the evolutionary response of wild radish ( raphanus ... · guidance in statistical analysis, in...

The evolutionary response of wild radish (Raphanus raphanistrum L.)

populations to selections in agriculture

Michael Brett Ashworth

B. Agribusiness (Hons)

Dip. Engineering Metallurgy

This thesis is presented for the degree of

Doctor of Philosophy

The University of Western Australia

School of Plant Biology

Faculty of Science

February 2015

THESIS DECLARATION

This thesis consists of a series of scientific papers that includes one published paper and four

manuscripts prepared for later publication. The bibliographical details of the work and where

the manuscripts appear in the thesis are outlined below:

Chapter 2: Ashworth, M. B., M. J. Walsh, K.C. Flower, S.B. Powles. (2014). Identification of the

first glyphosate-resistant wild radish (Raphanus raphanistrum L.) populations. Pest

Management Science 70(9): 1432–1436.

Chapter 3: The identification of eight glyphosate-resistant Lolium rigidum and one Raphanus

raphanistrum population from the first glyphosate-resistant canola plantings across the Western

Australian grainbelt.

Chapter 4: Wild radish (Raphanus raphanistrum L.) recurrent selection with glyphosate over four

generations results in low-level glyphosate and ALS herbicide resistance.

Chapter 5: Recurrent selection with reduced 2,4-D rates results in the rapid evolution of 2,4-D

and ALS resistance in wild radish (Raphanus raphanistrum L.).

Chapter 6: Recurrent selection on wild radish (Raphanus raphanistrum L.) flowering time leads

to rapid adaptation of reproductive phenology.

The work associated with the production of these manuscripts in this thesis is my own. The

contribution of the different co-authors is associated with the initial research directions and

editorial input in various versions of the drafts of the manuscripts.

PhD Candidate Co-ordinating Supervisor

Mr Michael Ashworth

Winthrop Professor Stephen Powles

Supervisor signature Supervisor signature

Associate Professor Michael Walsh

Assistant Professor Ken Flower

PAGE | i

PAGE | ii

ACKNOWLEDGEMENTS

This work was supported by funding from the Australian Government [Australian Postgraduate

Award (APA)] and the Grains Research and Development Corporation (GRDC) [Grains Research

Top-up Scholarship (GRS)]. Thank you to The University of Western Australia and the Australian

Herbicide Resistance Initiative (AHRI) for support and an excellent research and learning

environment.

I would like to express my sincere thanks to those who have patiently guided and supported me

during the entire span of this PhD study. Special acknowledgement is extended to my co-

ordinating supervisor Winthrop Professor Stephen Powles, who is the world leader in herbicide

resistance research. I am thankful to you for the opportunity to undertake this work and for the

skills and insight you taught to me regarding the evolution of herbicide resistance .Thank you to

Associate Professor Michael Walsh for his close supervision and guidance throughout this PhD.

Your knowledge of wild radish biology and control options was most helpful. We produced a lot

of generations in the past three years which provided great data, thank you. Finally thank you

to my third supervisor and close friend Assistant Professor Ken Flower. Thank you for all your

guidance in statistical analysis, in order to best quantify the readily observable responses in this

study. Also thanks for the numerous coffee discussions, where we weighed up the practical

ramifications of these responses. I look forward to working with you all in the future.

Within AHRI, I have a lot of people to thank. Mrs Lisa Mayer and Ms Brogan Micallef for your

friendliness, administrative support and 'ability to bend the rules'. Dr Danica Goggin, Associate

Professor Yu Qin, Dr Martin Vila-Aiub, Assistant Professor Roberto Busi and Dr Heping Han for

your kindness, particularly when a PhD student drowning in literature lobbed up for a quick chat.

A special thanks goes to my fellow PhD scholars, Adam Jalaludin, Goh Sou Sheng and Gayle

Somerville for your friendship and support during my time in the 'Post Grad Room'.

Outside AHRI, thank you to my mentor, close friend and ‘past’ pastor Bill Addison, who

continually supported me and reminded me that ‘all we do, is for the audience of one’.

Finally, I would like to extend my love and greatest gratitude to my wife Shantelle for all her

prayers, mentoring and support throughout this research. You have put up with so much and I

hope I can pay it all back. Special thanks also goes to the 'Fab Three', my fantastic children Abigail

(Abi), Matilda (Tilly) and Samuel (Sam) for their hugs, kisses and wrestles that means so much

after a hard day. Thanks for being so supportive and patient while dad was far too busy.

PAGE | iii

PAGE | iv

ABSTRACT

Herbicide-resistant weeds pose a significant threat to the sustainability of global grain

production. Of all weeds in Australian farming systems, wild radish (Raphanus raphanistrum L.)

is one of the most economically damaging, aggressively competing for water, nutrients and light.

As a result of herbicide over-reliance, wild radish populations in the Western Australian (WA)

grainbelt have evolved multiple resistances to inhibitors of acetolactate synthase (ALS),

phytoene desaturase (PDS), photosynthetic electron transport (PSII) and synthetic auxin

herbicides.

However, following 40 years of over-reliance on glyphosate, this PhD study is the first to report

glyphosate resistance in wild radish. Two wild radish populations from fallow fields near

Mingenew and Carnamah in the WA grainbelt exhibited a heritable, 3.2 (WARR37) and 4.5

(WARR38) fold resistance (LD50) to glyphosate. Both populations also had multiple resistances

to ALS inhibitors, PDS inhibitors and synthetic auxin herbicides, which is expected to intensify

the selection for glyphosate resistance in these populations.

The inspection of 24,000 ha of the first commercial transgenic glyphosate-resistant (GR) canola

plantings in the WA grainbelt (2010–2011) found large wild radish populations treated solely

with glyphosate, which often resulted in less than complete control. Glyphosate resistance in

wild radish however was rare, with only one additional GR population identified. This survey also

identified glyphosate resistance in eight annual ryegrass (Lolium rigidum L.) populations.

However no glyphosate resistance was identified in barley grass species (Hordeum sp.), brome

grass species (Bromus sp.), wild oat species (Avena sp.), capeweed (Arctotheca calendula L.) and

mallow (Malvia parviflora L.).

