Genetic Variation in Naturalized Wild Radish (Raphanus Raphanistrum L) Populations in the Mediterranean Climate of South-Western Australia

Genetic variation in naturalized wild radish (Raphanus raphanistrum L) populations in the mediterranean climate of south-western Australia

MUHAMMAD ALI BHATTI MSc. Agric. (Hons.)

This thesis is presented for the degree of Doctor of Philosophy of the University of Western Australia

School of Plant Biology Faculty of Natural & Agricultural Sciences The University of Western Australia

June 2004

Nothing in biology makes sense except in the light of evolution *(Dobzhansky, 1973)

Variation in different wild radish population pods collected during common garden experiments at University of Western Australia field station, Shenton Park, 2000. *Dobzhansky T (1973) Nothing in biology makes sense except in the light of evolution. American Biological Teacher 35, 125-129.

Abstract

Wild radish (Raphanus raphanistrum L.), an outcrossing annual plant, is one of the most widespread and successful colonising weeds in the Australian wheat belt. It was introduced accidentally during the latter part of the 19th century, apparently independently through the major ports of southern Australia. The widespread success of wild radish, and therefore the likelihood of distinct populations, gives us the opportunity to study the colonizing process with adaptation in annual outbreeding species, and to explore their genetic diversity.

The aims of this thesis were to study the genetic diversity of wild radish and to investigate which factors are important in its success. After an initial review of the literature (chapter 2), the thesis describes experiments where genetic variation between and within populations was compared at 55 sites in transects across the wheat belt and high rainfall zones of temperate Western Australia (chapter 3). In chapter 4, variation in life history traits was compared with variation in AFLP molecular markers. The role of seed dormancy in the survival of the species was examined in chapter 5, and variation in the oil content of seeds and their fatty acid composition was examined in chapter 6. Finally, the results were discussed in chapter 7 with special reference to the adaptive value of outcrossing in annual weeds.

The results suggest that wild radish has evolved to fit the Australian environment. However, measurement of 14 morphological and phenological characters showed that in most cases within site variation was much greater than that between sites. Most of the variation between sites was associated with geoclusters, a name given to zones of similar environmental conditions in regard to rainfall and temperature. Thus plants from areas with high rainfall and low temperature produced longer, wider pods with more segments, heavier seeds and flowered later than plants from more arid areas.

Four populations – from the geographically close wheat belt environments of Merredin and Nukarni, the geographically distant wheat belt environment of Mullewa and the high rainfall environment of Denmark – were compared in greater detail. The data collected as part of the transect mentioned above was supplemented with characterisation of DNA using AFLP markers. It was hypothesized that populations from nearby sites (Merredin and Nukarni) would show greater similarity in both DNA markers and life history traits than the more widely separated sites, and that populations from environmentally similar sites

(Mullewa, Merredin and Nukarni) would show greater similarity in life history traits than populations from an environmentally dissimilar site (Denmark).

Using both techniques, variation within populations was again higher than variation between populations. In terms of AFLP markers, principle coordinates and cluster analyses revealed that the Nukarni population appeared to be the most distinctive, while the Mullewa and Merredin populations showed the greatest level of similarity. The Denmark and Mullewa populations were most distinct in terms of life history traits. The Denmark population had the largest pods and seeds, while the Mullewa population had the smallest. Nukarni and Merredin were intermediate. It was concluded that the Nukarni population had a markedly different origin than the others but the similarity with Merredin in life history traits indicated that convergent evolution had taken place. The results suggest that evolution in the 150 years since its introduction is partly responsible for the widespread distribution of wild radish.

The seed dormancy status of two Western Australian populations (from a high rainfall site at Bunbury and a low rainfall site at Merredin) were investigated to see whether seed dormancy was contributing to ecological success. Short and long term dormancy patterns of buried seeds and seeds on the soil surface enclosed and not enclosed by pods, were investigated using seed produced in a common environment. Contrary to expectations, the two populations exhibited similar patterns of seed dormancy. Removal of the seeds from the pods significantly reduced dormancy: 97% of seeds in pods were dormant in the first year compared with only 25% of naked seeds. Burial increased germination of naked seeds but had little effect on dormancy. There is a suggestion in the literature however, that dormancy is imposed by the seed coat, not the pods, and it cannot be ruled out that the seed coats were damaged during removal from the pods.

The fatty acid composition of lipids in the whole seeds of four populations ranged from 38.7 to 42.2%. Within site variation accounted for 80% of total variation. The major fatty acids were erucic, oleic, linolenic, linoleic, eicosenic, and palmitic acids. The ratio of saturated to unsaturated fatty acids was approximately 1:10. The results indicate that there was little difference between sites in oil content and composition of wild radish. However, since only four populations were examined, the chances of finding genetic variation would be enhanced by a wider search. Nevertheless the results suggest that that the introduced germplasm showed relatively little diversity for fatty acids composition.

The results were discussed in terms of the adaptation of outbreeders. It was concluded that there was little evidence to suggest that inbreeding confers ecological advantage for colonizing exotic species, especially for crop weeds. It is possible that the constant environmental disruption facing crop weeds may result in ecological advantage for species able to adapt rapidly to changed conditions, which is likely to be the case with outbreeding species. The greater success of inbreeders in pastures, where environmental disruption is less frequent, would seem to support this hypothesis.

Finally, the four hypotheses put forward in chapter 2 were briefly reviewed.

III

Table of Contents Abstract ------I

Table of Contents------IV

List of Tables------VII

List of Figures------IX

List of Plates------XI

Acknowledgments ------XII

Dedication: ------XIV

Declaration ------XV

Chapter 1. General Introduction______1

1.1. Introduction...... 1

1.2. References...... 3

Chapter 2. Review of the literature ______7

2.1.The southern Australian habitat ...... 7 The Mediterranean climate of southern Australia 7 Soils types of southern Australia and in the Mediterranean basin 9 The natural vegetation in southern Australia and in the Mediterranean basin 9 Farming systems in southern Australia and in the Mediterranean basin 10 2.2.Invasion of alien plants from the Mediterranean basin into southern Australia ...... 11 The species and their origins 11 The special case of subterranean clover 13 2.3.Adaptation of native and alien plants to the mediterranean environment of southern Australia ...... 14 Perennials 14 Annuals 16 2.4.Genetic variation in populations of annual plants...... 19 Life history studies 19 Random genetic variation in plant populations 22 Genetic variation in the fatty acids composition of seed oils in wild populations 25

2.5.Wild radish (Raphanus raphanistrum L.) ...... 27 Taxonomic and morphological description 27 Origin and geographical description 29 History of introduction and spread in southern Australia 29 Preferred habitats 30 2.6. Conclusions and Research Objectives ...... 33

2.7.References...... 34

Chapter 3. Genetic variation between and within naturalized wild radish (Raphanus raphanistrum) populations in south-western Australia ______48

3.1 Abstract ...... 48

3.2 Introduction...... 49

3.3 Materials and methods ...... 50

3.4 Results...... 53

3.5 Discussion...... 59

3.6 References...... 64

Chapter 4. Genetic variation in naturalized wild radish (Raphanus raphanistrum) populations in south-western Australia: a comparison of life history traits and AFLP molecular markers ______67

4.1 Abstract ...... 67

4.2 Introduction...... 68

4.3 Materials and methods ...... 70

4.4 Results...... 74

4.5 Discussion...... 79

4.6 References...... 83

Chapter 5. Seed dormancy patterns contribute to the ecological success of wild radish (Raphanus raphanistrum L.) in Western Australia ______88

5.1 Abstract ...... 88

5.2 Introduction...... 88

5.3 Materials and methods ...... 90

5.4 Results...... 92

5.5 Discussion...... 99

5.6 References...... 103

Chapter 6. Genetic diversity in the composition of lipids in the seeds of south-western Australian populations of wild radish (Raphanus raphanistrum) ______105

6.1 Abstract ...... 105

6.2 Introduction...... 106

6.3 Materials and methods ...... 107

6.4 Results...... 109

6.5 Discussion...... 112

6.6 References...... 116

Chapter 7. General discussion ______118

7.1 References...... 124

Appendix I: DNA extraction methods ------128

Appendix II: Amplified Fragement Length Polymorphism procedure ------130

List of Tables

Table 3.1 Description of the variables measured on each plant...... 51 Table 3.2 The range of values, the coefficient of variation and the percentage of variation residing between and within sites for each plant trait. Significance level of the difference between sites (shown in the right hand column) is P<0.001 (***)...... 53 Table 3.3 The number of individual plants from the 55 populations in each flower colour class ...... 55 Table 3.4 Correlations between measured plant traits, five environmental variables and latitude and longitude. Significance is indicated as follows: P<0.001(***), P<0.005(**), P<0.05(*)...... 57 Table 3.5 Mean and standard errors for annual rainfall, and mean January and July maximum and minimum temperatures for each of the five geoclusters formed from k- means clustering (see text). Latitude and longitude are ranges...... 57 Table 3.6 Component loadings for plant traits in the principle component analysis, and the amount of variation accounted for by the first and second principal components...... 58 Table 3.7. Flowering times (days after sowing) for 55 populations of wild radish grown in the field at the University of Western Australia field Station at Shenton Park, Perth, WA...... 59 Table 3.8. Geocluster means of annual average rainfall, and means and standard deviations of flowering time (days after sowing), seed weight (g/100 seeds) and pod weight (g/10 pods)...... 61 Table 4.1 Mean maximum and minimum temperatures in January and July and mean annual rainfall at Merredin, Nukarni, Mullewa and Denmark1...... 72 Table 4.2 Description of the 11 life history traits measured on each plant from the four experimental sites...... 72 Table 4.3 For each plant trait, the mean trait value over all sites, the percentage of variation residing between and within sites and the significance level of the difference between populations P<0.001(***), P<0.005(**) and P<0.05(*)...... 75 Table 4.4. Component loadings for life history traits in the canonical variate analysis, as well as the amount of variation accounted for by the first and second canonical variants (CV1 and CV2)...... 76 Table 4.5 The number of polymorphic sites, the number of loci recorded, the percentage of polymorphic loci, the average gene diversity and the apportionment of variation between and within sites of wild radish populations collected at Merredin, Nukarni, Mullewa and Denmark ...... 77 Table 5.1. Date of sampling, and means and standard deviation of germinable, dormant, and dead threshed and naked seeds buried 2 cm beneath the soil surface and lying on the soil surface...... 97 Table 6.1. Mean maximum and minimum January and July temperatures (oC) and mean annual rainfall (mm) at Merredin, Nukarni, Mullewa and Denmark...... 107

VII

Table.6.2. Mean seed (g/100 seeds) and pod (g/10 pods) weights, number of pod segments and the range of pod widths and lengths (mm) of the plants collected from Merredin, Mullewa, Nukarni and Denmark when grown in a common garden at Shenton Park near Perth...... 108 Table 6.3. Chromatographic conditions used in analysis of wild radish fatty acid methyl esters by GC with a BPX-70 (50mx 0.22 mm) capillary...... 109 Table 6.4. Total lipid content and the content of each constituent fatty acid of the Merredin, Mullewa, Nukarni and Denmark populations. The Table also shows the amount of variation accounted for between populations and the coefficient of variation (CV %) of each fatty acid across all populations. Significance of the difference between populations is P<0.001 (***), P<0.005 (**), P<0.05 (*)...... 110 Table 6.5. Component loadings for plant fatty acids in the principal component analysis, as well as the amount of variation accounted for by the first and second principle components...... 111 Table 6.6 Component loading for plant fatty acid composition in the principle component analysis of the fatty acid content of nine oil crops and four populations of wild radish, as well as the amount of variation accounted for by the first and second principle components...... 115 Table 7.1. A comparison of within species variation recorded in various studies of inbreeders and outbreeders in annual and perennial plants...... 121

VIII

List of Figures

Figure 2.1 Mean monthly maximum and minimum temperatures and mean monthly rainfall at a) Geraldton in Western Australia and b) Jindiress in northern Syria...... 8 Figure 2.2 Origin and geographical distribution of wild radish. Wild radish is an abundant species in all continents and a serious weed in most...... 29 Figure 2.3 History of introduction and spread of wild radish in southern Australia. Each point on the map represents the origin of an herbarium specimen. Wild radish is most abundant in the wheatbelt and the adjacent high rainfall zone, and is absent from the tropics...... 29 Figure 3.1. Showing the sites at which the 55 populations of wild radish were collected around Western Australia. The legend indicates the localities of the five geo-clusters described in Table 3.5...... 56 Figure 3.2 Principal component analysis showing the relationship between plant traits and environment. Points represent the mean of PC scores for all sites within each geocluster. Error bars are ± 1 s.e. Biplots indicate the direction in which selected variables increase in value and are not proportional in magnitude to the data in Table 3.6...... 58 Figure, 4.1. Showing the sites at which the four populations of wild radish were collected around Western Australia. The legend indicates the localities of the five sites described in Table 4.1...... 71 Figure 4.2 A plot of canonical variate 1 (CV1) against canonical variate 2 (CV2) for each of the plants in which life history traits were measured. Plants with similar symbols originate from the same site as indicated on the figure (Merredin, Nukarni, Mullewa and Denmark)...... 76 Figure 4.3. Dendrogram from cluster analysis of wild radish (Raphanus raphanistrum) populations analysed for genetic variation using AFLP technique...... 78 Figure 4.4 A plot of principal coordinates 1 (PC1) against principal coordinate 2 (PC2) for each of the plants in which DNA was assayed using AFLP markers. Plants with similar symbols originate from the same site as indicated on the Fig. (Mullewa, Nukarni, Denmark and Merredin)...... 79 Figure 5.1. The percentage of germinable seeds (top), dormant seeds (bottom) and dead seeds (middle) in the un-threshed wild radish population of pods from Merredin when placed on the soil surface. Sampling commenced on 1 December 1999, when 100% of all seeds were dormant, and finished on 1st June 2001...... 93 Figure 5.2. The percentage of germinable seeds (top), dormant seeds (bottom) and dead seeds (middle) in the un-threshed wild radish population of pods from Bunbury when placed on the soil surface. Sampling commenced on 1 December 1999, when 100% of all seeds were dormant, and finished on 1st June 2001...... 93 Figure 5.3 Germinability expressed as a percentage of the viable seed populations from Merredin (solid line) and Bunbury (dotted line) of un-threshed wild radish seed on the

soil surface. Vertical bars associated with the amount of dormant seed represents ± 1 s.e...... 94 Figure 5.4. The percentage of germinable seeds, dormant seeds and dead seeds at four sampling dates between 15th January 2000 and 15th June 2001. The seeds were in un- threshed pods lying on the soil surface and were from the Merredin population...... 95 Figure 5.5. The percentage of germinable seeds, dormant seeds and dead seeds at four sampling dates between 15th January 2000 and 15th June 2001. The seeds were in un- threshed pods buried 2 cm beneath the soil surface and were from the Merredin population...... 95 Figure 5.6 The percentage of germinable seeds, dormant seeds and dead seeds at four sampling dates between 15th January 2000 and 15th June 2001. The threshed seeds were lying on the soil surface and were from the Merredin population...... 96 Figure 5.7 The percentage of germinable seeds, dormant seeds and dead seeds at four sampling dates between 15th January 2000 and 15th June 2001. The threshed seeds were buried 2 cm beneath the soil surface and were from the Merredin population...... 96 Figure 5.8 Germination of threshed and un-threshed wild radish seed after 14 days at temperatures ranging from 5-35 0C. The seed used was from the Merredin population. Vertical bars represent ± 1 s.e...... 98 Figure 5.9 Rate (expressed as time to reach 50% of final germination) at which threshed and un-threshed seed of wild radish germinated at temperatures between 5 and 35 0C. The seed was from the Merredin population. Vertical bars represent ± 1 s.e...... 98 Figure 5.10. Total weekly rainfall and cumulative number of seedling emergence at the experimental site, Shenton Park, WA...... 99 Figure 6.1. Fatty acid composition of wild radish seed based on mean of values of four sites sampled in southwestern Australia. Table 6.5 explains the descriptions of the fatty acids appearing in parentheses...... 110 Figure 6.2 Principal component analysis showing the relationship between plant fatty acids and environment. Points represent the mean of PC1 and PC2 scores for all plants within each site. Error bars ± I s.e...... 112 Figure 6.3 Principal component analysis showing the relative profiles of fatty acids for different vegetable oils. Points represent the mean of principal component scores for all crops ((AOCS) 1996) (see also Table 6.6) ...... 114

List of Plates

Plate 3.1.Fifty five wild radish populations at the rosette stage were grown in a common garden experimental site at the University Field Station, Shenton Park. The picture was taken in late July 2000...... 52 Plate 3.2. The same populations of wild radish at flowering at the University Field Station Shenton Park, in late September 2000...... 52 Plate 3.3. The wild radish population from York showing within population variation in morphological and physiological traits at the University Field Station, Shenton Park in September 2000...... 54 Plate 3.4. The wild radish population from Geraldton the at University Field Station Shenton Park, in early August 2000...... 60 Plate 3.5. The wild radish population from Denmark taken at the University Field Station Shenton Park, in late October 2000. The pictures in Plates 3.3 and 3.4 highlight the differences in flowering time and rate of development between the two populations.. 60

Acknowledgments

I am grateful to my supervisor, Professor Philip Cocks, for his generosity, patience, kindness and guidance during the period of this study, particularly for his encouragement and support. My frequent discussions with him changed my life and inspired my work. I am also grateful to his wife Margaret who, through her guidance and comments, helped build my confidence. I express sincere thanks to Professor Stan Kailis, for his supervision of the oil and fatty acid work in this thesis and for his generous hospitality. I gratefully acknowledge Dr Sarita Bennett, who supervised the molecular work and assisted me with statistics. Dr Ann Matthews help with the molecular work is also gratefully acknowledged. I also thank Dr Kioumars Ghamkhar for his advice in chapter 4.

I would like to thank my colleagues in the pasture ecology group: Ken Street, Jens Berger, Lingwen Zeng, Diana Fedorenko, Hayley Norman, Patrick Smith, Angelo Loi, Fiona Maley, Phil Nichols, Kathy Davies, David Ferris, Christopher Loo, Perry Dolling, Lindsay Bell, and especially Matthew Dunbabin and Margaret Campbell. We spent many hours discussing all aspects of pasture ecology, which has contributed to my understanding of ecology and genetics.

I also thank Professor Hans Lambers (Head) and the staff of the School of Plant Biology - Paul Christiansen, Derek Heptinstall, Jeremy Foster, Sandra Pickering, Rhonda Pride, Pandy Du Preez and Sheilagh Cairns - who provided a cheerful and friendly environment in which to work. Thanks also to the technical staff who assisted me at various times - Mike Blair, Rob Casey. I also thank the staff of the Biological Sciences library (especially Belinda Tiffen and Kael Driscoll), the International Student Centre (especially Daniel Renton and Rhonda Haskell), staff of the Postgraduate Research and Scholarships office and the English Language Centre.

While at UWA I also acknowledge help from the staff of the Centre for Legumes in Mediterranean Agriculture, especially Jon Clements and Mark Sweetingham. I also wish to acknowledge staff of the CRC for Plant-based Management of Dryland Salinity including my special friend and office mate, Richard Bennett.

XII

Finally, I thank my friends Tanveer Khan, Javed Hamid, Nazim Khan, Jalal Kamali, Mohammad Islam, Mohammad Latif, Mohammad Azam, Mudassar Qureshi, Mehtabh Ali, Hassan Zaheer, Farooq Rashid, Ali Raja, Mehmood Chattha, Syed Aqif Mukhtar, Khurram Zeeshan, Asif Mohammad, Zeeshan Ellahi, Choudhry Afzal Muhammad, Jehan Zeb Khan, Muhammad Amjad Awan, Muhammad Rashid Qureshi, Obaid Fakhar, Muhammad Shoeib Hashmi and Muhammad Naveed Aslam. Thanks also to Noman Hamid, Omar Hamid, Adis Duderija and their families. Thanks also to the Pakistani community in Perth and the Pakistan Academic Club of Australia.

Finally I would like to thank my wonderful family my father, mother, brother and sisters at home for their love and unwavering support during the period of my study in Australia and all the unknown, wonderful Australian people for their hospitality and friendly behaviour, which I will never forget. I will miss Australia for the rest of my life.

XIII

Dedication:

This humble effort is dedicated to my late grandmother,

Maryam Bhatti,

whose unselfish, unconditional love and affection inspired me to wholeheartedly embrace the entire humankind AND my late younger brother

Shaukat Ali Bhatti

who, after suffering from a long and tragic sickness, departed from this world at a very young age, the memory of whom, however, shall remain with our family forever.

XIV

Declaration

I declare that all sources are acknowledged and that this thesis is my own composition and the result of my own research

Muhammad Ali Bhatti

Chapter 1. General Introduction

1.1. Introduction

A large number of plants have colonized Australia from the Mediterranean basin, many of which are considered to be weeds (Parsons 1981). Most were introduced intentionally or accidentally as contaminants in animal feeds or on agricultural products. For example, prickly pear (Opuntia vulgaris Miller) in 1788 (Parsons 1981), blackberry (Rubus fruticosus L.) in 1843, soursob in 1840 (Oxalis pes-caprae L.), and Paterson’s Curse (Echium plantagineum L.) in 1890 were intentionally introduced as garden plants, or for medicinal and industrial purposes. Examples of plants introduced on animals or contaminating agricultural products are, in 1803, wild oats (Avena fatua L.) and in 1914, skeleton weed (Chondrilla juncea L.). Wild radish (Raphanus raphanistrum L.) seeds were probably introduced as contaminants of cereals and pastures(Woolls 1867). Woolls (1867) reported that even in 1867 wild radish was a successful weed of of corn and pastures in Europe. At the present time all of the species mentioned above are widespread, and appear to be extremely well suited to southern Australia, where they cover large areas

The breeding systems of colonizing species are known to play an important role in their success. It is well recognized for example, that among colonizing species, inbreeding is by far the most common system (Barrett and Richardson 1986; Henslow 1879; Price and Jain 1981). There are many factors that may contribute to this imbalance, such as i) the ability of isolated inbreeding plants to set seed following introduction (Baker 1967;1974), ii) the lower cost to inbreeders of seed production (Solbrig 1979), and iii) the inbreeding depression which can arise if predominantly outcrossing species are forced to self fertilise due to isolation following dispersal (Schemske 1985). Finally, it is recognized that the superior ability of inbreeding to maintain adapted genotypes makes a large contribution to the predominance of inbreeding among successful colonizers (Baker 1974).

However, Cocks (1999) states that many outbreeders also make successful colonizers, and that the genetic changes by which this is achieved are less clear. The best example in Australia is annual ryegrass (Lolium rigidum Gaud.), a widespread weed in the cereal belt of southern Australia. Not only has it colonised many different habitats but it also adapts rapidly to changed circumstances. For example it has developed resistance to some 20

1 herbicides and can do so within a few years of the first application of spray (Gill 1995; Hall et al. 1994; McAllister et al. 1995). Similarly, in Canada, wild mustard has developed resistance to several herbicides, including dicamba and MCPA (Webb and Hall 1995). Other outcrossing weeds that have successfully colonised large areas of southern Australia are salvation jane (Brown and Burdon 1983) and cape weed (Arctotheca calendula) (Dunbabin 2001).

Wild radish is widespread throughout the world (Cheam 1984). It is native to the mediterranean region and spread northwards through Europe in the pre–Christian era. It has been reported as a troublesome weed in cropping land in the British Isles (Bentham and Hooker 1954; Blackman and Templeman 1938; Butcher 1961; Clapham et al. 1952). It has been introduced to most of the temperate world and is now a common weed throughout Europe (Chater 1964), the north-east, central and pacific north-west United States and Canada (Mekenian and Willemsen 1975), New Zealand (Webb et al. 1988) and Turkey (Munz 1959).

In Australia, wild radish has been a troublesome weed since its accidental introduction during the latter part of the 19th century. It was probably introduced into Australia as a contaminant of agricultural produce (Donaldson 1986). It does not seem to have been intentionally introduced by the early settlers. It was reported as naturalised around Melbourne by 1856 soon after settlement (Hooker 1860), around Sydney by 1867 (Moore 1884; Woolls 1867), near Adelaide by 1875 (Schomburgk 1875) and in Queensland by 1913 (Bailey 1913). It is now well established in Western Australia (Gardner 1930; Meadly and Pearce 1954), South Australia (Black 1948; Black. 1909), Queensland (Everist 1913), NSW (Beadle et al. 1962; Burbidge and Gray 1970; Leigh and Mulham 1977), Victoria (Ewart 1930; Ewart and Tovey 1909; Goodman 1973; Willis 1972) and Tasmania (Curtis 1956; Hyde-Wyatt and Morris 1975). It is now a major weed of field crops over a wide range of southern Australian environments.

At present there is little information in the literature on the genetic diversity of wild radish or other outbreeding weeds in Australia. There is a need to investigate which factors are important in their success and how they differ from the inbreeding weed flora. The widespread success of wild radish and therefore the availability of distinct populations give us the opportunity to study the colonizing process in this annual outbreeding species. The results can be compared with the numerous studies of colonizing inbreeding species

available in both Australia (Cocks et al. 1976; Smith et al. 1996; Woodward and Morley 1974) and abroad (Allard and his coworkers in California). This thesis will

• Examine the literature on the colonization of weeds and other species

• Compare the characteristics of wild radish population collected along a transect in Western Australia

• Compare variation in life history traits with AFLP molecular markers of wild radish populations from four sources in Western Australia

• Examine the role of seed dormancy in the survival of the species and its genetic variation

• Use the oil content of seeds and their fatty acid composition to further study genetic variation among populations.

1.2. References Bailey FM (1913) 'Wild radish (Raphanus raphanistrum): In ‘Comprehensive Catalogue of Queensland Plants.' (Government Printer: Brisbane)

Baker HG (1967) Support of Baker's law-as a rule. Evolution 21, 853-856.

Baker HG (1974) The evolution of weeds. Annual Review of Ecology and Systematics 5, 1- 24.

Barrett SCH, Richardson BJ (1986) Ecology of biological invasions an Australian perspective. In 'Genetic attributes of invading species'. (Ed. RH Groves, JJ Burdon), pages 21-33. (Australian Academy of Science: Canberra)

Beadle NCW, Evans OD, Carolin RC (1962) 'Hand- book of the Vascular plants of the Sydney district and Blue Mountains.' (Beadle, et al Armidale NSW)

Bentham G, Hooker JD (1954) 'The British Flora.' (L. Reeve & Co. Ltd.)

Black JM (1948) 'Flora of South Australia.' (Government printer: Adelaide)

Black. JM (1909) 'The Naturalized Flora of South Australia.'(Government printer: Adelaide).

Blackman GE, Templeman WG (1938) The nature of the competition between cereal crops and annual weeds. Journal of Agriculture Science 28, 247-271.

Brown AHD, Burdon JJ (1983) Multilocus diversity in an outbreeding weed, Echium plantagineum L. Australian Journal of Biological Sciences 36, 503-509.

Burbidge NT, Gray M (1970) 'Flora of the Australian Capital Territory.' (National University Press: Canberra Australia)

Butcher RW (1961) 'A New Illustrated British Flora.' (L. Hill: London)

Chater AO (1964) Cruciferae 'Raphanus L. In Flora Europea. Pages, 260-346' (Cambridge University Press: Cambridge)

Cheam AH (1984) Coat imposed dormancy controlling germination in wild radish and fiddle dock seeds. Proceedings of the 7th Australian weeds Conference. 184-190.

Clapham AR, Tutin TG, Warburg EF (1952) 'Flora of the British Isles.' (Cambridge University Press: Cambridge)

Cocks PS (1999) Reproductive strategies and genetic structure of wild and naturalized legume populations. In 'Genetic resources of mediterranean pasture and forage legumes'. (Ed. SJ Bennett, Cocks, PS), pages. 20-31. (Kluwer Academic Publishers: Dordrecht, Boston, London)

Cocks PS, Boyce KG, Kloot PM (1976) The Hordeum murinum complex in Australia. Australian Journal of Botany 24, 651-662.

Curtis WM (1956) 'The Student's Flora of Tasmania.' (Government printer Hobart: Hobart)

Donaldson TW (1986) Wild radish (Raphanus raphanistrum L.): a rewiew of reserch on biology and control in Victoria 1976-1982. Plant Protection Quarterly 1: 160-162.

Dunbabin MT (2001) Genetic variation in the outbreeding coloniser capeweed in south- western Australia. PhD thesis, University of Western Australia.

Everist Sl (1981) The history of poisonous plants of Australia.'In Plants and Man in Australia (Eds DJ Carr and SGM Carr) pages, 223-255. (Academic Press: Sydney)

Ewart AJ, Tovey JR (1909) 'The weeds, poison plants and naturalized aliens of Victoria.' (Government Printer: Melbourne)

Gardner CA (1930) 'A systematic census of plants occurring in Western Australia.' (Government Printer: Perth)

Gill GS (1995) Development of herbicide resistance in annual ryegrass populations (Lolium rigidium Guad) in the cropping belt of Western Australia. Australian Journal of Experimental Agriculture 35, 67-72.

Goodman RD (1973) 'Honey Flora of Victoria.' (Department of Agriculture, Victoria: Melbourne)

Hall LM, Tardif FJ, Powles SB (1994) Mechanisms of cross and multiple resistance in Alopecurus myosuroides and Lolium rigidum. Phytoproduction 75, 17-23.

Henslow (1879) On the self-fertilization of plants. Transactions of the Linnean Society Series II Botany 1, 317-98.

Hooker JD (1860) 'Flora of Tasmaniae.' (L. Reeve: London)

Hyde-Wyatt BH, Morris DI (1975) 'Tasmanian Weed Handbook.' (Tasmanian department of agriculture)

Leigh JH, Mulham WE (1977) Vascular plants of the riverine plain of New South Wales with notes on distribution and pastoral use. Telopea 1(4), 225-293.

McAllister FM, Holtum JAM, Powles SB (1995) Dinitroaniline herbicide resistance in rigid ryegrass (Lolium rigidum). Weed Research 43, 55-62.

Meadly GRW, Pearce GA (1954) The chemical control of wild radish. Journal of Department Agriculture Western Australia 3(3), 309-315.

Mekenian MR, Willemsen RW (1975) Germination characteristics of Raphanus raphanistrum. I. Laboratory studies. Bulletin of the Torrey Botanical Club 102, 243- 252.

Moore C (1884) 'A census of the plants of New South Wales.' (Government Printer: Sydney)

Munz (1959) 'Flora of Turkey.' (University Press Edinburgh: Edinburgh)

Parsons WJ (1981) 'The history of introduced weeds.' In Plants and Man in Australia. (Eds DJ Carr and SGM.Carr) pages 179 -193. (Academic press: Sydney, Australia).

Price SC, Jain SK (1981) Are inbreeders better colonizers. Oecologia, 49, 283-286.

Schemske DW, Lande, R (1985) The evolution of self-fertilisation and inbreeding depression in plants. II . Empirical observations. Evolution 39, 41-52.

Schomburgk (1875) 'The Flora of South Australia.' (Government Printer: Adelaide)

Smith FP, Cocks PS, Ewing MA (1996) Short-term patterns of seed softening in Trifolium subterraneum, Trifolium glomeratum, and Medicago polymorpha. Australian Journal of Agricultural Research 47, 775-785.

Solbrig OT (1979) A cost-benefit analysis of recombination in plants. In 'Topics in plant population biology'. (Eds. Solbrig OT, Jain S, Johnson GB, and Raven PH.). (The MacMillan Press: New York) 5

Webb CJ, Sykes WR, Garnock-jones PJ (1988) 'Flora of NewZealand.' (DSIR Botany Division: Christchurch)

Webb SR, Hall JC (1995) Auxinic herbicide resistant and susceptible wild mustard (Sinapsis arvensis L.) biotypes: Effect of auxinic herbicides on seedling growth and auxin binding activity. Pesticide Biochemistry and Physiology 52, 137-148.

Willis JH (1972) 'A Handbook to Plants in Victoria.' (Melbourne University Press: Melbourne)

Woodward RG, Morley FHW (1974) Variation in Australian and European collections of Trifolium glomeratum L. and the provisional distribution of the species in southern Australia. Australian Journal of Agricultural Research 25, 73-88.

Woolls W (1867) Plants introduced accidentally (1866). In 'A Contribution to the Flora of Australia' pages. 136-152 (F. White: Sydney).

Chapter 2. Review of the literature

2.1.The southern Australian habitat

The Mediterranean climate of southern Australia

When applied to climate the term ‘mediterranean’ defines a climatic environment in which rain in the winter season is followed by summer drought. Aschmann (1973) defined the mediterranean climate as one in which annual precipitation ranges from 275-900mm, 65% of which falls during the winter months; in which the mean winter temperature remains below 15 0C, and where the minimum temperature does not fall below 0 0C for more than 3% of the total hours. Mediterranean climates occur in the transition zone between temperate and dry tropical climates and cover about 2% of the land surface of the globe (Thrower and Bradbury, 1973).

Areas with mediterranean climates tend to lie on the western side of continents around latitude 25-400. Examples of countries fitting this description are: the southern tip of South Africa, southern, especially south-western Australia, central Chile and most of coastal California. The mediterranean basin itself has the largest area of mediterranean climate, including much of North Africa, southeastern Europe, parts of southwestern Europe and west Asia from the Mediterranean coast through Iraq and parts of Iran to Baluchistan. Within this area the climate varies considerably from generally mild conditions on the coast to areas with cold winters, especially in the highlands of west Asia (Di Castri 1991).

In southern Australia rainfall is concentrated between May and October (Gentili 1971) while in the Mediterranean basin it falls between September and April. Fig. 2.1 compares the rainfall and temperature at Geraldton, Western Australia (123 years of records) and Jindiress in northern Syria. Although rainfall is comparable at the two sites Jindiress is much colder in winter and somewhat warmer in summer. In agricultural terms this results in more rapid crop and pasture growth at Geraldton, and the survival of species that are susceptible to frost.

