits much aneuploidy (sensu lato), including some agmatoploidy, whereas Juncus has almost no aneuploidy. Postreductional meiosis occurs in the Juncaceae according to Nordenskiold (1962). The four microspores all develop in the Juncaceae, but adhere together in a tetrad, which represents a condition reminiscent of that found in the Cyperaceae (Wulff 1939; Maheshwari 1949). The foregoing cytological and cytogenetic similarities between the Juncaceae and Cyperaceae, taken together with various morphological and anatomical similarities, indicate that the two families are much more closely related to one another than was assumed in the older systems of classification, where the Juncaceae were allied to the lilies and the Cyperaceae to the grasses (Wulff 1939; Maheshwari 1949; HSkansson 1954; Cronquist 1968).

PART EIGHT

Specialized Genetic Systems 29. Permanent Translocation Heterozygosity The Oenothera Genetic System Sources of Variation The Heterogamic Complex in the Oenothera hookeri-biennis Group Helercgamic Microspecies in the Oenothera biennis Group Heterogamic Microspecies and Complexes in Other Plant Groups 30. Permanent Odd Polyploidy The Rosa canina Genetic System The Helercgamic Complex in the Rosa canina Group Leu cop ogon juniperinus Cardamine insueta Evolutionary Potential of Helercgamic Complexes 31. Agamospermy Embryological Pathways Systematic Distribution Breeding System Hybridity and Agamospermy Formation of Gametophytic Apomixis Breeding Behavior Segregation Variation Pattern Advantageous Features cf Agamospermy 32. The Agamic Complex The Crepis occidentals Complex The European Blackberries Taraxacum The Rubus moriferus Pattern

379 380 385 387 394 Iff? 401 401 404 406 409 410 413 413 418 422 423 424 426 428 430 431 434 436 439 442 444

378

Specialized

Genetic

Systems

The Bouteloua curtipendula Complex Citrus Ptcris Stages of Development Evolutionary Potential 33. Vegetative Propagation of Hybrids Hybrid Clones Clonal Microspecies Clonal Complexes with Reproduction by Stem Offshoots Clonal Complexes with Reproduction by False Vivipary Evolutionary Potential 34. Natural Triploids Formation Maintenance Selected Examples

445 451 457 458 459 462 463 464 466 468 470 473 475 476 477

CHAPTER

TWENTY-NINE

Permanent Translocation Heterozygosity

Permanent translocation heterozygosity is a peculiar and highly specialized genetic system based on successive translocations and manifested in large chromosome rings. Reciprocal translocations are fairly common in many plant groups. Small rings (rings of 4 or 6 chromosomes) occur as polymorphic variants along with the corresponding bivalent-forming structural homozygotes in the natural populations of such plant groups. In some cases the small-ring-forming plants have a high frequency in the population as a result of a selective advantage of the heterozygotes. In a few cases large-ring-forming plants are maintained at high frequencies in the population by heterozygote advantage (see Grant 1975, ch. 22 for review). These cases approach but do not quite attain the condition of permanent translocation heterozygosity, which goes one step further. Permanent translocation heterozygotes are heterozygous for successive translocations involving at least several and usually most or all chromosomes of the complement. Consequently they form large or complete chromosome rings at meiosis. Genotypically they are diploids composed of two genomes of a special sort known as Renner complexes. The permanent translocation heterozygote breeds true by sexual means for its structural heterozygous condition, with little or no segregation of

380

Specialized Genetic Systems

homozygous individuals. It usually accomplishes this by the transmission of one Renner complex through the ovules and another through the pollen. The unilateral transmission is effected by a system of balanced lethals. The two Renner complexes are reassembled by fertilization, usually self-fertilization, and so a given heterozygous condition is perpetuated from one sexual generation to the next. Permanent translocation heterozygosity, then, is a genetic system characterized by the combination of several features: successive translocations, medium to large chromosome rings, balanced lethal systems, sexual reproduction usually by self-fertilization, and true-breeding for a particular structurally heterozygous condition. A permanent translocation heterozygote, if it has an adaptively valuable genetic constitution, can build up a population of identical individuals and spread throughout a favorable territory. It can develop into a hétérogamie microspecies. Hétérogamie microspecies, in turn, with further development, can become components of a larger assemblage, the hétérogamie complex. A hétérogamie complex is a hybrid complex composed of hétérogamie microspecies and their meiotically normal, bivalent-forming, ancestral species or semispecies.