The less than complete control of wild radish in the first GR canola crops grown in the WA

grainbelt was of concern, as four generations of glyphosate selection at rates that do not provide

full control, resulted in modest increases in glyphosate resistance (2.4-fold). Additionally this

selection also resulted in weak cross resistance to the ALS-inhibiting herbicides imazamox (4.7-

fold) and metosulam (3.7-fold). The biochemical basis of this weak resistance remains to be

investigated.

In response to the growing number of dicot species that have evolved glyphosate resistance in

the Americas, synthetic auxin tolerance has been incorporated into transgenic herbicide-

resistant crops. This study highlights that the sustainability of 2,4-D amine use on wild radish is

contingent on consistently achieving high levels of control. Following four generations of low-

dose 2,4-D amine selection, 2,4-D resistance increased 9.6-fold (LD50). Alarmingly along with 2,4-

PAGE | v

D resistance, cross resistance was also identified to the ALS-inhibiting herbicides metosulam

(3.5-fold), and chlorsulfuron (2.3-fold). The cross resistance to chlorsulfuron is considered

metabolism based, mediated by cytochrome P450 monooxygenases enzymes, as pre-treatment

with the known P450 inhibitor malathion restored chlorsulfuron to full efficacy. The

mechanism/s and genetic control of the auxinic and ALS cross resistance in this study remain to

be investigated.

With widespread multiple resistance in wild radish populations, non-herbicidal weed control

tactics are urgently required. Harvest weed seed control (HWSC) is a highly-effective non-

herbicidal weed control tool tactic that aims to intercept and treat weed seeds before they re-

enter the soil seed bank. The efficiency of HWSC however is contingent on the effective

collection of weed seeds at harvest, which may be shed through the evolution of early-maturing

biotypes. This study demonstrated that wild radish contains the phenotypic diversity to halve its

flowering time (FT) to 29 days after emergence (FD50) following 5 generations of early FT

selection and almost double flowering time to 114 days after emergence (FD50) following three

generations of late FT selection. With early FT selection, plant biomass and flowering height

were also significantly reduced making early FT selected populations potentially more

susceptible to crop competition.

These studies consistently demonstrated the adaptive capacity of wild radish populations. In

order to maintain the efficiency of weed control, highly-effective rates of herbicide and diverse

weed control tactics must be used to reduce the size of wild radish populations and therefore

reduce the genetic diversity that drives evolutionary shifts.

PAGE | vi

TABLE OF CONTENTS

THESIS DECLARATION I ACKNOWLEDGEMENTS III ABSTRACT V TABLE OF CONTENTS VII LIST OF FIGURES X LIST OF TABLES XII LIST OF PLATES XV LIST OF HERBICIDES XVII

CHAPTER 1. GENERAL INTRODUCTION 1 SELECTIONS HAVE CONSEQUENCES 2 SELECTIONS IN AGRICULTURE 4 Raphanus raphanistrum, A RESILIENT WEED IN AUSTRALIAN CROPPING SYSTEMS 6 BIOLOGY OF WILD RADISH 7 GENETIC DIVERSITY 8 EFFECT OF HERBICIDAL SELECTIONS ON WILD RADISH 8 “THE ARMS RACE AGAINST HERBICIDE RESISTANCE” 10 HARVEST WEED SEED CONTROL 12 CONCLUSION 12 RESEARCH OBJECTIVES 13 THESIS OUTLINE 13 REFERENCES 14

CHAPTER 2. IDENTIFICATION OF THE FIRST GLYPHOSATE-RESISTANT WILD RADISH (Raphanus raphanistrum L.) POPULATIONS 25

ABSTRACT 27 INTRODUCTION 27 MATERIALS AND METHODS 28

Collection of suspected glyphosate-resistant wild radish population 28 General procedures 28 Glyphosate dose–response on field-collected populations 28 Glyphosate dose–response on glyphosate-selected subpopulation 28 Multiple resistance profile 28 Data analysis 29

RESULTS AND DISCUSSION 29 Confirmation of glyphosate resistance 29 Evidence of multiple resistance 30

CONCLUSION 30 ACKNOWLEDGEMENTS 30 REFERENCES 30 SUPPORTING INFORMATION 32

CHAPTER 3. THE IDENTIFICATION OF GLYPHOSATE-RESISTANT Lolium rigidum AND Raphanus raphanistrum POPULATIONS FROM THE FIRST WESTERN AUSTRALIAN PLANTINGS OF GLYPHOSATE-RESISTANT CANOLA 33


Population collection 34

PAGE | vii

Resistance testing 35 2010 collected population resistance testing 36 2011 collected population resistance testing 37

Weed prevalence in 2011 GR canola fields 38 Data analysis 38

RESULTS AND DISCUSSION 39 Wild radish 39 Annual ryegrass 41 Other weed species 41

CONCLUSION 42 ACKNOWLEDGEMENTS 46 REFERENCES 46

CHAPTER 4. WILD RADISH (Raphanus raphanistrum L.) RECURRENT SELECTION WITH GLYPHOSATE OVER FOUR GENERATIONS RESULTS IN LOW-LEVEL GLYPHOSATE AND ALS HERBICIDE RESISTANCE 51


Plant material 53 General procedure for population selection 53 Final dose–response evaluations of recurrently-selected wild radish populations 54 Cross-resistance profiles of commencing susceptible (G0) and fourth generation glyphosate-selected progeny (S4) 54 Data analysis 55

RESULTS 57 Effect of sublethal glyphosate on G0 population 57 Cross-resistance pattern 57

DISCUSSION 67 Evolutionary response of wild radish to low-dose glyphosate selection 67


CHAPTER 5. RECURRENT SELECTION WITH REDUCED 2,4-D AMINE RATES RESULTS IN THE RAPID EVOLUTION OF 2,4-D AND ALS HERBICIDE RESISTANCE IN WILD RADISH (Raphanus raphanistrum L.) 75


Plant material 77 General procedure for population selection 77 Dose–response analysis 79 Cross-resistance profiles of commencing susceptible (G0) and fourth generation 2,4-D amine-selected progeny (S4) 79 Data analysis 80

RESULTS 80 Effect of sublethal 2,4-D amine selections on the commencing population (G0) 80 Cross-resistance to chlorsulfuron and metasulam 81

DISCUSSION 93 Wild radish response to recurrent low dose 2,4-D amine selection 93 Cross resistance 94 Implication of auxinic herbicide rate on resistance evolution 95