(a) Monthly rainfall & temperature recorded at Geraldton

M ax.t emp erat ure 35 M ini.t emperat ure 120 Rainfall

30 )

e 100 r m u t m m &

a 25

r 80 e ll ( mu a i 20 f

x mp in C) e

60 a 0 ( r ma f f 15

m t o o 40 n n mu a a 10 i e e n

M 20 M mi 5 0 0

n b r r n l g p t v c a ay Ju u e Ja Fe M Ap M JuMonthsA Se Oc No D

(b) Monthly rainfall and temperature recorded at (ICARDA) Jindiress Max: temperature

40 Mini: temperature 100 Rainfall C 90

0 35

& e 80 r m

u 30 mm) t u 70 a r ll ( m 25 e a

60 f axi in

mp 20 50 a e m f r 40 f

o 15 m t o

30 n a mu ean 10 i 20 e M 5 M

min 10 0 0 Ju Au S Oc N D Ja Fe Ma Ap Ma Ju ep ov ec l g t n b r r y n Months

Figure 2.1 Mean monthly maximum and minimum temperatures and mean monthly rainfall at a) Geraldton in Western Australia and b) Jindiress in northern Syria

In south-western Australia the highest rainfall occurs at the coast and falls rapidly inland. Rainfall is highest at the south-west tip and decreases in both a northerly and easterly

8 direction. In fact the distribution of rainfall isohyets are similar to the Mediterranean basin where, in both north Africa and west Asia, rainfall declines as distance from the coast increases.

In areas with mediterranean climate, plants face many challenges to survive. Frosts may occur in winter, length of the growing season varies from year to year, hot, dry summers are occasionally interrupted by heavy rains (especially in south-eastern Australia) and there are frequent seasonal droughts. Plants have evolved mechanisms to survive these challenges, and many of these will be explored in the remainder of this literature review and in later chapters.

Soils types of southern Australia and in the Mediterranean basin

Leeper (1966) states that the recurring theme of Australian soils is that they are nutrient poor and are derived from very old parent materials. Australia has only a small proportion of fertile soils, particularly in Western Australia, which has a remarkably large area of extremely poor soils. Most Australian soils are lacking in nitrogen and phosphorous and many are also deficient in minor nutrients such as copper, molybdenum and zinc. Sandy soils have poor water holding capacity, low cation exchange capacity and low organic matter content. Many are susceptible to wind and water erosion. Many southern Australian soils are acidic, inherently infertile, highly leached, heavily weathered and considered to be of great geological age (Di Castri 1991).

The soils of south-western Australian are moderately to very acidic, grey to brown, and often, though not invariably sandy. They cover extensive plains, valley floors, mid to lower hillslopes, and lower areas of the coastal plain, overlying parent material ranging from medium-textured alluvium to acid and intermediate igneous rocks (Moore 1998).

The Mediterranean basin is a large, heterogeneous region with a more complex geography than any of the other Mediterranean regions. Soils are mostly developed from limestone, although there are several areas with acidic soils in high rainfall areas and on some of the Mediterranean islands, which have granitic substrates (Di Castri 1991).

The natural vegetation in southern Australia and in the Mediterranean basin

In southern Australia there are five major plant families dominating the vegetation. These are the myrtle (Myrtaceae), pea (Fabaceae), protea (Proteaceae), southern heaths

(Epacridaceae) and grass (Poaceae) families. Two of the dominant genera are Eucalyptus and Acacia. The major vegetation types are sclerophyll woodlands, eucalypt shrubland (mallee), and heath land.

The vegetation of the Mediterranean basin has been disturbed by man’s activities for at least 10,000 years. More recently these activities include heavy grazing and cropping. In parts of Italy and France agriculture has withdrawn and there is re-invasion of weedy and woody perennials. In west Asia and north Africa it is not surprising that the native perennial vegetation has largely disappeared and been replaced by native annual grasses, legumes and herbs. Indeed, this flora is the source of much of the exotic Australian weed flora and many of its crops and pastures.

There are however, remnants of the older vegetation, particularly in mountainous areas. For example there are coniferous forest remnants in the Middle Atlas of Morocco (often Cedrus spp.) and in Syria and Lebanon, where Pinus is widespread. Unfortunately the cedars of Lebanon are now close to extinction.

Elsewhere there are large areas that were once evergreen oak and pistachio woodland. Functionally these were similar to the sclerophyll woodlands and shrublands of southern Australia. They are now dominated by herbaceous perennial and annual grasses with some legume and herb components. Under cropping there is very little vegetation, what there is being dominated by herbaceous weeds such as the Cruciferae.

Michael (1994) estimated that of the total number of species of vascular plants in Australia (between 15000 and 20000), about 2000 are aliens (Australian National Parks and Wild Life Service 1991). Of these about 35% belong to three families: Fabaceae (about 180 species), Asteraceae (about 230 species) and Poaceae (about 310 species), comprising around 15, 20 and 25%, respectively of the total number of Australian representatives in these three families. The proportion of alien plants, between 10 and 11% in Australia as a whole, varies considerably from state to state, within states and according to agricultural, ecological and environmental status and history.

Farming systems in southern Australia and in the Mediterranean basin

Australian farming systems depend on annual crops and pastures (Cocks, 2001) where cereals rotate with self-regenerating annual pastures (Puckridge and French 1983; Cocks 1997). The major and most important annual crop is wheat. The Western Australian

10 wheatbelt extends over a wide latitudinal range, from 28.5 0S to 34.5 0S. It is bounded by the 250 and 640 mm rainfall isohyets to the east and west respectively (Boyd 1976). In the Western Australian wheat-belt, pasture legumes and oilseeds are grown in rotation with wheat or barley. Pastures are based mainly on subterranean clover (Trifolium subterraneum L.), several species of annual medics (Medicago spp.) and yellow serradella (Ornithopus compressus), with a range of weedy grasses and herbs. The main crop legume is the narrow-leaved lupin (Lupinus angustifolius L.), which grows on some 1 million ha; field peas (Pisum sativum L.), faba beans (Vicia faba L.), chickpeas (Cicer arietinum) and lentils are also grown in restricted areas and the most important oilseed crop is canola (Brassica napus). Increasingly pasture-based farming systems are being replaced by continuous cropping or with rotations where the self-regenerating pastures are replaced with sown pastures in a phase farming system (Reeves and Ewing 1993).

In southern Australia, ley farming was developed using introduced annual legumes, mainly subterranean clover and a few medic species (Cocks, 1999). Ley farming is considered to be largely responsible for the 50% increase in wheat yields in Australia between 1950 and 1980 (Donald 1981).

Agriculture in Australia is a relatively recent development, especially in south-western Australia, where most farms were established from pre-existing natural vegetation in the early to mid 20th century. This replacement of the native perennial vegetation with exotic annuals has brought about significant water balance problems, which are behind the current salinity crisis (Hatton and Nulsen 1999).

2.2.Invasion of alien plants from the Mediterranean basin into southern Australia

The species and their origins

Watson (1847), defined aliens as those plants introduced from other countries, either accidentally or intentionally, which have become naturalized in the host country. Thellung (1912) stated that alien or exotic plants may be more or less naturalized plants present in a country or region which had not previously been mentioned in historic records, were introduced intentionally or accidentally (for example as contaminants in animal feeds or on agricultural products) and which increase and multiply without the direct help of man.

Alien plants in southern Australia originate from various broad region of the world, including northern Europe, North, South and tropical America, tropical and southern

Africa, east Asia and the Mediterranean regions of north Africa, west Asia and southern Europe (Darlington 1963). For example, Xanthium spinosum, a native of Chile and X. occidentale from North America were introduced to Australia with cotton seeds. X. orientale and X. cavanillesii from South America and Mexico respectively, were introduced to Australia as contaminants of animals or in grains in the first half of the nineteenth century (Widder 1923). Echinochloa colona, E. crus-galli, E. oryzoides and E. microstachya from North America and the Indo-Burma region were introduced to Australia in rice seed from California. Hordeum leporinum and H. glaucum, from the mediterranean basin (Cocks et al. 1976) probably contaminated hay and ballast in early shipments from Europe. Arctotheca calendula is a native of the Cape Province of South Africa (Adamson and Salter 1950), and it too, was probably a contaminant of hay and ballast.

In southern Australia, large numbers of aliens have colonized most of the land now used for agriculture. For example, Arctotheca calendula (Arnold et al. 1985; Wood 1994), at least four species of Hordeum, (Cocks et al. 1976), several Bromus species (Kon and Blacklow 1988), Trifolium subterraneum (Gladstones 1966; Cocks and Phillips 1979), other species of Trifolium (Smith et al. 1996; Gibberd and Cocks 1997; Norman et al. 1998; Norman et al. 2000) and Medicago (Cocks et al. 1979) have now colonized between 20 and 40 million ha of land in southern Australia. Species that are common in wasteland or have invaded natural vegetation include, in NSW, Cinnamomum camphora, Ailanthus altissima, Alternanthera philoxeroides, Ligustrum spp, and Hebenstretia dentata in Western Australia, Homeria spp, and Freesia refracta, and in South Australia Oxalis pes-caprae. The alien plants noted by Brown in 1802-4 near Sydney, include Urtica urens, Silene anglica, Anagallis arvensis, Euphorbia peplus, Poa annua and Gnaphalium pensylvanicum (Britten, 1906; Maiden, 1916), illustrating that plant invasion occurred very soon after European settlement.

Many of the aliens plants were introduced intentionally for grazing, as garden plants, for medicinal or industrial purposes or as food crops (Parsons 1981). Although many of these species have become weeds, others form the basis of Australian agriculture. Indeed the cost of weedy exotics, estimated to be as high as $3.3 billion/year (ARMCANZ et al. 1997), should be offset against the benefit to horticulture and agriculture of non weedy exotics, an exercise that is not always undertaken when formulating quarantine policy (eg Pheloung 1995). The particular case of subterranean clover is considered in the next section.

The special case of subterranean clover

The weedy aliens are pioneer species because of their ability to colonise disturbed or denuded land (Michael 1994). Parsons (1981) described how the native vegetation of Australia evolved under low grazing pressure from a relatively small population of herbivores such as kangaroos, wallabies and wombats. The exotic animals introduced by the early settlers, cattle and sheep, damaged the native vegetation, which in any event was unable to supply sufficient high quality feed to support viable animal industries. The alien plants were adapted to these conditions and were able to support the growing livestock industries.

Among the many species involved in the new industries were mediterranean grasses, annuals and herbs. The most important of these were the annual legumes, chief of which was subterranean clover (Trifolium subterraneum), which is the most important introduced legume in Australia, and which formed the basis of the livestock and cropping industries for the first 75 years of the twentieth century. Not only did subterranean clover provide a high protein diet for animals, it also controlled root diseases in rotation with cereals and, through its ability to fix nitrogen, improved soil fertility and increased yield of subsequent cereal crops (Puckridge and French 1983).

Subterranean clover was introduced accidentally during early European settlement (Gladstones and Collins 1983). Gladstones (1966) interprets historical and distributional evidence to show that there were probably two periods of importation of subterranean clover into Western Australia: from 1829 to 1842, and in the 1860s. It is widely assumed that subterranean clover in Australia originated from the Mediterranean basin including Greece, Portugal, Madeira and the Canary Islands as contaminants in animal feeds or on agricultural products.

Subterranean clover was introduced to agriculture by a grazier in the Mount Lofty Ranges in South Australia in 1889. However it was not until 1907 that a satisfactory method of harvesting its seed by means of a roller covered with sheepskin was developed. The high phosphorus requirements of subterranean clover were not realised until 1927, and it was only in the 1930s that significant areas of this pasture plant were established (Donald 1970). The clover collected by Howard, later named Mount Barker subterranean clover, is still one of the most widely distributed clovers (Cocks and Phillips 1979). It spread rapidly

13 on the neutral and acidic soils of the Adelaide Hills and South East of South Australia and soon after into Victoria, New South Wales and Western Australia, either as a result of deliberate sowing of seed or inadvertently, following the spread of hay and the introduction of livestock (Cocks 1993).

Subterranean clover was accompanied by other annual species of mediterranean origin. Many have become weeds, both of pastures and crops. Almost certainly wild radish, the subject of this thesis, was one of these species.

2.3.Adaptation of native and alien plants to the mediterranean environment of southern Australia

Perennials

Most of the native vegetation in southern Australia is perennial. Perennial plants have evolved mechanisms to survive the many challenges posed by the diverse mediterranean climate of the region, with its hot and dry summers, occasional heavy summer rains, frequent seasonal droughts, inland frosts and variable length of growing season. The first, and most important adaptation of the perennial vegetation is that it used all or nearly of the incident rainfall. Estimates of the amount of water penetrating to ground water systems under natural vegetation rarely exceed 5 mm (Dunin 2002), while, under introduced exotic annuals up to 100 mm or even more of the incident rainfall may escape (Smettem 1998).

Not only does the native vegetation use more water but it is also adapted to withstand water stress. For example, native plants have tough, drought resistant leaves, which are thick and have a coating or cuticle that protects them from dehydration (Dallman 1998). Drought deciduous plants drop most or all of their leaves during summer drought. Many native shrubs and trees can ‘hydraulically lift’ water from deep in the soil profile and release it nearer the surface where the plant obtains its nutrients (Pate et al. 1998). Seedlings develop their tap roots rapidly and older plants grow a mat of finer roots close to the soil surface (Dallman 1998). Nutrient poor soils are especially widespread in southern Australia, where plant adaptations to these conditions are prominent. In the Proteacea, cluster roots form, which enhance nutrient uptake in impoverished soils (Lambers and Poot 2003; Shane et al. 2003). Dry heat at the end of the summer and in the early autumn creates conditions for wildfires, which were frequent before European settlement and continue to be frequent. In common with mediterranean plants in general, Australian natives are typically adapted to

14 fire. Many native species have lignotubers in which the top of the plant can be completely destroyed, usually by fire, and sprout again from the buried lignotuber. Most eucalypts and other Myrtacea fall into this category but many other Australian natives, including many Proteacea, respond to fire in this way (Bradstock et al. 2002; Wooller and Wooler 2001; Enright and Goldblum 1999; Graham et al. 1998).

Lignotubers are produced by plants occurring naturally in other mediterranean environments. For example, Erica arborea (Canadell and Lopez-Soria 1998), and Erica australis (Cruz et al. 2003) both natives of the Mediterranean basin, two Erica species from South Africa (Verdaguer and Ojeda 2002) and several Mediterranean Quercus species including Q. pyrenaica and Q. coccifera (Kummerow et al. 1990; Calvo et al. 2003) respond to fire in this way. Resprouting from lignotubers is a widespread trait in plants of the mexical shrublands of Mexico and California (Lloret et al. 1999).

For those species which are obligate seeders and do not have the capacity to reproduce from lignotubers, fire often plays a role in stimulating germination. In a study of regeneration after fire of Eucalyptus baxteri woodland, more than 12,000 seedlings germinated in response to smoke and more than 7,000 in response to heat (Enright and Kintrup 2001). Many of the Proteacea retain seeds in the fruits until they are released as a result of fire (Wooller and Wooler 2001). Fire almost certainly plays a role in the germination of hardseeded legumes, both annual and perennial, originating from Australia and from the Mediterranean basin (eg Hansen et al. 1992).

Indeed many of the other mechanisms by which Australian natives cope with mediterranean conditions are common to perennials native to other mediterranean regions (Dallman 1998). For example, lucerne explores a far greater depth of soil than do mediterranean annuals, rivalling species such as the native Banksia prionotes (Pate et al. 1998; Latta et al. 2001; Latta et al. 2002).

Bulbs and other geophytes retain their food and water stores underground during the drought season. They are most common in South Africa and the Mediterranean basin and progressively less common in California and Australia. Nevertheless many species are geophytes in all mediterranean climates.

Most, but not all geophytes grow in winter and spring and are dormant in summer. Some, for example Crocus and many species of Cyclamen flower in early autumn using nutrient

15 and water reserves stored from the previous winter. Most however, for example Narcissus, Iris, Gladiolus, flower following a short period in winter/spring (Dallman 1998).

Annuals

Most of the survival strategies of annual species adapted to mediterranean climates are also common to the various regions, although the majority of annuals have as their origin the Mediterranean basin itself. Unlike the perennials, which have evolved to tolerate drought, the annuals have evolved to avoid drought. These strategies centre around reproductive strategies and include seed dormancy, rapid life cycle and seed dispersal. Of these, rapid life cycle is the most responsive to environmental change and is the strategy in which most species show ecotypic differentiation (Ehrman and Cocks 1996). For example, Trifolium glomeratum (a native of the mediterranean basin) has evolved later flowering time in response to exposure to long growing season (Woodward and Morley 1974; Smith et al. 1995). Capeweed (Arctotheca calendula), a native of southern Africa, responds to longer growing season in a similar way (Dunbabin 2001), as do two Hordeum species (Cocks et al. 1976), various Bromus species of Mediterranean origin (Kon and Blacklow 1988) and many clovers, including subterranean clover (Ehrman and Cocks 1996; Norman 2002).

Mediterranean annual plants survive and maintain their seed bank from year to year through various seed dormancy mechanisms. Seed dormancy assists the survival of plants in several ways, categorised by Baskin and Baskin (1998) as i) ensuring the persistence of species in risky environments, ii) enabling germination to coincide with the season most favourable for plant growth, iii) preventing seedlings from competing with their siblings or the mother plant, iv) allowing plants to survive periods of the year unfavourable for growth, and v) interacting with other life cycle traits in helping the plant to maintain itself.

The seed dormancy mechanisms of mediterranean annuals varies widely. The legumes rely on hardseededness, which is the development of a seed coat that is impermeable to water. This is normally a mechanism for long term seed dormancy in which seeds survive from season to season and from year to year. For example, Medicago noeana survived in the seed bank at Aleppo, north Syria for more than five years, and several genotypes of Medicago rigidula survived almost as long (Cocks 1988). In Australia, Taylor and Ewing (1996) demonstrated that even the relatively soft-seeded subterranean clover will live for many years if the seed is buried. In mediterranean legumes hardseededness is almost

16 universal, one of only a few exceptions being some genotypes of French serradella (Ornithopus sativus) (Snowball 1996). Other species, including Trifolium, Medicago, Biserrula, other species of Ornithopus, and Hymenocarpus exhibit hardseededness to varying degrees (Ehrman and Cocks 1996).

Legumes lose hardseededness in two stages (Taylor 1984; Taylor and Revell 2002). Firstly, there is a requirement for the seeds to experience a long period of high temperature. This is the reason why seed burial, which insulates the seeds from temperature extremes, prolongs the life of dormant legume seeds. Secondly, the seeds require a relatively short period of alternating temperatures, which dictates the time of the year in which the seeds germinate. Thus species with low temperature requirements for the second stage will not germinate until late autumn (eg Medicago polymorpha) (Taylor 1996), while seeds softening at higher temperatures will germinate much earlier (eg subterranean clover) (Taylor 1984).

Evidence is emerging that some legumes adapted to cold conditions lose their hardseededness in winter. For example, it appears that populations of Medicago lupulina in Canada require low temperature as part of the first stage of softening, followed by fluctuating temperatures for the second stage (MH Entz pers. comm). It is even possible that the low temperature requirement can replace the high temperature requirement in species normally reacting to high temperatures since M. rigidula, a species native to west Asia and north Africa, is able to survive in the prairies of the United States.

The two stage softening process in legumes has led to the concept of half life – the time taken for half of the dormant seeds to soften in any year (Smith et al. 1996). A low half life means that the seeds germinate in early autumn, while a high half life means that softening is deferred until late autumn. Species that typically have short half lives include subterranean clover and Trifolium angustifolium, species with very long half lives include T. spumosum, T. tomentosum, T. lappaceum and T. purpureum, while T. cherleri, T. glomeratum, T. hirtum, T. clypeatum, T. argutum and T. clusii are intermediate (Norman et al. 1998; Norman et al. 2000).

The common grasses and herbs do not use hardseededness to impose seed dormancy. In general, the grasses have only short-term dormancy, which prevents germination as the result of summer rains and to varying degrees defers germination until autumn. Echium plantagineum (variously known as Paterson’s Curse and Salvation Jane depending on your

17 view of its value) germinates even earlier in the growing season than does subterranean clover (Piggin 1976) and appears to have much shorter lived seed than subterranean clover. Capeweed, on the other hand, has much longer-term seed dormancy and defers much of its germination until seasonal rains are assured (Dunbabin and Cocks 1999). Most of the grasses behave like Echium plantagineum and vary only to the extent to which they defer germination until autumn. For example in Tasmania, 95% of Bromus diandrus seed germinated within 27 days of shedding (Harradine 1986). In Western Australia, Gill and Blacklow (1985) observed that there was ecotypic differentiation for seed dormancy in B. diandrus, with the duration of dormancy positively correlated with the duration of the rain- free summers at the sites where the seed was collected. Short though the duration of dormancy in B. diandrus may be, Chapman et al. (1999) suggest that it is even shorter for Hordeum leporinum. The dormancy of barley grass seeds varies with species, with that found in the driest environments (Hordeum glaucum) having much longer dormancy than the species from the wettest and mildest environment (H. murinum) (Popay 1981). There is some evidence that the lemma and palea assist in the imposition of dormancy in the barley grasses, especially H. spontaneum (Gudkova 1976). In common with most mediterranean annuals, barley grass germinates fastest at about 200 C, and will not germinate above 300 C (Cocks and Donald 1973; Popay 1975)

Cocks (1999) defined seed dispersal as the mechanism used by the mother plant to distribute seeds. Some species bury their seeds (e.g. subterranean clover, V. sativa subsp. amphicarpa, Lathyru ciliolatus), in which case dispersal is limited to the base of the mother plant (Christiansen et al. 1996). Others produce inflorescences that are light enough to be dispersed by wind (e.g. T. tomentosum). Yet others are dispersed by animals (e.g. most species of Medicago are dispersed in wool and fur). Several plant species exhibit amphicarpy (flower above and below the ground) whereby the below ground seeds are protected from predation, fire and desiccation (Cheplick and Quinn 1987), while there is the opportunity for the above ground seeds to be more widely dispersed (Cheplick 1996). Some authors believe that amphicarpy is the result of selection in very patchy environments (Wilson 1983). Seed dispersal through wind (Schismus arabicus) and water (Gymnarrhena micrantha) (Gutterman 1994) are also common. In these cases the seed itself is made more buoyant by very small size (Gutterman 1994) or the possession of wings (Baker and

O’Dowd 1982; Wilson 1983). Some seeds are distributed ballistically, after the seed pod shatters (Wilson 1983).

Small seed size is also important for the survival of seeds after ingestion by ruminants. For example, Thompson et al. (1990) found that 60% of legume seeds smaller than 1 mg survived ingestion by Awassi sheep, while seeds larger than 2 mg were almost entirely digested. As a result most small-seeded clovers (eg Trifolium tomentosum, T. glomeratum) are dispersed in the faeces of sheep while subterranean clover and most annual medics are not. In Syria, Ghassali et al. (1997) used this concept to deliberately disperse legume seeds to degraded rangeland with considerable success.

Survival strategies of annuals and perennials differ significantly, with drought tolerance important in perennials and drought avoidance in annuals. The strategies of annual plants centre around their reproduction and include time to complete the life cycle, seed dormancy and seed dispersal. All of these are likely to show genetic variation both within and between populations and between species. The next section examines genetic variation in mediterranean annuals with emphasis on their reproductive strategies.

2.4.Genetic variation in populations of annual plants

The discussion in the previous section suggests that genetic variation in annual plants will be a key to their colonisation of widespread areas. The mean flowering time of many populations varies according to environmental conditions with later flowering ecotypes more commonly found in areas with longer growing seasons. There have been fewer studies of variation within populations, mostly relating to life history traits. In this section these studies are discussed, and compared where possible with studies using enzymes or molecular markers.

The comparison is important because variation in life history traits reflects in most cases the outcome of selection processes, while that of enzymes and molecular markers are random or, at the most represent genetic drift.

Life history studies

Plant biologists and ecologists have measured the life history traits of plants in common gardens for some time. The common garden eliminates, or nearly so, the effect of environment and allows for the expression of genetic differences. The variables obtained

19 from living plants or seeds are collected from a variety of natural habitats can be compared in this way.

It is widely recognised that the majority of life history traits expressed by plants are adaptive and thus acted upon by natural selection (Marshall and Allard 1970; Lande 1977; Barrett 1982; Barrett 1988). Though the adaptive significance of a polymorphism may be unclear in some cases, a more thorough investigation often reveals its significance. For example, New (1958), and New (1959) studied Spergula arvensis in the United Kingdom and demonstrated variation in seed coat morphology between populations, the significance of which was initially unclear. However New and Herriot (1981) demonstrated that a particular seed coat morphology has direct adaptive value in allowing seeds to germinate under moisture regimes typical of the areas in which they exist. Thus the patterns of variation found in life history traits are likely to be the result of adaptation to prevailing environmental conditions.

One of the first life history studies of genetic variation in inbreeding colonisers was that of Imam and Allard (1965). Studying Avena fatua in California, they identified distinct differentiation between widely spaced populations, as well as differentiation between more closely situated sites. They concluded that the variation adapts each population to its particular environment. Similar differentiation is evident in Australian populations of Bromus diandrus. Kon and Blacklow (1988), investigated eight populations from across Australia and found differentiation in a number of characters including flowering time and panicle number, and were able to relate these differences to the environment from which the plants were collected. Other inbreeding colonisers exhibiting marked ecotypic differentiation include Poa annua (Lush 1989) and Capsella bursa-pastoris (Neuffer and Meyer-Walf 1996). Another interesting example of adaptation through ecotypic differentiation is the variation in frost tolerance of some annual medics collected in north Syria (Cocks and Ehrman 1987).

Observations such as these have led Allard (1975) to conclude that self pollination leads to specific advantages in colonising species. He argued that self-pollination fixes certain adapted genotypes in the various micro-environments that are present in any habitat. He saw that these fixed genotypes, which he picturesquely termed ‘supergenes’ confer high fitness and hence increased adaptation to local environments. Outbreeding populations are unlikely to be able to do this and are therefore less fitted to being colonising species

(Barrett 1982), supporting the previously held view that the predominance of self pollinators among successful colonisers is due to the need to insure against failure to pollinate where plant density is low (Stebbins 1950) and the existence of barriers in local populations (Kannenberg and Allard 1967).

Although studies of outbreeding colonisers are less common, ecotypic differentiation has been demonstrated in some species. On a broad scale, Wood and Degabriele (1985) studied patterns of variation in life history traits of the outbreeding coloniser Echium plantagineum. Using material from 8 sites across south-eastern Australia, they found significant between- site variation for 15 of the 27 characters measured. Variation in some of these characters was significantly correlated with environmental variables at the site of collection, indicating that ecotypic differentiation had occurred in this species. Similarly, widespread herbicide resistance in the outbreeder Lolium rigidum in Australia provides evidence of ecotype differentiation in response to a continued application of herbicide (Gill 1995). The latter case is of special interest because it demonstrates that ecotypic differentiation occurs within a few years of the selection pressure being applied. Most other examples of evolutionary change appear to take longer, although in classical evolutionary terms most modern examples of ecotypic differentiation represent a rapid response to evolutionary pressures.

With the possible exception of herbicide resistance in Lolium rigidum, the above studies on ecotypic differentiation were conducted on widely separated populations. However, ecotypic differentiation also occurs on closely adjacent populations. For example, in South Australia, Cocks (1992) found a significant relationship between flowering time and elevation along a number of transects in a pasture containing subterranean clover (Trifolium subterraneum). He attributed this to the changing water relationships from the top to the bottom of the slope. The changes in flowering time often occurred over a few metres, a result confirming earlier research by Aston and Bradshaw (1966), McNeilly (1968) and Snaydon and Davies (1976). These classical papers reported results on the European outbreeding perennial grasses Agrostis stolonifera and Anthoxanthum odoratum where tolerance to heavy metals and high pH changed over boundaries of only a few metres. These cases are of special interest because they indicate that the environmental variation imposing the selection pressures are stronger than the flow of pollen and seed over these relatively short distances. Hamrick and Allard (1972), working with the annual Avena

21 barbata in southern California also noted rapid changes in gene frequencies over short distances.

The special advantages that inbreeding gives to colonisers is open to some question on pragmatic grounds. For example, in southern Australia some of the most successful colonisers are outbreeders. Lolium rigidum has already been mentioned: it is probably the most widespread and successful annual grass. Capeweed (Artotheca calendula) is also widespread (Arnold et al. 1985) and may constitute as much as 50% of the biomass of pastures in the Western Australian wheat belt. Capeweed is an obligate outbreeder, or almost so (Van de Loo 1987; Dunbabin 2001). Wild radish, a widespread and successful weed and the subject of this thesis is also an outbreeding species.

Random genetic variation in plant populations

Isozymes or variable forms of the same enzymes were first recognised in 1957 (Stebbins 1989). They provide a rapid and simple method of assessing genetic variation at multiple loci within living organisms. Since their discovery, isozymes have been used to investigate genetic variation in hundreds of plant and animal species. They provide an opportunity to investigate the arrangement of genetic variation in species with different breeding systems.

There are a number of isozyme studies of inbreeding species. For example, Bosbach and Hurka (1981) studied the inbreeding weed Capsella bursapastoris and found that although a large number of polymorphic isozyme loci exist across the species as a whole, individual populations are often monomorphic for certain loci, demonstrating a high degree of differentiation between populations. Likewise, populations of the inbreeding weed Emex spinosa in Australia tend to be genetically uniform, such that all variation exists between populations (Marshall and Weiss 1982). Finally Warwick (1990) summarised studies of five colonising species in Canada, all of which exhibit substantial inter-population differentiation and low within-population variation. These results tend to support Allard’s (1975) conclusions discussed earlier, where adapted genotypes are likely to be fixed in successful populations.

On the other hand, little differentiation for isozyme loci is observed between populations of the outbreeding colonisers Trifolium pratense (Hagen and Hamrick 1998), Secale montanum (Sun and Corke 1992), Calamagrostis canadensis (MacDonald and Lieffers 1991), Centaurea solstitialis (Sun 1997), Apera spica-venti (Warwick 1990) and

Alopecurus myosuroides (Chauvel and Gasquez 1994). In these species there were large amounts of variation within individual populations,

Another way of assessing genetic diversity is the use of molecular markers, which offer a number of advantages over morphological and physiological life history measurements. Firstly their expression, particularly DNA markers, is relatively unaffected by the environment in which the plant grows, unlike the majority of life history traits which are affected by environment (Marshall and Brown 1975). Thus any variation measured can be confidently assumed to be genetic, and not the product of environmental heterogeneity. Secondly, procedures for assessing molecular diversity are relatively inexpensive and rapid ( Brown and Weir 1983; Westman and Kresovich 1997), and do not require large numbers of plants. Indeed, tissue samples for molecular analysis can often be harvested directly from natural populations. Thirdly, molecular markers lend themselves to a large range of statistical techniques which can be used to assess the amount and arrangement of genetic diversity (Brown and Weir 1983; Hamrick and Godt 1990; Weir 1990; Hoelzel and Dover 1991; Kremer et al. 1998). This allows the comparison of results of one study with those from another (Westman and Kresovich 1997). Of particular interest in colonising species are non-random associations between loci. The build up of complexes of well-adapted genes is part of the adaptation process of colonising species, particularly inbreeders (Allard 1975; Allard et al. 1993). Molecular markers lend themselves to the investigation of this type of adaptation. Fourthly, some categories of molecular markers are inherited in a co- dominant fashion, allowing levels of heterozygosity to generate genetic diversity information, which is not easily obtainable in other ways. Breeding system can be assessed in this way (Layton and Ganders 1984; Burdon and Brown 1986). Finally, analysis of molecular markers has the potential to assess a large and random portion of the genome, whereas life history traits reflect only that part of the genome, which is expressed phenotypically (Barrett 1982).

While the assessment of a large and random fraction of the genome may be desirable in some circumstances (Westman and Kresovich 1997; Petersen and Seberg 1998), the random nature of the markers under examination is a distinct disadvantage in assessing the adaptation of a species to its environment. Unlike life history traits, the majority of DNA polymorphisms (Hartl 1988; Nissen et al. 1995) and probably isozymes (Powell 1983; Spitze 1993; Isabel et al. 1995) are selectively neutral. Thus, the patterns of genetic

23 diversity they reveal may not be the result of adaptation of the species to its environment, but may instead reflect random processes associated with gene flow and genetic drift (Hurka 1990). That there are often differences in the patterns of genetic variation exhibited by molecular markers and life history traits (Moran et al 1981; Warwick and Black 1986; Schut et al. 1997) reflects the difference in forces controlling these traits (Knapp and Rice 1998; Venable et al 1998).

A number of studies of genetic diversity in colonizing species have been conducted using molecular techniques. Restriction fragment length polymorphism (RFLP) (Nissen et al. 1992; Richard et al. 1995; Rowe et al 1997), random amplified polymorphic DNA (RAPD) (Miller et al. 1996; Moodie et al.1997; Okoli et al. 1997; Rowe et al. 1997; Meikle et al. 1999), amplified fragment length polymorphism (AFLP) (Andrews et al. 1998; Dunbabin et al. 2001), and microsatellites or single series repeats (SSR) (Meikle et al. 1999) are DNA techniques which have been used to document genetic diversity in plant species, including colonisers.

For example, Andrews et al. (1998) studied herbicide resistance in the inbreeding weed Avena fatua, and detected distinct population differentiation within single fields using AFLP markers. Similarly, Rowe et al. (1997), found nearly 50% of all RFLP variation resides between populations of the inbreeding coloniser Euphorbia esula in North America. Pakniyat et al. (1997) studying the inbreeder Hordeum spontaneum, found sharp genetic differences between closely spaced populations.