The Oenothera

Genetic

System

The numerous heterogamic microspecies belonging to the Oenothera biennis group in North America and Europe are herbaceous diploids (2n = 14) with a predominantly self-pollinating breeding system. They are permanent translocation heterozygotes which form complete rings of 14 chromosomes in many microspecies and smaller rings in others. A ringforming Oenothera plant, following self-fertilization, yields all ringforming progeny. The expected classes of structural homozygotes are absent in the progeny of the translocation heterozygote. The latter breeds true by sexual means for its hybrid constitution. The cytogenetics of Oenothera has been worked out by a succession of investigators. The anomalous breeding behavior of these plants was discovered by DeVries and explained by Renner in the classical period. The analysis was then greatly extended by Cleland and his students during a period of more than forty years. Good reviews of the early work will be

Permanent

Translocation

Heterozygosity

381

found in Cleland (1936) and Darlington (1937a). Much factual evidence is summarized in the later monograph of Cleland and co-workers (19S0). A review with numerous literature references is given by Cleland (1962). The most comprehensive treatment is the book-monograph by Cleland (1972). For a concise recent review see Grant (1975). It is convenient to describe the genetic system of permanent translocation heterozygosity here in terms of the classical subject, Oenothera lamarckiana. Oenothera lamarckiana is widely distributed in Europe and North America (figure 29.1) (Munz 1949, 1965). The chromosomes of Oenothera lamarckiana pair to form a ring of 12 and 1 bivalent at meiosis (figure 29.2). This species is thus heterozygous for successive translocations on 12 of its 14 chromosomes. The end arrangement of the Oenothera lamarckiana chromosomes can be expressed in terms of the arbitrary standard end arrangement found in the related structurally homozygous species, Oe. hookeri. The 14 arms of the 7 chromosomes in a standard race of Oe. hookeri are assigned consecutive numbers, as shown in table 29.1. The two haploid sets are structurally homologous in most individuals of Oe. hookeri, which consequently exhibit regular bivalent formation at meiosis. The haploid sets in Oenothera lamarckiana differ from that in the standard or DeVries race of Oe. hookeri by several translocations (table 29.1). Furthermore, the two sets in Oe. lamarckiana differ from one another with respect to successive translocations on all chromosomes except those with the 1 - 2 ends (table 29.1). Pairing of homologous arms at meiosis thus accounts for the observed configuration of a ring of 12 and 1 bivalent (figure 29.2B). At metaphase I, the ring-forming chromosomes align themselves on the spindle in a zigzag fashion for alternate disjunction (figure 29.2B). At anaphase I, neighboring chromosomes in the ring separate to opposite poles and, conversely, the alternating chromosomes pass to the same poles. In this way the two differentiated genomes are reassembled intact at the end of meiosis (Cleland 1936, 1962). The special types of genomes found in Oenothera are known as Renner complexes and are designated by names. The two Renner complexes in Oenothera lamarckiana are velans and gaudens (table 29.1). Oenothera lamarckiana thus has the heterozygous genomic constitution velans/gaudens. The result of alternate disjunction of the ring-forming chromosomes at meiosis is the segregation of the whole velans complex

382

Specialized

Figure 2 9 . 1 . Oenothera (Munz 1949)

lamarckiana

Genetic

( = 0e.