PAGE | viii


CHAPTER 6. RECURRENT SELECTION ON WILD RADISH (Raphanus raphanistrum L.) FLOWERING TIME LEADS TO RAPID CHANGES IN GENOTYPE 107


Population 109 Initial flowering date selection procedure 110 Subsequent selections general procedure 110 Analysis of selection and crossing lines 114 Data analysis 114

RESULTS 115 Effect of recurrent early-flowering time selection 115 Effect of recurrent late-flowering time selection 116

DISCUSSION 125 Early selection 125 Adaptability of wild radish populations to selection 125

Implications for harvest weed seed control resistance 127 CONCLUSION 128 ACKNOWLEDGEMENTS 131 REFERENCES 131

CHAPTER 7. GENERAL CONCLUSIONS 137 KEY OBSERVATIONS AND RESISTANCE MANAGEMENT IMPLICATIONS 138 FUTURE RESEARCH PRIORITIES 142 FINAL REMARK 143 REFERENCES 143

PAGE | ix

LIST OF FIGURES

FIGURE 1.1. DARWIN’S INITIAL [NOTE: “I THINK.”] 1837 SKETCH OF THE “TREE OF LIFE” WHERE DARWIN PROPOSED THAT ALL SPECIES DIVERSIFIED FROM A SINGLE ORIGIN (DARWIN 1938). 3 FIGURE 1.2. SHIFT IN A POPULATION AS A RESULT OF SELECTING ONE PHENOTYPIC EXTREME [ADAPTED FROM NEVE (2013)]. 4 FIGURE 1.3. A: THE GLOBAL INCREASE IN HERBICIDE-RESISTANT CASES. B: NUMBER OF SPECIES RESISTANT TO DIFFERENT HERBICIDE MODE OF ACTION CLASSES (HRAC CODES) (HEAP 2014). 5 FIGURE 1.4. A: WORLDWIDE DISTRIBUTION OF WILD RADISH [REPRODUCED FROM HOLM (1997)]. B: THE DISTRIBUTION OF WILD RADISH WITHIN AUSTRALIA [REPRODUCED FROM PARSONS AND CUTHBERTSON (1992)]. 6 FIGURE 2.1. SURVIVAL DOSE–RESPONSE CURVES FOR THE SUSCEPTIBLE WILD RADISH POPULATION S1 (…○…) AND FIELD-COLLECTED GLYPHOSATE-RESISTANT POPULATIONS R1 (_ _•_ _) AND R2 (__ ▲_ _ ). EACH SYMBOL REPRESENTS THE MEAN OF FIVE TREATMENTS FOR THE SUSCEPTIBLE POPULATION AND SIX TREATMENTS FOR THE RESISTANT POPULATIONS. THE PLOTTED LINES ARE PREDICTED SURVIVAL CURVES USING A THREE-PARAMETER LOG-LOGISTIC MODEL [EQUATION (1)]. VERTICAL BARS REPRESENT THE MEAN ± SE (N=4). 29

FIGURE 2.2. SURVIVAL DOSE–RESPONSE CURVES FOR THE POOLED SUSCEPTIBLE WILD RADISH POPULATION S1, S2 AND S3 (…○…) AND PROGENY SUBPOPULATION OF GLYPHOSATE-RESISTANT R1 (_ _•_ _) AND R2 (__ ▲_ _ ). EACH SYMBOL REPRESENTS THE MEAN OF SEVEN TREATMENTS FOR THE SUSCEPTIBLE POPULATIONS AND EIGHT TREATMENTS FOR THE RESISTANT POPULATIONS. THE PLOTTED LINES ARE PREDICTED SURVIVAL CURVES USING A THREE-PARAMETER LOG-LOGISTIC MODEL [EQUATION (1)]. VERTICAL BARS REPRESENT THE MEAN ± SE (N=5). 30

FIGURE 3.1. LOCATIONS OF SURVEYED GR CANOLA FIELDS ACROSS THE WA GRAINBELT IN 2010 AND 2011. AVERAGE RAINFALL ISOHYETS SHOWN, HIGH: 450–750 MM, MEDIUM: 324–450 MM AND LOW: <325 MM. 35 FIGURE 4.1. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE LOW DOSE GLYPHOSATE SELECTED GENERATIONS S1 (…□…), S2 (…∆…), S3 (…◊…) AND S4 (…○…). EACH SYMBOL REPRESENTS THE MEAN OF EIGHT RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 SE OF THE MEAN. 59 FIGURE 4.2. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE UNSELECTED CONTROL GENERATIONS C1 (…□…), C2 (…∆…), C3 (…◊…) AND C4 (…○…), DERIVED FROM CROSSING 20 RANDOMLY-SELECTED PLANTS IN THE ABSENCE OF SELECTION. EACH SYMBOL REPRESENTS THE MEAN OF EIGHT RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 SE OF THE MEAN. 60 FIGURE 4.3. DOSE–RESPONSE SURVIVAL CURVES FOR IMAZAMOX (A) AND METOSULAM (B) FOR THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE GLYPHOSATE-SELECTED GENERATION S4 (…○…). EACH SYMBOL REPRESENTS THE MEAN OF SIX RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED SURVIVAL CURVES USING A THREE PARAMETER LOG LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 SE OF THE MEAN. 64 FIGURE 5.1. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE LOW-DOSE 2,4-D AMINE-SELECTED GENERATIONS S1 (…□…), S2 (…∆…), S3 (…◊… ) AND S4 (…○…). EACH SYMBOL REPRESENTS THE MEAN OF NINE RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT SE OF THE MEAN. 83 FIGURE 5.2. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE UNSELECTED CONTROL GENERATIONS C1 (…□…), C2 (…∆…), C3 (…◊…) AND C4 (…○…), DERIVED FROM CROSSING 20 RANDOMLY-SELECTED PROGENY. EACH SYMBOL REPRESENTS THE MEAN OF EIGHT RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 SE OF THE MEAN. 84 FIGURE 5.3. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION (G0) (__•__) AND THE FOURTH LOW DOSE 2,4-D AMINE SELECTED GENERATION (S4) (…∆…)