Though similar DNA marker studies of outbreeding colonisers are rare, studies of non- colonising species show a lack of population differentiation. Martin et al. (1997) studied genetic diversity in three closely spaced populations of the outbreeder Erodium paularense using RAPD markers and found only 15-20% of all variation residing between sites. Wolff et al. (1994) studying two species of Plantago using minisatellite DNA, found high variation within and low variation between populations in the outcrossing P. lanceolata, while the converse trend was evident in the selfing P. major.

Both isozyme and molecular techniques have therefore confirmed that inbreeders are more likely to show variation between populations than within populations, while the reverse is true of outbreeding populations.

Genetic variation in the fatty acids composition of seed oils in wild populations

In addition to life history traits and molecular markers, genetic variation is likely to exist in the chemical composition of the vegetative and reproductive parts of plants. For example, subterranean clover contains isoflavones in its leaves, which vary both in the total amount of isoflavones and their constituents (Francis and Millington 1965; Cocks and Phillips 1979; Cocks 1992). Similarly, some Medicago, Trigonella and Melilotus species produce coumarins in older leaves (Sudaric et al. 1976; Ingham 1981; Dreyer et al. 1987). The adaptive significance of both groups of compounds, if any, is unclear, although they may have a role in disease and insect resistance. In many other groups of plants, including the Brassicacaea, seed oils are widely present and show genetic variation.

Variation in the seed oil composition of lipids has been measured in natural populations of Thymus vulgaris (Granger and Passet, 1973), Rosmarinus officinalis (Dellacasa et al. 1999), Salvia fruticosa (Skoula et al. 1999), and Ocimum gratissimum (Vieira et al. 2001), (all Labiatae). All populations displayed significant intraspecific variation in fatty acid composition. Over the last 10 years, the crucifer Arabidopsis thaliana has been developed as a model system for the genetic analysis of lipid biosynthesis (Browse and Somervillie, 1994) opening the door for studies on genetic variation in the Brassicaceae. Millar and Kunst (1999) concluded that the composition of seed oils of different ecotypes of A. thaliana did not differ despite the fact that they originate from diverse geographical locations, although the relative amounts changed. The observed differences are probably due to altered levels or activities of the relevant fatty acid biosynthetic enzymes. However, because no new fatty acids were detected, their collection of ecotypes did not contain fatty acid biosynthetic enzymes with novel substrate specificities.

In a study of south Brazilian populations of Cunila galioides (Labiatae), Echeverrigaray et al. (2003) found that oil composition of Cunila galioides is under genetic rather than environmental control, a fact that has been determined in several other plant species (Hay and Waterman 1993). In the case of C. galioides the oil composition did vary between ecotypes: for example populations from the northeast plateaue of Rio Grande do Sul contained citral whereas populations from the high altitude grasslands of of the Atlantic Ranges contained ocimene. Oils of the menthene group were found in transition populations . The difference in chemical composition appeared to be genetically determined. Echeverrigaray et al. (2003) suggested that genetic variation of the essential 25 oils in C. galioides indicate considerable potential for the domestication and genetic improvement of this aromatic plant. He considered that the geographical distribution of the oil chemotypes can be explained by localized inbreeding, associated with low levels of gene flow between populations and a low incidence of recessive gene expression. It is also likely that variation in the essential oil composition may help to protect plants against pathogens or extreme climatic conditions suggesting that natural selection is operating.

Turning to the Brassicaceae, results from Egypt indicate that there are significant differences between species: landraces of Brassica nigra have the highest oil content and those of Brassica napus the lowest oil content (Tahoun, et al 1999). Wang et al. (1998) assayed 25 strains and five cultivars of Brassica rapa, B. nigra, B. napus, B. juncea, B. carinata and Sinapis alba and found genetic variation in the sinapine content of seed oils, especially of Sinapis alba. Sinapine content per seed was positively correlated with seed size.

In a study of Ethiopian or Abyssinian mustard (Brassica carinata) Alemayehu et al. (2001) found wide variation in fatty acids, primarily erucic acid, which varied from 6 to 51% of the total, and oleic acid, which varied from 5 to 34%. Linoleic and linolenic acids were less variable. The high oleic acid genotypes exhibited not only low erucic but also higher linoleic (25%) and considerably lower linolenic acid (8%) contents. It was possible to classify the F2 populations with the lowest erucic acid into three distinct classes: the first class had an erucic acid content of 6-12%, while the second and third classes had 18-32% and 36-42%, respectively. The existence of a multiple allelic series will make possible the breeding of lines with zero levels of erucic acid without the need for interspecific crossing.

Rebetzke (1998) investigated the genetic basis for reduced palmitic and stearic acid in the seed oils of soybean (Glycine max). Hybridisation of low palmitic lines with locally adapted lines revealed frequencies of reduced and normal palmitic acid among F2 progeny consistent with segregation at a single major locus. Repeatability of this variation was examined in 87 reduced and normal palmitic F5:7 lines randomly sampled from each cross. Reduced palmitic acid lines ranged between 54 and 72 g/kg, and normal palmitic acid lines between 90 and 119 g/kg for both crosses. The large genotypic ranges observed for both F2 and F5:7 generations may be explained by an undetermined number of genetic modifiers associated with the major palmitic acid locus. High narrow-sense heritabilities (>80%) for

26 palmitic and stearic acid contents suggest that total saturates may be reduced by selection in a few environments for major and modifier genes controlling reduced palmitic acid content.

Aggarwal et al. (2003) mapped the loci influencing the content of six fatty acids in mustard. Transgressive variation was evident for all six fatty acids irrespective of the levels of differences between the parents. The frequency distribution was normal for linolenic, linoleic and stearic acids, while deviation from normality was observed for three other fatty acids. The content of erucic acid was negatively correlated with the contents of all other fatty acids, which were themselves positively correlated. Based on single marker analysis and interval mapping, two loci each for linoleic, linolenic and erucic acids were mapped to marker intervals on three linkage groups. Position of log of odds ratio peaks suggested the presence of common, linked and independently segregating loci. The percentage of phenotypic variance explained by individual quantitative trait loci ranged from 11 to 20%, whereas the cumulative action of loci detected for different traits accounted for 16 to 28% of the variance. The additive effect for an individual locus ranged from 1.1 to 4.3. Presence of the favourable alleles at both the contributing loci in most of the recombinant inbred lines with a high linolenic acid content and of the unfavourable alleles in the lines with a low linolenic acid content indicated the possibility of pyramiding useful genes from phenotypically similar parental lines.

These genetic studies indicate that variation of seed oils in natural populations of many species in the Brassicaceae is possible, as it is in other plant species. Seed oils and their constituent fatty acids in wild radish and other Brassicaceae may be useful as indicators of overall genetic variation in naturalised populations occurring in southern Australia.

2.5.Wild radish (Raphanus raphanistrum L.)

Taxonomic and morphological description

Wild radish (Raphanus raphanistrum) is a member of the Brassicaceae (syn. Cruciferae) or mustard family. It is an outcrossing, diploid species with 2n = 18 chromosomes. Alternate names are jointed charlock (USA), ramnas (South Africa), runch (United Kingdom), white charlock and wild kale. Raphanus is the latinised form of the Greek word Raphanos, (a vegetable grown from antiquity) derived from ‘ra’, meaning quickly and ‘phainomai’ to appear. This was the name given to cultivated radish (Raphanus sativus) because of its rapid germination and emergence after sowing. The specific name, raphanistrum, is formed

27 by the addition of ‘istrum’, which means ‘in the form of’ indicating that the plant is similar to the cultivated species. Raphanus is taxonomically distinct in having fruits that are a siliquas and indehiscent (Gill et al. 1980).

Wild radish is an erect (up to 1.5 m tall) annual weed (Cheam 1986), reproducing via seed. The seedlings have cotyledons of eight to 15 by 10 to 20mm. A short hypocotyl is present. The young plant develops as a flat rosette in which the leaves are rough, the lower deeply lobed, with a much enlarged terminal segment. The mature plant has a large basal rosette of leaves, with an erect or spreading, much branched stem of 30 to 100 cm, sometimes higher, covered with short, stiff bristle hairs. Mature stem leaves are entire or with slight indentations eight to 20 cm long and five to 10 cm wide at the apex, with a scattering of short stiff bristles that makes them rough to the touch. The uppermost stem leaves are narrow, shorter and often undivided. The flower perianth has four free segments, the flowers are in long terminal, corymbose racemes, and the petal colour varies from pale yellow to white with purple veins. Flowers are one to 2 cm across, with pedicels of one to 2 cm. The siliqua or fruit is a fleshy pod, spongy, long, narrow, cylindrical, five to 10 mm in diameter, two to 7 cm long, terminating in a one to 2 cm pointed beak, longitudinally ribbed. The siliqua is indehiscent with two to 10 seeds, constricted but without a septum between seeds. The lower pods are often small and seedless, while the constrictions allow the pod to be broken into one-seeded sections when ripe. Seeds are globular, ovoid, reddish to dark-brown, two to 4 mm in diameter, and covered with a fine network of veins with shallow interspaces (Hyde-Wyatt and Morris 1975; Auld and Medd 1987; Cunningham et al. 1992; Parsons and Cuthbertson 1992).

The root system is extensive and fibrous to a depth of 20 cm. The roots spread horizontally to 80 cm in all directions. Two strong taproots reach to a depth of about 70 cm at which point they divide into three to four branches and continue to descend. Another fibrous array of roots spreads through an area of 80 cm in diameter at 120-160 cm depth (Kutschera 1960).

Wild radish is widespread throughout the world (Cheam 1984). It is native to the mediterranean region and spread northwards through Europe in the pre–Christian era. It has been reported as a troublesome weed in cropping land in the British Isles (Blackman and Templeman 1938; Clapham et al. 1952; Bentham and Hooker 1954; Butcher 1961). In the intervening years it has been introduced to most of the temperate world and is now a 28

common weed throughout Europe (Chater 1964), the north-east, central and Pacific north- west of the United States and Canada (Mekenian and Willemsen 1975), and is frequent in New Zealand (Webb et al. 1988) and Turkey (Munz 1959).

Origin and geographical description

Figure 2.2 Origin and geographical distribution of wild radish. Wild radish is an abundant species in all continents and a serious weed in most (Piggin et al 1978).

Distribution of wild radish in southern Australia.

Figure 2.3 Distribution of wild radish in southern Australia. Each point on the map represents the origin of a herbarium specimen. Wild radish is most abundant in the wheatbelt and the adjacent high rainfall zone, and is absent from the tropics. (Piggin et al 1978)

In Australia, wild radish has been a troublesome weed since its accidental introduction during the latter part of the last century. It was probably introduced into Australia as a contaminant of agricultural produce (Donaldson 1986). It does not seem to have been intentionally introduced by the early settlers. It was reported as naturalised around Melbourne by 1856 soon after settlement (Hooker 1860), around Sydney by 1867 (Woolls 1867; Moore 1884), near Adelaide by 1875 (Schomburgk 1875) and in Queensland by 1913 (Bailey 1913). It is now well established in Western Australia (Gardner 1930; Meadly and Pearce 1954; Meadly 1965), South Australia (Black 1909; Black 1948), Queensland (Everist 1913), NSW (Burbidge and Gray 1970; Beadle et al. 1962; Leigh and Mulham 1977), Victoria (Ewart and Tovey 1909; Ewart 1930; Willis 1972; Goodman 1973) and Tasmania (Curtis 1956; Hyde-Wyatt and Morris 1975).

Preferred habitats

Soil types

Wild radish is adapted to soils ranging from heavy, fertile loams to sands and black soils that are at times eroded or excessively wet (Borowiec et al. 1972). It is known to grow on chalk soils, although it is often thought of as an indicator of acid soils (Vogel 1926).

The impact of wild radish on crops

Weed control is essential for maximum yield. Weeds are successful largely due to characteristics that confer superior colonising and competitive abilities. Plant attributes particularly important to weeds include adaptations for effective seed dispersal, seed dormancy, long-lived seed bank, rapid growth and phenotypic plasticity. Most weeds also produce copious quantities of seeds and/or have means of vigorous vegetative reproduction. Weeds represent a unique category of vegetation that plays a critical role in the dynamics of agricultural ecosystems (Altieri and Liebman 1988).

In the case of wild radish, its success as a weed can be attributed to its specific germination ecology, and its widespread occurrence as a contaminant of grain seeds. This is especially important in canola, where the seeds of wild radish are similar in shape and size, making it difficult to clean contaminated seeds, increasing the risk of spreading. When contaminating canola, wild radish seeds decrease oil quality because they increase the level of erucic acid in extracted oil and increase the glucosinolate content of the remaining meal

Wild radish is extremely competitive in crops. Its competition with wheat has been well researched (Code et al. 1978). In wheat, a wild radish density of 200 plants/m2 will cause crop losses of 50%. Even with a wild radish density of only 7 plants/m2, 10% yield losses have been reported (Code and Reeves 1981). In a similar study in Western Australia (Moore 1979) reported that 25 plants/m2 of wild radish emerging with the wheat crop resulted in 7-11% yield reduction; 50 plants resulted in 15-20% reduction, 75 plants in 19- 26% reduction and 100 plants in 25-33% reduction. In the central west of NSW, a dense infestation of 160 plants/m2 reduced wheat yield by 70-80% (Dellow and Milne 1987). Yield reductions of this magnitude rank wild radish as one of the most damaging weeds of cereals (Poole and Gill 1987). Its competition with wheat mainly occurs during the early growth stages of the crop. Significant yield increases have resulted when wild radish was controlled by herbicide application during the two-to five leaf stage of the crop (Dellow and Milne 1987), with yields often five times higher than when the weed is controlled after the crop tillers (Cheam et al 1995).

Canola also suffers yield losses due to wild radish competition: in fact uncontrolled populations reduce both yield and quality. Green wild radish segments give off toxins that inhibit germination and metabolism, and the affected grain usually dies (Cheam 1996). Those seeds that emerge produce abnormal seedlings. The fibrous stems of wild radish choke the header comb, making harvest difficult. The green stems of wild radish are high in moisture, which may raise the moisture content of grain above acceptable storage levels. This usually occurs in years with late rains when wild radish continues to grow and remains green after crop maturity (Meadly 1965).

The seeds of wild radish are the most toxic part of the plant. Its poisonous principle is allyl isothiocynate, derived from the glucosinolate, sinigrin. Apart from poisoning, wild radish has caused a taint in milk if grazed by dairy cows (McBarron 1977). The seeds are harmful to animals and may prove toxic when animals are confined to areas where it is abundant. Lambs grazing the flowering plant may contract jaundice - haemoglobinuria - and evidence of liver damage. Similar poisoning of cattle has been reported when the beasts were crowded on to wild radish to eat down an unusually luxuriant growth of the weed (Cheam et al. 1995). The affected animals initially lose appetite and most of them show lassitude, stupor and paralysis, while some are excitable. The dung is coated with mucus and as the disease progresses there is watery blood-stained diarrhoea. Abortion is noted in some cows.

Seed production and dispersal

Wild radish is a prolific seeder, producing more than 17,000 seeds/m2 from a weed stand of 52 plants/m2 (Reeves et al.1981). Up to 45,000 seeds/m2 were found in some plots. About 20 to 40% of the seed produced in any season are still viable one year later. This suggests that the seed bank remains high even under the worst conditions for seed production (Piggin et al. 1978). Seeds can remain viable in the soil for 15 to 20 years (Kurth 1967). Germination tests on seed buried at 0, 1, 5 and 10 cm revealed that viability of seeds at the surface was reduced to 43, 18 and 5% at 6, 12 and 24 months respectively, partly due to germination. There was a much slower decline at the 10 cm depth (76-53%) during the 6- to 24- month period. Only 16% of the seed placed at 1 cm was viable after 24 months. In general, seeds placed near the surface deteriorated most rapidly. Because some of the viable seed does not germinate promptly some seedlings will survive even if many are killed by adverse conditions (eg moisture stress) during establishment. As a result early cultivation and herbicide application will not control seedlings that germinate relatively later. Late treatments may give clean grain but yields may be lowered because of weed competition or spray injury . In bad years more than one treatment may be needed.

Germination and seed dormancy

When tested in the laboratory wild radish seed germinates over a wide range of temperatures (constant or alternating in light or dark) but is most responsive to wide temperature fluctuations (Piggin et al. 1978). Perhaps this explains why the weed germinates so well in the field after autumn rains, when there is normally great temperature fluctuation. In the laboratory, seed vernalized at 5° C in the light and then subject to a fluctuating temperature of 25 to 5°C in the dark increased germination compared with vernalisation and subsequent temperature fluctuation in constant light (Piggin et al. 1978). This light-dark effect may help to explain the large increases in germination after cultivation.

In the United States, Mekenian and Willemsen (1975) studied the germination of seeds from an uncultivated area. Their experiments covered the after-ripening of seeds held at room temperature for 6, 8 and 10 months following leaching at 21° C for 8, 16 and 24 hr and stratification in moist sand at 5° C for periods from 3 to 15 weeks. Germination was studied at four alternating temperatures from 5/15°C to 20/30°C. After - ripening was

32 completed in less than 6 months in their studies. Germination was generally higher in the dark at all alternating temperature and storage times. Light germination increased with increasing temperatures and storage times. Stratification generally decreased germination and all leaching treatments gave about the same result, but with indications of some improvement in the dark and with alternating temperatures. Germination of freshly harvested seed was low, in agreement with early innate seed dormancy found in so many summer and winter annuals. Here, as in Australia, the pods that overwinter on the ground seem to have broken and deteriorated to the extent that the seed is released from chemical and mechanical inhibition.

Mekenian and Wilson (1975) observed that the pod walls appeared to inhibit germination.

2.6. Conclusions and Research Objectives

The review suggests that the formation of genetically distinct populations is a common strategy in colonising plants, which enables their populations to adapt to a range of environments. This phenomenon has been studied extensively for inbreeders. To date, there is very little information concerning the role that ecotype formation has on the ability of outbreeding species to adapt to different environments within their naturalized range. Wild radish, as a widespread and successful outbreeding coloniser provides an ideal opportunity to investigate this phenomenon. That it is unlikely to have gone through a major genetic bottleneck during its introduction, and that it exists across a wide range of different environments, suggests that it has the necessary pre-requisites for ecotype formation.

The following hypotheses have been developed for testing.

Hypothesis 1 (tested in chapter 3)

• Wild radish populations are highly variable both within and between sites.

• At least some of the variability is related to environmental variability and this results in the formation of ecotypes.

Hypothesis 2 (tested in chapter 4)

• Populations from nearby sites show greater similarity than more widely separated sites, and populations from environmentally similar sites show greater similarity than populations from dissimilar sites.

• Genetic variation in both AFLP markers and life history traits is greatest within and least between closely adjacent sites.

Hypothesis 3 (tested in chapter 5)

• Seed dormancy declines with time both within and between seasons and is due principally to chemical inhibitors in the pod.

• Buried seeds will lose their dormancy at the same rate as seed lying on the soil surface

• Seed populations will vary in dormancy, with those from the dry areas showing the most dormancy. Seeds germinate most rapidly at temperatures between 15o C and 25o C.

Hypothesis 4 (tested in chapter 6)

• Wild radish populations exhibit genetic variation in the quantity and quality of seed oils

• The yield and quality of oils and their fatty acids will provide future useful genetic material, either in its own right or in the genetic improvement of related species.

2.7.References

Adamson RS, Salter TM (1950) 'Flora of the Cape Peninsula.' (Juta and Co: Cape Town) Aggarwal RAK, Sharma R, Kumar R, Mohapatra T, Sharma RP (2003) Molecular mapping of loci affecting the contents of three major fatty acids in Indian mustard (Brassica juncea L). Journal of Plant Biochemistry and Biotechnology 12, 131-137. Alemayehu N, Becker HC (2001) Variation and inheritance of erucic acid content in Brassica carinata germplasm collections from Ethiopia. Plant Breeding 120, 331- 335. Allard RW (1975) The mating system and microevolution. Genetics 79, 115-126.

Allard RW, Garcia P, Saenz-de-Miera LE, Perez de la Vega M (1993) Evolution of multilocus genetic structure in Avena hirtula and Avena barbata. Genetics 135, 1125-1139. Altieri MA, Liebman (1988) 'The Impact, Uses and Ecological Role of Weeds in Agroecosystems.' (CRC Press: Boca Raton, Florida) Andrews TS, Morrison IN, Penner GA (1998) Monitoring the spread of ACCase inhibitor resistance among wild oat (Avena fatua) patches using AFLP analysis. Weed Science 46, 196-199. ARMCANZ (1997) 'The National Weeds Strategy: A strategic approach to weed problems of national significance.' (Agriculture and Resource Management Council of Australia and New Zealand: Canberra, Australia). Arnold GW, Ozanne PG, Galbraith KA (1985) The capeweed content of pastures in south- west Western Australia. Australian Journal of Experimental Agriculture 25, 117- 123. Aschmann H (1973) Distribution and peculiarity of mediterranean ecosystems. In ‘Mediterranean Type Ecosystems. Origin and Structure’. (Eds F Di Castri, H Mooney). (Springer Verlag,: New York) Aston JL, Bradshaw AD (1966) Evolution in closely adjacent plant populations. II. Agrostis stolonifera in maritime habitats. Heredity 21, 649-664. Auld BA, Medd RW (1987) 'Weeds.' (Inkata Press: Melbourne) Australian National Parks and Wildlife Service (1991) Plant Invasions. In 'The Incidence of Environmental Weeds in Australia'. (National Parks and Wild Life Service: Canberra, Australia) Bailey FM (1913) Wild Radish (Raphanus raphanistrum): In ‘Comprehensive Catalogue of Queensland Plants.' (Government Printer: Brisbane) Baker GA, O'Dowd DJ (1982) Effects of plant density on the production of achene types in the annual Hypochoeris glabra. Journal of Ecology 70, 201-215. Barrett SCH (1982) Genetic variation in weeds. In 'Biological Control of Weeds with Plant Pathogens', pages 93-98. (John Wiley and Sons: New York) Barrett SCH (1988) Genetics and evolution of agricultural weeds. In ‘Weed Management in Agroecosystems: Ecological Approaches.' (CRC Press: Boca Raton, Florida, USA) Baskin CC, Baskin JM (1998) Germination ecology of seeds with non deep physiological dormancy. In ‘Seeds. Ecology, Biogeography, and Evolution of Dormancy and Germination’, pages 49-85. (Academic Press: San Diego) Beadle NCW, Evans OD, Carolin RC (1962) 'Handbook of the Vascular Plants of the Sydney District and Blue Mountains .' (Beadle et al.: Armidale, NSW) Bentham G, Hooker JD (1954) 'Handbook of The British Flora.' (L Reeve & Co. Ltd) Black JM (1948) 'Flora of South Australia.' (Government Printer: Adelaide) Black. JM (1909) 'The Naturalized Flora of South Australia.' (Government Printer: Adelaide)

Blackman GE, Templeman WG (1938) The nature of the competition between cereal crops and annual weeds. Journal of Agriculture Science 28, 247-271. Borowiec S (1972) The influence of soil conditions and kinds of crops on the constancy of occurrence of weeds. Ekologia Polska 20, 199-271. Bosbach K, Hurka H (1981) Biosystematic studies on Capsella bursa-pastoris (Brassicaceae): enzyme polymorphism in natural populations. Plant Systematics and Evolution 137, 73-94. Bradstock RA, Williams JE, Gill AM (2002) 'Flammable Australia: the fire regimes and biodiversity of a continent.' (Cambridge University Press: Cambridge, UK) Britten J (1906) Introduced Plants at Sydney. Journal of Botany (British Foreign) 44, 234- 235. Brown AHD, Weir BS (1983) Measuring genetic diversity in plant populations. In 'Isozymes in Plant Genetics and Breeding, Part A.' (Eds SD Tanksley, TD Orton), pages 219-238. (Elsevier: Amsterdam) Browse JA, Somerville CR (1994) In E.M. Meyerowitz, & C.R. Somervillie. Arabidopsis, 881-912. Burbidge NT, Gray M (1970) 'Flora of the Australian Capital Territory.' (Australian National University Press: Canberra, Australia) Burdon J, Brown A (1986) Population genetics of Echium plantagineum L.: target weed for biocontrol. Australian Journal of Biological Science 39, 369-378. Butcher RW (1961) 'A New Illustrated British Flora.' (L Hill: London) Calvo L, Santalla S, Marcos E, Valbuena L, Tarrega R, Luis E (2003) Regeneration after wildfire in communities dominated by Pinus pinaster, an obligate seeder, and in others dominated by Quercus pyrenaica, a typical resprouter. Forest Ecology & Management 184, 209-223. Canadell J, Lopez-Soria L (1998) Lignotuber reserves support regrowth following clipping of two Mediterranean shrubs. Functional Ecology 12, 31-38. Chapman R, Bolger TP, Coultre, IF le (1999) Seed dormancy release in three common pasture grasses from a Mediterranean-type environment under contrasting conditions. Australian Journal of Experimental Agriculture 39, 143-147. Chater AO (1964) Cruciferae, Raphanus L. In ‘Flora Europea.', pages 260-346. (Cambridge University Press: Cambridge) Chauvel B, Gasquez J (1994) Relationships between genetic polymorphism and herbicide resistance within Alopecurus myosuroides Huds. Heredity 72, 336-344. Cheam AH (1984) Coat imposed dormancy controlling germination in wild radish and fiddle dock seeds. In ‘Proceedings of the 7th Australian Weeds Conference’, pages 184-190. Cheam AH (1986) Seed production and seed dormancy in wild radish (Raphanus raphanistrum) and some possibilities for improving control. Weed Research 26, 405-413.

Cheam AH (1996) Toxicity of green wild radish pods on crop seeds. Farmnote No. 91/96. (Agriculture Western Australia: Perth, WA). Cheam AH, Code GR (1995) The biology of Australian weeds. Wild radish (Raphanus raphanistrum L.). Plant Protection Quarterly 10, 2-13. Cheplick GP (1996) Do seed germination patterns in cleistogamous annual grasses reduce the risk of sibling competition. Journal of Ecology 84, 247-255. Cheplick GP, Quinn JA (1987) The role of seed depth litter and fire in the seedling establishment of amphicarpic peanutgrass, Amphicarpum purshii. Oecologia 73, 459-465. Christiansen S, Abd El Moneim AM, Cocks PS, Singh M (1996) Seed yield and hardseededness of two amphicarpic pasture legumes (Vicia sativa ssp. amphicarpa and Lathyrus ciliolatus) and two annual medics (Medicago rigidula and M. noeana). Journal of Agricultural Science 126, 421-427. Clapham AR, Tutin TG, Warburg EF (1952) 'Flora of the British Isles.' (Cambridge University Press: Cambridge) Cocks PS (1988) Seed production and seed survival under grazing of annual medics Medicago spp. in north Syria. Journal of Agricultural Science 110, 455-464. Cocks PS (1992) Evolution in sown populations of subterranean clover (Trifolium subterraneum L.) in South Australia. Australian Journal of Agricultural Research 43, 1583-1595. Cocks PS (1993) Legumes from the Mediterranean basin: a continuing source of agricultural wealth for southern Australia. (Occasional Publication 1, Co-operative Research Centre for Legumes in Mediterranean Agriculture: Perth, WA). Cocks PS (1997) Integrated cereal/livestock systems for small farms in the dry areas of west asia and north africa. Agro's Annual Review of Crop Ecology 1, 100-150. Cocks PS (1999) Reproductive strategies and genetic structure of wild and naturalized legume populations. In ‘Genetic Resources of Mediterranean Pasture and Forage Legumes’. (Eds. SJ Bennett, PS Cocks), pages 20-31. (Kluwer Academic Publishers: Dordrecht) Cocks PS (2001) Ecology of herbaceous perennial legumes: a review of characteristics that may provide management options for the control of salinity and waterlogging in dryland cropping systems. Australian Journal of Agricultural Research 52, 137- 151. Cocks PS, Bennett SJ (1999) Introduction: role of pasture and forage legumes in mediterranean farming systems. In ‘Genetic Resources of Mediterranean Pasture and Forage Legumes’. (Eds SJ Bennett, PS Cocks), pages 9-19. (Kluwer Academic Publishers: Dordrecht) Cocks PS, Boyce KG, Kloot PM (1976) The Hordeum murinum complex in Australia. Australian Journal of Botany 24, 651-662. Cocks PS, Donald CM (1973) The early vegetative growth of two annual pasture grasses (Hordeum leporinum Link and Lolium rigidum Gaud.). Australian Journal of Agricultural Research 24, 11-19.

Cocks PS, Phillips JR (1979) Evolution of subterranean clover in South Australia. 1. The strains and their distribution. Australian Journal of Agricultural Research 30, 1035- 1052. Code G, Reeves T, Brook H, Piggin C (1978) The herbicidal control of wild radish (Raphanus raphanistrum). In 'Proceedings of the 1st Conference of the Council of Australian Weed Science Societies', Melbourne, pages 241-247. Code GR, Reeves TG (1981) Chemical control of wild radish in wheat. In '6th Australian Weeds Conference', pages 59-63. Cruz A, Perez B, Velasco A, Moreno JM (2003) Variability in seed germination at the interpopulation, intrapopulation and intraindividual levels of the shrub Erica australis in responce to fire-related cues. Plant Ecology 169, 93-103. Cunningham GM, Mulham WE, Milthorpe PL, Leigh JH (1992) 'Plants of Western New South Wales.' (Inkata Press: Sydney & Melbourne) Curtis WM (1956) 'The Student's Flora of Tasmania,.' (Government printer Hobart: Hobart) Dallman PR (1998) 'Plant Life in the World's Mediterranean Climates.' (University of California Press: Berkeley, California). Darlington CD (1963) 'Chromosome Botany and the Origins of Cultivated Plants.' (George Allen & Unwin: London) Dellacasssa E, Lorenzo D, Moyna P, Frizzo CD, Serafini LA, Dugo P (1999) Rosmarinus officinalis L. (Labiatae) essential oils from the south of Brazil and Uruguay. Journal of Essential Oil Research 11, 27-30. Dellow JJ, Milne BR (1987) Wild radish (Raphanus raphanistrum). Department of Agriculture NSW Agfact, P7. 6.6. Di Castri F (1991) 'An Ecological Overview of the Five Regions of the World with a Mediterranean Climate.' (University Press Cambridge: Cambridge, UK) Donald C (1981) 'Innovation in Australian Agriculture.' (Sydney University Press: Sydney, Australia) Donald CM (1970) The pastures of southern Australia. In ‘The Australian Environment. (Ed. GW Leeper). (CSIRO and Melbourne University Press: Melbourne) Donaldson TW (1986) Wild radish (Raphanus raphanistrum L.): a review of research on biology and control in Victoria 1976-1982. Plant Protection Quarterly 1: 160-162. Dreyer DL, Jones KC, Jurd L (1987). Lack of chemical factors in host-plant resistance of alfalfa towards alfalfa weevil: ovicidal activity of some coumarin derivatives. Journal of Chemical Ecology 13, 917-924. Dunbabin MT (2001) Genetic variation in the outbreeding coloniser capeweed in south- western Australia. PhD thesis, University of Western Australia. Dunbabin MT, Cocks PS (1999) Ecotypic variation for seed dormancy contributes to the success of capeweed (Arctotheca calendula) in Western Australia. Australian Journal of Agricultural Research 50, 1451-1458.