Systems

erythrosepala,

Oe.

glazioviana).

to one daughter nucleus and the gaudens complex to the other (figure 29.2B). The v elans /gaudens heterozygote, in other words, normally produces just two classes of gametes, velans and gaudens, as far as the structural arrangement is concerned. Self-fertilization or sib crossing of Oenothera lamarckiana would be expected to yield three classes of progeny—velans/velans,

Permanent

Translocation

383

Heterozygosity

A

B

Figure 29.2. Chromosome pairing and disjunction in Oenothera lamarckiana (2n = 14). This species is a velans/gaudens heterozygote. (A) Diakinesis, showing a ring of 12 chromosomes and I bivalent. (B) Metaphase I, with ring aligned on spindle for alternate disjunction at anaphase. (Emerson 1935)

Table 29.1. End arrangement of Renner complexes in Oenothera lamarckiana compared with standard arrangement. (Cleland 1950, 1972) End arrangement of 7 chromosomes in haploid set

Species

Renner complex

Oe. hookeri, DeVries race

standard hookeri

1-2 3-4

Oe. lamarckiana

velans gaudens

1 - 2 3 - 4 5 - 8 7 - 6 9 - 1 0 11-12 1 3 - 1 4 1-2 3-12 5 - 6 7-11 9 - 4 8 - 1 4 13-10

5-6 7-8

9 - 1 0 11-12 13-14

velans/gaudens, and gaudens I gaudens—in a Mendelian ratio of 1 : 2 : 1 . Instead, only the heterozygous type is obtained, as noted earlier. The absence of the expected classes of structural homogygotes is due to the operation of a system of balanced lethals (Renner, 1925; Cleland, 1936, 1962). The gene system consists of complementary lethal factors balanced in heterozygous condition. The double heterozygote

/ +/2 \ L + I is viable; but

384

Specialized

Genetic

Systems

the homozygous recombinations

are not. Each

set of complementary lethal factors is linked to one Renner complex. Taking the balanced lethals into consideration, the constitution of Oenothera

lamarckiana

can be written now as

velans + / 2 gaudens / , +

The

velans/velans and gaudens/gaudens types are homozygous for lethal factors and die in the zygote stage. Not only the lethal factors but also genes determining various phenotypic traits are linked to the diverse Renner complexes (see Cleland 1962, 1972; also Grant 1975 for concise review). Furthermore, the genes borne on different chromosomes belonging to the same Renner complex are transmitted in a single linkage group. The translocation hybrid breeds true, therefore, not only for its particular structurally heterozygous constitution, but also and at the same time for a particular ensemble of phenotypic characters. The several components in the genetic system of permanent translocation heterozygosity vary from species to species within the Oenothera biennis group. One variable is the size of the ring. The ring of 12 chromosomes found in Oe. lamarckiana is not typical of the group as a whole. A population of Oe. irrigua (=Oe. hookeri hewettii) in New Mexico regularly forms a ring of 8 + 3 bivalents. The most common configuration, however, in microspecies throughout the central and eastern United States, is a ring of 14 chromosomes (Cleland 1950, 1972). The balanced lethal system also shows variation. Lethals are absent or sporadic in western members of the group. Balanced lethals are characteristic, however, of the heterogamic microspecies in central and eastern North America and in Europe. Two types of balanced lethal system occur (figure 29.3). In the first type, as exemplified by Oenothera lamarckiana, the lethals act in the zygote stage; in the other, which is widespread in North America, they operate in the gametophytes. Accordingly, the Oenothera plants with zygote lethals are semisterile as to seeds, and those with gametophyte lethals are semisterile as to pollen (Cleland 1936, 1950, 1962, 1972). A further consequence of a system of gametophyte lethals is that a given heterogamic microspecies regularly transmits one

A A

M

B

AB

Permanent

Translocation Heterozygosity

B

A

AB

385

K

*

B

AB

Figure 29.3. Two types of balanced lethal system in Oenothera. Left, system employing zygote lethals. Right, system with gametophyte lethals. (Cleland 1964)

of its Renner complexes through the pollen and the opposite complex through the eggs.