PAGE | x

FOLLOWING CHLORSULFURON TREATMENT. EACH SYMBOL REPRESENTS THE MEAN OF FOUR REPLICATES. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 SE OF THE MEAN. 88 FIGURE 5.4. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION (G0) (__•__) AND THE FOURTH LOW DOSE 2,4-D AMINE SELECTED GENERATION (S4) (…∆…) FOLLOWING METOSULAM TREATMENT. EACH SYMBOL REPRESENTS THE MEAN OF FOUR REPLICATES. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 SE OF THE MEAN. 89 FIGURE 5.5. DOSE–RESPONSE CURVES OF SURVIVAL (A) AND BIOMASS (B) FOR MALATHION PRE-TREATMENT PRIOR TO CHLORSULFURON. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG LOGISTIC MODEL [1] . DASHED LINES AND CIRCULAR SYMBOLS (- - - - -) DENOTE THE COMMENCING WILD RADISH POPULATION (G0); SOLID LINES AND TRIANGULAR SYMBOLS (_____) DENOTE THE FOURTH LOW DOSE 2,4-D AMINE SELECTED GENERATION (S4). SOLID SYMBOLS (▲, •) DENOTE NO MALATHION PRE-TREATMENT. HOLLOW SYMBOLS (∆,◦) DENOTE MALATHION PRE-TREATMENT. VALUES ARE THE MEAN OF FOUR REPLICATES WITH VERTICAL BARS REPRESENTING 1 SE OF THE MEAN. 91 FIGURE 6.1. HIERARCHY OF FLOWERING TIME SELECTION APPLIED TO THE COMMENCING WILD RADISH POPULATION (G0). 112 FIGURE 6.2. THE OBSERVED POPULATION RESPONSE TO EARLY FLOWERING TIME SELECTION AGAINST THE UNSELECTED COMMENCING WILD RADISH POPULATION G0 (__•__). EARLY FLOWERING TIME SELECTED GENERATIONS EF1 (…∆…), EF2 (…X…), EF3 (…□… ), EF4 (…◊…) AND EF5 (…○…). EACH SYMBOL REPRESENTS CUMULATIVE DATA POINTS OF 75 REPLICATE PLANTS. THE PLOTTED LINES ARE PREDICTED CUMULATIVE FLOWERING DATE CURVES FITTED TO A FOUR PARAMETER LOGISTIC MODEL [1]. 117 FIGURE 6.3. THE OBSERVED POPULATION RESPONSE TO LATE FLOWERING TIME SELECTION AGAINST THE UNSELECTED COMMENCING WILD RADISH POPULATION G0 (__•__). LATE SELECTED GENERATIONS LF1 (…X…), LF2 (…∆…) AND LF3 (…□…). EACH SYMBOL REPRESENTS CUMULATIVE DATA POINTS OF 75 REPLICATE PLANTS (EXCEPT LF3 = 65 PLANTS (MAX 86%)). THE PLOTTED LINES ARE PREDICTED CUMULATIVE FLOWERING DATE CURVES FITTED TO A FOUR-PARAMETER LOGISTIC MODEL [1]. 119 FIGURE 6.4. POPULATION BIOMASS RESPONSE AT FLOWERING FOR THE COMMENCING (G0), EARLY SELECTED (EF1–EF5) AND LATE SELECTED (LF1–LF3) GENERATIONS. EACH SYMBOL REPRESENTS THE POPULATION MEAN OF THE SELECTED POPULATION WITH THE VERTICAL AND HORIZONTAL BARS REPRESENTING 1 SD OF THE MEAN (N=75; LF3 N=64). 123 FIGURE 6.5. POPULATION HEIGHT RESPONSE AT FLOWERING FOR THE COMMENCING (G0), EARLY SELECTED (EF1–EF5) AND LATE SELECTED (LF1–LF3) GENERATIONS. EACH SYMBOL REPRESENTS THE POPULATION MEAN WITH THE VERTICAL BARS REPRESENTING 1 SD OF THE MEAN (N=75; LF3 N=64). 124

PAGE | xi

LIST OF TABLES

TABLE 1.1. CHANGE IN HERBICIDE RESISTANCE STATUS FOR WILD RADISH BETWEEN 2003 AND 2010 FOR WILD RADISH POPULATIONS FROM THE WESTERN AUSTRALIAN GRAINBELT. R: RESISTANT POPULATION (SURVIVAL GREATER THAN 20%), D: DEVELOPING RESISTANCE (SURVIVAL LESS THAN 20%), S: SUSCEPTIBLE POPULATION (CONTROL IS EQUIVALENT TO A SUSCEPTIBLE CONTROL BIOTYPE). 10 TABLE 2.1. LOCATION AND DATE OF COLLECTION OF GLYPHOSATE-RESISTANT AND GLYPHOSATE-SUSCEPTIBLE WILD RADISH POPULATIONS. 28 TABLE 2.2. MULTIPLE RESISTANCE TRAITS IN FIELD-COLLECTED (G0) WILD RADISH POPULATIONS R1 AND R2 (ZERO INDICATES FULLY SUSCEPTIBLE POPULATIONS, 1–19% DEVELOPING RESISTANCE AND >20% SURVIVAL RESULTS IN CLASSIFICATION AS RESISTANT POPULATION). STANDARD ERRORS ARE IN PARENTHESES. 29 TABLE 2.3. PARAMETER ESTIMATES FOR SURVIVAL FROM THE THREE-PARAMETER LOG-LOGISTIC MODEL [EQUATION (1)] USED TO CALCULATE LD50 AND R/S VALUES FOR WILD RADISH CONTROL POPULATIONS (S1, S2 AND S3) AND THE FIELD-COLLECTED (G0) AND THE PROGENY SUBSET GLYPHOSATE-RESISTANT POPULATIONS (G1) (R1 AND R2) TREATED WITH A RANGE OF GLYPHOSATE DOSES (STANDARD ERRORS IN PARENTHESES)A. 30