Dunin FX (2002) Integrating agroforestry and perennial pastures to mitigate waterlogging and secondary salinity. Agricultural Water Management 53, 259-270. Echeverrigaray S, Fracaro F, Ana Cristina AS, Paroul N, Wasum R, Luciana AS (2003) Essential oil composition of south Brazilian populations of Cunila galioides and its relation with the geographic distribution. Biochemical Systematics and Ecology 31, 467-475. Ehrman T, Cocks PS (1996) Reproductive patterns in annual legume species on an aridity gradient. Vegetatio 122, 47-59. Enright N, Goldblum D (1999) Demography of a non-sprouting Hakea species (Proteaceae) in fire-prone Eucalyptus woodlands of southern Australia in relation to stand age, drought and disease. Plant Ecology 144, 71-82. Enright N, Kintrup A (2001) Effects of smoke, heat and charred wood on the germination of dormant soil-stored seeds from a Eucalyptus baxteri heathy-woodland in Victoria, south-eastern Australia. Austral Ecology 26, 132-141. Everist Sl (1981) The history of poisonous plants in Australia. In ‘Plants and Man in Australia’ (Eds DJ Carr, SGM Carr), pages 223-255. (Academic Press: Sydney) Ewart (1930) 'Flora of Victoria.' (Government Press Melbourne: Melbourne) Ewart AJ, Tovey JR (1909) 'The Weeds, Poison Plants and Naturalized Aliens of Victoria.' (Government Printer: Melbourne) Francis, CM, Millington AJ (1965) Varietal variation in isoflavone content, its estimation by microtechnique. Australian Journal of Agricultural Research 15, 557-564. Gardner CA (1930) 'A Systematic Census of Plants Occurring in Western Australia.' (Government Printer: Perth) Gentili JE (1971) 'Climates of Australia and New Zealand. World Survey of Climatology.' (Elsevier: Amsterdam) Ghassali F, Cocks PS, Osman AE, Gintzburger G, Christiansen S, Semaan A, Leybourne M (1999) Rehabilitation of degraded grasslands in north Syria: Use of farmer participatory research to encourage the sowing of annual pasture legumes. Experimental Agriculture 35, 489-506. Gibberd MR, Cocks PS (1997) Effects of waterlogging and soil pH on the micro- distribution of naturalised annual legumes. Australian Journal of Agricultural Research 48, 223-229. Gill GS (1995) Development of herbicide resistance in annual ryegrass populations (Lolium rigidum Gaud.) in the cropping belt of Western Australia. Australian Journal of Experimental Agriculture 35, 67-72. Gill GS, Blacklow WM (1985) Variations in seed dormancy and rates of development of great brome, Bromus diandrus Roth as adaptations to the climates of southern Australia and implications for weed control. Australian Journal of Agricultural Research 36, 295-304. Gill NT, Vear KC, Barnard DJ (1980) 'Cruciferacae.' (The Garden City Press Limited: Letchworth, Hertfordshire, UK)

Gladstones JS (1966) Naturalized subterranean clover (Trifolium subterraneum L.) in Western Australia: the strains, their distribution, characteristics and possible origins. Australian Journal of Botany 14, 329-354. Gladstones JS, Collins WJ (1983) Subterranean clover in Australia. Journal of the Australian Institute of Agricultural Science 49, 191-202. Goodman RD (1973) 'Honey Flora of Victoria.' (Department of Agriculture, Victoria: Melbourne) Graham AW, Wallwork MA, Sedgley M (1998) Lignotuber bud development in Eucalyptus cinerea (F. Muell. ex Benth). International Journal of Plant Sciences 159, 979-988. Granger R, Passet J (1973) Thymus vulgaris spontane de France: races chimiques e chemotaxonomie. Phytochemistry 12, 1683-1691. Gudkova GN (1976) On seed dormancy periods of wild species of Hordeum and the effect of floral scales on their germination. Byulleten Vsesoyuznogo Ordena Lenina i Ordena Druzhby Narodov Instituta Rastenievodstva Imeni N.I. Vavilova Number 60, pages 25-26.. Gutterman Y (1994) Strategies of seed dispersal and germination in plants inhabiting deserts. The Botanical Review 60, 373-425. Hagen MJ, Hamrick JL (1998) Genetic variation and population genetic structure in Trifolium pratense. Journal of Heredity 89, 178-181. Hamrick JL, Allard RW (1972) Microgeographical variation in allozyme frequencies in Avena barbata. In 'Proceedings of the National Academy of Sciences of the United States of America', pages 2100-2104. Hamrick JL, Godt MJW (1990) Allozyme diversity in plant species. In ‘Plant Population Genetics, Breeding and Genetic Resources. (Eds AHD Brown, T Clegg, AL Kahler and BS Weir), pages 43-63. (Sinauer Associates: Sunderland Massachusetts, USA) Hansen A, Pate J, Hansen AP (1992) Growth reproductive performance and resource allocation of the herbaceous obligate seeder Gompholobium marginatum R. Br. (Fabaceae). Oecologia 90, 158-166. Harradine AR (1986) Seed longevity and seedling establishment of Bromus diandrus Roth. Weed reserch 26, 173-180. Hartl DL (1988) 'A Primer of Population Genetics.' (Sinauer Associates: Massachusetts, USA). Hatton TJ, Nulsen RA (1999) Towards achieving functional ecosystem mimicry with respect to water cycling in southern Australian agriculture. Agroforestry Systems 45, 203-214. Hay RKM, Waterman PG (1993) 'Volatile Oil Crops: Their Biology, Biochemistry and Production'. (Longman Scientific & Technical: Harlow, UK). Hoelzel AR, Dover GA (1991) 'Molecular Genetic Ecology.' (Oxford University Press: Oxford). Hooker JD (1860) 'Flora of Tasmania.' (L Reeve: London)

Hurka H (1990) Differentiation and adaptation in the genus Capsella (Brassicaceae). In 'Biological Approaches and Evolutionary Trends in Plants.' (Academic Press: London). Hyde-Wyatt BH, Morris DI (1975) 'Tasmanian Weed handbook.' (Tasmanian Department of Agriculture: Hobart) Imam AG, Allard RW (1965) Population studies in predominantly self-pollinated species. VI. Genetic variability between and within natural populations of wild oats from differing habitats in California. Genetics 51, 49-62. Ingham JL (1981) Phytoalexin induction and its chemosystematic significance in the genus Trigonella. Biochemical Systematics and Ecology 9, 275-281. Isabel N, Beaulieu, J, Bousquet, J. (1995) Complete congruence between gene diversity estimates derived from genotypic data at enzyme and random amplified polymorphic DNA loci in black spruce. In 'Proceedings of the National Academy of Sciences of the United Stated of America', pages 6369-6373. Kannenberg LW, Allard RW (1967) Population studies in predominantly self-pollinated species. VIII. Genetic variability in the Festuca microstachys complex. Evolution 21, 227-40. Knapp EE, Rice KJ (1998) Comparison of isozymes and quantitative traits for evaluating patterns of genetic variation in purple needlegrass (Nassella pulchra). Conservation Biology 12, 1031-1041. Kon KF, Blacklow WM (1988) Identification, distribution and population variability of great brome Bromus diandrus Roth and rigid brome Bromus rigidus Roth. Australian Journal of Agricultural Research 39, 1039-1050. Kremer A, Petit RJ, Pons O (1998) Measures of polymorphism within and among populations. In ‘Molecular Tools for Screening Biodiversity”. (Eds A Karp, PG Isaac and DS Ingram). (Chapman and Hall: London) Kummerow J, Kummerow M, Trabaud L (1990) Root biomass root distribution and the fine-root growth dynamics of quercus-coccifera L. in the garrigue of southern France. Vegetatio 87, 37-44. Kurth H (1967) The germinative behavior of weeds. SYS Reporter 3, 6-11. Kutschera (1960) 'Root Atlas of the Middle-European Weed Fields and Plant Culture.' (Frankfurt, Germany). Lambers H, Poot P (2003) Structure and functioning of cluster roots and plant responses to phosphate deficiency. Plant & Soil (Special Issue) 248, 9. Lande R (1977) The influence of the mating system on the maintenance of genetic variability in polygenic characters. Genetics 86, 485-498. Latta RA, Blacklow LJ, Cocks PS (2001) Comparative soil water, pasture production, and crop yields in phase farming systems with lucerne and annual pasture in Western Australia. Australian Journal of Agricultural Research 52, 295-303. Latta RA, Cocks PS, Matthews C (2002) Lucerne pastures to sustain agricultural production in south-western Australia. Agricultural Water Management. 53, 99- 109.

Layton CR, Ganders FR (1984) The genetic cinsequences of contrasting breeding systems in Plectritis (Valerianaceae). Evolution 38, 1308-1325. Leeper GW (1966) 'The Australian Environment.' (CSIRO and Melbourne University Press: Melbourne) Leigh JH, Mulham WE (1977) Vascular plants of the riverine plain of New South Wales with notes on distribution and pastoral use. Telopea 1, 225-293. Lloret F, Verdu M, Flores-Hernandez N, Valiente-Banuet (1999) Fire and resprouting in Mediterranean ecosystems: insights from an external biogeographical region, the mexical shrubland. American Journal of Botany 86, 1655-1661. Lush WM (1989) Adaptation and differentiation of golf course populations of annual bluegrass Poa annua. Weed Science 37, 54-59. MacDonald SE, Lieffers VJ (1991) Population variation, crossing and colonization of disturbed areas by Calamagrostic canadensis: evidence from allozyme analysis. American Journal of Botany 78, 1123-1129. Maiden JH (1916) 'Weeds at Sydney in 1802-4.' Agricultural Gazette, New South Wales. Marshall DR, Allard RW (1970) Isozyme polymorphisms in natural populations of Avena fatua and A.barbata. Heredity 25, 373-382. Marshall DR, Brown AHD (1975) Optimum sampling strategies in genetic conservation. In Crop Genetic Resources for Today and Tomorrow. (Eds OH Frankel, JG Hawkes), pages 53-80. (Cambridge University Press: Cambridge, UK) Marshall DR, Weiss PW (1982) Isozyme variation within and among Australian populations of Emex spinosa (L) Campd. Australian Journal of Biological Science 35, 327-32. Martin C, Gonalez-Benito ME, Iriondo JM (1997) Genetic diversity within and among populations of a threatened species: Erodium paularense Fern. Gonz. & Izco. Molecular Ecology 6, 813-820. McBarron EJ (1977) Medical and Veterinary Aspects of Plants Poisons in NSW . (NSW Department of Agriculture Sydney). McNeilly T (1968) Evolution in closely adjacent plant populations. III. Agrostis tenuis on a small copper mine. Heredity 23, 99-108. Meadly GRW (1965) 'Weeds of Western Australia.' (Government Printer: Perth) Meadly GRW, Pearce GA (1954) The chemical control of wild radish. Journal of Department Agriculture Western Australia 3, 309-315. Meikle A, Finch RP, McRoberts N, Marshall G (1999) A molecular genetic assessment of herbicide resistant Sinapsis arvensis. Weed Research 39, 149-158. Mekenian MR, Willemsen RW (1975) Germination characteristics of Raphanus raphanistrum. I. Laboratory studies. Bulletin of the Torrey Botanical Club 102, 243- 252. Michael PW (1994) Alien plants. In ‘Australian Vegetation’. (Ed. RH Groves), pages 57- 83. (Cambridge University Press: Cambridge, UK)

Millar AA, Kunst L (1999) The natural genetic variation of the fatty-acyl composition of seed oils in different ecotypes of Arabidopsis thaliana. Phytochemistry 52, 1029- 1033. Miller ACE, Brookes CP, Loxdale HD, Cussans GW (1996) Using RAPD markers to identify genets of an arable grass weed, Arrhenatherum elatius spp. bulbosum. Annals of Applied Biology 129, 71-82. Moodie M, Finch RP, Marshall G (1997) Analysis of genetic variation in wild mustard (Sinapis arvensis) using molecular markers. Weed Science 45, 102-107. Moore C (1884) 'A Census of the Plants of New South Wales.' (Government Printer: Sydney) Moore G (Ed.) (1998) Soils of south-western Australia. In ‘Soil Guide. A Handbook for Understanding and Managing Agricultural Soils. Agriculture Western Australia Bulletin No. 4343. (Department of Agriculture: Perth, Western Australia) Moore JH (1979) Influence of weed species and density on the yield of crops. In 'Western Australian weeds conference'. Muresk. 92-94 Moran GF, Marshall DR, Muller WJ (1981) Phenotypic variation and plasticity in the colonizing species Xanthium strumarium L. (Noogoora Burr). Australian Journal of Biological Science 34, 639-648. Munz (1959) 'Flora of Turkey.' (University Press Edinburgh: Edinburgh, UK) Neuffer B, Meyer-Walf M (1996) Ecotypic variation in relation to man made habitats in Capsella: field and trampling area. Flora (Jena) 191, 49-57. New JK (1958) A population study of Spergula arvensis L. Annals of Botany 22, 457-477. New JK (1959) A population study of Spergula arvensis 2. Annals of Botany 23, 22-33. New JK, Herriot JC (1981) Moisture for germination as a factor affecting the distribution of the seedcoat morphs of Spergula arvensis L. Watsonia 13, 323-324. Nissen SJ, Masters RA, Lee DJ, Row ML (1992) Comparison of restriction fragment length polymorphisms in chloroplast DNA of five leafy spurge (Euphorbia spp.) accessions. Weed Science 40, 63-67. Nissen SJ, Masters RA, Lee DJ, Row ML (1995) DNA-based marker systems to determine genetic diversity of weedy species and their application to biocontrol. Weed Research 43, 504-513. Norman HC, Cocks PS, Galwey NW (2002) Hardseededness in annual clovers: variation between populations from wet and dry environments. Australian Journal of Agricultural Research 53, 821-829. Norman HC, Cocks PS, Galwey NW, Osman AE (2000) Variability in short and long-term seed dormancy in clover plants from a single population. Cahiers Options Mediterraneennes 45, 369-372. Norman HC, Cocks PS, Smith FP, Nutt BJ (1998) Reproductive strategies in Mediterranean annual clovers: Germination and hardseededness. Australian Journal of Agricultural Research 49, 973-982.

Okoli CAN, Shilling DG, Smith RL, Bewick TA (1997) Genetic diversity in purple nutsedge (Cyperus rotundus L.) and yellow nutsedge (Cyperus esculentus L.). Biological Control 8, 111-118. Pakniyat H, Powell W, Baird E, Handley LL, Robinson D, Scrimgeour CM, Nevo E, Hackett CA, Caligari PDS, Forster BP (1997) AFLP variation in wild barley (Hordeum spontaneum C. Koch) with reference to salt tolerance and associated ecogeography. Genome 40, 332-341. Parsons WJ (1981) 'The history of introduced weeds.' In Plants and Man in Australia. (Eds DJ Carr and SGM.Carr) pages 179 -193. (Academic press: Sydney, Australia). Parsons WT, Cuthbertson EG (1992) 'Noxious Weeds of Australia.' (Inkata Press: Melbourne) Pate JS, Unkovich MJ, Erskine PD, Stewart GR (1998) Australian mulga ecosystems-13C and 15N natural abundances of biota components and their ecophysiological significance. Plant, Cell & Environment 21, 1231-1242. Petersen G, Seberg O (1998) Molecules vs morphology. In 'Molecular Tools for Screening Biodiversity.' (Chapman and Hall: London) Pheloung P (1995) Determining the weed potential of new plant introduction to Australia. Draft Report to the Australian Weeds Committee and the Plant Industries Committee. (Agriculture Protection Board: Western Australia). Piggin CM (1976) Factors affecting seed germination of Echium plantagineum L and Trifolium subterraneum L. Weed Research 16, 337-344. Piggin CM, Reeves TG, Brooke HD, Code GR (1978) Germination of wild radish (Raphanus raphanistrum). Proceeding of the 1st Conference of Australia Weed Science Society, pages 233-240. Poole ML, Gill GS (1987) Competition between crops and weeds in southern Australia. Plant protection quarterly 2, 86-96. Popay AI (1975) Laboratory germination of barley grass. In 'Proceedings of the 28th New Zealand Weed and Pest Control Conference', pages 7-11. (New Zealand Weed and Pest Control Society: Hamilton, New Zealand.) Popay AI (1981) Germination of seeds of five annual species of barley grass. Journal of Applied Ecology 18, 547-557. Powell JR (1983) Molecular approaches to studying founder effects. In 'Genetics and Conservation. A Reference Manual for Managing Wild Animal and Plant Populations.’ (The Benjamin/Cummings Publishing Company: California, USA) Puckridge D, French R (1983) The annual legume pasture in cereal ley farming systems of southern Australia: a review. Agriculture Ecosystems and Environment 9, 229-267. Rebetzke GJ, Burton JW, Carter TE, Jr; Wilson RF (1998) Genetic variation for modifiers controlling reduced saturated fatty acid content in soyabean. Crop Science 38, 303- 308. Reeves TG, Ewing MA (1993) Is ley-farming in Mediterranean zones just a passing phase? In 'XVII International Grasslands Congress'. (Ed. M Baker), pages 2169-2177. (New Zealand Grassland Association: Palmerston North, New Zealand.)

Reeves TG, Code GR, Piggin CM (1981) Seed production and longevity, seasonal emergence, and phenology of wild radish, (Raphanus raphanistrum). Australian Journal of Experimental Agriculture and Animal Husbandry 21, 524-530. Richard M, Jubier MF, Bajon PH, Lejeune B (1995) A new hypothesis for the origin of pentaploid Holcus from diploid Holcus lanatus L. and tetraploid Holcus mollis L. in France. Molecular Ecology 4, 29-38. Rowe ML, Lee DJ, Nissen SJ, Masters RA (1997) Genetic variation in North American leafy spurge (Euphorbia esula) determined by DNA markers. Weed Science 45, 446-454. Schomburgk (1875) 'The Flora of South Australia.' (Government Printer: Adelaide) Schut JW, Qi, X., and Stam, P. (1997) Association between relationship measures based on AFLP markers, pedigree data and morphological traits in barley. Theoretical and Applied Genetics 95, 1161-1168. Shane MW, Vos M de, Roock S de, Cawthray GR, Lambers H (2003) Effects of external phosphorus supply on internal phosphorus concentration and the initiation, growth and exudation of cluster roots in Hakea prostrata R.Br. Plant & Soil, 209-219. Skoula M, El Hilali I, Makris AM (1999) Evaluation of the genetic diversity of Salvia fruticosa Mill. clones using RAPD markers and comparison with the essential oil profiles. Biochemical Systematics and Ecology 27, 559-568. Smettem K (1998) Deep drainage and nitrate losses under native vegetation and agricultural systems in the mediterranean climate region of Australia. Land and Water Resources Research and Development Corporation Report, RAPPSO 2/98, Canberra, Australia. Smith FP, Cocks PS, Ewing MA (1995) Variation in the morphology and flowering time of cluster clover (Trifolium glomeratum L.) and its relationship to distribution in southern Australia. Australian Journal of Agricultural Research 46, 1027-1038. Smith FP, Cocks PS, Ewing MA (1996) Short-term patterns of seed softening in Trifolium subterraneum, Trifolium glomeratum, and Medicago polymorpha. Australian Journal of Agricultural Research 47, 775-785. Snaydon RW, Davies MS (1976) Rapid population differentiation in a mosaic environment. VI. Populations of Anthoxanthum odoratum at sharp boundries. Heredity 37, 9-25. Snowball R (1996) The Australian germplasm of pink flowering serradella species. In 'Proceedings of the 8th Australian Agronomy Conference,' pages 596-509. (Australian society of Agronomy Inc: Toowoomba, Australia) Spitze K (1993) Population structure in Daphnia obtusa: quantitative genetic and allozyme variation. Genetics 135, 367-374. Stebbins GL (1950) 'Variation and Evolution in Plants.' (Columbia University Press: New York) Stebbins GL (1989) Introduction. In ‘Isozymes in Plant Biology’. (Eds DE Soltis, PS Soltis). (Chapman and Hall: London) Sudaric F, Nevjestic A, Rukavina L, Stancic S (1976) Dicoumarol poisoning of calves fed poor quality clover. Veterinarski Glasnik 30, 905-910.

Sun M (1997) Population genetic structure of yellow starthistle (Centaurea solstitialis), a colonizing weed in the western United States. Canadian Journal of Botany 75, 1470-1478. Sun M, Corke H (1992) Population genetics of colonizing success of weedy rye in northern California. Theoretical and Applied Genetics 83, 321-329. Tahoun MK, Abdel-Halim I, Ghazy AEA (1999) Oil and erucic acid contents of five Brassica species grown for four successive years in Egypt. In 'New Horizons for an Old Crop: Proceedings of the 10th International Rapeseed Congress', Canberra, Australia Taylor GB (1984) Effect of burial on the softening of hard seeds of subterranean clover. Australian Journal of Agricultural Research 35, 201-210. Taylor GB (1996b) Incidence and measurement of autumn seed softening within Medicago polymorpha L. Australian Journal of Agricultural Research 47, 575-586. Taylor GB, Ewing MA (1996) Effects of extended (4-12 years) burial on seed softening in subterranean clover and annual medics. Australian Journal of Experimental Agriculture 36, 145-150. Taylor GB, Revell CK (2002) Seed softening, imbibition time, and seedling establishment in yellow serradella. Australian Journal of Agricultural Research 53, 1011-1018. Thellung A (1912) La flore adventice de Montpellier. Memoires de la Nationale des Sciences Naturelles et Mathematiques de Cherbourg 38, 57-728. Thomson EF, Rihawi S, Cocks PS, Osman AE, Russi L (1990) Recovery and germination rates of seeds of mediterranean medics and clovers offered to sheep at a single meal or continuously. Journal of Agricultural Science 114, 295-300. Thrower NJW, Bradbury DE (1973) The physiography of the mediterranean lands with special emphasis on California and Chile. In ‘Mediterranean-Type Ecosystems: Origin and Structure. (Eds F Di Castri, HA Mooney). (Springer Verlag: New York) Van de Loo FJ (1987) Resistance of Arctotheca calendula to bipyridyl herbicides. Honours thesis, The University of Adelaide. Venable LD, Dyreson E, Pinero D, Becerra JX (1998) Seed morphologies and adaptive geographic differentiation. Evolution 52, 344-354. Verdaguer D, Ojeda F (2002) Root starch storage and allocation patterns in seeder and resprouter seedlings of two Cape Erica (Ericaceae) species. American Journal of Botany 2002, 1189-1196. Vieira RF, Grayer RJ, Paton A, Simon JE (2001) Genetic diversity of Ocimum gratissimum L. based on volatile oil constituents, flavonoids and RAPD markers. Biochemical Systematics and Ecology 29, 287-304. Vogel F (1926) 'Contribution to the Knowledge of the Growing Place Requirements of Wild Radish and Sinapis arvensis.' (Bovarian Farming Year Book: Frankfurt, Germany) Wang SX, Oomah BD, McGregor DI, Downey RK (1998) Genetic and seasonal variation in the sinapine content of seed from Brassica and Sinapis species. Canadian Journal of Plant Science 78, 395-400.

Warwick SI (1990) Allozyme and life history variation in five northwardly colonizing North American weed species. Plant Systematic Evolution. 16, 41-54. Warwick SI, Black LD (1986) Genecological variation in recently established populations of Abutilon theophrasti (velvetleaf). Canadian Journal of Botany 64, 1632-1643. Watson HC (1847) 'Cybele Britannica.' (Longman & Co: London) Webb CJ, Sykes WR, Garnock-jones PJ (1988) 'Flora of New Zealand.' (DSIR Botany Division: Christchurch, New Zealand) Weir BS (1990) 'Genetic data analysis.' (Sinauer Associates: Sunderland) Westman AL, Kresovich S (1997) Use of molecular marker techniques for description of plant genetic variation. In 'Biotechnology and Plant Genetic Resources: Conservation and Use’, pages 9-48 (CAB International: New York) Widder FJ (1923) Die Arten der Gattung Xanthium. In Feddes Repertorium Beihefte.Bd. 20) Willis JH (1972) 'A Handbook to Plants in Victoria.' (Melbourne University Press: Melbourne) Wilson MF (1983) 'Plant Reproductive Ecology.' (John Wiley & Sons: New York) Wolff K, Rogstad SH, Schaal BA (1994) Population and species variation of minisatellite DNA in Plantago. Theoretical and Applied Genetics 87, 733-740. Wood H (1994) The introduction and spread of capeweed, Arctotheca calendula Levyns (Asteraceae) in Australia. Plant Protection Quarterly 9, 94-100. Wood H, Degabriele R (1985) Genetic variation and phenotypic plasticity in populations of Patterson's Curse (Echium plantagineum L.) in south-eastern Australia. Australian Journal of Botany 33, 677-685. Woodward RG, Morley FHW (1974) Variation in Australian and European collections of Trifolium glomeratum L. and the provisional distribution of the species in southern Australia. Australian Journal of Agricultural Research 25, 73-88. Wooller SJ, Wooller RD (2001) Seed set in two sympatric banksias, Banksia attenuata and B. baxteri. Australian Journal of Botany 49, 597-602. Wools W (1867) Plants introduced accidentally. In 'A Contribution to the Flora of Australia', pages 136-152 (F White: Sydney)

Chapter 3. Genetic variation between and within naturalized wild radish (Raphanus raphanistrum) populations in south-western Australia

3.1 Abstract

Genetic variation between and within populations of the outbreeding annual weed, wild radish (Raphanus raphanistrum), was studied using seed collected from 55 sites in the Western Australian wheatbelt and adjacent high rainfall sites. Seed samples were collected every 15 to 20 km in transects from Denmark to Southern Cross and Geraldton to Mullewa. Each site (20 m2 in area) was sampled by randomly collecting seed from 30 plants. The material was grown at the University of Western Australia field station at Shenton Park, Perth, WA. Fourteen morphological and phenological characters were measured and the results analyzed using analysis of variance, principal component analysis and cluster analysis.

The results showed a large amount of variation for all the measured plant traits. The number of pod segments varied from 1 to 12, seed weight from 2.6 to 16.1 mg and flowering time from 96 to 173 days. All 14 traits measured were variable and all differed significantly between sites. For all traits, within site variation was greater than between sites. Seed weight (47%), flowering time (47%) and pod weight (39%) showed the greatest between site variation, while leafiness (9%) and the number of pod segments (12%) showed the least. Most of the variation between sites was associated with geoclusters based on rainfall and temperature. Plants from areas with high rainfall and low temperature produced longer and wider pods, larger seeds and pods with more segments, than did plants from low rainfall, high temperature sites. These plants also tended to flower later than those from hotter, drier sites.

The results clearly demonstrate that wild radish has formed genetically distinct populations within Western Australia in the 150 or so years since it was introduced. In this respect it is similar to many self-pollinating annuals, including the naturalised legumes and grasses. It is also similar to the obligate outbreeder, Arctotheca calendula (capeweed), although capeweed displays a greater association between flowering time and rainfall than does wild radish. As with most of these species it is the reproductive traits that display the greatest between site variation and hence adaptive and evolutionary significance.

3.2 Introduction

Genetic variation in populations is the expression of differences within and between populations for the many traits concerned and is usually measured in a common environment. There are a number of factors responsible for creating genetic variation, such as mutation, recombination, genetic drift, mating system, gene flow and selection pressure (Slatkin 1987). Among the literature for colonizing species, there are a number of studies demonstrating genetic variation within and between populations in outbreeding and self- pollinating plants (Barrett and Shore 1990; Brown and Burdon 1983; Cocks 1999; Cocks et al. 1976; Dunbabin 2001; Price and Jain 1981; Schemske and Lande 1985; Warwick 1990b; Woodward and Morley 1974). The breeding systems of colonizing species are known to play an important role in their success. It is well recognized for example, that among colonizing species, inbreeding is by far the more common system (Barrett and Richardson 1986; Henslow 1879; Price and Jain 1981). However, Cocks (1999) states that many outbreeders also make successful colonizers, and that the genetic changes by which this is achieved are less clear. The best example in Australia is annual ryegrass (Lolium rigidum Gaud.), a widespread weed in the cereal belt of southern Australia. Not only has it colonized many different habitats but it also adapts rapidly to changed circumstances. For example, it has developed resistance to some 20 herbicides and can do so within a few years of the first application of spray (Gill 1995; Hall et al. 1994; McAllister et al. 1995). Similarly, in Canada, wild mustard has developed resistance to several herbicides, including dicamba and MCPA (Webb and Hall 1995). Other outcrossing weeds that have successfully colonized large areas of southern Australia are salvation jane (Echium plantagineum) (Brown and Burdon 1983) and cape weed (Arctotheca calendula) (Dunbabin 2001, Dunbabin and Cocks 1999).

In Australia wild radish has been a troublesome weed since its accidental introduction during the latter part of the last century. It was probably introduced into Australia as a contaminant of agricultural produce (Donaldson 1986), and since that time has successfully colonized a wide range of environments. It is an obligate out breeder, which can respond rapidly to environmental changes.

Despite its success however, little is known about genetic variation in wild radish. The main aim of the study reported in this chapter is to investigate genetic variation within and between populations of wild radish, to determine the traits that have contributed to its

49 success in Western Australia and to relate variation to environmental factors. The hypotheses tested in this chapter are that:

o being an outbreeder, there will be more within population variation for most traits than there is between population variation;

o reproductive traits, especially flowering time, will show the greatest between population variation; and

o some reproductive traits, especially flowering time, will be associated with environmental variables.

3.3 Materials and methods

Seed collection

Wild radish seed was collected along transects from Denmark to Southern Cross and Geraldton to Mullewa (Fig. 3.1). The collection was carried out in 1999 between late December and early February. The sites were selected because they include a range of annual rainfalls, temperatures and growing season lengths in a relatively short distance.

Samples were collected approximately every 15-20 km in sites of 20 m2 in area. Each site was sampled by randomly collecting the seed from 20 to 30 plants, and the samples were placed in labelled paper bags. Vegetation details, habitat type (crop, pasture or roadside) were noted at each site. Latitude and longitude were measured using GPS model 315 manufactured by MAGELLAN (AGD 84 Datum). Subsequent data such as average rainfall, evaporation and mean temperatures were collected for localities near each site from data supplied by the Commonwealth Bureau of Meteorology, Canberra, Australia.

Seed processing

The collected material was dried for approximately 48 hours at 60 0C to remove all moisture. After drying in the oven the samples were further dried in a glass house because some material remained damp and caution was exercised to ensure that the seeds were not damaged by heat.

The dried samples were threshed with a hand grinder and the seed placed over a stack of sieves to separate the sample into a number of fractions to make cleaning easier. Both loose seed and intact pods were retained from the sample. The seed was bulked for each site, placed in seed envelopes, weighed and stored.

Plant establishment

The clean seeds were directly sown into the field on 19th April 2000. Where possible 10 plants from each site were planted into two blocks, which were arranged in a randomized complete block design. Seeds were sown in rows with 50 cm between each plant and 150 cm between rows (See plate number 3.1 and 3.2). Plants were weeded by hand as required and irrigated as necessary throughout the season. Before planting a basal treatment of super potash at a rate of 350 kg/ha was applied. Later, in early May after the experiment was established the entire area received an application of NPK fertilizer at a rate of 350 kg/ha. Glyphosate was applied prior to sowing to control weeds at a rate of 2.5 L/ha to the entire planting area.

Measurement

A number of variables were measured on each plant. These were number of leaves, length of leaf, pod length, pod width, number of segments/10 pods, weight of 10 pods, number of primary branches, plant height, plant width, leaf area, date of flowering, flower colours, growth habit, and 100 seed weight (Table 3.1).

Table 3.1 Description of the variables measured on each plant.

Plant trait Abbreviation Description Flowering time flwrtime The number of days from planting until the first flower opened Flower colour flcolor 10 types: 1, white; 2, yellow; 3, purple, 4,creamy white, 5,brownish yellow, 6,purple patches on white ground, 7pink patches on white ground, 8, yellow patches on white ground, 9,pink patches on yellow ground, 10,brownish white Number of leaves nleaves At the time when the first flower opened, the total number of leaves on the main stem leaves Plant height pltht Length of the main stem from ground to shoot apex at harvesting (mm) Plant width plwid Plant width in north –south and east - west directions (mm) Leaf area larea Area of leaf subtending the first flowering branch (cm2) Number of branches npbranchs The number of branches at harvest Pod length pdtht Mean length of 10 pods per plant (mm) Pod width pdwid Mean width of 10 pods per plant (µm) Number of pod segmennsegmnt Mean of 10 pods per plant Seed weight seedwt 100 seed weight per plant (g) Pod weight podwt Weight of 10 pods per plant (g)

Plate 3.2. The same populations of wild radish at flowering at the University Field Station Shenton Park, in late September 2000.

Analysis of the results Data were analyzed in six steps; first the data was checked for normality using Microsoft Excel and Genstat. Next, in order to determine if means differed significantly between sites and to allocate variation between and within sites a general analysis of variance was performed on each variable individually. Thirdly, the sites were grouped using K-means clustering of standardized environmental variables (variables standardized using Z-scores). In this way 5‘geo-clusters’ were generated, each cluster consisting of environmentally similar sites (Berger et al. 2003).

Fourthly, a principal component analysis was performed on the plant traits with the mean values of each trait at each site used as variates. Mean scores and their standard deviations were calculated for each of the geoclusters. Finally, mean PC1 scores in each geo-cluster were plotted against mean PC2 scores for each geo-cluster.

3.4 Results

Wild radish displayed a large amount of variation for all measured traits. Table 3.2 shows the range of values for each trait over all sites, and the total population coefficient of variation (CV), which gives an indication of the relative amount of variation exhibited by each trait. The most variable traits were leaf area (CV = 62%) and plant height (CV = 46%), while the traits displaying the least variation were days to flowering (CV = 5%), seed weight (CV = 19%) and plant width (CV = 24%).

Table 3.2 The range of values, the coefficient of variation and the percentage of variation residing between and within sites for each plant trait. Significance level of the difference between sites (shown in the right hand column) is P<0.001 (***).

Plant trait Range of values CV% % Variation % Variation within sites between sites flwrtim e 96 to 173 (days) 4.7 54.0 46.0*** flcolor 1 to 10 (colors) 70 88.0 12.0*** nl eaves 1 to 17 (numbers) 42 83.0 17.0*** pltht 3 to 97 (cm) 46 79.0 21.0*** pl twid 7 to 94 (cm) 24 72.0 28.0*** larea 6 to 259 (cm2) 62 80.0 20.0*** lratio 1.14 to 6.0 25 90.0 10.0*** nb ranh 1 to 26 (numbers) 46.9 81.0 19.0*** pdlegth 4 to 90 (mm) 23 85.0 15.0*** pd width 0.1 to 0.7 (µm) 30.2 67.0 33.0*** nsegmnt 1 to 12 (numbers) 29 88.4 11.6*** seed wt 0.26 to 1.61 (g) 19.4 53.0 47.0*** Podwt 0.51 to 4.49 (g) 26.9 61.0 39.0***

Results of the one-way ANOVA are shown in Table 3.2. All traits were significantly different (P<0.001) between sites. The proportion of total variation residing between and within sites differed from trait to trait. Seed weight (47%), flowering time (47%) and pod weight showed the largest proportion of variation between sites, while traits with a low proportion of variation between sites were number of pod segments (12%), proportion of leaf (9%) and pod length (16%). The proportion of variation occurring within sites is a mirror image of that between sites, and ranges from 90% (leaf proportion) to 53% (seed weight) (See plate number 3.3).

Plate 3.3. The wild radish population from York showing within population variation in morphological and physiological traits at the University Field Station, Shenton Park in September 2000.

Table 3.3 shows that the predominant flower colours are yellow, white and creamy white, although approximately 20% of the Western Australian population falls into the remaining seven flower colour classes. Almost all plants could be classified as rosettes with upright flowering stems (data not shown).