Sources of Variation The causes of hereditary variation in the ring-forming Oenotheras are diverse. Occasional outcrossing between ordinarily inbreeding microspecies is one important source of new forms. Other processes operate to produce new variations within an inbreeding line itself. For, although Oenothera lamarckiana and its relatives generally yield uniform progeny after self-pollination, they do not do so exclusively, but give rise to a certain low proportion of new segregate forms, which comprise the mutations of DeVries. In order to discuss the mutability of Oenothera it is necessary first to recognize that some genes are linked to a Renner complex in a ringforming heterozygote, whereas other genes are not. The latter segregate in the normal way. In an incomplete translocation heterozygote like Oe. lamarckiana, the bivalent-forming chromosomes segregate independently of the ring-forming chromosomes, and the genes on the bivalent chromosomes are consequently not a part of the Renner complexes. Genes borne on the terminal pairing segments of the ring-forming chromosomes can cross more or less freely from one Renner complex to the other in a complex heterozygote. Such genes are also independent of any particular Renner complex (see Cleland 1962). These genes which recombine freely do not of course give rise to new "mutants" in Oenothera.

386

Specialized Genetic Systems

The genes determining the characteristic phenotypic traits transmitted in a block through any given Renner complex, and passed on uniformly in inheritance by any given complex heterozygote, must be located on the nonpairing differential segments of the translocation chromosomes (Darlington 1937a, Cleland, 1962). These are the proximal chromosome regions in the neighborhood of the translocation breaks; for an explanation of the chromosome mechanics involved see Grant (1975). The balanced lethal factors must also be located in these differential segments. The translocation heterozygote then breeds true for the particular heterozygous gene combination in this part of the genome by means of the genetic system described in the preceding section. Cytogenetic events occur at regular but rare intervals, however, to separate one or more genes on the differential segments from their respective Renner complexes and expose them to expression in new homozygous combinations. One such process is rare crossing-over and exchange of factors between Renner complexes. A gene may be crossed over occasionally into the opposite Renner complex. Self-pollination then yields some exceptional progeny which are homozygous for this gene and appear as new stable mutant types. Another process is the occurrence of new reciprocal translocations within a given Renner complex. This can break up a large ring in a complex heterozygote into smaller rings and bivalents. Whole chromosomes and blocks of genes can then form homozygous segregates after self-pollination. The exceptional homozygous segregates arising by the above and other means constitute the diploid mutations in Oenothera (see Renner 1941; Cleland 1962, 1972). The ring-forming chromosomes are subject to irregularities of separation at anaphase I. Nondisjunction occurs occasionally and gives some 8chromosome gametes. They in turn yield some trisomic progeny with altered phenotypes. The majority of the mutants in Oenothera lamarckiana have turned out to be trisomies with In = 15 (Cleland 1962, 1972). Hybridization between different complex heterozygotes is the second main source of new variations in Oenothera. Two heterogamic microspecies, each of which normally yields one class of offspring on selfing, often give rise to four classes of progeny in their first hybrid generation. Thus the cross of Oe. lamarckiana (velans/gaudens) x Oe. strigosa (deprimenslstringens) yields the four possible diploid combina-

Permanent

Translocation

Heterozygosity

387

velansldeprimens, velans/stringens, gaudens/deprimens, and gaudens/stringens (Darlington 1937a; Cleland 1962). tions:

Many of these new genome combinations, in the above and in other hybrid crosses, have more bivalents or small rings than either parent (Darlington 1937a; Cleland 1962). Some gene pairs which are linked to Renner complexes in the parental microspecies, and are maintained there in a permanently heterozygous condition, will then become independently assorting in the intermicrospecies hybrids. Such genes can go on to segregate into new homozygous types in later generations. And, where the homozygotes are viable, these new segregates can become permanent additions to the variation pool of the hybrid complex. Some first-generation hybrids between two heterogamic microspecies have complete or nearly complete rings associated with a new combination of balanced lethal factors. Such hybrids will breed true. They represent the potential beginnings of new hybrid microspecies originating as direct products of crossing between preexisting hybrid microspecies (Cleland 1962).