TABLE 3.1. FREQUENCY AND DENSITY OF WEED SPECIES PRESENT IN 166 GR CANOLA FIELDS IN 2011. FIGURES DENOTE MEAN PLANT DENSITY OF TEN REPLICATE (250 MM X 250 MM) QUADRANT RATINGS. PLANT DENSITY PRIOR TO GLYPHOSATE TREATMENT DENOTED BY SURVIVING AND DAMAGED PLANTS AT ADVANCED GROWTH STAGE (>3 LEAF FOR GRASSES, >2 LEAF FOR DICOTS). PARENTHESES INDICATE SPECIES PLANT DENSITY AS A PERCENTAGE OF THE TOTAL NUMBER OF FIELDS REGISTERING THAT SPECIES. 43 TABLE 3.2. FREQUENCY AND DENSITY OF WEED SPECIES PRESENT FOLLOWING GLYPHOSATE TREATMENT IN 166 GR CANOLA FIELDS IN 2011. FIGURES DENOTE MEAN PLANT DENSITY OF TEN REPLICATE (250 MM X 250 MM) QUADRANT RATINGS. PARENTHESES INDICATE SPECIFIC PLANT DENSITIES AS A PERCENTAGE OF THE TOTAL NUMBER OF FIELDS REGISTERING THAT SPECIES. PLANT DENSITY SURVIVING GLYPHOSATE APPLICATION DENOTED BY SURVIVING PLANTS AT ADVANCED GROWTH STAGE (>2 LEAF FOR GRASSES, >1 LEAF FOR DICOTS). LATE-EMERGING SPECIES (<3 LEAF FOR GRASSES, <2 LEAF FOR DICOTS). 44 TABLE 3.3. PARAMETER ESTIMATES FROM THREE PARAMETER LOG LOGISTIC MODEL [1] USED TO CALCULATE LD50 AND R/S VALUES FOR SUSCEPTIBLE WILD RADISH POPULATIONS (S1–S3) AND FIELD COLLECTED (R) AND FIELD COLLECTED PROGENY (R1) POPULATIONS. STANDARD ERRORS IN PARENTHESES. 45 TABLE 4.1. SEED PROGENY COLLECTED FROM BASAL WILD RADISH POPULATION WARR 7 [G0]. SELECTED PLANTS (N) AT A SPECIFIC SUBLETHAL DOSE OF GLYPHOSATE (G HA–1). GLYPHOSATE RECURRENT SELECTED GENERATIONS S1–S4. RECURRENTLY UNSELECTED GENERATIONS C1–C4. 56 TABLE 4.2. PARAMETER ESTIMATES FOR SURVIVAL FROM THE THREE PARAMETER LOG LOGISTIC MODEL [1] USED TO ESTIMATE LD50 PARAMETERS FOR PERCENT SURVIVAL BY NON-LINEAR REGRESSION ANALYSIS. STANDARD ERRORS FOR EACH PARAMETER ESTIMATE ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF LD50 FOR THE COMMENCING (G0) AND RESPECTIVE GLYPHOSATE SELECTED (S1–S4) OR UNSELECTED CONTROL (C1–C4) POPULATIONS. 61 TABLE 4.3. PARAMETER ESTIMATES FOR BIOMASS FOR EACH GLYPHOSATE SELECTED AND UNSELECTED POPULATION USING THE THREE PARAMETER LOG LOGISTIC MODEL [1] TO ESTIMATE GR50 PARAMETERS. STANDARD ERRORS FOR EACH PARAMETER ESTIMATE ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF GR50 FOR THE COMMENCING (G0) AND THE SELECTED (S1–S4) OR CONTROL (C1–C4) POPULATIONS. 62 TABLE 4.4. POPULATION SURVIVAL FOR THE COMMENCING WILD RADISH POPULATION (G0), GLYPHOSATE-SELECTED GENERATION (S4) AND HERBICIDE-SELECTED (S4) PROGENY WHEN SPRAYED AT THE RECOMMENDED RATE OF A RANGE OF HERBICIDES WITH KNOWN ACTIVITY AGAINST WILD RADISH. EACH RESULT IS THE MEAN OF FOUR REPLICATES. STANDARD ERRORS OF THE MEANS ARE IN PARENTHESES. 63 TABLE 4.5. PARAMETER ESTIMATES FOR SURVIVAL FOR IMAZAMOX, METOSULAM AND CHLORSULFURON USING THE THREE PARAMETER LOG LOGISTIC MODEL [1] TO ESTIMATE LD50 PARAMETERS FOR THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE GLYPHOSATE-SELECTED GENERATION (S4) (…○…). STANDARD ERRORS FOR EACH PARAMETER ESTIMATE ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF LD50 FOR THE COMMENCING POPULATION (G0) AND THE GLYPHOSATE-SELECTED GENERATION (S4). 65