Table 3.3 The number of individual plants from the 55 populations in each flower colour class

Flower colour (code in Table 3.1) Number of plants White 371 (33%) Yellow 385 (35%) Purple 61 (5%) Creamy white 140 (13%) Yellow with brownish dot 42 (4%) White with purple dot 44 (4%) Pink with white dot 42 (4%) Yellow with white dot 24 (2%) Pink with yellow dot 5 (<1%) Brownish with white dot 1 (<1%) Total 1115 (100%)

Pearson’s correlation matrix (Table 3.4) showed that three plant traits (plant width, number of branches and flowering time) were significantly correlated with annual rainfall and maximum summer temperature. Few other plant traits were significantly correlated with the environmental variables.

Five geo-clusters were generated by K-means clustering, each containing between 4 and 27 sites. The mean and standard errors for rainfall, mean maximum and minimum January and July temperatures and the range of latitudes and longitudes for each geo-cluster are shown in Table 3.5. Geo-cluster 1 has no standard error for the climatic variables because each of the sites used the same meteorological station for its nearest observation. Distribution of the geo-clusters are shown in 3.1.

The results of the principal component analysis appear in Table 3.6 and Fig. 3.2. The first PC accounts for 36% of all variation and the second for 21% (Table 3.6). In Fig. 3.2 the mean values of the principal component scores for each of the geoclusters is plotted against PC1 and PC2, the bars representing standard errors. The biplots are an indication of the direction in which selected variables increase in size. Fig. 3.2 shows that for geoclusters 1, 4 and 5 (high rainfall and low temperature), the plants tend to be larger and later flowering than plants from the more arid geoclusters 2 and 3. Flowering time is later than would be expected for the rainfall in geocluster 1, although in absolute terms plants from high rainfall areas flower later than those from low rainfall areas (Table 3.6 and Fig. 3.2). Plants in the large geocluster 2, representing plants from the central and eastern wheatbelt, appear to be

55 significantly different from plants collected elsewhere in the state. These plants are mainly early flowering, smaller in stature and less leafy than other plants. Table 3.7 shows the name of each of the 55 populations ( the nearest town to where they were collected) and the range of flowering times.

Number of geo cluster zones #Y 1 Ñ 2 $T 3 U% 4 N \& 5

#Y ÑÑ #Y Ñ #Y Ñ Ñ ÑÑ Ñ

$T $T $T $T $T $T $T$T $T$T $T $T $T$T$T $T $T $T $T$T$T $T$T

%U%U %U &\ &\ %U%U%U &\ &\ &\ &\&\ &\\

500 0 500 1000 Kilometers

Figure 3.1. Showing the sites at which the 55 populations of wild radish were collected around Western Australia. The legend indicates the localities of the five geo-clusters described in Table 3.5

Table 3.4 Correlations between measured plant traits, five environmental variables and latitude and longitude. Significance is indicated as follows: P<0.001(***), P<0.005(**), P<0.05(*).

January July .January July Annual/ maximum maximum minimum minimum Plant traits rainfall Lattitude Longitude temperature temperature temperature temperature

Flowring 0.442*** -0.447*** -0.223 -0.416** 0.023 -0.37** -0.018 time Number of -0.034 -0.097 0.178 0.061 0.338* 0.093 -0.125 leaves Plant height -0.007 -0.087 0.248 0.07 0.427*** 0.164 -0.219 Plant width 0.524*** -0.589*** -0.312* -0.537*** 0.05 -0.49*** 0.035 Leaf ratio 0.153 -0.214 -0.113 -0.128 0.093 -0.149 0.129 Leaf area -0.178 0.133 0 -0.078 -0.219 0.066 -0.084 Number of 0.465*** -0.505*** -0.375** -0.539*** -0.178 -0.502*** 0.183 branches Pod length -0.074 0.009 -0.134 -0.025 -0.097 -0.039 0.075 Pod width 0.203 -0.153 0.22 -0.038 0.329* 0.091 -0.403** Number of -0.004 -0.015 -0.179 -0.07 -0.117 -0.101 0.095 segment Seed weight -0.114 0.146 0.298* 0.226 0.311 0.312* -0.329 Pod weight 0.145 -0.151 -0.052 -0.16 0.028 -0.08 -0.166

Table 3.5 Mean and standard errors for annual rainfall, and mean January and July maximum and minimum temperatures for each of the five geoclusters formed from k- means clustering (see text). Latitude and longitude are ranges.

Number of cluster 1 2 3 4 5 Number of sites in cluster 4 9 27 6 9 Annual average rainfall (mm) 376 ± 55 467± 0.00 357± 25 550 ± 73 945 ±117 Maxi.average temp.Jan 34 ± 0.3 32 ± 0.00 36 ± 0.3 31 ± 0.8 27 ± 2.0 Mini.average temp.Jan 17 ± 0.44 18 ± 0.00 19 ± 0.43 14 ± 0.11 13 ± 0.41 Maxi.average temp.July 16 ± 0.44 19 ± 0.00 18 ± 0.30 15 ± 0.16 15 ± 1.05 Mini.average temp.July 6 ± 0.59 9 ± 0.00 7 ± 0.54 5 ± 0.66 6 ± 0.57 Latitude 0S 280 11 to 290 15 280 33 to 290 31 300 02 to 310 53 320 34 to 330 51 340 58 to 33038 1140 39 to 1150 20 to 1150 1160 32 to 1170 Longitude 0E 1140.39 45 1160 0 to 1190 0 35 1150 to 1170 29 57

Table 3.6 Component loadings for plant traits in the principle component analysis, and the amount of variation accounted for by the first and second principal components

Plant traits Latent Vectors PC1 PC2

% Variation 36 21

Flowering time -0.307 -0.310 Number of leaves -0.370 -0.239 -0.333 -0.387 Plant height Plant width -0.328 -0.331 Leaf ratio 0.010 0.002 Leaf area -0.050 -0.016 Pod length -0.289 0.457 Pod width -0.374 0.085 Number of pod segments -0.153 0.502 Seed weight -0.375 0.154 Pod we ight -0.401 0.308

Number of pod segments 1.000 Pod weight

Seed weight 0.500

Pod length

0.000

PC1 -3.000 -2.000 -1.000 0.000 1.000 2.000 3.000 4.000

-0.500

Geocluster 1

-1.000 Geocluster 2

Number of leaves Geocluster 3

-1.500 Geocluster 4

Geocluster 5

Plant height -2.000 PC2

Flowering time -2.500

Figure 3.2 Principal component analysis showing the relationship between plant traits and environment. Points represent the mean of PC scores for all sites within each geocluster. Error bars are ± 1 s.e. Biplots indicate the direction in which selected variables increase in value and are not proportional in magnitude to the data in Table 3.6.

Table 3.7. Flowering times (days after sowing) for 55 populations of wild radish grown in the field at the University of Western Australia field Station at Shenton Park, Perth, WA.

First Mean First Mean Name of Flowering Name of Flowering range flowering days to flowering days to Populations range (days) Populations (days date flower date flower North Bannister 107 121 135 Morawa 113 127 141 Wandering 101 123 146 Morawa 100 113 126 Williams 96 116 136 Mullewa 95 116 136 Katanning 92 118 145 Mullewa 102 119 136 Katanning 114 138 163 Mullewa 105 124 143 Broomehill 89 116 143 Geraldton 120 133 146 York 117 126 136 Geraldton 106 128 150 York 99 113 127 Dongara 121 135 149 York 92 109 127 Midlands 107 121 135 Cunderdin 110 129 148 Mingenew 111 133 156 Cunderdin 124 138 152 Three Springs 127 136 146 Tammin 106 117 128 Carnamah 107 123 139 Doodlakine 106 118 130 Moora 94 114 135 Merredin 107 126 145 New Norcia 113 131 150 Noongal 102 115 129 New Norcia 130 141 153 Moorine Rock 109 122 136 Donnybrook 125 138 151 Old Nukarni 102 116 130 Ballingup 126 141 157 Nukarni 84 105 127 Bridgetown 123 136 150 Nungarin 95 106 117 Manjimup 103 126 150 Trayning 117 129 142 Mount Barker 117 131 145 Korrelocking 119 135 151 Mount Barker 113 136 158 Wyalkatchem 108 124 140 Avondale 122 136 150 Wyalkatchem 98 120 141 Merredin 102 122 142 Dowerin 114 133 151 Denmark 135 145 156 Wongan Hills 121 130 139 Denmark 138 150 162 Wongan Hills 118 132 147 Denmark 121 133 145 Pithara 122 136 151 Geraldton 98 117 137 Perenjori 118 122 126

3.5 Discussion

The formation of genetically distinct populations adapted to different environmental conditions has probably contributed to the widespread success of this colonizing species (See plate number 3.4 and 3.5).

Plate 3.4. The wild radish population from Geraldton the at University Field Station Shenton Park, in early August 2000.

Plate 3.5. The wild radish population from Denmark at the University Field Station Shenton Park, in late October 2000. The pictures in Plates 3.4 and 3.5 highlight the differences in flowering time and rate of development between the two populations.

Among other annual colonizing species, the trait most commonly demonstrating an association with the environment is flowering time: within species early flowering is associated with short growing season (Cocks 1999; Ehrman and Cocks 1996; Woodward and Morley 1974). For example, Woodward and Morley (1974) concluded that variation in flowering time contributed to the success of Trifolium glomeratum in both south-western and south-eastern Australia. Smith et al. (1995), also working with the same species, also found that flowering time was associated with length of growing season, but that other traits were largely homogeneous within two described morphs. The control of flowering time in most Mediterranean annuals is similar, in that they respond to vernalization and photoperiod (Woodward and Morley 1974). One would expect therefore that flowering time would play an important role in the adaptation of wild radish to Western Australia.

The data in Fig. 3.2 indicate that the story may not be quite so straightforward. Although there is a clear relationship between flowering time and rainfall (Table 3.4), and there are genetically distinct populations in regard to flowering time, the data suggest that other factors may be involved. For example, Fig. 3.2 suggests that geo-cluster 2, with mean annual rainfall of 357 mm, flowers later than geo-cluster 5, with mean annual rainfall of 945 mm. In fact the mean flowering times of the geo-clusters do not differ significantly (Table 3.8), suggesting that although the relationship shown in Table 3.4 is significant, the 20% of variation accounted for by this relationship is not the single reason for adaptation of the species.

It is likely that the impact of flowering time is modified by that of seed and pod size. Where pod and seed size are small, as is the case in geocluster 3, the time taken from flowering to maturity is likely to be less (See plate number 3.5).

Table 3.8. Geocluster means of annual average rainfall, and means and standard deviations of flowering time (days after sowing), seed weight (g/100 seeds) and pod weight (g/10 pods).

Geocluster Mean Flowering Seed Pod zones rainfall time weight weight 2 467 238 ± 7 0.76 ± 0.12 1.68 ± 0.34 3 357 232 ± 8 0.76 ± 0.06 1.78 ± 0.23 1 374 234 ± 9 0.65 ± 0.13 1.57 ± 0.39 4 550 237 ± 10 0.66 ± 0.09 1.82 ± 0.43 5 945 246 ± 8 0.64 ± 0.09 1.82 ± 0.43

The relatively early flowering time in this geo-cluster coupled with its small seeds and pods suggests that the species is indeed adapted to the drier areas. Had seed size been larger it is likely that flowering time would have been even earlier. The plants in geo-cluster 3 are also smaller in stature and therefore less demanding of available water, whereas plants in most other geo-clusters are more demanding of the environment. Geo-cluster 1, although early flowering, has large seeds and large leafy plants; geocluster 2 has large seeds and much smaller and less leafy plants; geo-cluster 4, although similar to geo-cluster 2 in many respects, produces much larger pods, and geocluster 5 from the wettest areas, is later flowering, and has small seeds but large pods.

The adaptation of wild radish to Western Australia is therefore complex. In this it is similar to colonizing plants in non Mediterranean environments, where a wide range of life history traits and ecological factors play a role in determining the patterns of genetic variation within and among populations (Barrett and Shore 1990). For example, the adaptation of Lolium rigidum in mediterranean and non-mediterranean areas is strongly influenced by its resistance to herbicides. Some temperate perennial grasses adapt rapidly to changes in heavy metal content (Karataglis et al. 1986; Symeonidis et al. 1985) and grazing regime (Snaydon 1970). The reviews of Barrett and Richardson (986) and Warwick (1990a) give further illustrations.

The amount of variation between populations is also greater than expected for an outbreeding annual weed, where pollen and seed dispersal are by wind and as a contaminant of agricultural produce respectively. Allard (1975) and Allard (1988) have proposed that one of the reasons that most colonizers are inbreeding is that inbreeding allows the populations to ‘fix’ desirable combinations of genes without the risk of their dilution by outcrossing. The results proposed here and by Dunbabin (2001) for capeweed suggest that this advantage may be much less than previously supposed. Indeed, the results for wild radish mirror those obtained for capeweed (Arctotheca calendula) by Dunbabin, who found that capeweed showed almost as much genetic variation between populations as did Hordeum leporinum, a self-pollinating annual grass. In Dunbabin’s comparison however, the distinction between populations was much less if molecular markers were used instead of life history traits, an issue that will be explored in a later chapter. It is possible that the genetic variation in wild radish is partly the result of multiple

62 introductions of the species (Wools 1867), an issue that will also be explored in a later chapter.

However, there is one case where multiple introductions are clearly at work. Cocks et al. (1976) found that of two closely related Hordeum species in the H. murinum complex, H. glaucum was more successful in the drier areas of South Australia and Victoria while H. leporinum was more successful in the wetter areas. Indeed the distributions of the two grasses seems discrete, with the 425 mm rainfall isohyet dividing their habitats. Along this isohyet the species frequently occur together, as they do at Adelaide and Murray Bridge, near where they were likely to have been introduced.

In capeweed Dunbabin (2001) found a negative correlation between plant size and length of growing season, which he considered to be an adaptation to the more intensive grazing likely to occur in high rainfall areas. In high rainfall areas such as Denmark and Mount Barker, land is more frequently used for grazing than in areas with shorter growing seasons, such as Merredin, Geraldton and Mullewa, where cropping is more intensively practiced. Mysterud and Mysterud (2000), in Norway stated that “it is well known that grazing may affect plant morphology.” They went on to note that “intense grazing induces shoot formation, which may lead to mats of short grass”, and that “rapidly growing plants have greater nutrient requirements and less defense against herbivores.” For this reason plants that have evolved under intense grazing are likely to be slow growers. On the other hand, rapidly growing plants may still dominate after intense grazing, because fast regrowth is an alternative strategy to combat grazing. Whether grazing leads to dominance of slow- growing plants of low nutritive value depends on the combination of the plants' ability to regrow (plant tolerance) and the selectivity of the grazer. Sheep are selective grazers, and it has been shown that sheep often change vegetation composition. For example, a study of lawn weeds by Warwick and Briggs (1979) found that in environments where cutting and grazing were severe, selection favoured shorter and smaller plants, leading to genetic differences between populations in characters such as leaf size and inflorescence length. Watson (1969) studying Potentilla erecta in Scotland, found a trend towards genetically smaller plants in grazed environments, where a more extensive spread of leaves is beneficial.

There is no suggestion that grazing has had any of these effects in this study of wild radish. Indeed, the results suggest that plants from geoclusters 1, 4 and 5 are more likely to be

63 larger and leafier than those from the more arid wheatbelt. The explanation is that, even where rainfall is high, wild radish is more likely to be a weed of crops. In high rainfall areas crops are more dense and will provide a greater intensity of competition leading to the evolution of larger and leafier weeds. Where cropping is unusual, such as in south coastal Denmark, most of the wild radish was collected from roadsides and disturbed native vegetation, neither of which is intensively grazed.

3.6 References Allard RW (1975) The mating system and microevolution. Genetics 79, 115-26.

Allard RW (1988) Genetic changes associated with the evolution of adaptedness in cultivated plants and their wild progenitors. The Journal of Heredity 79, 225-38.

Barrett SCH, Richardson BJ (1986) Ecology of biological invasions an Australian perspective. In Genetic Attributes of Invading Species. (Eds RH Groves and JJ Burdon) 21-33. (Australian Academy of Science: Canberra)

Barrett SCH, Shore JS (1990) Isozymes in plant biology. In ‘Isozyme Variation in Colonizing Plants’. (Eds DE Soltis and PS Soltis) 106-126. (Portland, Oregon, USA: Discorides Press,)

Brown AHD, Burdon JJ (1983) Multilocus diversity in an outbreeding weed, Echium plantagineum L. Australian Journal of Biological Sciences 36, 503-509.

Cocks PS (1999) Reproductive strategies and genetic structure of wild and naturalized legume populations. In Genetic Resources of Mediterranean Pasture and Forage Legumes. (Eds SJ Bennett and PS Cocks), pages 20-31. (Kluwer Academic Publishers: Dordrecht)

Cocks PS, Boyce KG, Kloot PM (1976) The Hordeum murinum complex in Australia. Australian Journal of Botany 24, 651-62.

Cocks PS, Phillips JR (1979) Evolution of subterranean clover in South Australia. 1. The strains and their distribution. Australian Journal of Agricultural Research 30, 1035- 52.

Dunbabin MT (2001) Genetic variation in the outbreeding coloniser capeweed in south- western Australia. PhD thesis, University of Western Australia.

Ehrman T, Cocks PS (1996) Reproductive patterns in annual legume species on an aridity gradient. Vegetatio 122, 47-59.

Gill GS (1995) Development of herbicide resistance in annual ryegrass populations (Lolium rigidium Gaud.) in the cropping belt of Western Australia. Australian Journal of Experimental Agriculture 35, 67-72.

Gladstones JS (1966) Naturalized subterranean clover (Trifolium subterraneum L.) in Western Australia: the strains, their distributions, characteristics, and possible origins. Australian Journal of Botany 14, 329-354.

Hall L, Tardif F, Powles S (1994) Mechanisms of cross and multiple resistance in Alopecurus myosuroides and Lolium rigidum. Phytoproduction 75, 17-23.

Henslow (1879) On the self-fertilization of plants. Transactions of the Linnean Society Series II Botany 1, 317-98.

Karataglis SS, McNeilly T, Bradshaw AD (1986) Lead and zinc tolerance of Agrostis capillaris and Festuca rubra across a mine pasture boundary at Minera, North Wales UK. Phyton 26, 64-72.

McAllister F, Holtum JAM, Powles SB (1995) Dinitroaniline herbicide resistance in rigid ryegrass (Lolium rigidum). Weed Research 43, 55-62.

Mysterud A, Mysterud I (2000) Ecological effects of domestic ruminant grazing. II. Effects of grazing on vegetation. Fauna (Oslo) 53, 80-105.

Price SC, Jain SK (1981) Are inbreeders better colonizers? Oecologia 49, 283-286.

Schemske DW, Lande R (1985) The evolution of self-fertilisation and inbreeding depression in plants. II. Empirical observations. Evolution 39, 41-52.

Slatkin M (1987) Gene flow and the geographic structure of natural populations. Science 236, 787-92.

Smith FP, Cocks PS, Ewing MA (1995) Variation in the morphology and flowering time of cluster clover (Trifolium glomeratum L.) and its relationship to distribution in southern Australia. Australian Journal of Agricultural Research 46, 1027-1038.

Snaydon RW (1970) Rapid population differentiation in a mosaic environment. I. The response of Anthoxanthum odoratum populations to soils. Evolution 24, 257-69.

Symeonidis L, McNeilly T, Bradshaw AD (1985) Differential tolerance of three cultivars of Agrostis capillaris to cadmium, copper, lead, nickel and zinc. New Phytologist 101, 309-316.

Warwick SI (1990a) Biological Approaches and Evolutionary Trends in Plants. In ‘Genetic Variation in Weeds- with Particular Reference to Canadian Agricultural Weeds. (Ed. S Kawano)’pages 3-18. (Academic Press: London)

Warwick SI (1990b) Allozyme and life history variation in five northwardly colonizing North American weed species. Plant Systematic Evolution. 16, 41-54. 65

Warwick SI, Briggs D (1979) The genecology of lawn weeds. III. Cultivation experiments with Achillea millefolium L., Bellis perennis L., Plantago lanceolata and contrasting grassland habitats. New Phytologist 83, 509-36.

Watson PJ (1969) Evolution in closely adjacent plant populations VI. An entomophilous species, Potentilla erecta, in two contrasting habitats. Heredity 24, 407-22.

Webb SR, Hall JC (1995) Auxinic herbicide resistant and susceptible wild mustard (Sinapis arvensis L.) biotypes: effect of auxinic herbicides on seedling growth and auxin binding activity. Pesticide Biochemistry and Physiology 52, 137-148.

Wools W (1867) Plants introduced accidentally. In ‘A Contribution to the Flora of Australia’, pages 136-152. (F. White: Sydney)

Chapter 4. Genetic variation in naturalized wild radish (Raphanus raphanistrum) populations in south-western Australia: a comparison of life history traits and AFLP molecular markers

4.1 Abstract

Populations of wild radish (Raphanus raphanistrum) were sampled at Merredin, Nukarni, and Mullewa in the Western Australian wheat belt and at Denmark in the high rainfall, south coast. After growing randomly selected plants at Perth, DNA was extracted from young, fresh leaves and genetic variation within and between populations was measured using amplified fragment length polymorphism (AFLP). The results were analysed using the ARLEQUIN and Genstat statistical programs, and multivariate analyses were used to verify the relationships. Life history traits were also measured and analyzed using canonical variate analysis. It was hypothesized that populations from nearby sites (eg Merredin and Nukarni) would show greater similarity than more widely separated sites, and that populations from environmentally similar sites (eg Mullewa and Merredin) would show greater similarity than populations from dissimilar sites (eg Merredin and Denmark), and that this would be most marked in life history traits.

Despite its breeding system, wild radish has formed genetically distinct populations across Western Australia. The primer combination EcoRl AAC/MseI CAA resulted in a total of 448 loci, of which 99% of the Merredin, 84% of the Nukarni, 93% of the Mullewa and 88% of the Denmark loci were polymorphic. Principal coordinates analysis and cluster analyses showed that variation within populations (71.7%, Fst= 0.71) was higher than variation between populations (28.3%, Fst=0.28). The Nukarni population appeared to be the most distinctive. The Mullewa and Merredin populations showed the greatest level of similarity.

The Denmark and Mullewa populations were most distinct in terms of life history traits. The Denmark population had the largest pods and seeds, while the Mullewa population had the smallest. Nukarni and Merredin were intermediate. Flowering time was not related to either of the first two canonical variates, which together accounted for 98% of variation. This suggests that the role of flowering time in the distribution of this species is obscure, a conclusion supporting the results of previous work with this species. Hierarchical cluster

67 analysis broadly supported these conclusions, although some individuals from Mullewa were similar to the Denmark cluster.

The results from the life history study support the hypothesis that environment is acting strongly on these traits, especially seed and pod size. The genetic markers however, despite separating the biotypes, did not support the hypothesis that the greatest differences would arise from populations separated most in terms of space and environment. Indeed, the results suggest that wild radish was introduced to Western Australia on several occasions, with the likelihood that the Merredin and Nukarni populations represent separate introductions. Nevertheless, the life history traits reveal that evolution since introduction is convergent where environmental conditions are similar and divergent where they differ.

4.2 Introduction

Wild radish (Raphanus raphanistrum), a cross pollinating species of Mediterranean origin, has widely colonized in Western Australia where it is considered to be a serious weed of cereals and canola. Seed dormancy varies among naturalized populations. For example, (Cheam 1986) found that populations collected in wet areas have greater seed dormancy than those from drier areas, a result that differs from capeweed (Artcotheca calendula) in the same region (Dunbabin and Cocks 1999). However, little is known of genetic variation over all and there have been no measures of variation in DNA among Western Australian populations.

Genetic variation in self-pollinating annual species has been measured by many authors. Woodward and Morley (1974) found that time for flowering in Trifolium glomeratum is related to length of growing season in south-eastern Australian populations. In a more detailed study, Smith et al. (1995) observed a similar relationship for the same species in Western Australia. Most other self-pollinating species that have been examined behave in the same way; for example Hordeum leporinum and H. glaucum (Cocks et al. 1976), Bromus diandrus (Kon and Blacklow 1989) and most annual legumes native to Syria (Ehrman and Cocks 1996). Other traits known to vary between populations are seed dormancy (Dunbabin and Cocks 1999; Groves et al. 1982), response of seedlings to temperature (Groves 1975), and isozymes (Burdon et al. 1980; Panetta 1990). Chapter 3 outlines some results obtained from wild radish indicating that time to flowering and seed size show strong between site variation. 68

Inbreeding species are, for a variety of reasons, likely to demonstrate significant between population differentiation (Allard 1965; Allard 1975; Kon and Blacklow 1989; Neuffer and Meyer-Walf 1996). Outbreeders have much less potential to do this and generally show less between population differentiations. However, Dunbabin (2001), in his study of the outbreeding capeweed and the inbreeding barley grass (Hordeum leporinum), did not detect different variation between populations using life history traits, although he did using genetic markers. For outcrossing wild radish one might expect that differentiation between populations in life history traits and in genetic markers will depend on the distance between them.

To measure genetic markers the amplified fragment length polymorphism (AFLP) method was used. In coming to this decision four other methods were considered. Firstly, isozymes were rejected because of the relatively low frequency of polymorphisms using this technique (Moodie 1997). Restriction fragment length polymorphisms (RFLP) (Tanksley et al. 1989) were rejected because assays require relatively large amounts of pure DNA, they are costly, are relatively slow and because, for wild or weedy species, primers are often not available. Thirdly, randomly amplified polymorphic DNA (RAPD) markers (Williams et al. 1990) were rejected because they show typical dominant inheritance heterozygosity (Moodie et al. 1997) and are potentially non-reproducible. Finally, simple sequence repeat (SSR) markers (Akkaya et al. 1992) were rejected because they are labour intensive, knowledge of the sequence is needed and there are a limited number that are useful for detecting molecular markers (Liu 1998).

Amplified fragment length polymorphisms (AFLP) combine the restriction site aspect of RFLP with the experimental amplification aspects of PCR-based DNA markers (Matthes et al. 1998; Vos et al. 1995). Amplification is performed in two steps, known as pre-selective amplification and selective amplification. For the pre-selective step, double stranded adapter sequences are ligated to the ends of the restriction sites. These adapter sequences subsequently serve as universal binding sites for primer annealing in a PCR reaction. For the selective step, new primer sequences of bases that extend into the restriction fragment are used. This results in selective amplification of those fragments in which the primer extensions match the nucleotides flanking the restriction sites. The number of fragments that can be produced ranges between 50 and 100 on a denaturing polyacrylamide gel.

Furthermore, with the use of fluorescent-labeled primers, one can use three pairs of primers, which would allow detection of even more polymorphisms on a single gel.

AFLP is increasingly used as the DNA profiling technique of choice, because it is faster, less labour intensive and identifies numerous markers. It is also reproducible and is able to detect one base pair difference between bands with no prior sequence information (Matthes et al. 1998; Parkash 1998). Zheng and Lapitan (1998) noted that the AFLP technique detects a large number of independent genetic loci in a single reaction and due to the stringent PCR conditions of the technique, highly specific primary reliability is ensured.

In this study, AFLPs were used to determine the genetic variation between and within populations of wild radish. The results were compared with variation in life history traits. Four populations were chosen: one from Denmark on the south coast of Western Australia (high rainfall, geographically isolated from the other test populations), two from near Merredin in the eastern wheat belt (low rainfall, geographically close to each other) and one from Mullewa in the northern wheat belt (low rainfall, geographically isolated from the other test populations). The hypothesis tested was that genetic variation in both AFLP markers and life history traits would be greatest within and least between populations at the eastern wheat belt sites, and greatest between population variation would occur between the isolated populations at Denmark and Mullewa.

4.3 Materials and methods

Origin of plant populations In December 1999 wild radish populations were collected at Merredin (310 22’S, 1180 32’E), Nukarni (30 km south of Merredin 310 23’S, 1180 15’E ), Mullewa ( 280 33’S, 1150 29’E ) and Denmark ( 340 58'S, 1170 39’E) (Fig.4.1). These populations were selected because they represent a range of annual rainfall, temperature and growing season lengths (Table 4.1).

Collection Site Locations %U Merredin $T Nukarni #S Mullewa N &V Denmark

#S Mullewa

Merredin $T%U Nukarni

Denmark &V 200 0 200 400 Kilometers

Figure, 4.1. Showing the sites at which the four populations of wild radish were collected around Western Australia. The legend indicates the localities of the four sites described in Table 4.1

Table 4.1 Mean maximum and minimum temperatures in January and July and mean annual rainfall at Merredin, Nukarni, Mullewa and Denmark1.

Site Max. January Max. July Min. Min. July Annual mean Temp.(0C) Temp(0C) January Temp Rainfall locations Temp(0C) (mm/yr) (0C) Merredin 34 16 17 5 314 Nukarni 34 16 18 6 328 Mullewa 37 19 19 7 340 Denmark 25 16 13 7 1001 1 Meteorological data provided by Commonwealth Bureau of Meteorology, Canberra Australia

Measurement of life history traits

The seeds were directly sown into the field on 19th April 2000. Where possible 10 plants from each site were planted in a randomized complete block design with two replicates (a total of 20 plants). The number of plants surviving to maturity from each site were: Merredin 16, Nukarni 20, Mullewa 18 and Denmark 18. The plots were weeded and fertilized as required and irrigated when necessary (see chapter 3). A number of variables were measured on each surviving plant (Table 4.3). In all analyses means of the two replicates were used; that is plant 1 in replicate 1 was meaned with plant 1 in replicate 2 and so on.

Table 4.2 lists the life history traits measured.

Table 4.2 Description of the 11 life history traits measured on each plant from the four experimental sites.

Plant trait Abbreviation Description Units

Flowering time flwrtime The number of days from sowing until anthesis of the first flower d Number of leaves nleaves Number of leaves on the main stem when the first flowers opened Plant height pltht Length of the main stem from ground to shoot apex mm Plant width plwid Mean of plant width in the north –south, east - west directions mm Leaf area larea Area of the leaf subtending the first flowering branch cm2 Ratio of leaf length to width lratio Ratio of leaf length to width Number of primary branches npbranchs The number of branches at maturity Pod length pdtht Mean length of 10 pods per plant mm Pod width pdwid mean width of 10 pods µm Number of pod segments nsegmnt Mean number of pod segments/pod in 10 pods Seed weight seedwt Weight of 100 seeds per plant g Pods weight podwt Weight of 10 pods per plant g

The data were analysed using the method of residual maximum likelihood (REML) (Peterson and Thompson 1971) and canonical variate analysis. REML analysis was used to apportion genetic variation between and within sites. A general analysis of variance was used to check levels of significance on the life history traits between populations. Canonical variate analysis was used to separate the four groups. Further details are in chapter 3.

DNA extraction

Twenty five plants of each population were grown in a glass house, and 300 mg of fresh young leaf material was sampled from each plant for the DNA extraction. It was not always possible to extract DNA successfully; in our case DNA was successfully extracted from 18 Merredin plants, 15 Nukarni plants, 21 Mullewa plants and 20 Denmark plants. The reason for the difficulty may be associated with compounds in the leaf tissue, which are often found in broad-leafed plants, making extraction of high quality DNA difficult (Kim et al. 1997). Three extraction methods were tested; DNAzol from Life Technologies Australia Pty Ltd (Chomczynski et al. 1997), Sarkosyl Ultrapure (GIBCO) and the modified CTAB method of Doyle and Doyle (1987). The best method was found to be the modified CTAB method, which resulted in high quality DNA. A full description of the modified CTAB technique including an extra phenol/chloroform extraction, is given in Appendix 1.

AFLP molecular markers

The GIBCO BRL Life Technologies Kit was used for the AFLP analysis. The full procedure is detailed in Appendix 2. For wild radish, the only modification required was that the undiluted pre-amplification template was used at the selective amplification step instead of the dilution listed. The rest of the Life Technologies procedure was followed according to the instructions

A number of primer combinations were screened to determine which gave the clearest AFLP fingerprints. One pair of combinations was selected which showed bands enough to score. As AFLP markers are dominant, alleles are scored as either presence or absence at each AFLP fragment. A DNA ladder (30-700bp) was included on each gel to estimate the size of the fragments. Where bands were scored outside this range their size was estimated by extrapolation from the ladder.

In order to assess the reliability of the AFLP procedure, one DNA sample from each site was selected for parallel AFLP analysis with one primer combination. Identical AFLP analysis fingerprints were obtained each time the same DNA template was used, indicating consistent and reliable results from the technique.

The AFLP data were analysed using the University of Geneva's Arlequin programme (Committee 1998; Schneider et al. 2000).

Within population variation was estimated using two methods. The percentage of polymorphic loci is a measure of allelic richness within each site. The average gene diversity is the probability that homologous loci, randomly chosen from two individuals, are different. This latter statistic reflects the overall diversity of the loci, not simply the number of loci which are polymorphic. The division of total variation within and between populations was measured using the analysis of molecular variance (AMOVA).

Principal coordinates analysis was used to group individuals. A similarity matrix was generated from all the recorded loci for each individual, calculated using the simple matching method.

Cluster analysis was calculated using AFLP data of all the individuals at the four sites. For this method a similarity matrix was generated using the Euclidean distance method, and the plants clustered using the group average method. The results are presented as a dendrogram (Figure 4.3).

4.4 Results

Life history traits

Table 4.3 shows the percentage of variation both between and within populations for all life history traits. REML analysis revealed that variation between sites was greater than within sites for seed weight, pod weight and flowering date; for the remainder of variables, within site variation was substantially greater than that between sites. However, apart from leaf shape and leaf area, the traits differed significantly between sites.

The first canonical variate accounted for 84.5% of the variation, and the second for 14.0%. Only insignificant amounts of variation were accounted for by higher order variates. The first canonical variate consisted of inputs from variables associated with pod and seed size,

74 the variates becoming larger as pod and seed size became larger (Table 4.4). The second variate was negatively associated with pod width and seed size and included an element of leaf shape.