The Heterogamic

Complex in the Oenothera biennis Group

Hookeri-

The North American species of Oenothera subgenus Oenothera are tall perennial to annual herbs, often weedy, with yellow vespertine flowers. They are all diploids with In = 14. The group has been extensively studied from the cytogenetic standpoint by Cleland and his school, and from the taxonomic standpoint by Munz (Cleland 1949, 1950, 1962, 1964, 1972; Munz 1949, 1965). The North American Euoenotheras are taxonomically difficult owing to the large range of variations and the absence of clear-cut interspecific boundary lines. In his revision of this assemblage, Munz (1965) recognizes ten taxonomic species, many of which are polytypic with two or more subspecies. Cleland (1962, 1964, 1972) recognizes a similar array of taxonomic species, or "groups of r a c e s " as he terms them. The systems of Munz and of Cleland, though not identical, are similar and well coordinated. The two systems of classification are summarized and correlated in table 29.2.

388

Specialized

Genetic

Systems

(Since this chapter was written a new taxonomic treatment of the Euoenotheras has been presented by Raven, Dietrich, and Stubbe [ 1 9 7 9 1980], Their system differs substantially from that o f Munz. It is summarized and compared with the Munz and Cleland systems in table 2 9 . 2 . ) Most of the genetic work has been reported in connection with Cleland's nomenclature. It will save confusion to use that nomenclature in the following discussion. The taxonomic equivalents of Cleland's units can be found in table 2 9 . 2 . Oenothera

lamarckiana

arose in Europe and is adventive in North

America. We are primarily concerned here with the other nine, native North American species, comprising the Oe. hookeri-biennis

group.

The geographical distribution of the nine native species is shown in Table 29.2. North American taxa of Oenothera subgenus Oenothera by Munz (1965), Cleland (1972), and Raven et al. (1979- 1980) Munz's system

Cleland's names

recognized

Raven et al.

Oe. hookeri (9 subspp.)

Oe. hookeri

Oe. elata (hookeri sunk in elata)

Oe. longissima (2 subspp.)

Oe. hookeri

Oe. longissima

Oe. jamesii

Oe. hookeri

Oe. jamesii

Oe. elata

Oe. elata

Oe. elata (expanded concept) Oe. wolfii (segregated from hookeri)

Oe. argillicola

Oe. argillicola

Oe. argillicola

Oe. grandiflora L'Heritier (not de Vries)

Oe. grandiflora

Oe. grandiflora

Oe. parviflora II

Oe. oakesiana

Oe. parviflora subsp. parviflora

Oe. parviflora I

Oe. parviflora

Oe. biennis subsp. caeciarum

Oe. biennis II

Oe. biennis

subsp. centralis

Oe. biennis I

Oe. biennis

subsp. austromontana

Oe. biennis III

Oe. austromontana

Oe. strigosa (3 subspp.)

Oe. strigosa

Oe. villosa

Oe. erythrosepala (naturalized in N.A.)