PAGE | xii

TABLE 4.6. PARAMETER ESTIMATES FOR BIOMASS FOR IMAZAMOX, METOSULAM AND CHLORSULFURON USING THE THREE PARAMETER LOG LOGISTIC MODEL [1] TO ESTIMATE GR50 PARAMETERS FOR THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE GLYPHOSATE-SELECTED GENERATION (S4) (…○…). STANDARD ERRORS FOR EACH PARAMETER ESTIMATE ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF GR50 FOR THE COMMENCING POPULATION (G0) AND THE GLYPHOSATE-SELECTED GENERATION (S4). 66 TABLE 5.1. SEED PROGENY COLLECTED FROM BASAL WILD RADISH POPULATION WARR 7 [G0]. SELECTED PLANTS (N) AT A SPECIFIC SUBLETHAL DOSE OF 2,4-D AMINE (G HA–1). 2,4-D AMINE RECURRENT SELECTED GENERATIONS S1–S4. RECURRENTLY UNSELECTED GENERATIONS C1–C4. 78 TABLE 5.2. PARAMETER ESTIMATES FROM THE THREE PARAMETER LOG LOGISTIC MODEL [1] USED TO ESTIMATE LD50 PARAMETERS FOR PERCENT SURVIVAL BY NON-LINEAR REGRESSION ANALYSIS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF LD50 FOR THE COMMENCING POPULATION (G0) AND RESPECTIVE 2,4-D AMINE-SELECTED (S1–S4) OR CONTROL (C1–C4) POPULATIONS. 85 TABLE 5.3. PARAMETER ESTIMATES FROM THE THREE PARAMETER LOG LOGISTIC MODEL [1] USED TO ESTIMATE GR50 PARAMETERS FOR BIOMASS IN THE FORM OF PERCENT MEAN CONTROL CALCULATED BY NON-LINEAR REGRESSION ANALYSIS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF GR50 FOR THE COMMENCING POPULATION (G0) AND RESPECTIVE 2,4-D AMINE-SELECTED (S1–S4) OR CONTROL (C1–C4) POPULATIONS. 86 TABLE 5.4. POPULATION SURVIVAL FOR THE COMMENCING WILD RADISH POPULATION (G0) AND THE FOURTH LOW DOSE 2, 4-D AMINE-SELECTED GENERATION (S4) WHEN SPRAYED AT THE RECOMMENDED RATE WITH A RANGE OF HERBICIDES WITH KNOWN ACTIVITY AGAINST WILD RADISH. EACH RESULT IS THE MEAN OF FOUR REPLICATES. STANDARD ERRORS OF THE MEANS ARE IN PARENTHESES. 87 TABLE 5.5. PARAMETER ESTIMATES FOR CHLORSULFURON, METOSULAM, IMAZAMOX AND SULFOMETURON-METHYL CALCULATED BY NON-LINEAR REGRESSION ANALYSIS USING A THREE PARAMETER LOG LOGISTIC MODEL [1] TO ESTIMATE LD50 PARAMETERS FOR SURVIVAL AND GR50 PARAMETERS FOR BIOMASS (% OF MEAN CONTROL). STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF LD50 OR GR50 BETWEEN THE COMMENCING POPULATION (G0) AND THE FOURTH LOW DOSE 2,4-D AMINE-SELECTED GENERATION (S4). 90 TABLE 5.6. PARAMETER ESTIMATES FOR MALATHION PRE-TREATMENT PRIOR TO CHLORSULFURON, CALCULATED BY NON-LINEAR REGRESSION ANALYSIS USING A THREE PARAMETER LOG LOGISTIC MODEL [1] TO ESTIMATE LD50 PARAMETERS FOR SURVIVAL (A) AND GR50 PARAMETERS FOR PLANT BIOMASS (% UNTREATED CONTROL) (B). STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. PRE-TREATMENT RATIOS WERE CALCULATED AS THE RATIO OF LD50 OR GR50 BETWEEN THE MALATHION PRE-TREATED AND NON-PRE-TREATED PLANTS WITHIN BOTH THE COMMENCING POPULATION (G0) OR THE FOURTH LOW DOSE 2,4-D AMINE SELECTED GENERATION (S4). 92 TABLE 6.1. FLOWERING TIME ADVANCEMENT (DAYS TO FLOWERING AND CUMULATIVE GROWING DAY DEGREES (GDD)) OF EACH WILD RADISH ACCESSION DURING SELECTION. 113 TABLE 6.2. PARAMETER ESTIMATES (DAYS TO FLOWERING) FOLLOWING EARLY FLOWERING TIME SELECTION USING THE FOUR PARAMETER LOGISTIC MODEL [1] USED TO ESTIMATE FD50 PARAMETERS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. SELECTION RATIOS WERE CALCULATED BASED ON FD50 VALUES FOR THE UNSELECTED COMMENCING POPULATION (G0) AND RESPECTIVE SELECTED OR UNSELECTED CONTROL PROGENY. 118 TABLE 6.3. PARAMETER ESTIMATES FOR THE DAYS TO FLOWERING FOLLOWING LATE FLOWERING SELECTION USING THE FOUR PARAMETER LOGISTIC MODEL [1] USED TO ESTIMATE FD50 PARAMETERS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. SELECTION RATIOS WERE CALCULATED BASED ON FD50 VALUES FOR THE UNSELECTED COMMENCING POPULATION (G0) AND RESPECTIVE SELECTED OR UNSELECTED CONTROL PROGENY. 120 TABLE 6.4. PARAMETER ESTIMATES OF THE CUMULATIVE DAY DEGREES (GDD) UNTIL FLOWERING FOLLOWING EARLY FLOWERING DATE SELECTION USING THE THREE PARAMETER LOGISTIC MODEL [1] TO ESTIMATE FD50 GDD PARAMETERS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. SELECTION RATIOS WERE CALCULATED BASED ON FD50 VALUES FOR THE UNSELECTED COMMENCING POPULATION (G0) AND RESPECTIVE SELECTED OR UNSELECTED CONTROL PROGENY. 121

PAGE | xiii

TABLE 6.5. PARAMETER ESTIMATES OF THE CUMULATIVE DAY DEGREES (GDD) UNTIL FLOWERING FOLLOWING LATE FLOWERING SELECTION USING THE FOUR PARAMETER LOGISTIC MODEL [1] USED TO ESTIMATE FD50 GDD PARAMETERS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. SELECTION RATIOS WERE CALCULATED BASED ON FD50 VALUES FOR THE UNSELECTED COMMENCING POPULATION (G0) AND RESPECTIVE SELECTED OR UNSELECTED CONTROL PROGENY. 122

PAGE | xiv

LIST OF PLATES

PLATE 4.1. PHOTOGRAPHIC REPRESENTATION OF THE RESPONSE OF THE GLYPHOSATE-SELECTED GENERATION TO GLYPHOSATE. G0 REPRESENTS THE UNSELECTED, COMMENCING WILD RADISH POPULATION. S1–S4 REPRESENT FOUR GLYPHOSATE-SELECTED GENERATIONS (S1–S4). 69 PLATE 5.1. PHOTOGRAPHIC REPRESENTATION OF THE 2,4-D AMINE DOSE–RESPONSE QUANTIFIED IN FIGURE 5.1. G0 REPRESENTS THE UNSELECTED, SUSCEPTIBLE BASE POPULATION. GENERATIONS S1–S4 REPRESENT THE POPULATION RESPONSE TO FOUR GENERATIONS OF 2,4-D AMINE SELECTION. 97 PLATE 5.2. PHOTOGRAPHIC REPRESENTATION OF THE CHLORSULFURON CROSS RESISTANCE RESPONSE TO MALATHION PRE-TREATMENT IN FIGURE 5.5. G0 REPRESENTS THE UNSELECTED, SUSCEPTIBLE BASE POPULATION. S4 REPRESENTS THE FOURTH GENERATION OF 2,4-D SELECTION. MALATHION TREATMENT WAS APPLIED 30MIN BEFORE CHLORSULFURON. 98 PLATE 6.1. PHOTOGRAPHIC EXAMPLES OF FT SELECTED POPULATION BIOMASS AND HEIGHT. LEFT: FIVE TIMES EARLY FT SELECTED GENERATION EF5 (BIOMASS 2.4 G, FLOWERING HEIGHT 17 CM). RIGHT: THREE TIME LATE FT SELECTED GENERATION LF3 PRIOR TO FLOWERING (BIOMASS 72 G, PLANT HEIGHT 146 CM) 126 PLATE 6.2. VISUAL DIFFERENCES IN WILD RADISH REPRODUCTIVE CAPACITY (UNMEASURED) WITH EARLY FT SELECTION. LEFT: FOUR GENERATIONS OF EARLY FT SELECTION. RIGHT: UNSELECTED COMMENCING POPULATION G0. 129 PLATE 6.3. PROSTRATE GROWTH IN THE EF5 GENERATION DUE TO THE EFFECTS OF LOW BIOMASS ACCUMULATION PRIOR TO FLOWERING. 129 PLATE 6.4. GLASSHOUSE AND POT IRRIGATION LAYOUT DURING FINAL COMPARISON OF ALL SELECTED AND CONTROL LINES. 130