In Figure 4.2 plants from the four populations grouped together, with plants from the geographically close sites, Merredin and Nukarni, showing the greatest similarity. The figure shows plants from Denmark produced larger pods than plants from Mullewa in particular, with the sites at Merredin and Nukarni intermediate. Plants from Nukarni and Denmark produced the largest seeds.

Table 4.3 For each plant trait, the mean trait value over all sites, the percentage of variation residing between and within sites and the significance level of the difference between populations P<0.001(***), P<0.005(**) and P<0.05(*).

Mean trait values over all %variation Plant trait % variation within sites sites between sites Flowering time 236.2 55.40*** 44.60 Number of leaves 3.4 18.87** 81.13 Plant height 23.4 6.92* 93.08 Plant width 41.1 1.44* 98.56 Ratio of leaf length to 2.3 0.00n.s 100.00 width Leaf area 35.9 0.00n.s 100.00 Number of primary 8.1 20.69** 79.31 branches Pod length 38.9 41.61*** 58.39 Pod width 0.2 27.85*** 72.15 Number of pod 5.0 25.31** 74.69 segments 100 Seed weight 0.6 77.69*** 22.31 10 pods weight 1.5 67.51*** 32.49

Table 4.4. Component loadings for life history traits in the canonical variate analysis, as well as the amount of variation accounted for by the first and second canonical variants (CV1 and CV2).

Plant traits CV1 CV2 % Variation 84.5 14.0 Number of primary branches -0.421 -0.148 Flowering time -0.064 -0.057 Plant height -0.04 0.094 Ratio of leaf length to width 1.19 1.153 Leaf area 0.014 0.021 Number o. of leaves 1.107 0.059 Number of pod segments 0.862 -0.221 Pod width 8.028 -2.332 Pod length -0.052 0.008 10 pods weight 2.205 0.261 100 seed weigh 2.086 -7.133 Plant width -0.006 -0.007

4 Merredin Nukarni 3 Mullewa 2 Denmark

1 CV1 0 -8 -6 -4 -2 0 2 4 6 8

-1

-2

-3 CV2

-4

Figure 4.2 A plot of canonical variate 1 (CV1) against canonical variate 2 (CV2) for each of the plants in which life history traits were measured. Plants with similar symbols originate from the same site as indicated on the figure (Merredin, Nukarni, Mullewa and Denmark).

AFLP molecular markers

The combination EcoRlAAC/MseI CAA was the best primer combination. This showed the clearest bands in all four populations resulting in 112 scoreable bands at Merredin, 82 at Nukarni, 100 at Mullewa, and 110 at Denmark.

Table 4.5 The number of polymorphic sites, the number of loci recorded, the percentage of polymorphic loci, the average gene diversity and the apportionment of variation between and within sites of wild radish populations collected at Merredin, Nukarni, Mullewa and Denmark

Merredin Nukarni Mullewa Denmark No. of polymorphic sites 110 70 85 97 No. of loci recorded 112 112 112 112 No. of usable loci 111 83 91 110 Average gene diversity1 0.39 ± 0.20 0.35 ± 0.18 0.36 ± 0.19 0.32 ± 0.16 % variation between sites2 28.3 % variation within sites2 71.7 1Probability that two randomly chosen homologous nucleotides are different (± LSD). 2Measured by AMOVA. Table 4.5 shows the results of the gene diversity indices and the analysis of molecular variance (AMOVA) for each population. The results indicate that up to 99% of the loci were polymorphic with average gene diversity ranging from 0.32 ± 0.16 at Denmark to 0.39 ± 0.20 at Merredin. Overall, 72% of the variation was found to be within sites and 28% between sites.

Figure 4.3. Dendrogram from cluster analysis of wild radish (Raphanus raphanistrum) populations analysed for genetic variation using AFLP technique.

The dendrogram of the cluster analysis is shown in Figure 4.3, where it can be seen that almost all individuals group at the site levels. The majority of the population from Mullewa separated from the majority of the Denmark population at the 56% similarity level (but one of the Mullewa population differed from the remainder at the 52% level and two individuals from the Denmark population separated from the remainder at the 49% level); the population from Merredin differed from the Mullewa and Denmark populations at the 48% similarity level, and the Nukarni population differed from all other populations at the 40% similarity level. One individual from Merredin separated from the Nukarni population at the 47% level and from the remainder of the Merredin population at the 40% level. In other words the cluster analysis revealed that the two most closely related populations are from Denmark and Mullewa, and the population from Nukarni is the least closely related to the others.

In general, the principal coordinates analysis supports the cluster analysis (Figure 4.4). The first three coordinate loadings contain respectively, PC1: 22.8% of the total variation PC2: 12.4% of the total variation and PC3: 10.4%. Individuals are grouped according to their sites of origins, Nukarni is genetically the most distinct; and Merredin and Mullewa are genetically the closest. In contrast to the cluster analysis the principal coordinate analysis separates the Denmark population more clearly from the Merredin population. However, one individual from Denmark lies within the Merredin population.

0.4

Mullewa 0.3 Nukarni Denmark 0.2 Merredin 0.1 1 PC 0 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4

-0.1

-0.2

-0.3

-0.4

PC2 -0.5

Figure 4.4 A plot of principal coordinates 1 (PC1) against principal coordinate 2 (PC2) for each of the plants in which DNA was assayed using AFLP markers. Plants with similar symbols originate from the same site as indicated on the Fig. (Mullewa, Nukarni, Denmark and Merredin).

4.5 Discussion

It was hypothesized that populations from nearby sites (eg Merredin and Nukarni) would show greater similarity than more widely separated sites, and that populations from environmentally similar sites (eg Mullewa and Merredin) would show greater similarity than populations from dissimilar sites (Merredin and Denmark). The results from the life history study support the hypothesis, showing that environment is acting strongly on these traits, especially seed and pod size. The genetic markers, although separating the biotypes, did not support the hypothesis that the greatest differences would arise from populations separated most in terms of space and environment. The explanation may lie in the random nature of genes selected during the DNA analysis compared with the strong selective pressures acting on the life history traits.

The distinctive molecular patterns of the populations from Merredin and Nukarni (Fig. 4.3 and 4.4) suggest that they have different origins and were introduced to the central wheat

79 belt on separate and unrelated occasions. Such an explanation of the results agrees with the interpretation of Gladstones (1966), who used leaf markers to hypothesise that there were a number of separate introductions of subterranean clover to Western Australia early in the colony’s history. Leaf markers are unlikely to possess any adaptive value and will therefore behave in a similar way to the molecular markers used here. Gladstones’ results demonstrate that subterranean clover was introduced from different parts of mediterranean and northern Europe in the 50 years leading to the building of the railway network in the wheatbelt. Evolution of subterranean clover since that time appears to have been through hybridisation and natural selection, with relatively few leaf markers predominating, but with the distinctly different ecophysiological behaviour of selected genotypes strongly related to environmental conditions at the sites of collection (Cocks et al. 1982; Cocks and Phillips 1979)

Annual pasture plants and weeds are common in southern Australia, and this fact itself indicates that there have been repeated introductions. For example, at least two species of barley grass (Hordeum leporinum and H. glaucum) are present in different parts of South Australia (Cocks et al. 1976). These barley grasses come from different parts of the mediterranean basin. Capeweed (Arctotheca calendula), which is found throughout southern Australia (Arnold et al. 1985), is a native of South Africa. Numerous annual legumes, mainly from the mediterranean basin, are also found in southern Australia. It is inconceivable that these are the results of single introductions.

It is widely recognized that the majority of life history traits expressed by plants are adaptive and thus acted upon by natural selection (Barrett 1982; Barrett 1988; Lande 1977; Marshall and Allard 1970). Though the adaptive significance of a polymorphism may be unclear in some cases, a more thorough investigation often reveals its significance. For example, New (1958) studied Spergula arvensis in the United Kingdom and demonstrated variation in seed coat morphology between populations, the significance of which was initially unclear. However, New and Herriot (1981) demonstrated that a particular seed coat morphology has direct adaptive value in allowing seeds to germinate under moisture regimes typical of the areas in which they exist. Thus the patterns of variation found in life history traits are likely to be the result of adaptation to prevailing environmental conditions.

The results presented here clearly demonstrate that adaptive evolution has occurred since wild radish was introduced to Western Australia. Thus the genetically dissimilar populations at Nukarni and Merredin have converged to form similar populations in terms of life history traits, while the genetically more similar populations at Denmark and Mullewa have diverged dramatically in terms of their life history traits. Evolution has been a potent force on these populations since their introduction, although insufficient time has elapsed for this to be expressed in terms of molecular markers, where genetic drift rather than selection would act to differentiate the populations.

Evolution has resulted in ecotypic differentiation of many other annual species introduced to Australia since early in the 19th century. These include Trifolium glomeratum (Woodward and Morley 1974), Hordeum leporinum and H. glaucum (Cocks et al. 1976), Trifolium subterraneum (Cocks and Phillips 1979), Artotheca calendula (Dunbabin 2001) and others. In the case of Trifolium subterraneum, differentiation occurred over 30 years or less (Cocks 1992), with the suggestion that, where hybridisation is possible (this species is mainly self fertilizing), ecotypic differentiation takes place almost immediately. It is therefore not surprising that wild radish shows such strong differentiation in populations as distant as that of Denmark from the wheat belt populations.

Among most of these annual species, one of the most commonly demonstrated ecotypic differences between populations is for flowering time (Bennett 1997; Cocks and Phillips 1979; Ehrman and Cocks 1996; Woodward and Morley 1974). At a site where the growing season is short, flowering and seed set must occur early to ensure seed is set before the dry summer begins. Where the season is longer, a greater vegetative period provides the maximum opportunity for dry matter accumulation and hence potential for the largest seed set. In the present study wild radish from areas with high rainfall and low temperature flowers later than strains from low rainfall, high temperature areas. These aspects of the results were also highlighted in Chapter 3 and will be discussed further in the General Discussion (Chapter 7)

The results of both approaches used here (AFLPs and life history traits) indicate that the amount of within population variation is much greater than between populations. Kercher and Conner (1996) studied isozymes in wild radish and also found that the amount of within population variation was much greater than between populations. It is likely that 81 out-crossing species will contain significantly more within and less between population variation than inbreeding species (Levin 1977). For example, an isozyme study of the outbreeding colonizer Silene vulgaris by Runyeon and Prentice (1996) found that only 2% of the genetic variation occurred between populations. In another study, Sun (1997) found that Centaurea solstitialis in North America exhibited only 1% of its variation between populations. Other outbreeders which exhibit this arrangement of genetic diversity include weedy rye (Sun and Corke 1992), and Echium plantagineum (Barrett 1982).

Variation in the genetic structure of plant populations is the result of a balance between gene flow, genetic drift and selection pressure (Hayward et al. 1997). Selection pressure is expressed in plant life history traits, which are of direct survival value to the plant (Barrett 1982; Barrett 1988; Lande 1977; Marshall and Allard 1970). This helps balance the high gene flow between the outcrossing wild radish populations, reducing within population variation and increasing between population variation. However, AFLPs and other molecular markers are assumed to be selectively neutral or nearly so (Hartl and Seefelder 1998; Isabel et al. 1995; Nissen et al. 1995; Powell 1983; Spitze 1993), so their distribution will be determined largely by gene flow and drift. The high within population variation of wild radish populations is the result of high levels of gene flow, which also limits between population differentiation. There is little or no selection acting to balance this high gene flow.

Although the assessment of a large and random fraction of the genome is desirable in some circumstances (Petersen and Seberg 1998; Westman and Kresovich 1997), the random nature of the markers under examination is a distinct disadvantage in assessing the adaptation of a species to its environment. Unlike life history traits, the majority of DNA polymorphisms (Hartl and Seefelder 1998; Nissen et al. 1995) and proably isozymes (Isabel et al. 1995; Powell 1983; Spitze 1993) are selectively neutral. Thus, the patterns of genetic diversity they reveal may not be the result of adaptation of the species to its environment, but may instead reflect random processes associated with gene flow and genetic drift (Hurka 1990). That there are often differences in the patterns of genetic variation exhibited by molecular markers and life history traits (Moran et al. 1981; Schut 1997; Warwick and Black 1986) reflects the difference in forces controlling these traits (Knapp and Rice 1998; Venable et al. 1998).

4.6 References Akkaya MS, Bhagwat AA, Cregan PB (1992) Length polymorphism of simple sequence repeat DNA in soyabean. Genetics 132, 1131-1139.

Allard RW (1965) Genetic systems associated with colonizing ability in predominantly self-pollinating species. In 'Proceedings of the First International Union of Biological Sciences Symposium on General Biology - The Genetics of Colonising Species', pages 49-75. (Ácademic Press: New York)

Allard RW (1975) The mating system and microevolution. Genetics 79, 115-126.

Arnold GW, Ozanne PG, Galbraith KA (1985) The capeweed content of pastures in south- west Western Australia. Australian Journal of Experimental Agriculture 25, 117- 123.

Barrett SCH (1982) Genetic variation in weeds. In 'Biological Control of Weeds with Plant Pathogens.' (John Wiley and Sons: New York)

Barrett SCH (1988) Genetics and evolution of agricultural weeds. In ‘Weed Management in Agroecosystems: Ecological Approaches (Ed. Alteri.)’, pages 57-76. (CRC Press: Baton Rouge, Florida, USA)

Bennett SJ (1997) Genetic variation between and within two populations of Trifolium glomeratum (cluster clover) in Western Australia. Australian Journal of Agricultural Research 48, 969-976.

Burdon J, Marshall D, Groves R (1980) Isozyme variation in Chondrilla juncea L. in Australia. Australian Journal of Botany 28, 193-198.

Cheam AH (1986) Seed production and seed dormancy in wild radish Raphanus raphanistrum and some possibilities for improving control. Weed Research 26, 405- 414.

Chomczynski P, Mackey K, Drews R, Wilfinger W (1997) DNAzol: a reagent for the rapid isolation of genomic DNA. Biotechniques 22, 550-553.

Cocks PS (1992) Evolution in sown populations of subterranean clover Trifolium subterraneum L. in South Australia. Australian Journal of Agricultural Research 43, 1583-1595.

Cocks PS, Boyce KG, Kloot PM (1976) The Hordeum murinum complex in Australia. Australian Journal of Botany 24, 651-662.

Cocks PS, Phillips JR (1979) Evolution of subterranean clover in South Australia. 1. The strains and their distribution. Australian Journal of Agricultural Research 30, 1035- 1052.

Committee G (1998) Genstat 5 v. 4.1. Lawes Agricultural Trust, Rothampstead, UK.

Doyle J, Doyle JL (1987) A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19, 11-15.

Dunbabin MT, Cocks PS (1999) Ecotypic variation for seed dormancy contributes to the success of capeweed (Arctotheca calendula) in Western Australia. Australian Journal of Agricultural Research 50, 1451-1458.

Dunbabin MT (2001) Genetic variation in the outbreeding coloniser capeweed in south- western Australia. PhD thesis, University of Western Australia.

Ehrman T, Cocks PS (1996) Reproductive patterns in annual legume species on an aridity gradient. Vegetatio 122, 47-59.

Groves RH (1975) Growth and development of five populations of Themeda australis in response to temperature. Australian Journal of Botany 23, 951-963.

Groves RH, Hagon MW, Ramakrishnan PS (1982) Dormancy and germination of seed of eight populations of Themeda australis. Australian Journal of Botany 30, 373-386.

Hartl L, Seefelder S (1998) Diversity of selected hop cultivars detected by fluorescent AFLPs. Theoretical and Applied Genetics 96, 112-116.

Hayward MD, Sackville-Hamilton NR (1997) Genetic diversity, population structure and conservation. In 'Biotechnology and Plant Genetic Resources.’ (CAB International: New York)

Hurka H (1990) 'Differentiation and adaptation in the genus Capsella (Brassicaceae). In 'Biological Approaches and Evolutionary Trends in Plants.' (Academic Press: London)

Isabel N, Beaulieu J, Bousquet J (1995) Complete congruence between gene diversity estimates derived from genotypic data at enzyme and random amplified polymorphic DNA loci in black spruce. In 'Proceedings of the National Academy of Sciences of the United Stated of America', pages 6369-6373

Kercher S, Conner JK (1996) Patterns of genetic variability within and among populations of wild radish, Raphanus raphanistrum (Brassicaceae). American Journal of Botany 83, 1416-1421.

Kim CS, Lee CH, Shin JS, Chung YS, Hyung NI (1997) A simple and rapid method for isolation of high quality genomic DNA from fruit trees and conifers using PVP. Nucleic Acids Research 25, 1085-1086.

Knapp EE, Rice KJ (1998) Comparison of isozymes and quantitative traits for evaluating patterns of genetic variation in purple needlegrass (Nassella pulchra). Conservation Biology 12, 1031-1041.

Kon KF, Blacklow WM (1989) Identification, distribution and population variability of great brome Bromus diandrus Roth and rigid brome Bromus rigidus Roth. Australian Journal of Agricultural Research 39, 1039-1050.

Lande R (1977) The influence of the mating system on the maintenance of genetic variability in polygenic characters. Genetics 86, 485-498.

Levin DA (1977) The organisation of genetic variability in Phlox drummondii. Evolution 31, 477-494.

Liu BH (1998) Statistical Genomics: Linkage, Mapping and QTL Analysis. CRC, Boca Raton, Florida.

Marshall DR, Allard RW (1970) Isozyme polymorphisms in natural populations of Avena fatua and A. barbata. Heredity 25, 373-382.

Matthes MC, Daly A, Edwards KJ (1998) 'Amplified Fragment Length Polymorphism (AFLP).' (Chapman and Hall: London)

Moodie M, Finch RP, Marshall G (1997) Analysis of genetic variation in wild mustard (Sinapis arvensis) using molecular markers. Weed Science 45, 102-107.

Moran GF, Marshall DR, Muller WJ (1981) Phenotypic variation and plasticity in the colonizing species Xanthium strumarium L. (Noogoora Burr). Australian Journal of Biological Science 34, 639-648.

Neuffer B, Meyer-Walf M (1996) Ecotypic variation in relation to man made habitats in Capsella: field and trampling area. Flora (Jena) 191, 49-57.

New JK (1958) A population study of Spergula arvensis L. Annals of Botany 22, 457-477.

New JK, Herriot JC (1981) Moisture for germination as a factor affecting the distribution of the seedcoat morphs of Spergula arvensis L. Watsonia 13, 323-324.

Nissen SJ, Masters RA, Lee DJ, Row ML (1995) DNA-based marker systems to determine genetic diversity of weedy species and their application to biocontrol. Weed Research 43, 504-513.

Panetta FD (1990) Isozyme variation in Australian and South African populations of Emex australis Steinh. Australian Journal of Botany 38, 161-168.

Parkash CS (1998) 'AFLP Marker: A New Powerful Tool for Genome Analysis.' (Center for Plant Biotechnology Research: Tuskegee, University of Alabama).

Petersen G, Seberg O (1998) Molecules vs morphology. In 'Molecular Tools for Screening Biodiversity.' (Chapman and Hall: London)

Peterson HD, Thompson R (1971) Recovery of inter-block information when block sizes are unequal. Biometrika 58, 545-554.

Powell JR (1983) Molecular approaches to studying founder effects. In 'Genetics and Conservation. A Reference Manual for Managing Wild Animal and Plant Populations. (The Benjamin/Cummings Publishing Company: California)

Runyeon H, Prentice HC (1996) Genetic structure in the species pair Silene vulgaris and S. uniflora (Caryophyllaceae) on the Baltic island of Oland. Ecography 19, 181-193.

Schneider S, Rosseli D, Excoffier L (2000) A software for population genetics data analysis. (University of Geneva: Geneva Switzerland)

Schut JW, Qi X, Stam P (1997) Association between relationship measures based on AFLP markers, pedigree data and morphological traits in barley. Theoretical and Applied Genetics 95, 1161-1168.

Spitze K (1993) Population structure in Daphnia obtusa: quantitative genetic and allozyme variation. Genetics 135, 367-374.

Sun M (1997) Population genetic structure of yellow starthistle (Centaurea solstitialis), a colonizing weed in the western United States. Canadian Journal of Botany 75, 1470-1478.

Sun M, Corke H (1992) Population genetics of colonizing success of weedy rye in northern California. Theoretical and Applied Genetics 83, 321-329.

Tanksley SD, Young ND, Paterson AH, Bonierbal MW (1989) RFLP mapping in plant breeding: new tools for an old science. Biotechnology 7, 257-264.

Venable LD, Dyreson E, Pinero D, Becerra JX (1998) Seed morphologies and adaptive geographic differentiation. Evolution 52, 344-354.

Vos P, Hogers R, Bleeker M, Reijians M, Van de Lee, Hornes T, Frijters M, Pot AJ, Peleman J, Kuiper M, Zabeau M (1995) AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research 23, 4407-4414.

Warwick SI, Black LD (1986) Genecological variation in recently established populations of Abutilon theophrasti (velvetleaf). Canadian Journal of Botany 64, 1632-1643. 86

Westman AL, Kresovich S (1997) Use of molecular marker techniques for description of plant genetic variation. In 'Biotechnology and Plant Genetic Resources, Conservation and Use. (CAB International: New York)

Williams JGK, Kubelik AR, Livak KJ, Rafalski JA, Tingey SV (1990) DNA polymorphism amplified by arbitrary primers are useful as genetic markers. Nucleic Acids Research 18, 6531-6535.

Zheng QM, Lapitan NLV (1998) A comparison of amplified and restriction fragment length polymorphism in wheat. Cereal Research Communication 26, 7-13.

Chapter 5. Seed dormancy patterns contribute to the ecological success of wild radish (Raphanus raphanistrum L.) in Western Australia

5.1 Abstract

The seed dormancy status of two Western Australian wild radish populations - from Merredin (322 mm rainfall) and Bunbury (870 mm) - were investigated to determine the contribution of seed dormancy to ecological success. Short and long term dormancy patterns of buried seeds and seeds on the soil surface enclosed and not enclosed by pods, were examined using seed produced in a common environment.

Contrary to expectations, the two populations exhibited similar patterns of seed dormancy. On the soil surface the average number of dormant seeds in the first year was 73% from Merredin and 74% from Bunbury. In the second year, 48% of Merredin seeds remained dormant and 36% of the Bunbury seeds. Removal of the seeds from the pods significantly reduced dormancy: 97% of seeds in pods were dormant in the first year compared with only 25% of naked seeds. By June of the second year, 60% of the seeds in pods remained dormant while only 5% of those that had been removed from pods remained dormant. Burial increased germination of naked seeds but had little effect on dormancy. Cycling of dormancy was observed in both populations, particularly for seeds from the Merredin population.

These features of the dormancy of wild radish seeds may make this species more resilient to seasonal variability and help explain why wild radish is so widespread and abundant across southern Australia.

5.2 Introduction

Wild radish is a widespread weed throughout the world (Parsons and Cuthbertson 1992), including in the Mediterranean cropping lands of southern Australia. It is known to compete strongly with wheat (Code et al. 1978; Poole and Gill 1987), reducing grain yields, contaminating harvested grain, and making harvesting more difficult. Control of wild radish in canola is difficult, because of its close botanical relationship to canola and the lack of herbicide alternatives (currently, atrazine tolerant canola is available and it is widely used for the control wild radish. Genetically manipulated canola may become available in

88 the future) (Byrdwell and Neff 2001; Dunfield and Germida 2001; James et al. 2003; Petukhov et al. 1999a;1999b)

In Mediterranean areas a successfully naturalised annual species adapted to mediterranean environments needs to survive the seasonal summer stress in a dormant state. Having done so the seeds need to germinate when growing conditions become favourable in late autumn and winter. So far our understanding of the role of seed dormancy and how it contributes to the success of wild radish is limited to a single paper by Cheam (1986), although there are a number of reported studies on seed dormancy mechanisms in other annual species (Dunbabin and Cocks 1999; Gill and Blacklow 1985; Norman et al. 2002; Revell et al. 1999; Taylor 1996a;Taylor 1996b; Taylor and Ewing 1996). In wild radish, attributes contributing to persistence are flexible germination, high reproductive capacity, long seed life and appropriate dormancy (Cheam and Code 1995; Holm et al. 1977). The proportion of wild radish germinating from the seed bank between seed dispersal and the break of the season in autumn may vary from 10 to 30% (Cheam and Code 1995) depending on depth of seed burial (Code et al. 1987), diurnal temperature fluctuations and soil moisture (Piggin et al. 1978). Consequently, a large proportion of seed remains dormant within the soil, providing the primary source of new infestations each year. Further study is needed to understand the pattern and duration of seed dormancy in the field.

The aim of the series of experiments reported in this chapter was to investigate the within and between years changes in the seed dormancy patterns of wild radish. Dormancy of two populations of wild radish was compared to determine if there is ecotypic differentiation in seed dormancy and, if so, how these differences are related to the environment from which the population originated. Dormancy of buried seeds and seeds on the soil surface were compared, and the effect of temperature on germination of seeds protected and unprotected by the pods was measured. Importantly, the results of all experiments are integrated to provide an overview of the complete seed dormancy strategy of wild radish and how this contributes to the success of the species as a weed.

Two hypotheses were tested:

1) that populations from lower rainfall environments produce seeds that have deeper and more prolonged dormancy than seeds from higher rainfall populations, and

2) that dormancy in wild radish seeds is imposed by the seed pods.

5.3 Materials and methods

Wild radish seeds were collected from two contrasting regions of Western Australia before the pods fell on the ground in November 1998. The first site was near Bunbury (320 45’40”S, 1150 53’11”E, rainfall 870 mm), a high rainfall coastal environment, and the second site was near Merredin (310 29’14”S, 1180 12’52”E, rainfall 322 mm), typical of the eastern wheatbelt of Western Australia. These sites were chosen to cover the widespread distribution of wild radish in Western Australia and to represent the extremes of climatic zones where wild radish is found.

To eliminate possible environmental effects, the experimental seeds were produced at the same location: the University of Western Australia Field Station at Shenton Park. In May 1999, 50 m2 plots of each population were established. Each plot contained 100 plants, the space between each plant was 40 cm, and surrounding each plot were two rows of canola to limit pollen transfer between populations. All experimental plants were marked, and any weeds, wild radish or otherwise, were hand weeded. The plots were well fertilised and watered when necessary. Wild radish plants were eliminated from the vicinity of the plots. In early December whole plants from the centre of each plot were dried and threshed. Plants from the outside rows were not harvested to further reduce the opportunity for cross- pollination. After threshing, the seeds were thoroughly mixed and stored in paper bags until the start of each experiment. There were three replicates.

After harvesting, the seeds were tested for viability using the tetrazolium test.

Experiment 1: changes in seed dormancy of the two populations over two years

Pods obtained from each replicate of each population were placed inside a fine mesh nylon packet and secured to the soil surface under a rain protection (fibre glass) sheet at the Field Station. There were 100 pods in each packet.

Samples of the Bunbury and Merredin populations were taken each month from December to June each year; in December, January, February and June at approximately 30 days intervals and March, April, May at 15 days intervals. No samples were taken between June and December. Sampling commenced on December 1st 1999.

Three replicates of 100 seeds from each packet (50 x 50 mm) were placed onto filter paper in a 90 mm Petri dish, moistened with deionised water and placed in an incubator at 150C. Germinated seeds were counted and removed every day for 14 days. After 14 days, all un- 90 germinated seeds were tested for viability, using tetrazolium (Ellis et al. 1985). The threshed and unthreshed seeds were cut to ensure the penetration of the 1% tetrazolium solution and the seeds were stained for 3 hours at 300C. All seeds that reacted with the tetrazolium were regarded as viable.

The germination status of each seed in the sample was classified as either a) dead – the seed had died in the field and failed to pass the tetrazolium test, b) dormant – the seed failed to germinate during the germination test but was proven to be viable after the tetrazolium test, or c) germinable – germinated during the germination test.

Experiment 2: effect of pods on seed dormancy

Seeds from the Merredin population were used in this experiment and the following treatments applied:

1. Un-threshed pods were placed on the soil surface

2. Pods were threshed and the naked seeds were placed on the soil surface

3. Un-threshed pods were buried 2 cm in the soil

4. Naked seeds were buried 2 cm in the soil

Each replicate was placed inside a fine mesh nylon packet (see experiment 1) and either pegged to the surface or buried at the designated depth.

Germination was tested each year in January and June. One hundred seeds from randomly selected packets in each replicate and treatment were counted and the seeds placed onto filter paper in a 90 mm Petri dish where they were moistened with deionised water. Seeds were germinated at 150C and the germinated seeds counted and removed every day for14 days. After 14 days, all un-germinated seeds were tested for viability using a 1% solution of tetrazolium (Ellis et al. 1985).

The seeds were classified as dead, dormant or germinable as described for experiment 1.

Experiment 3: effect of temperature on germination of fresh seeds

Seeds collected from Merredin were used directly to test the response of germination to temperature. Three replicates of 100 seeds were counted and placed on filter paper in a 90

91 mm Petri dish where they were moistened with deionised water. A range of different temperatures were imposed: 5, 11, 15, 20, 25, 30, and 350C. Germinated seeds were counted and removed every day for14 days. After 14 days, all un-germinated seeds were tested for viability using a 1% tetrazolium solution as before (Ellis et al. 1985). Both naked seeds and un-threshed pods were used in the experiment as separate treatments.

Experiment 4: pattern of seedling establishment

In nature the pods of wild radish fall directly on to the soil surface from where, when conditions are right, they germinate and produce seedlings. This experiment examines this under experimental conditions at Shenton Park.

The pods used were from the Merredin population established at Shenton Park. Six replicates of 100 pods were spread in plots (3 x 1 m) in two separate blocks, originally established to provide two watering treatments. When this became impossible (due to early rainfall), the experiment was continued with 12 replicates. The pods were placed onto bare soil surface and the number of germinated seedlings were counted every fortnight. After counting, all germinated seedlings were removed.

Statistical tests

Microsoft Excel was used to calculate means and standard errors of dead, dormant and germinable seeds from each observation. Rate of germination was calculated as the number of days to 50% germination at each temperature.

5.4 Results

Experiment 1: changes in seed dormancy of the two populations over two years

The seeds from both sites were 100% dormant and 100% viable immediately after harvest in early December.

By December 30, 1999, 92% of the Merredin (Fig. 5.1) seed population and 83% of the Bunbury population (Fig. 5.2) remained dormant on the soil surface. From January, dormancy of the Merredin population fell from 98% to a low point of 63% in March and dormancy of the Bunbury population from 88% in January to a low point of 62% in May. Average dormancy during the first year was 73% at Merredin with 20% germinated and 7% dead and 74% at Bunbury, with 15% germinated and 10% dead.

Merredin Dormant Dead Germinable

100

Seed (%) 40

9 0 /00 /00 /00 /0 /00 /00 /00 /00 /00 00 /01 /01 /01 /01 /01 /01 5 3 0/4 1/3/01 1/4 1/5/01 1/6/01 30/1 29/2 15/3 30/3 15/4 3 15/5 30/ 15/6 15/1 15/2 15/ 15/4 15/5 30/12/9 18/12/ Date of Sampling

Figure 5.1. The percentage of germinable seeds (top), dormant seeds (bottom) and dead seeds (middle) in the un-threshed wild radish population of pods from Merredin when placed on the soil surface. Sampling commenced on 1 December 1999, when 100% of all seeds were dormant, and finished on 1st June 2001.

Bunbury Dormant Dead Germinable

100

50 Seed (%) 40

0 0 1 1 /00 /0 /00 /00 /00 /0 /00 /00 /00 /0 /01 /01 /0 /01 /3 6 3 /5 5/3 0/5 5/2 1/3/01 1/4/01 1/5/01 1/6/01 30/1 29/2 1 30 15/4 30/4 15/5 3 15/ 15/1 1 15/ 15/4 15 30/12/99 18/12/00 Date of Sampling

Figure 5.2. The percentage of germinable seeds (top), dormant seeds (bottom) and dead seeds (middle) in the un-threshed wild radish population of pods from Bunbury when placed on the soil surface. Sampling commenced on 1 December 1999, when 100% of all seeds were dormant, and finished on 1st June 2001.

100 First year observation Second year observation Merredin 90

80 Bunbury

Germinable seeds % 30

0 0 0 0 1 1 /99 /00 /00 /0 /00 /00 /0 /00 /00 /0 /01 /0 /01 /01 01 /01 0 /01 /01 2 /1 /2 /3 3 /4 /4 5 /5 /6 2/0 /2 /3 /4/ /5/ /5 0 9 5 0/ 5 0 5/ 5 1 5/1 5 1/3 1 5/4 1 5 1/6 0/1 3 2 1 3 1 3 1 30 1 1 1 15 1 1 3 18/ Date of Sampling

Figure 5.3 Germinability expressed as a percentage of the viable seed populations from Merredin (solid line) and Bunbury (dotted line) of un-threshed wild radish seed on the soil surface. Vertical bars associated with the amount of dormant seed represents ± 1 s.e.

During the second year the maximum dormancy of the Merredin population was 61%, recorded in February 2001 and the minimum of 31% was recorded in December 2000. Dormancy of the Bunbury population was at its highest in December 2000 (56%) and lowest in April 2001 (19%). Average dormancy during the second year was 48% at Merredin, with 13% germinable and 39% dead, and 36% at Bunbury, with 11% germinable and 53% dead. Germinability expressed as a percentage (Fig. 5.3) of the viable seed populations from Merredin and Bunbury of unthreshed wild radish seed on the soil surface.

Experiment 2: effect of pods on seed dormancy

Germination of un-threshed seeds Fig. 5.4 shows that in January 2000, 97% of un-threshed seeds in the pods remained dormant when placed on the soil surface. By June 2000 germinability had increased to 20%, but it fell slightly by January 2001 and remained below the first year high at least until June.