Oe. lamarckiana de Vries

Oe. glazioviana

subsp. angustissima

Permanent

Translocation

Heterozygosity

389

figures 29.4 and 29.5. Good distribution maps have not been published by Cleland or Munz. The distribution areas shown in our maps are therefore approximate only, but are based on the published accounts of these authors. The taxonomic species fall into two series on type of genetic system, as noted by Cleland (1950, 1962, 1964, 1972). One subgroup has a normal diploid genetic system. The plants have large flowers, are open-pollinated, form bivalents at meiosis (with some exceptions), and lack balanced lethal factors. The second subgroup consists of permanent translocation heterozygotes. The plants have large chromosome rings (mostly rings of 14), predominant self-pollination, and balanced lethals. The bivalent-forming species are. Oe. hookeri, argillicola, grandiflora, and elata. The derived ring-forming species are: Oe. parviflora, biennis, strigosa, and lamarckiana. Some populations of Oe. hookeri, Oe. jamesii, and Oe. longissima have small or medium-sized rings and, in some cases, also balanced lethals, and are thus transitional between the two main subgroups (Cleland 1962, 1975; Munz 1949, 1965). The phylogenetic roots of the ring-forming species are clearly to be sought in the bivalent-forming species. The latter have of course the normal sexual genetic system. They have mesic ecological preferences which are considered to be primitive for this plant group. Some of the bivalentforming species are disjunct and relictual, indicating a fairly old age. By contrast, the ring-forming species with their derived genetic system are aggressive and often weedy; they occupy disturbed fields in the middle west and extend into northern areas recently under ice. It is very significant, therefore, that one particular Renner complex known as hookeri Johansen is common and widespread in Oe. hookeri and occurs again in Oe. grandiflora and Oe. argillicola. The end arrangement of hookeri Johansen is 1 - 2 3 - 4 5 - 6 7 - 1 0 9 - 8 11-12 1314. This is not the only Renner complex found in Oe. hookeri but it is the most common one. Furthermore, the individual chromosomes in the hookeri Johansen complex reappear frequently in other complexes. For these reasons the hookeri Johansen complex is regarded as an ancestral end arrangement in the group (Steiner 1951, 1952; Stinson 1953; Cleland 1972). The three or four taxonomic species of ring-forming Euoenotheras are poorly distinguished and very variable, both phenotypically and genomically. Phenotypically they range from mesophytic types with broad thin

Cleland 1972)

Figure 29.5. Geographical distribution of the ring-forming species of subgenus Oenothera native in North America. (Based on data of Munz 1965; and Cleland 1972)

392

Specialized

Genetic

Systems

leaves to xerophytic types with narrow thick leaves. Genomically, there are about 130 known Renner complexes in Oe. parviflora, biennis, and strigosa (this number is based on a count of the different complexes listed in Cleland 1975, appendix). These complexes are combined in pairs to form innumerable heterogamic microspecies, some local, some wide-ranging. The phylogenetic relationships of the derived ring-formers to the ancestral bivalent-formers are consequently not simple and clear-cut. Some phylogenetic relationships can be seen in broad outline, but large gaps also occur. In general, the phylogeny can be said to be reticulate, since the ring-forming taxonomic species and microspecies are products of hybridization between different pairs of parental species. Cleland's (1964, 1972) reconstruction of the phylogeny is based on evidence of morphology, geography, chromosome end arrangements, plastids, and 5 factors or self-incompatibility factors that may be transformed into pollen lethals in the balanced lethal systems. The various lines of evidence do not always agree, and this leads to uncertainties in some parts of the phylogeny. Cleland's (1964, 1972) phylogenetic hypothesis is summarized in figure 29.6 and in the following additional comments. Cleland suggests that the Oe. hookeri-biennis group had its center of origin in Mexico and Central America. The primitive forms migrated from this center into temperate North America in five successive waves, giving rise to five species or semispecies designated Populations I, II, III, IV, and V respectively. Population I is the immediate ancestor of Oe. argillicola. Population II of Oe. grandiflora, and Population V of Oe. hookeri. Populations III and IV are hypothetical. Populations H V are the ancestors of the ring-forming Euoenotheras. They hybridized in different combinations to give rise to the ring-forming types as indicated in figure 29.6. Furthermore, translocations accumulated in the various lines over the course of time. The assumptions in this hypothesis regarding center of origin and migration waves are unnecessary. Also, Oe. hookeri proper supposedly does not contribute to the derived types, but this seems unlikely in view of its wide geographical sympatry with Oe. strigosa. It might be better, therefore, to drop some of these assumptions and simply postulate a divergence into several ancestral species in North America, of which some are known and some unknown. We can then go on to the better documented