PAGE | xv

PAGE | xvi

LIST OF HERBICIDES

COMMON AND CHEMICAL NAMES OF HERBICIDES DISCUSSED IN THIS THESIS (WSSA 2010).

Common name Chemical name

Atrazine 6-chloro-N-ethyl-N9-(1-methylethyl)-1,3,5-triazine-2,4-diamine

Bromoxynil 3,5-dibromo-4-hydroxybenzonitrile

Chlorsulfuron 2-chloro-N-[[(4-methoxy-6-methyl-1,3,5-triazin-2-

yl)amino]carbonyl]benzenesulfonamide

Diflufenican N-(2,4-difluorophenyl)-2-[3-(trifluoromethyl)phenoxy]-3-

pyridinecarboxamide

Diuron N-(3,4-dichlorophenyl)-N,N-dimethylurea

Diquat 6,7-dihydrodipyrido[1,2-a:29,19-c]pyrazinedium ion

Glyphosate N-(phosphonomethyl)glycine

Imazapyr 2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl]-3-

pyridinecarboxylic acid

Imazamox 2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl]-5-

(methoxymethyl)-3-pyridinecarboxylic acid

MCPA 4-chloro-2-methylphenoxyacetic acid

Metosulam N-(2,6-dichloro-3-methylphenyl)-5,7-dimethoxy[1,2,4]triazolo[1,5-

a]pyrimidine-2-sulfonamide

Metribuzin 4-amino-6-(1,1-dimethylethyl)-3-(methylthio)-1,2,4-triazin-5(4H)-one

Metsulfuron 2-[[[[(4-methoxy-6-methyl-1,3,5-triazin-2-

yl)amino]carbonyl]amino]sulfonyl]benzoic acid

Paraquat 1,19-dimethyl-4,49-bipyridinium ion

Pyrasulfatole 5-hydroxy-1,3-dimethyl-1H-pyrazol-4-yl)[2-(methylsulfonyl)-4-

(trifluoromethyl)phenyl]methanone

Sulfometuron-

methyl

2-[[[[(4,6-dimethyl-2-pyrimidinyl)amino]carbonyl]amino]sulfonyl]benzoic

acid

2,4-D (2,4-dichlorophenoxy)acetic acid

PAGE | xvii

PAGE | xviii

CHAPTER 1. GENERAL INTRODUCTION

CHAPTER 1.

GENERAL INTRODUCTION

"It takes all the running you can do, to keep in the same place"

Lewis Carroll (1960) The Garden of Live Flowers (Chapter 2)

Alice's Adventures in Wonderland and Through the Looking-Glass

Illustrated by John Tenniel

The New American Library, New York

The advent of agriculture some ten thousand years ago allowed mankind to efficiently produce

food surpluses for the first time. This in turn released constraints on population size (Rosenberg

1990), resulting in the world welcoming its seven billionth inhabitant in October 2011. However

the world’s population continues to climb, and is projected to increase to between 8.1 billion

and 10.6 billion people by 2050 (Jabeen 2014). Together with increased affluence and this

rapidly increasing population, the issues surrounding food security have never been greater

(Godfray et al. 2010). Of all global food commodities, field grains have made one of the greatest

contributions towards alleviating worldwide hunger, by providing a readily storable and

transportable source of food (Borlaug 2002; Sakschewski et al. 2014). However, the security of

world grain production is continually threatened by the presence of crop-infesting, competitive

weed species (Powles and Yu 2010).

In order to address this annual threat of crop weeds, the agrichemical industry developed a

range of highly-effective herbicidal products (starting with the auxinic herbicide, 2,4-D and

MCPA in 1946 (Weintraub 1953)) that efficiently remove weeds from fields. In Australia, the

introduction of herbicides enabled growers to increase the frequency and scale of dryland

cropping (Powles et al. 1997) with 24 million ha (or just over 3% of the Australian continent)

now devoted to grain production (ABARE 2013; BRS 2013). With the relatively low gross margins

in Australian grain cropping, Australian family farming properties are large (average >3000 ha)

(ABARES 2013). In order to efficiently control weeds over such large areas, grain growers rely on

herbicides for efficient weed control. However as stated by Powles and Yu (2010), “Despite the

PAGE | 1


efficiency of herbicides, they have not led to the extinction of weeds. Rather, evolutionary forces

have acted upon the genetic diversity that exists within large weed populations, to select for

gene traits that endow herbicide resistance”. Following 50 years of herbicide over-reliance in

Australia, 38 weed species have evolved herbicide resistance to 11 of the 14 commonly-used

herbicide modes of action. This widespread resistance poses a significant threat to the

sustainability of grain production in Australia, which is one of the world’s largest grains exporting

nations (ABARES 2014).

SELECTIONS HAVE CONSEQUENCES ON POPULATIONS

The concept that the selection of advantageous traits has consequences was first conceived by

the ancient Greek philosopher Anaxiamander (611–547 B.C.) who predicted that the nature of

species change over time in order to thrive in continually-changing environments (Wright 1984).

However it was not until the beginning of the 19th century, when the first mechanistic concept

driving evolutionary change in species was presented by Jean Baptiste de Lamarck (1744–1829)

in the book [translated] ‘Zoological Philosophy: Exposition with Regard to the Natural History of

Animals’ (Lamarck 1809). In this work, Lamarck (1809) conceptualised that characteristics in

species could be directly inherited by progeny. In Lamarck’s (1809) theory, inherited traits were

‘acquired’ through an individual’s interaction with the environment. Even though Lamarck’s

theories have been widely disputed, they were the first real attempt to link the observed

diversity in species to external stresses namely the environment (Wright 1984; Kutschera and

Niklas 2004).