Unthreshed seeds (Surface) Dead Dormant 100% Germinable

80%

60%

40% % of seeds 20%

0% 15/01/00 15/6/00 15/01/01 15/6/01 Date of Sampling

Figure 5.4. The percentage of germinable seeds, dormant seeds and dead seeds at four sampling dates between 15th January 2000 and 15th June 2001. The seeds were in un- threshed pods lying on the soil surface and were from the Merredin population.

Dead Unthreshed seeds (Buried) Dormant

100% Germinable 80%

60%

40% % of Seeds 20%

0% 15/01/00 15/6/00 15/01/01 15/6/01 Date of Sampling

Figure 5.5. The percentage of germinable seeds, dormant seeds and dead seeds at four sampling dates between 15th January 2000 and 15th June 2001. The seeds were in un- threshed pods buried 2 cm beneath the soil surface and were from the Merredin population.

The results in Figs 5.4 and 5.5 show that dormancy of un-threshed seeds is similar regardless of whether the seeds were on the surface or buried. The results suggest that a greater number of the buried seeds may have died.

Germination of threshed seeds 95

Threshed seeds (Surface) Dead Dormant Germinable 100%

80%

60%

40% of seeds % 20%

0% 15/01/00 15/6/00 15/01/01 15/6/01 Date of Sampling

Figure 5.6 The percentage of germinable seeds, dormant seeds and dead seeds at four sampling dates between 15th January 2000 and 15th June 2001. The threshed seeds were lying on the soil surface and were from the Merredin population.

Dead Threshed seeds (Buried) Dormant

100% Germinable 80% 60%

40% of Seeds % 20% 0% 15/01/00 15/6/00 15/01/01 15/6/01 Date of Sampling

Figure 5.7 The percentage of germinable seeds, dormant seeds and dead seeds at four sampling dates between 15th January 2000 and 15th June 2001. The threshed seeds were buried 2 cm beneath the soil surface and were from the Merredin population.

Fig. 5.6 shows that give a 16% of the threshed seeds lying on the soil surface remained germinable in the first year. By the second year viability had decreased to the extent that

96 there were no germinable seeds remaining in June. A small amount of seed remained dormant. In contrast, buried seed remained viable and non dormant over the experimental period (Fig. 5.7).

Means and standard errors of all treatments are in Table 5.1.

Table 5.1. Date of sampling, and means and standard deviation of germinable, dormant, and dead threshed and naked seeds buried 2 cm beneath the soil surface and lying on the soil surface.

Threshed seeds Unthreshed seeds Date of Treatment Germinable Dormant Germinable Dormant Dead seed Dead seed sampling seed seed seed seed

th 15 Jan 00 Buried 1.7±2.1 94.7±3.5 3.7±1.5 80.7±9.7 14.7±9.3 4.7±0.6 st 1 Jun 00 Buried 11.0±6.2 83.3±7.6 5.7±1.5 93.7±4.9 4.7±2.9 1.7±2.1 th 15 Jan 01 Buried 4.0±2.0 74.7±8.6 21.3±3.5 69.7±8.5 12.0±7.8 18.3±4.5 st 1 Jun 01 Buried 13±2.6 54.3±3.1 32.7±4.5 70.7±9.6 14.3±4.7 15.0±5.0 th 15 Jan 00 Surface 58.7±21.6 29.3±14.7 12.0±6.9 0±0 97.7±0.6 2.3±0.6 st 1 Jun 00 Surface 57.0±16.6 38.3±16.1 4.7±0.6 18.7±6.7 75.0±7.0 6.3±0.6 th 15 Jan 01 Surface 18.3±9.3 15.7±1.2 66.0±8.9 6.7±3.5 52.7±6.4 40.7±4.0 st 1 Jun 01 Surface 0.3±0.6 6.3±1.5 93.7±1.5 13.3±9.5 47.7±6.8 39.0±16.1

Experiment 3: Effect of different temperatures on germination of threshed and un- threshed seeds

Temperature affected both the amount and the rate of germination. Maximum germination occurred for both threshed and un-threshed seeds at 150C (Fig. 5.8). There was no germination of un-threshed seeds at 5 0C or at 30 0C or above, although some of the threshed seed germinated even at these temperatures. More threshed seeds germinated at all temperatures than did un-threshed seeds. Similarly the rate of germination for both threshed and un-threshed seeds, peaking at 15 0C or higher for both threshed and un-threshed (Fig. 5.9)

Germination rate was higher for un-threshed seeds than for threshed seeds. However the seeds remained viable indicating that unfavourable temperatures for germination resulted in secondary dormancy. The difference between threshed and unthreshed seeds was highest at the extreme temperatures of 50C and 350C.

100 Threshed seeds

90 Unthreshed seeds

40 Germinable seeds % 30

0 5 111520253035

Temperature (0C)

Figure 5.8 Germination of threshed and un-threshed wild radish seed after 14 days at temperatures ranging from 5-35 0C. The seed used was from the Merredin population. Vertical bars represent ± 1 s.e.

12 Threshed seeds

Unthreshed seeds 10

6 50% germinable seeds o

Days t 4

0 5 111520253035 Germination temperature (0C)

Figure 5.9 Rate (expressed as time to reach 50% of final germination) at which threshed and un-threshed seed of wild radish germinated at temperatures between 5 and 35 0C. The seed was from the Merredin population. Vertical bars represent ± 1 s.e.

Experiment 4: Pattern of seedling establishment

There was some germination following light rains in early March (Fig. 5.10). However, most germination occurred between late April and mid May following heavy rains in the second half of April. Germination continued in June, July and August at relatively low levels.

Weekly rainfall

100 Cumulative seedling emergence 140 90 120 80 70 100

60 80 50 40 60 30 40 20 Total weekly rainfall (mm) 20

10 Cumulative number of seedlings 0 0

0 00 /00 /00 /00 00 /00 00 /00 00 /0 /00 2/ 3 3 4/ 5/ 6/ 6 7 3/ 4/ 3/ 1/ 6/5 1/ 1/ 18/2 19/ 18/4 1 16/ Date of sampling

Figure 5.10. Total weekly rainfall and cumulative number of seedling emergence at the experimental site, Shenton Park, WA.

5.5 Discussion

The first, and perhaps most important aspect of the results, is that wild radish has an effective and long lasting dormancy mechanism. In this it is similar to the successful exotic annual legumes, most of which have dormancy mechanisms lasting several years. In the case of these legumes, the mechanism is located in the seed coat, and is generally referred to as hardseededness (Taylor 1981; Taylor 1996a; Taylor 1996b; Taylor and Ewing 1996). In its dormant state the seed coat is impermeable to water and appears to become permeable in response to firstly, high temperatures and secondly alternating temperatures. Legume seeds can remain dormant for several years (Taylor and Ewing 1988).

The present study does not indicate the longevity of seed dormancy, although the rate of seed death suggests that the seeds will not live as long as those of the annual legumes. Code

99 et al. (1987) however, conducted an experiment where pods were buried at 0-10 cm and emergence was observed for six years. Their results indicate that 93% of the plants emerging over six years did so in the first year, and a further 3.6% emerged in the second year. Very few seedlings emerged from the third year onwards, indicating that, to all intents and purposes the seed of wild radish are short lived. When using a technique similar to the one used here (a rain out shelter) however, Code and his colleagues observed that the seeds lived somewhat longer. Thus, in keeping with the results in Fig. 5.3, the seed population buried at 5 cm remained 50% dormant after two years. Indeed, 7% buried pods of the population was viable even after six years. It seems likely therefore, that in the absence of soil moisture wild radish is long lived but in its presence (the normal situation) the seed will maintain significant populations for only one year, which is much less than most hard seeded annual legume populations.

The results clearly show that the factor responsible for seed dormancy is located in the pods, and is removed once the seeds are threshed (Figs 5.6 to 5.9). This confirms the results of Mekenian and Willemsen (1975). Cheam (1986) also reported that dormancy of wild radish seed is controlled by the pod: in June Cheam’s populations 93% were dormant when the seeds were protected by pods and only 23% were dormant in the absence of pods. The former authors believe that dormancy is controlled by a non-leachable inhibitor present in the pod, which is broken down by high temperature. They also formed the opinion that dormancy may be the result of mechanical restriction by the pod, but this seems unlikely. It is possible that leachable inhibitors are also involved since they appear to control short term dormancy in several annual grasses, such as the annual species of Hordeum (Cocks and Donald 1973) and Bromus (Gill and Blacklow 1985). At least for Bromus, germination is stimulated by gibberellic acid.

In their review of the biology of wild radish Cheam and Code (1995) put forward strong evidence of genetic differentiation between populations in seed dormancy. They quote a study by Cheam (1986) which clearly shows that southern populations show greater dormancy than northern populations within the agricultural areas of Western Australia. Surprisingly, the southern populations are the more dormant, a result that contrasts with Dunbabin and Cocks (1999) who found that, for capeweed (Arctotheca calendula), the northern seed populations were the more dormant ones. Interestingly, two of Cheam’s wild

100 radish populations and both populations of Dunbabin and Cocks originated in the same two localities: at Chapman east of Geraldton and at Mount Barker.

The populations used here also originated from contrasting rainfall environments – 320 mm at Merredin and 870 mm at Bunbury, but the transect is east/west and not north/south. Although there are some differences between the Bunbury and Merredin seed populations, especially in the second year, these are likely to be associated with experimental error. For example, the increase in dead seeds in the Merredin population on two occasions in the second year of the experiment are associated with large standard errors and are clearly not real (Fig. 5.1). Fig. 5.3, where germinability is expressed as a percentage of the living seed gives a better picture. Here the seed reaches its maximum germinability rather later at Bunbury than it does at Merredin, a result which seems to support Cheam (1986).

If these results are to be believed, and there is no reason not to believe them, then wild radish behaves significantly differently from other annual species. Capeweed has already been mentioned but the best evidence comes from the annual legumes. Smith et al. (1996) examined the hard seed breakdown of two populations of Trifolium glomeratum – from the eastern wheat belt (accession 30) and from Busselton (accession 42), not far from Bunbury. When tested at Shenton Park, accession 42 lost all of its dormancy in one year, while accession 30 lost less than half, with 60% remaining dormant. Similar results have been obtained for subterranean clover (Gladstones 1966), and a wide range of annual legumes collected in Syria (Ehrman and Cocks 1996), where high levels of hardseededness were associated with aridity. However, in a study of 34 lines of M. polymorpha from different origins, Taylor (1996a) found that, although the incidence of autumn seed softening varies widely, it seems not to be associated with any particular geographical region.

The most comprehensive study of these issues is that of Norman et al. (2002) who examined, amongst several other reproductive traits, the residual hardseededness of six clover species (T. cherleri, T. glomeratum, T. hirtum, T. spumosum, T. subterraneum and T. tomentosum) after 6 months exposure on the soil surface. In all six species there were more seeds remaining hard in accessions from arid areas compared with accessions from wetter areas, although the differences were often not significant.

Figs 5.4 to 5.7 support earlier conclusions that germination is enhanced by shallow burial. (Cheam 1986; Code et al. 1987). In the present study, burial of un-threshed seeds appears

101 to increase dormancy and reduce seed death, while burial of naked seeds retains germinability. While this is in general agreement with other species, the effect of burial on the longevity of subterranean clover, for example, is powerful, extending the life of the seeds by several years (Taylor and Ewing 1992).

The optimum temperature for germination (15 0C) is rather less than that quoted by Piggin et al. (1978), probably because wild radish was found by these authors to respond to alternating temperatures.

There is one final point to make about technique. While great care was taken to ensure that the Merredin and Bunbury populations retained their genetic identity – rows of canola were grown around the edges of the plots, and seed samples were taken only from the centre – it is possible that the similarity in seed dormancy between the populations is due to cross pollination at Shenton Park. This is not considered likely but could explain the apparent differences between our results and those for other species recorded by other authors.

NOTE APPENDED AFTER THE COMPLETION OF THIS DISCUSSION

In a thesis completed at the University of Melbourne and which has recently been drawn to the author’s attention, Ken Young (2001) appears to put forward evidence that the dormancy of wild radish is controlled by the seed coat and not by the pod. Any disruption of the seed coat apparently releases dormancy even though dormancy itself is not caused by an impermeable seed coat. The role of the pod appears to be limited to ‘environmental control’, where the pod restricts the entry of water slowing the rate of emergence.

Young’s results depend heavily on a microscopic examination of the seed coat for evidence of disruption. Where disrupted, even if only slightly, germination was considerably enhanced.

The evidence of Cheam (1986), Mekenian and Willemsen (1975) and this thesis, that the pod is the source of dormancy, is cast into doubt by Young’s results. In all cases separation of the pods and seeds may have been accompanied by disruption of the seed coat, even though great care was taken to leave the seed coat intact. The question however, remains open until an explanation of the mechanism of seed dormancy is put forward. For example, it would appear unlikely that a leachable inhibitor is involved, since that would remain when the seed coat is disrupted.

Of interest is the fact that Young supports the view that cycling of dormancy takes place. 102

5.6 References

Byrdwell WC, Neff W (2001) Autoxidation products of normal and genetically modified canola oil varieties determined using liquid chromatography with mass spectrometric detection. Journal of Chromatography. A. 905, 85-102. Cheam AH (1986) Seed production and seed dormancy in wild radish (Raphanus raphanistrum) and some possibilities for improving control. Weed Research 26, 405-413. Cheam AH, Code GR (1995) The biology of Australian weeds. Wild radish (Raphanus raphanistrum L.). Plant Protection Quarterly 10, 2-13. Cocks PS, Donald CM (1973) The germination and establishment of two annual pasture grasses (Hordeum leporinum Link and Lolium rigidum Gaud.). Australian Journal of Agricultural Research 24, 1, 1-10. Code GR, Reeves TG, Brook HD, Piggin CM (1978) The herbicidal control of wild radish (Raphanus raphanistrum). In '1st Conference of the Council of Australian Weed Science Societies', Melbourne, pages 241-247 Code GR, Reeves TG, Gales BC (1987) The effect of various crop rotations on wild radish populations. In 'Proceedings of the 8th Australian Weeds Conference', pages 360- 363 Dunbabin MT, Cocks PS (1999) Ecotypic variation for seed dormancy contributes to the success of capeweed (Arctotheca calendula) in Western Australia. Australian Journal of Agricultural Research 50, 1451-1458. Dunfield KE, Germida JJ (2001) Diversity of bacterial communities in the rhizosphere and root interior of field-grown genetically modified Brassica napus. Microbiology Ecology 38, 1-9. Ehrman T, Cocks PS (1996) Reproductive patterns in annual legume species on an aridity gradient. Vegetatio 122, 47-59. Ellis RH, Hong TD, Roberts EH (1985) 'Handbook of Seed Technology for Genebanks.' (International Board for Plant Genetics: Rome) Gill GS, Blacklow WM (1985) Variations in seed dormancy and rates of development of great brome, Bromus diandrus Roth as adaptations to the climates of southern Australia and implications for weed control. Australian Journal of Agricultural Research 36, 295-304. Gladstones JS (1966) Naturalized subterranean clover (Trifolium subterraneum L.) in Western Australia: the strains, their distribution, characteristics and possible origins. Australian Journal of Botany 14, 329-354. Holm L, Plucknett D, Pancho J, Herberger J (1977) 'Distribution and Biology.' (University Press of Hawaii: Hawaii, USA) James D, Schmidt A-M, Wall E, Green M, Masri S (2003) Reliable detection and identification of genetically modified maize, soybean, and canola by multiplex PCR analysis. Journal of Agricultural & Food Chemistry 51 5829-5834. Mekenian MR, Willemsen RW (1975) Germination characteristics of Raphanus raphanistrum. I. Laboratory studies. Bulletin of the Torrey Botanical Club 102, 243- 103

252. Norman HC, Cocks PS, Galwey NW (2002) Hardseededness in annual clovers: variation between populations from wet and dry environments. Australian Journal of Agricultural Research 53, 821-829. Parsons WT, Cuthbertson EG (1992) 'Australian Weed Control Hand Book.' (Inkata Press: Sydney and Melbourne) Petukhov I, Malcolmson LJ, Przybylski R, Armstrong L (1999a) Frying performance of genetically modified canola oils. Journal of the American Oil Chemists Society 76, 627-632. Petukhov I, Malcolmson LJ, Przybylski R, Armstrong L (1999b) Storage stability of potato chips fried in genetically modified canola oils. Journal of the American Oil Chemists Society 76, 889-896. Piggin CM, Reeves TG, Brooke HD, Code GR (1978) Germination of wild radish (Raphanus raphanistrum). Proceeding of 1st Conference Council of Australia Weed Science Society, Melbourne, pages 233-240. Poole ML, Gill GS (1987) Competition between crops and weeds in southern Australia. Plant Protection Quarterly 2, 86-96. Revell CK, Taylor GB, Cocks PS (1999) Effect of length of growing season on development of hard seeds in yellow serradella and their subsequent softening at various depths of burial. Australian Journal of Agricultural Research 50, 1211- 1223. Smith FP, Cocks PS, Ewing MA (1996) Short term patterns of seed softening in Trifolium subterraneum, T. glomeratum and Medicago polymorpha. Australian Journal of Agricultural Research 47, 775-85. Taylor GB (1981) Effect of constant temperature treatments followed by fluctuating temperatures on the softening of hard seeds. Australian Journal of Plant Physiology 8, 547-558. Taylor GB (1996a) Effect of the environment in which seeds are grown and softened on the incidence of autumn seed softening in two species of annual medics. Australian Journal of Agricultural Research 47, 141-159. Taylor GB (1996b) Incidence and measurement of autumn seed softening within Medicago polymorpha L. Australian Journal of Agricultural Research 47, 575-586. Taylor GB, Ewing MA (1988) Effect of depth of burial on the longevity of hard seeds of subterranean clover and annual medics. Australian Journal of Experimental Agriculture 28, 77-81. Taylor GB, Ewing MA (1992) Long-term patterns of seed softening in some annual pasture legumes in a low rainfall environment. Australian Journal of Experimental Agriculture 32, 331-337. Taylor GB, Ewing MA (1996) Effects of extended (4-12 years) burial on seed softening in subterranean clover and annual medics. Australian Journal of Experimental Agriculture 36, 145-150. Young KR (2001) Germination and emergence of wild radish (Raphanus raphanistrum L.). PhD thesis, University of Melbourne.

104

Chapter 6. Genetic diversity in the composition of lipids in the seeds of south-western Australian populations of wild radish (Raphanus raphanistrum)

6.1 Abstract

The fatty acid composition of lipids was measured in wild radish (Raphanus raphanistrum) populations from Merredin (310 22’S, 1180 32’E), Nukarni (310 23’S, 1180 15’E), and Mullewa (280 33’S, 1150 29’E) in the Western Australian wheat belt and Denmark (340 58'S, 1170 39’E) in the high rainfall, south coast. Clean seeds from each site were directly sown in April 2000 at the University of Western Australia Field Station and harvested in November. Lipids from the whole seeds were extracted using a Soxhlet apparatus and the fatty acid composition determined by gas chromatography.

The total lipids (oil) content of the whole seeds ranged from 38.7 to 42.2%. Within site variation accounted for 80% and between site variation for 20% of the total variation. The major fatty acid elements in the seeds were erucic, oleic, linolenic, linoleic, eicosenic and palmitic acids. The fatty acids ranged from 8.9 to 10.3% saturated, 83.3 to 87.4% unsaturated and the ratio of saturated/unsaturated ranged from around 1:8 to1:10. Stearic acid showed the biggest variation between populations (44%).

The results indicate that there is little difference between sites in the total lipid content and fatty acid composition of wild radish, at least in Western Australia. However, only four populations were examined and the chances of finding genetic variation for these traits would be enhanced by a wider search. Nevertheless the results suggest that, in the 150 years since wild radish was introduced, there has been little change in these traits. Furthermore, it seems possible that the introduced germplasm showed relatively little diversity for lipid content and the composition of fatty acids. At least in south-western Australia the use of lipid and fatty acid analysis to determine genetic variation did not prove useful. It may however, prove useful in older populations or where greater diversity had been introduced in the first place.

The possible role of wild radish in the development of oil seed crops is discussed.

105

6.2 Introduction

Genetic variation in wild populations has been analyzed in a number of ways. The majority of studies used isozymes (Hamrick et al. 1979; Layton and Ganders 1984), morphological and physiological life history traits (Bennett 1997; Cocks et al. 1982; Dunbabin and Cocks 1999) and DNA-based molecular markers (Bennett and Mathews 2003; Lindqvist-Kreuze et al. 2003; Park et al. 2003). However, most members of the Brassicaceae contain seed oils, and Raphanus spp. are no exception (Kimber and McGregor 1995). While the seed oils are potentially of economic significance, they may also be used to assess genetic variation in wild populations. For example, there is some variation in the erucic acid content of Brassica napus which has enabled the selection of low erucic acid lines of oil seed rape (Nuttall and Moulin 1992; Salisbury and Ballinger 1995). The widespread distribution and associated genetic variation of wild radish in Western Australia (see earlier chapters) suggest that variation in oil content and composition is possible. However, wild radish has only been present in the region for about 150 years so variation, if present, will suggest either that there were many introductions with different fatty acid content or that natural selection in south-western Australia strongly favoured particular fatty acid patterns in particular locations.

Wild radish is a member of the Brassicaceae, a family which has contributed a number of important crops. The most important in southern Australia is canola, which refers to cultivars of oil seed rape producing seed with <2% erucic acid and <40 mol/l of total glucosinolates (Nuttall and Moulin 1992; Salisbury and Ballinger 1995). These traits were derived from the Canadian cultivar “Zephyr” (low erucic acid) and the European cultivar “Erglu” (low glucosinolates) (Roy 1984). Because of its quality, canola oil is recognised as nutritionally superior to most other edible oils. In addition to its low erucic acid, canola is characterized by low levels of saturated fatty acids (less than 4% palmitic acid ) and relatively high levels of oleic (55–60%) and linolenic acids (8–10%) (McDonald 1995). Wild radish is an important economic weed in canola, partly because it raises the erucic acid levels in canola oil and hence reduces oil quality.

Although there are many reasons why wild radish would be difficult to domesticate the presence of useful oils may stimulate interest in either the use of wild radish in breeding programs for other members of the Brassicaceae or domestication of the species itself. It is

106 therefore useful to know more of the pattern of fatty acids in this species and especially any genetic variation that occurs. This chapter describes the fatty acid content of the oils of wild radish collected at four contrasting sites in south-western Australia. Total oil content was measured in plants grown in Perth and the fatty acid content of the oils determined using gas chromatography. The results were related to variation in other aspects of the south- western populations and discussed in terms of their evolutionary and economic significance. In this chapter the hypotheses tested were that 1) wild radish populations exhibit genetic variation in the quality of seed oils, and 2) the yield and quality of oils and their fatty acids will provide future useful genetic material, either in their own right or in the genetic improvement of related species.

6.3 Materials and methods

Wild radish seeds were collected from Merredin (310 22’S, 1180 32’E), Nukarni (310 23’ S, 1180 15’E), Mullewa (280 33’S, 1150 29’E) and Denmark (340 58'S, 1170 39’E). Clean seeds from each site were directly sown on 19th April 2000 at the University of Western Australia Field Station and harvested on 28th November 2000. Maximum and minimum January and July temperatures and annual rainfall at each collection site are shown in Table 6.1. Details of plot management at the University Field Station are described in chapter 3. A brief description of the plant material is in Table 6.2.

Table 6.1. Mean maximum and minimum January and July temperatures (oC) and mean annual rainfall (mm) at Merredin, Nukarni, Mullewa and Denmark.

Site January July January July minimum Mean annual locations maximum maximum minimum rainfall Merredin 34 16 17 5 314 Nukarni 34 16 18 6 328 Mullewa 37 19 19 7 340 Denmark 25 17 14 7 1001

107

Sample ID Seed weight Pod weight Number of pod Pod width Pod segments length Merredin 0.70 1.88 3 to 10 0.1 to 0.5 23 - 61 Mullewa 1.29 3.49 1 to 12 0.2 to 0.7 32 - 79 Nukarni 0.46 1.05 2 to 7 0.1 to 0.3 20 - 56 Denmark 0.47 1.15 2 to 6 0.1 to 0.4 19 - 51

Lipids from the whole seeds were extracted using a Soxhlet apparatus (Gunstone 1996). Whole seeds (5 g) were ground into a fine powder with a coffee grinder. Three replicates (1 g samples) of each population were extracted for six hours with 20 ml of hexane (99.5%).

The fatty acid composition was determined by gas chromatography (GC) on the liquid samples by converting the extracted liquids to fatty acid methyl esters (Alonso et al. 1997). A known amount of each lipid extract (15-30 mg) was dissolved in 4 ml hexane in a stoppered tube. One ml of methanolic potassium hydroxide (2M) was then added as a trans-esterification agent. After thorough mixing, the mixture stood for 20 minutes, after which the hexane layer was removed and transferred into glass vials. GC separation of the fatty acid methyl esters (FAMES) was performed on a GC-17A V3 (SHIMADZU, Japan) equipped with flame-ionization detector and flow splitter. The GC conditions are described in Table 6.3.

One µl of the hexane fraction was injected into the apparatus. For quantitative analysis, nonadecanoic acid methyl ester was used to establish calibration curves as an external standard. Each test run had its own calibration curve. The peaks had a clear base line and were identified on the basis of their retention times and standards.

Data analysis In order to determine variation between and within sites for the total lipid content and for each fatty acid, a general analysis of variance was performed separately on each variable. The effect of site environment on fatty acid composition was assessed using principal

108 component analysis. Plants were grouped according to sites and the mean PC1 score plotted against the mean PC2 score for each group.

Table 6.3. Chromatographic conditions used in analysis of wild radish fatty acid methyl esters by GC with a BPX-70 (50mx 0.22 mm) capillary.

Parameter Setting Column head pressure 200 Kpa Split flow ratio 35 Oven temperature Initial temperature 1700C Initial time 35 min Temperature rate 50C/min Final temperature 2200C Final time 5 min Injector temperature 2500C Detector temperature 2850C Carrier gas Helium Detector gas Hydrogen

The fatty acid composition of wild radish was compared with a number of oil seed crops using principal component analysis in a similar way.

6.4 Results

Lipid content The total lipid content of the whole seeds ranged from 38.7% at Nukarni to 42.2% at Merredin (Table 6.4). Most of the variation occurred within populations (80%).

Fatty acid profile

The mean fatty acid profile of all plants tested is shown in Figure 6.1. The Figure shows that the dominant fatty acids are erucic acid, oleic acid, linoleic acid, linolenic acid and eicosenic acid. For each site the fatty acid profiles are shown in Table 6.4. The results indicate that the fatty acid profiles from all four sites are broadly similar, although 6 of the 9 main fatty acids differed significantly between sites.

Furthermore the fatty acids were composed of approximately 8.9 to 10.3% saturated and 83.3 to 87.4% unsaturated. The ratio of saturated/unsaturated fatty acids ranged from 1:8 to1:10.

109

Others Palmitic (c16:0) 2% Palmitic (c16:0) Lignoceric (c24:0) 5% Stearic (c18:0) 0% 2% Stearic (c18:0) Oleic (c18:1) Oleic (c18:1) 17% Vaccenic (c18:1) Linoleic (c18:2) Erucic (c22:1) 37% Linolenic(c18:3) Vaccenic (c18:1) Arachidic (c20:0) 1% Eicosenic (c20:1)

Linoleic (c18:2) Behenic (c22:0) 11% Erucic (c22:1)

Behenic (c22:0) Lignoceric (c24:0) 1% Others

Eicosenic (c20:1) Linolenic(c18:3) 10% 13% Arachidic (c20:0) 1%

Figure 6.1. Fatty acid composition of wild radish seed based on mean of values of four sites sampled in southwestern Australia. Table 6.5 explains the descriptions of the fatty acids appearing in parentheses.

Table 6.4. Total lipid content and the content of each constituent fatty acid of the Merredin, Mullewa, Nukarni and Denmark populations. The Table also shows the amount of variation accounted for between populations and the coefficient of variation (CV %) of each fatty acid across all populations. Significance of the difference between populations is P<0.001 (***), P<0.005 (**), P<0.05 (*).

%variation Lipids & fatty acids Merredin Mullewa Nukarni Denmark CV (%) B/sites Total lipids 42.2 41.9 38.7 41.2 6.9 20.4* Palmitic acid 5.51 4.93 5.55 5.35 11.6 7.7 n.s Stearic acid 2.38 2.06 2.23 1.72 13.6 44.4*** Oleic acid 17.39 14.83 15.24 17.05 12.5 27.1*** Linoleic acid 11.65 10.25 11.73 11.58 12.2 8.2** Linolenic acid 12.78 12.63 13.19 13.45 12.2 7.0 n.s Arachidic acid 1.3 1.03 1.33 0.87 20.7 38.1*** Eicosenoic acid 10.05 9.35 9.53 9.11 12.8 7.3 n.s Behenic acid 1.06 1.18 1.17 0.94 16.8 18.5*** Erucic acid 33.21 36.28 36.12 36.24 12.2 8.1* Others 3.62 3.14 3.91 3.64 90.2 0.24 n.s Saturated 10.25 9.2 10.28 8.88 Unsaturated 85.08 83.34 85.81 87.43 Saturated/Unsaturated 1:8.30 1:9.06 1:8.35 1:9.85

110

Variation between and within sites

Stearic acid (44%) showed the greatest between site variation, while arachidic (38%), oleic (27%), linolenic (7%) and eicosenic (7%) acids showed the least. The acids with the greatest within site variation were those with the least between site variation: linolenic (93%) and eicosenoic (93%) acids had the greatest within site variation, while stearic (56%), arachidic (62%) and oleic (73%) acids displayed the least within site variation.

Table 6.5. Component loadings for plant fatty acids in the principal component analysis, as well as the amount of variation accounted for by the first and second principle components.

Latent Vectors Fatty acids PC1 PC2 % Variation 41 24 1 Palmitic (c16:0) 0.464 0.111 Stearic (c18:0) 0.371 -0.419

Oleic (c18:1) n9 0.286 0.350 Linoleic (c18:2) 0.430 0.173 Linolenic (c18:3) 0.195 0.308 Arachidic (c20:0) 0.394 -0.419 Eicosenic (c20:1) 0.427 0.060 Behenic (c22:0) 0.000 -0.608 Erucic (c22:1) 0.073 0.126

1The figures in parentheses indicate the number of carbon atoms in each fatty acid molecules and the number of double bonds, itself an indicator of saturation. Thus c16:0 indicates a fatty acid with 16 carbon atoms and zero double bonds.

Component loadings arising from the principal component analysis are shown in Table 6.5. The first PC accounts for 40.5% of all variation and the second for 23.6%. In Fig. 6.2 populations are grouped according to where they were collected and plotted against their average scores. The Figure indicates that the populations differ in respect to PC2 (low values of PC2 accompany high values of behenic, stearic and arachidic acids) with populations from Denmark and Mullewa having strikingly similar fatty acid profiles. Within the range recorded here, Merredin and Nukarni populations tend to have lower values for the three fatty acids than do the Mullewa and Denmark populations.

111

PC1 comprises a set of fatty acids which do not appear to differ significantly between collection sites. In particular, the Mullewa population exhibited a great deal of within site variability for this set of fatty acid parameters.

6.5 Discussion

This chapter explores the use of lipid and fatty acid analysis to study genetic variation in Western Australian populations of wild radish. Although the populations were collected from different sites (Table 6.1) with very different environments, their seed oils and fatty acids compositions are markedly similar. All populations of wild radish seed in this study contain high levels of erucic, oleic, linoleic and linolenic acids. Unlike the molecular analysis reported in chapter 4 there appears to be little evidence of a significant difference between the Nukarni and Merredin sites. The data in Table 6.4 does indicate however, that there are significant differences between populations in fatty acid composition, although they are relatively minor. The data also show that within site variation is much greater than between site variation, suggesting that there has been very little evolution in the fatty acid content and composition of this species since its introduction to Western Australia.

3.5

2.5 Merredin Mullewa Nukarni 1.5 Denmark

0.5 PC1 -3.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 3.5 -0.5

-1.5

-2.5

-3.5 PC2

Figure 6.2 Principal component analysis showing the relationship between plant fatty acids and environment. Points represent the mean of PC1 and PC2 scores for all plants within each site. Error bars ± 1 s.e.

112

This result is not surprising because, in outbreeding species such as wild radish, strong selection pressure is needed for populations to develop differences in the short time that wild radish has been in Western Australia. For example Hamrick et al. (1979) in reviewing the literature of plant species with different breeding systems, found that for most traits outbreeding species have a greater proportion of their variation within populations and less between populations.

The similarity in the seed oils of wild radish populations differs from some studies of naturally occurring oils found in the literature. For example, in the Lamiaceae, genetic variation is widespread in Thymus vulgaris (Granger and Passett 1973), Rosmarinus officinalus (Dellacasa et al. 1999) and many other species. In particular, in Cunila galioides, Echeverrigaray et al. (2003) have demonstrated that oils and some of their constituents are under genetic control and show ecotypic differentiation, with certain chemotypes restricted to particular geographical areas. In the Brassicaceae, similar variation occurs in the composition of oils, with oleic and erucic acids varying widely in Brassica carinata (Alemayehu et al. 2001).

In the light of this evidence it is likely that variation might be found in wild radish if enough populations were sampled from enough localities. The fact that only four populations were sampled, all from Western Australia, severely limits the opportunities for genetic variation to be expressed. If populations from throughout Australia were sampled the opportunities for finding variation would be greatly increased. However, given that we know of no reason why there would be an evolutionary advantage for different oil compositions and that there is no evidence of separate introductions with different fatty acid compositions, it is likely that the species has been in Australia for an insufficient period for genetic drift and mutation to have developed distinct chemotypes. Therefore, it would probably be necessary to go beyond Australia if the genetic variation in oil content of wild radish is to be fully explored.