Permanent Translocation Heterozygosity

393

Figure 29.6. Postulated phylogeny of the native North American Euoenotheras. (redrawn with minor modifications from Cleland 1964, 1972) relationships of the ring-forming types. The superstructure of the phylogenetic network (in figure 29.6) is better known than the base. The egg-transmitted (or so-called alpha complex) of Oe. biennis I has the hookeri Johansen end arrangement in two known populations, and an arrangement one translocation removed from hookeri Johansen in many other populaions over a wide area, along with still other more derived arrangements. Oenothera biennis I and II are very similar phenotypically and have parallel ancestries from the same two parental species. One of these was which furnished the hookeri Johansen complex to Oe. Oe. grandiflora, biennis I (but not to Oe. biennis II). On the other side Oe. biennis I and II share a common ancestry with Oe. strigosa. But the s/ngtwa-like characters are carried in the pollen-transmitted (beta) complex in Oe biennis I and in the female (alpha) complex in Oe. biennis II. The alpha and beta complexes are reversed in Oe. biennis I and II. Oenothera biennis III is a product of biennis I 9 x biennis II c? (Cleland 1972; figure 29.6).

394

Specialized

Genetic

Systems

The Renner complexes in Oe. parviflora, strigosa, and biennis (with the exception of the alpha complex of biennis I mentioned above) are quite different from the ancestral hookeri Johansen complex, being three or more translocations removed from the latter. The two Renner complexes in Oe. parviflora I are both far removed from those in the putatively ancestral Oe. argillicola and Oe. grandiflora. Oenothera parviflora II is also genomically distant from its putative ancestor, Oe. argillicola. Oenothera parviflora I has a common ancestry with Oe. biennis as regards its alpha complexes. Oenothera parviflora II, unlike parviflora I, carries sfr/gaya-like characters in its alpha complexes and apparently shares a common ancestry with Oe. strigosa (figure 29.6).

Hétérogamie

Microspecies in the Biennis Group

Oenothera

A diverse array of Renner complexes on both the egg-transmitted and pollen-transmitted side exists within the ring-forming species. The number of alpha and beta complexes listed by Cleland (1972, appendix) is shown in table 29.3, and some examples are given in table 29.4. The possible number of diploid combinations of these alpha and beta complexes clearly runs into the scores or hundreds within each ring-forming taxon. A diploid combination of two Renner complexes defines a given hétérogamie microspecies. Hétérogamie microspecies are the basic population units of the hétérogamie complexes and the raison d'etre of the Oenothera genetic system, since they are true-breeding hybrids heterozygous T a b l e 2 9 . 3 . T h e n u m b e r of k n o w n Renner complexes in Oenothera parviflora, biennis, and strigosa. (From count of complexes listed in Cleland 1972, appendix)

Taxon Oe. Oe. Oe. Oe. Oe. Oe.

parviflora I parviflora II biennis 1 biennis II biennis III strigosa

No. alpha complexes

No. beta complexes

9 6 14 12 8 27

2 4 27 7 4 11

0 1o O .£ « Oo Mea '5 (3 4>cu P® 0 e0*>.O a §> > E.z •tíco •g o-•8i •C -C ü B «e e "uE I8l8 S M ii • IE UC "oo •35 11 o u £ o -

f N f•o M»

•

&

r —T OO,•»_ oo «o "i « ml (iN lf*l mm NMNNNMNM Ti i i 7 7 i i o 00 00 — rs ^I ^I9I ^I ^ I J A O — o— o— o Tt— o 00 I IIIIII I r-r-r-Ttr^r-r^-o ^ 'O NO waTjw 00 ^ I IIIIII I »nv'ivivivivimv-i o — I IIIIII I ftcicirtcifimci 00

o •oc o e O

M

I V

"o. Ë o c e u OC Co /3 »s « ja eg H

e — C C ° •/> uî 1/1 u u O O f! jefsJÍ2 ? et««»* 00 00 8 « 'fi

396

Specialized

Genetic

Systems

for a set of successive translocations and for the genes borne on the translocated chromosomes. A particular genie heterozygote, if adaptively valuable, can multiply its heterozygous constitution by means of the Oenothera genetic system, and spread as a heterogamic microspecies of local or regional extent. Cleland did not list the complex heterozygotes or heterogamic microspecies as such, but it is possible to reconstruct them from his data by combining the alpha and beta complexes recorded separately from the same locality (in Cleland 1972, appendix). We count the following numbers of known heterogamic microspecies in each ring-forming species: Oe. paniflora, strigosa,

4 microspecies; Oe. biennis,

90 microspecies; and Oe.

41 microspecies.

The genomic constitution of several

heterogamic

microspecies

is

shown for purposes of illustration in table 29.5. Although the areal extent of these and most other microspecies in the group is unknown, many of them are probably local in distribution. Some, however, are widespread. For example, the microspecies of Oe. biennis

II labeled Northeast in table

29.5 is recorded from sites in Pennsylvania, New York, Ontario, New Brunswick, and Quebec; it obviously represents a successful heterozygous combination. In recent years electrophoretic methods have been brought to bear on the analysis of microspecies in the Oe. biennis

group by Levin and co-

workers. A population of Oe. biennis

II in a field in Connecticut was assayed

for 19 enzyme loci. Five of the 19 loci were heterozygous, the mean heterozygosity being 26% (5/19), a very high level. Moreover, all individuals in the population have the same heterozygous genotype for the enzyme loci. In other words, the level of polymorphism is low. O n e particular heterogamic microspecies characterizes this population and probably a series of populations throughout a large area in Connecticut (Levin, Howland and Steiner 1972). Levin (1975b) assayed 44 populations of Oe. biennis

I in Illinois for

20 enzyme loci. The mean heterozygosity was 4 . 5 % of the loci. The levels of polymorphism in the populations were as follows: 1 2

genotype in 59% of populations genotypes in 27% of populations

397

Permanent Translocation Heterozygosity

3-5

genotypes in 16% of populations

The main type of variation here is between populations rather than within populations. The mean heterozygosity for enzyme loci in a larger series of populations in the Oe. biennis group is 9.5% in Oe. biennis proper, 14.9% in Oe. parviflora, and 2.8% in Oe. strigosa. This can be compared with 8.0% heterozygosity in the bivalent-forming Oe. argillicola (Levy and Levin 1975; Levin, Ritter, and Ellstrand 1979). It has been stated that there is less genie heterozygosity in the Oe. biennis group than might be expected in permanent heterozygotes (Levin 1975b). The function of the Oenothera genetic system, however, is not necessarily to maintain high levels of heterozygosity, but rather to replicate a particular heterozygous gene combination.

Hétérogamie

Microspecies

and Complexes Groups

in Other

Plant

Oenothera subgenus Oenothera is not native but naturalized in Europe; but it has evolved there in the two or three centuries since the original introductions from North America. The diverse strains of Oenothera introduced into Europe, usually in ship's ballast, hybridized to produce an array of new indigenous hétérogamie microspecies. Renner (1942) recognized 18 ring-forming species of subgenus Oenothera in Europe. A few of these such as the European Oe. biennis apparently arose in North America, but most of the 18 European species arose de novo by hybridization in Europe. The most famous of these is Oe. lamarckiana (velanslgaudens). The velans complex is close to Oe. hookeri and the gaudens complex is within the range of variation of Oe. biennis. The hybridization between Oe. biennis and Oe. hookeri took place in Europe, and the hybrid product, Oe. lamarckiana, spread across Europe and later to North America. Other well-known European species are Oe. suaveolens (albicans/flavens), Oe. mollis (simulonsIplanans), and Oe. rubricaulis (tingens/rubens) (Cleland, 1972). Oenothera subgenus Oenothera is represented in South America by a

«s ro

•o § V U e o T3 U a OQ

^i Cii TCl i(N il C Cl 90 il 00 ni Ct fNrNÍMfMOCNClíN — — — — — — — ~ 00ÍN— Il II II II II - —m —m 00 I Î7I " oI OO — VO 00 00 —O O o o I I Ilr- IIr- rII - r- •t r» o — >© -Cl a

fN I m I Cl Cl CiÏCl7Ci C Ci m ( O Tt — t •

u >> < S O

< z> X

y r a. O t s
co O

> a