The results also suggest that, as far as oils and their composition are concerned, wild radish, when introduced to Western Australia, showed little genetic variation. This is in contrast to the molecular variation described in chapter 4, where the Nukarni population appeared to be distinct. If we assume that there were multiple introductions of wild radish and that, for example, the Nukarni population represents a different introduction from the other three sites (chapter 4), the lack of variation here suggests that there may not be significant

113 variation in the Australian or indeed the World population. Such a conclusion however, would need confirmation from further, more widely based surveys.

The comparison with other species containing oil is interesting because of the possible commercial exploitation of wild radish. Wild radish oil contains six commercially important fatty acids: palmitic, stearic, oleic, linoleic, linolenic and erucic acids. With its high erucic acid content it could not be used for food purposes; however, it could be used for non-food purposes, such as diesel fuel, lubricating oils and surface coatings. For food its use would be limited, like mustard seed, to food condiments.

To provide comparisons with wild radish oil, principal component analysis was performed on the fatty acid composition of different vegetable oils. Data for these oils were extracted from published sources ((AOCS) 1996). Table 6.6 shows the component loadings for individual fatty acids of the different oils and the amount of variation accounted for by PC1 (34% of the total) and PC2 (17%). Fig. 6.3 shows clearly that the wild radish oils are similar to mustard seed oils, but, because of its high erucic acid, less so to other common vegetable oils.

5 Jojoba oil 4

1 White mustard oil Yellow mustard oil Linseed oil Peanut oil Corn oil 0 PC1 -5 -4 -3 -2 -1 0 1 2 3 4 5 Palm oil -1 Brown mustard oil Olive oil All four wild radish populations oil -2 Canola oil -3

-4

-5 PC2

Figure 6.3 Principal component analysis showing the relative profiles of fatty acids for different vegetable oils. Points represent the mean of principal component scores for all crops ((AOCS) 1996) (see also Table 6.6) .

114

This conclusion is supported by the fact that canola oil, obtained from modified rape seed, separates in the analysis from wild radish mainly because of its low erucic acid content, highlighting the problem that contamination with wild radish can cause unacceptably high levels of erucic acid in commercial canola oil.

The economic importance of commercial oil crops such as canola depend not only on their oil contents but also on the quality and amount of the protein meal remaining after oil extraction. This aspect was not examined during this study. However, it is unlikely that wild radish protein would be significantly inferior to canola protein. Further work in this area is needed if a full understanding of the commercial potential of wild radish is to be realised.

Of course, the possession of some interesting traits is hardly likely, of itself, to lead to the domestication of a weedy species such as wild radish. Nevertheless, the genetic variation that exists within and between populations suggests that selections of wild radish for disease resistance, drought tolerance, resistance to pod shattering, and uniform flowering time may be possible through plant breeding. The domestication of wild radish might then become possible for niches where other oil seeds are unreliable because of low rainfall or inhospitable soil. Perhaps more likely is the use of these traits in wild radish for wide crosses with closely related species such as canola or mustard.

Table 6.6 Component loading for plant fatty acid composition in the principle component analysis of the fatty acid content of nine oil crops and four populations of wild radish, as well as the amount of variation accounted for by the first and second principle components.

Plant fatty acid Latent Vectors compositions PC1 PC2 % variation 34.1 16.9

Arachidic (c20:0) 0.41 -0.24 Behenic (c22:0) 0.36 -0.25 Eicosenoic (c20:1) 0.10 0.55 Erucic (c22:1) 0.43 0.05 Lignoceric (c24:0) 0.08 0.11 Linoleic (c18:2) -0.17 -0.16 Linolenic (c18:3) 0.04 -0.04 Oleic (c18:1) -0.30 -0.37 Palmitic (c16:0) -0.31 -0.11 Stearic (c18:0) -0.26 -0.28 Others 0.27 -0.07 115

Outcrossing and self-fertilization are common within the genus Brassica (Bateman 1955; Virtue 1996). Canola is known to be predominantly self-fertile with the potential of about 30% outcrossing (Rakow and Woods 1987; Williams 1986), which nevertheless results in the free and widespread dispersal of genes. Chevre et al. (1994) demonstrated that hybrids could be produced between wild radish and canola. Spontaneous hybridization between male-sterile canola and wild radish generated 3,734 hybrid seeds/m2. High chromosome pairing was observed and the hybrid pollen fertility was measured at 0-30% (Eber et al. 1994). Further research on the spontaneous outcrossing and seed production between various male-sterile and transgenic lines of canola with wild radish led to the discovery that the seeds produced showed size dimorphism (Baranger et al. 1995). The use of wild radish genes in canola is therefore possible and, where useful genes can be identified, desirable.

However, there is a down side to the relatively straightforward hybridisation between these species. Herbicide susceptible wild radish is a target for control within the herbicide- resistant crop, but the probability of hybridisation with canola and the possibility that the resulting hybrids may be fertile runs the risk of transferring the resistance gene to wild radish.

6.6 References

Alonso L, Fontecha J, Lozada L, Juarez M (1997) Determination of mixtures in vegetable oils and milk fat by analysis of sterol fraction by gas chromatography. Journal of American Oil Chemist's Society 72, 131-135. AOCS (1996) Official Methods and Recommended Practices of the American Oilseed Chemists Society. (American Oil Chemist's Society: Champaign, Illinois, USA, 5th edition). Baranger A, Chevre A, Eber F, Renard M (1995) Effect of oilseed rape genotype on the spontaneous hybridisation rate with a weedy species-an assessment of transgenic dispersal. Theoretical and Applied Genetics 91, 956-963. Bateman AJ (1955) Self-incompatibility systems in angiosperms III. Cruciferae. Heredity 9, 53-68. Bennett SJ (1997) Genetic variation between and within two populations of Trifolium glomeratum (cluster clover) in Western Australia. Australian Journal of Agricultural Research 48, 969-976. Bennett SJ, Mathews A (2003) Assessment of genetic diversity in clover species from Sardinia, Italy, using AFLP analysis. Plant Breeding 122, 362-367. Chevre A, Eber F, Baranger A, Kerlan M, Barret P, Festoc G, Vallee P, Renard M (1994) Interspecific gene flow as a component of risk assessment for transgenic Brassicas. Acta Horticulturae 407, 169-179.

116

Cocks PS, Craig AD, Kenyon RV (1982) Evolution of subterranean clover in South Australia. II. Change in genetic composition of a mixed population after 19 years' grazing on a commercial farm. Australian Journal of Agricultural Research 33, 679-695. Dunbabin MT, Cocks PS (1999) Ecotypic variation for seed dormancy contributes to the success of capeweed (Arctotheca calendula) in Western Australia. Australian Journal of Agricultural Research 50, 1451-1458. Eber F, Tanguay X, Chevre AM, Baranger A, Vallee P, Renard M (1994) Spontaneous hybridization between a male sterile oilseed rape and two weeds. Theoretical and Applied Genetics 88, 362-368. Gunstone FD (1996) Fatty Acids and Lipid Chemistry. (Blackwell Academic & Professional: London). Hamrick JL, Linhart YB, Mitton JB (1979) Relationships between life history characteristics and electrophoretically detectable genetic variation in plants. Annual Review of Ecology and Systematics 10, 173-200. Kimber DS, McGregor DI (1995) The species and their origin, cultivation and world production. In ‘Brassica Oilseeds Production and Utilization’. (Eds DS Kimber, DI McGregor). (CAB International University Press: Cambridge, UK) Layton CR, Ganders FR (1984) The genetic consequences of contrasting breeding systems in Plectritis (Valerianaceae). Evolution 38, 1308-1325. Lindqvist-Kreuze H, Koponen H, Valkonen JPT (2003) Genetic diversity of arctic bramble (Rubus arcticus L. subsp. arcticus) as measured by amplified fragment length polymorphism. Canadian Journal of Botany 81, 805-813. McDonald BE (1995) Oil properties of importance in human nutrition. In ‘Brassica Oilseeds Production and Utilization’. (Eds DS Kimber, DI McGregor). (CAB International University Press: Cambridge UK) Nuttall WF, Moulin AP (1992) Yield response of canola to nitrogen, phosphorus, precipitation and temperature. Agronomy Journal 84, 765-768. Park K-C, Lee JK, Kim N-H, Shin Y-B, Lee J-H, Kim N-S (2003) Genetic variation in Oryza species detected by MITE-AFLP. Genes & Genetic Systems 78, 235-243. Rakow G, Woods DL (1987) Outcrossing in rape and mustard under Saskatchewan Prairie conditions. Canadian Journal of Plant Science 67, 147-151. Roy NN (1984) Interspecific transfer of Brassica juncea-type high blackleg resistance to Brassica napus. Euphytica 33, 295-303. Salisbury PA, Ballinger DJ (1995) Blackleg disease on oilseed Brassica in Australia: a review. Australian Journal of Experimental Agriculture 35, 665-672. Virtue JG (1996) Improving the assessment of new weed threats: developing techniques with cruciferous weeds of cropping. In 'Proceedings of the Eleventh Australian Weeds Conference'. Melbourne, Australia, pages 85-88. (Weed Society of Victoria: Frankston, Victoria.) Williams I, Martin A, White R (1986) The pollination requirements of oil-seed rape. The Journal of Agricultural Science 106, 27-30.

117

Chapter 7. General discussion

Three important topics emerge from this thesis. Firstly, the results raise the question about the importance of inbreeding for colonising species. Secondly, they pose the question as to what is the relative value of molecular markers and life history traits in understanding genetic diversity. Finally, they give some indications about why wild radish is so beautifully adapted to the Western Australian wheat belt. The chapter concludes with a brief look at what future research should be conducted to clarify these and other issues.

The widespread success of wild radish and some other outbreeding species in colonising the agricultural areas of Western Australia challenges the widely held belief that inbreeding confers an advantage during the colonisation process (Allard 1975). Inbreeding is thought to be advantageous because:

o Single plants can establish at new sites without the need for pollination by a second plant;

o Traits that lead to adaptation in environmental niches can be fixed, and not subject to the dilution that may occur in outbreeding species

In most cases more than one individual is likely to be involved in colonisation. Speculation on how exotic species arrived in Australia usually revolves around ballast, often taken on board in South Africa, and hay to feed livestock (Gladstones 1966). To this we can add contamination of the coats of animals by seeds with awns or other means of attaching themselves to livestock. Wool in particular, is a good vehicle for transfer in this way (Cocks and Phillips 1979). It is therefore highly unlikely that only one individual of a species would be so transported – the advantage that inbreeders have in not requiring a second individual would thus occur only occasionally. This would apply even more so where species were colonising new sites after their original establishment in Australia.

Evidence for multiple introductions to Australia includes the many different strains of subterranean clover that have been observed around Perth, Adelaide and Melbourne (Gladstones 1966, Cocks and Phillips 1979). Where two strains differ in their leaf markings it is likely that they have arisen from separate introduction. The leaf markings of those arising from hybridisation, and there are a great number of these (Cocks and Phillips 1979), resemble one or other of their parents. Thus the majority of subterranean clover strains in 118

South Australia resemble the earliest cultivars, Mount Barker and Dwalganup, which are therefore their putative parents. In Western Australia however, Gladstones (1966) observed a great number of apparently unrelated strains, which he considered were likely to have arisen from separate introductions. Therefore it is likely that even obligate outbreeders will have the ability to outcross from their first introduction, and this must have occurred with capeweed (Arctotheca calendula).

The argument that inbreeders have the ability to fix sets of alleles leading to adaptation for micro-environments, an ability that outbreeders lack, has been well argued and is convincing (Allard 1975). Indeed, Antonovics (1971) argues that inbreeding evolved for the purpose of fixing desirable genes or gene combinations. Clearly, where environments vary in close proximity, a method of sustaining genetic integrity should give inbreeders a significant advantage over outbreeders. Environments often differ significantly, even over distances of a metre or less, and it is here that one would expect inbreeding to offer the greatest advantage. A good example is the population of subterranean clover on Kangaroo Island, South Australia, where time of flowering was related to elevation. The various genotypes within that population were hybrids of the cultivars Mount Barker and Dwalganup mentioned previously. Homozygous lines had colonised the various parts of the slope, apparently linked to water relationships: in the higher and steeper parts of the slope the population was earlier flowering, where the soil was drier in late spring (Cocks 1992). These populations displayed ecotypic differentiation over distances of only a few metres.

Yet inbreeders by no means dominate the successful colonisers in southern Australia. While many annual pasture species and weeds are inbreeders many are not. Capeweed, one of the most successful invaders of annual pastures in southern Australia (Arnold et al. 1985), is an obligate outbreeder (Dunbabin 2001). It appears to have the same capacity to evolve ecotypes as does barley grass (Hordeum leporinum), a widespread, inbreeding, annual grass. Both species produce similar amounts of within and between site variation in life history traits, although barley grass shows greater between site variation in molecular markers. And, of course, wild radish has most successfully developed between site variation in life history traits (chapter 3).

Outbreeding perennial grasses are also able to develop ecotypes over very short distances and in relatively short periods. For example, populations of sweet vernal (Anthoxanthum odoratum) were genetically different even when they were separated by less than 1 m,

119 apparently associated with soil pH and grazing pressure (Snaydon and Davies 1976). Although gene flow as a result of any one flowering was restricted to 2 m, this distance exceeded the boundaries over which ecotypic differentiation was observed. Evolution of tolerance to heavy metals on old mine sites and on roadsides has been recorded on a number of occasions: for sweet vernal on zinc-contaminated mines (Antonovics 1972), for plantain (Plantago lanceolata) on lead-contaminated road sides (Wu and Antonovics 1976), for copper tolerance in Agrostis stolinifera (Wu et al. 1975), for copper and zinc tolerance in Agrostis tenuis (Whalley et al. 1974), and for five perennial grasses (including Deschampia cespitosa and Festuca ovina) on zinc-contaminated soils (Al-Hiyaly et al. 1990). In all cases evolution of tolerance to the heavy metal had occurred over short distances and short periods.

These results suggest that for outbreeders selection for adaptive genes is as effective as is selection for adaptive combinations of genes in inbreeding annuals, suggesting that outbreeding is unlikely to impact on the ability of outbreeding annual species such as wild radish to adapt during the colonising process. They demonstrate that wild radish has readily evolved genetically differentiated populations as a means of adapting to different environments across south-western Australia. Though wild radish was used as a case study, there is no reason to expect that other widespread and successful outbreeding colonisers will not exhibit similar differentiation in response to environmental variation.

Allard (1965) and others have predicted that genetic variation within populations is likely to be more between populations for inbreeders plants than it is for outbreeders. While this does not appear to be the case for life histories in southern Australian colonisers (Dunbabin 2001), it may be for molecular markers. This contention is supported by the results of this study where, although there were no inbreeders for comparison, life history traits reflected environmental differences while molecular markers appeared to reflect differences in the origins of the populations (chapter 4). Table 7.1 illustrates that in the studies examined, within site variation measured using molecular markers was far greater in outbreeding species than it was for inbreeding species (between site variation is thus far greater for the inbreeders). In a direct comparison of an inbreeder with an outbreeder – both populations were collected at the same sites - Dunbabin (2001) found that molecular markers in the inbreeder displayed far greater between site variations than did the outbreeder. It is therefore clear that if we wish to measure evolution of adaptation we should examine life

120 history traits, while if we are interested in genetic variation and the origin of populations we should use molecular markers.

Table 7.1. A comparison of within species variation recorded in various studies of inbreeders and outbreeders in annual and perennial plants

Species Family Annual Breeding Within Reference or system variation perennial Silene Caryophyllaceae Biennial Out 18-231 Mengoni et al. (2000) paradoxa Scutellaria Labiatae Perennial In 81, 2 Sun (1999) indica Symplocos Symplocaceae Perennial Out 543 Deshpande et al. (2001) laurina Sinapis Brassicaceae Annual Out 601 Moodie et al. (1997) arvensis Glycine soja Fabaceae Annual In (13% 92 Fujita et al. (1997) out) Argyroxiphium Asteraceae Perennial Out 144 Friar et al. (2001) kauense Sideritis Lamiaceae Perennial Out 891 Vazquez et al. (1999) pusilla Camellia Theaceae Perennial Out 841 Kaundun et al. (2002) sinensis Asimina triloba Annonaceae Perennial Out 882 Huang et al. (1998) Bromus Poaceae Perennial Out 84-96 Diaby et al. (2003) inermis 1 RAPD markers 2 Allozymes 3 PCR 4 Microsatellites It is interesting to reflect that the majority of crop weeds in southern Australia are outbreeders, although many pasture weeds are not. The best example is annual ryegrass (Lolium rigidum), a widespread weed in the cereal belt of southern Australia that outcrosses freely. Not only has it colonised many different habitats but it has also adapted rapidly to changed circumstances. For example, it has developed resistance to some 20 herbicides, and can do so within a few years of the first application of the spray (Gill 1995; Hall et al. 1994; McAllister et al. 1995). Similarly in Canada, wild mustard (Sinapsis arvensis) has

121 developed resistance to several herbicides, including dicamba and MCPA (Webb and Hall 1995). These results suggest that outbreeders may have the capacity to adapt rapidly to changed circumstances, a situation that frequently arises in crop-based farming systems using herbicides, crop rotations, tillage and the removal of seeds during harvest. Other widespread weeds that have successfully colonised southern Australia are two pasture weeds - salvation jane (Echium plantagineum), capeweed (Arctotheca calendula) and the subject of this thesis, wild radish (Raphanus raphanistrum).

It seems likely therefore that breeding system of itself has only limited effect on the ecological success of colonising species and that life history traits play a far more important role. The next section of this discussion examines the adaptation of wild radish to the environment of south-west Australia and compares it with other exotic species colonised in the region.

It is clear that wild radish has formed genetically distinct populations, especially for the reproductive traits of flowering time, seed size and pod mass (chapter 3). While these were related broadly to environmental variation the relationship was not strong. The results in chapter 4 confirmed these conclusions, suggesting that, although the populations were distinct, the environmental variables measured may not have been the ones responsible for population differentiation. Similarly, in chapter 5, although seed dormancy was studied in populations from markedly contrasting sites, seed behaviour was strikingly similar.

There have been a number of studies in southern Australia over the past 25 years investigating genetic variation in colonising species. Most demonstrated that there were few traits that could be linked directly to environmental variables. For example, Smith et al. (1995) found that, of 9 life history traits in the inbreeder Trifolium glomeratum, just two (flowering time and seed mass) were weakly related to any environmental variable. The remainder showed no discernible ecotypic differentiation. Similarly, Fortune et al. (1995) noted little variation in flowering time in a host of self-pollinating naturalized legumes in Western Australia and suggested this may be due to introduction of only limited genetic diversity. In eastern Australia, Young et al. (1992) detected no genetic variation in populations of the inbreeder, Medicago laciniata. Norman et al. (1998) found that, in a wide array of legumes from Australia and the Mediterranean basin, only flowering time was weakly related to any environmental variable, with no other traits displaying consistent trends. Kon and Blacklow (1988) found that flowering time and panicle emergence of the

122 inbreeder Broumus diandrus were related to growing season length, but that there was no relationship for seeds per panicle, seed mass or seeds per plant. Fedorenko (2000) found, for the inbreeder Medicago minima, very little genetic variation in a number of south- western Australian populations. It seems that many colonisers in southern Australia demonstrate little ability to undergo wide scale ecotypic differentiation.

The above species are inbreeders. In some cases outbreeding may have resulted in greater ecotypic differentiation. For example, of the 15 traits exhibiting significant between site variation in Echium plantagineum, six were correlated with one or more environmental variable (Wood and Degabriele 1985). The authors concluded that an ecocline had developed in this species since its introduction. Clearly however, wild radish has not developed this level of ecotypic differentiation and is comparable in this respect with inbreeding annuals.

It is possible that environmental variation is being inadequately measured in most of these studies. The strong relationship between elevation and flowering time in a single field on Kangaroo Island has already been mentioned (Cocks 1992). In that case it is likely that elevation was a precise measure of length of growing season, much better than the broad generalisations associated with rainfall measurements in most other studies, including this one. The measurements of toxic levels of heavy metals is also precise and explains the good relationships obtained in the many studies of tolerance in perennial grasses. Future studies of genetic variation in wild radish and other colonising species should look at environmental variables much more closely. It would also be helpful if variation was examined in closely adjacent sites, where pollen had a real chance to move between populations. In this way a better idea of the ability of wild radish to form distinct populations where gene flow was not restricted would be possible. Finally, comparison with other inbreeding and outbreeding weeds would clarify many of these relationships.

In conclusion the four hypotheses posed at the end of chapter 2 are re-examined in terms of genetic variation:

1. Wild radish populations are genetically variable and at least some of this variation can be related to environmental variables.

123

As discussed above the genetic variation in both life history traits and molecular markers is considerable. However, it was not closely related to environmental variables possibly for the reasons outlined above.

2. Populations from nearby sites are genetically similar while those from more distant sites are genetically less similar.

This appeared to be the case for life history traits but was not so for molecular markers. As discussed in chapter 4 this indicates that the life history traits have evolved to meet environmental conditions while the molecular markers are a much better indicator of the historic genetic relationships between populations.

3. Seed dormancy varies between populations with those from dry areas more dormant than those from wetter areas.

This hypothesis was not supported. There are two explanations – either dormancy was broken during seed treatment, and there is some support for this conclusion in the recent literature, or environment is not exerting a selection pressure over the environments assessed.

4. Wild radish populations exhibit genetic variation in the quality and quantity of seed oils.

This hypothesis was not supported. Although there was some variation in seed oil quality and quantity it was low and unrelated to other variation. However, as was pointed out in chapter 6, the Western Australian population may be limited in respect to seed oils, and if further exploration of the gene pool of wild radish was undertaken it is likely that greater genetic variation would be found.

7.1 References Al-Hiyaly SAK, McNeilly T, Bradshaw AD (1990) The effect of zinc contamination from electricity pylons. Contrasting patterns of evolution in five grass species. New Phytologist 114, 183-190.

Allard RW (1975) The mating system and microevolution. Genetics 79, 115-26.

Antonovics J (1971) The effects of a heterogeneous environment on the genetics of natural populations. American Scientist 59, 593-599.

124

Antonovics J (1972) Population dynamics of the grass Anthoxanthum odoratum on a zinc mine. Journal of Ecology 60, 351-365.

Arnold GW, Ozanne PG, Galbraith KA (1985) The capeweed content of pastures in south- west Western Australia. Australian Journal of Experimenta Agriculture 25, 117-23.

Cocks PS (1992) Evolution in sown populations of subterranean clover (Trifolium- Subterraneum L.) in South Australia. Australian Journal of Agricultural Research 43, 1583-1595.

Cocks PS, Phillips JR (1979) Evolution of subterranean clover in South Australia. 1. The strains and their distribution. Australian Journal of Agricultural Research 30, 1035- 1052.

Deshpande AU, Apte GS, Bahulikar RA, Lagu MD, Kulkarni BG, Suresh HS, Singh NP, Rao MKV, Gupta VS, Pant A, Ranjekar PK (2001) Genetic diversity across natural populations of three montane plant species from the Western Ghats, India revealed by intersimple sequence repeats. Molecular Ecology 10, 2397-2408.

Diaby M, Casler MD (2003) RAPD marker variation among smooth Brome grass cultivars. Crop Science 43, 1538-1547.

Dunbabin MT (2001) Genetic variation in the outbreeding coloniser capeweed in south- western Australia. PhD thesis, University of Western Australia.

Fedorenko FDE (2000) Ecology and genetic variation of Medicago minima (L.) Bart. PhD thesis, The University of Western Australia.

Fortune JA, Cocks PS, Macfarlane CK, Smith FP (1995) Distribution and abundance of annual legume seeds in the wheatbelt of Western Australia. Australian Journal of Experimental Agriculture 35, 189-197.

Friar EA, Boose DL, Ladoux T, Roalson EH, Robichaux RH (2001) Population structure in the endangered Mauna Loa silversword, Argyroxiphium kauense (Asteraceae), and its bearing on reintroduction. Molecular Ecology 10, 1657-1663.

Fujita R, Ohara M, Okazaki K, Shimamoto Y (1997) The extent of natural cross-pollination in wild soyabean (Glycine soja). Journal of Heredity 88, 124-128.

Hall LM, Tardif FJ, Powles SB (1994) Mechanisms of cross and multiple resistance in Alopecurus myosuroides and Lolium rigidum. Phytoproduction 75, 17-23. 125

Huang H, Layne DR, Riemenschneider DE (1998) Genetic diversity and geographic differentiation in pawpaw (Asimina triloba (L.) Dunal) populations from nine states as revealed by allozyme analysis. Journal of the American Society for Horticultural Science 123, 635-641.

Kaundun SS, Young-Goo P (2002) Genetic structure of six Korean tea populations as revealed by RAPD-PCR markers. Crop Science 42, 594-601.

Kon KF, Blacklow WM (1988) Identification distribution and population variability of great brome Bromus-diandrus Roth and rigid brome Bromus-rigidus Roth. Australian Journal of Agricultural Research 39, 1039-1050.

McAllister FM, Holtum JAM, Powles SB (1995) Dinitroaniline herbicide resistance in rigid ryegrass (Lolium rigidum). Weed Research 43, 55-62.

Mengoni A, Gonnelli C, Galardi F, Gabbrielli R, Bazzicalupo M (2000) Genetic diversity and heavy metal tolerance in populations of Silene paradoxa L. (Caryophyllaceae): a random amplified polymorphic DNA analysis. Molecular Ecology 9, 1319-1324.

Moodie M, Finch RP, Marshall G (1997) Analysis of genetic variation in wild mustard (Sinapis arvensis) using molecular markers. Weed Science 45, 102-107.

Norman HC, Cocks PS, Smith FP (1998) Reproductive strategies in Mediterranean annual clovers. I. Germination and hardseededness. Australian Journal of Agricultural Research 49, 973-982.

Snaydon RW, Davies MS (1976) Rapid population differentiation in a mosaic environment. IV. Populations of Anthoxanthum odoratum at sharp boundaries. Heredity 37, 9-25.

Sun M (1999) Cleistogamy in Scutellaria indica (Labiatae): Effective mating system and population genetic structure. Molecular Ecology 8, 1285-1295.

Vazquez JL, Gomez-Mercado F, Guerrero JLG, Rodriguez-Garcia I, Garcia-Maroto F (1999) Genetic relationships and population structure within taxa of the endemic Sideritis pusilla (Lamiaceae) assessed using RAPDs. Botanical Journal of the Linnean Society 129, 345-358.

Whalley KA, Khan MSI, Bradshaw AD (1974) The potential for evolution of heavy metal tolerance in plants. 1. Copper and zinc tolerance in Agrostis tenuis. Heredity 1974, 309-319. 126

Wood H, Degabriele R (1985) Genetic variation and phenotypic plasticity in populations of Patterson's curse (Echium plantagineum L.) in south-eastern Australia. Australian Journal of Botany 33, 677-685.

Wu L, Antonovics J (1976) Experimental ecological genetics in Plantago 11. Lead tolerance in Plantago lanceolata and Cynodon dactylon from a roadside. Ecology 57, 205-208.

Wu L, Bradshaw AD, Thurman DA (1975) The potential for evolution of heavy metal tolerance in plants. 111. The rapid evolution of copper tolerance in Agrostis stolonifera. Heredity 34, 165-187.

Young RR, Croft PH, Sandral GA (1992) Variation in flowering times and agronomic characteristics of Medicage laciniata (L.) Miller collected from diverse locations in New South Wales. Australian Journal of Experimenta Agriculture 32, 59-63.

127

Appendix I: DNA extraction methods Genomic DNA extraction of wild radish using modified CTAB method

This procedure yields between 30 and 120 mg of high quality (OD = 1.6-2.0) DNA.

• Preheat 7.5 ml of CTAB buffer in a 10 ml tube at 60 0C in a water bath. Add 0.2%(15µl) mercaptol-ethanol to the buffer.

• Pre-cool a mortar and pestle with liquid nitrogen.

• Grind 3g of fresh leaf tissue with a pestle and mortar using liquid nitrogen.

• Scrape powder immediately into the preheated CTAB buffer.

• Place the tube in a 600C water bath and swirl the tube regularly over the next 60 minutes.

• Add 7ml chloroform-isoamyl alcohol (24:1) and mix gently, but thoroughly.

• Centrifuge at low speed (1600x g) for 10 minutes to separate the phases.

• Remove the aqueous phase with a wide-bore pipette and transfer to a clean tube to add 5ml of cold isopropanol. Mix gently to precipitate the nucleic acids.

• Spool out DNA with a glass rod and transfer to 10ml 76% ethanol containing 10mM ammonium acetate and leave in fridge for overnight.

• Spin at high speed for 3 mins. Discard the ethanol.

• Allow most of the moisture dry off and add 0.7ml of T.E to dissolve the pellet.

128

• Add 10µl Rnase to a final concentration of 10µg/ml and incubate at 370C for 30 mins.

• Add 700µl phenol chloroform. Mix thoroughly and spin for 3 minutes to separate the phases.

• Add 700µl chloroform. Mix thoroughly and spin for 3 minutes to separate the phase and remove the aqueous phase into another Eppendorf tube.

• Add 50µl of 2M sodium acetate (pH 5,0) and fill the tube with ethanol. Invert tube to thoroughly mix contents and recover DNA by centrifugation at high speed for 10 minutes.

• Discard the ethanol, air dry completely and wash the DNA with 800µl 70% alcohol, gently mix and vortex and centrifugation for 3 minutes.

• Discard the ethanol, air dry completely and suspend DNA in water.

129

Appendix II: Amplified Fragement Length Polymorphism procedure

Amplified Restriction Fragment Polymorphism (AFLP) was performed using the AFLP Analysis System 1 from Life Technologies Australia Pty.Ltd. The AFLP procedure is described below. For wild radish, the only modification required was that the undiluted preamplification template was used at the selective amplification step instead of the dilution listed. The rest of the Life Technologies procedure was followed according to the instructions.

1. Restriction Digestion of Genomic DNA

a) Add the following to a 1.5 ml microcentrifuge tube

1. Sample DNA (250 ng in ≤18µl) ≤18µl 2. 5x reaction buffer 5 µl 3. EcoR I/Mse I 2 µl 4. AFLP grade water (Distilled water) to 25 µl

b) Mix gentely and collect the reaction by brief centrifugation. Incubate at 370C for 2 hours.

c) Incubate the mixture for 15 minutes at 700C to inactivate the restriction endonucleases. Place tube on ice and collect contents by brief centrifugation.

2. Ligation of Adapters

a) To the eppendorf tube add 1. Adapter ligation solution 24 µl 2. T4 DNA ligase 1 µl b) Mix gently at room temperature, centrifuge briefly to collect contents, and incubate at 200C 20C for 2 hours

130

c) Perform a 1:10 dilution of the ligation mixture by adding 10µl to 90 µl of TE buffer. d) The unused portion may be stored at –200C.

3. Preamplification Reaction

a) Add the following to a 0.2ml PCR tube

4. Diluted template DNA 5 µl 5. Pre-amp primer mix 40 µl 6. 10xPCR buffer plus Mg 5 µl 7. Taq DNA polymerase (1unit/ µl) 1 µl

b) Mix gently and centrifuge briefly to collect reaction. Perform 20 PCR cycles at 94 0C for 30 seconds, 56 0C for 60 seconds, 72 0C for 60 seconds. Soak temperature is 4 0C.

c) Perform a 1:5 dilution by transferring 3 µl to a microcentrifuge tube containing 147 µl TE buffer.

d) Both diluted and undiluted portions can be stored at –200C.

4. Primer Labelling

Primer labelling is performed by phosphorylating the 5’end of the EcoR I primers with (γ- 33P) ATP and T4 kinase. EcoR I AAC / MseI CAA primer combination was used.

a) add the following components to a 1.5ml microcentrifuge tube.

1. EcoRI primer (select 1) 18 µl 2. AFLP- grade water 10 µl 3. 5x kinase buffer 10 µl 4. (γ-33P) ATP (2,000 Ci/mmol) 10 µl 5. T4 kinase 2 µl

131

b) Mix gently and centrifuge briefly to collect contents of tube. Incubate at 370C for 1 hour.

c) Inactivate the enzyme at 700C for 10 minutes. Centrifuge briefly to collect the reaction contents.

5. Selective AFLP Amplification

a) For each primer pair, add the following components to a 1.5 ml microcentrifuge tube and label it “Mix 1”.

1. Labelled EcoR I primer 5 µl 2. Mse 1 primer 45 µl

b) To a 1.5 ml microcentrifuge tube labelled “Mix 2”, add

1. AFLP grade water 79 µl 2. 10x PCR buffer for AFLP 20 µl 3. Taq DNA polymerase 1 µl

c) Each AFLP amplification is assembled by combining the following in a 0.2 ml microcentrifuge tube.

1. Diluted template DNA 2. Mix 1 3. Mix 2

d) Mix gently and centrifuge briefly to collect reaction.

e) Perform

1. One cycle at 94 0C for 30 sec, 65 0C for 30 sec, and 72 0C for 60 sec. 2. Lowere the annealing temperature each cycle 0.7 0C for 12 cycles, giving a touch down phase of 13 cycles.

132

3. 23 cycles at 94 0C for 30 sec, 56 0C for 30 sec, and 72 0C for 60 sec.

6. Gel Analysis 1. After PCR, add an equal volume (20 µl) of formamide dye (98% formamide, 10m M EDTA, bromophenol blue, xylene cyanol) to each reaction. Heat for 3 minutes at 90 0C and immediately place on ice. 2. Pour a 5% denaturing polyacrylamide gel (20:1 acrylamide: bis, 7.5 M urea, 1x TBE buffer) with 0.4mm spacers and a sharks-tooth comb. 3. Pre-electrophorese the gel at 70 W for 20 minutes. 4. Load 5 µl of each sample.

133