Soon after Lamarck (1809), the first widely-accepted theories outlining a mechanism-driving

evolutionary adaptation of species was independently co-postulated by Alfred Russel Wallace

(1823–1913) in ‘On the laws which has regulated the introduction of species‘ in 1855 and Charles

Robert Darwin (1809–1882) in the ‘On the origins of species’ in 1859 (Wallace 1855; Darwin

1859). At the time, Darwin (1859) referred to the diversity present in nature as “the mystery of

mysteries”, with both Darwin and Wallace’s work linking the source of this observed diversity to

the selection of advantageous traits (Wallace 1855; Darwin and Wallace 1858; Darwin 1859).

However Darwin’s (1859) theory emphasised the role of competition between individuals,

whereas Wallace (1855) emphasised the effect of environmental stress on a whole species,

forcing them to adapt. Darwin (1859) most aptly stated that: “any extremely slight modification

in structure nor habit within one individual would give it an advantage over another”. In Darwin’s

(1859) theory, characteristics that provide either increased survival or allow for the production

of more offspring would increase in frequency and in turn become increasingly dominant,

leading to continuous adaptation (Wright 1984) or micro-evolution (Dawkins and Krebs 1979).

PAGE | 2


Darwin (1859) went on to conceptualise that all life may have descended from a single primitive

form (Figure 1.1). While this concept was highly controversial, it highlights the evolutionary

potential of Darwin’s and Wallace’s theories to evolve an almost endless array of specific

adaptations in response to highly specific stresses (Darwin and Wallace 1858; Dawkins and Krebs

1979).

FIGURE 1.1. DARWIN’S INITIAL [NOTE: “I THINK.”] 1837 SKETCH OF THE “TREE OF LIFE” WHERE DARWIN

PROPOSED THAT ALL SPECIES DIVERSIFIED FROM A SINGLE ORIGIN (DARWIN 1938).

The genetics underpinning Darwin’s (1859) evolutionary theory was provided by the Augustinian

friar, Gregor Johann Mendel (1822–1884), in his classical paper presented at the Brünn Natural

History Society in 1865. At the time, Mendel’s theory was initially dismissed and went

unacknowledged for more than three decades (Mendel 1866). Mendel’s observations on the

heredity of flowering traits in pea plants (Pisum sativum L.) formed the basis of two fundamental

principles of genetics: (1) the principle of segregation and (2) the principle of individual

assortment. The principle of segregation suggests that for any particular trait, a pair of alleles

are presented by each parent, which in turn separate with only one allele from each parent

passing on to the progeny as a matter of chance. According to the principle of independent

assortment, different pairs of these alleles can be passed onto different progeny independently

(Mendel 1866; Fisher 1936). Through the diverse recombination patterns of genes, Mendel

PAGE | 3


indicated that progeny can exhibit traits that were not expressed in either parent (Fairbanks and

Rytting 2001).

SELECTIONS IN AGRICULTURE

In Darwin's (1859) and Wallace’s (1855) theories of evolution, selective forces were based on

large populations competing for limited resources (e.g. food). In the context of this thesis these

selective forces can be in the form of singular catastrophic events such as a herbicide (or other

weed control tactics) apply a strong directional selective force, recurrently selecting for one

phenotypic extreme, outside the range of susceptible phenotypes (Figure 1.2) (Powles and Yu

2010). Soon after herbicides were developed, the plausibility that weeds may evolve resistance

was first raised by Blackman (1950) and Abel (1954), with Harper (1956) stating that “the most

effective intensity of selection is one that reduces a population to a small resistant residue,

capable of rapid reproduction”.

FIGURE 1.2. SHIFT IN A POPULATION AS A RESULT OF SELECTING ONE PHENOTYPIC EXTREME [ADAPTED FROM NEVE

(2013)].

Since Harper’s prediction (1956), the recurrent application of agrichemicals has led to the

evolution of organisms resistant to herbicides (Holt et al. 1993; Powles and Yu 2010), insecticides

(Melander 1914; Roush and McKenzie 1987) and fungicides (Delp 1980; Ma and Michailides

2005). In plants, herbicide resistance is defined as “the inherited ability of a plant to survive and

reproduce following exposure to a dose of herbicide normally lethal to the wild type” (Vencill et

al. 2012). Just 11 years after the first herbicide, 2,4-D (2,4-dichlorophenoxyacetic acid) was

introduced, resistance was first identified in fireweed (Erechtites hieracifolia L.) (Hilton 1957)

and wild carrot (Daucus carota L.) (Switzer 1957; Whitehead and Switzer 1963). Following 60

years of herbicide over-reliance, herbicide resistance is now a rapidly escalating agronomic

problem (Figure 1.3 A.), confirmed in 235 plant species worldwide (comprising of 138 dicots and

97 monocots) to 21 of the 25 known herbicide modes of action (Figure 1.3 B.)(Heap 2014).

PAGE | 4


A. B.

FIGURE 1.3. A: THE GLOBAL INCREASE IN HERBICIDE-RESISTANT CASES. B: NUMBER OF SPECIES RESISTANT TO

DIFFERENT HERBICIDE MODE OF ACTION CLASSES (HRAC CODES) (HEAP 2014).

In most of the described cases of herbicide resistance, resistance has been endowed by the

selection of monogenic gene traits (Darmency 1994; Powles and Yu 2010). This is thought to be

the result of strong herbicide selection continually acting outside the phenotypic range of

susceptible phenotypes, selecting only strongly-resistant individuals that possess a rare gene

mutation (Powles 2010). The selection of these individuals results in large, sudden changes in

the resistance status of a population (Powles and Yu 2010). The evolutionary dynamics of

monogenic resistance is affected by mutation frequency, inheritance (nuclear vs. maternal),

dominance of the trait, population size and dispersal of resistance alleles within and between

weed populations (Darmency 1994; Jasieniuk 1996) .

However herbicide resistance can also be selected at doses that do not provide full control.

When selection acts within the range of the population's standing genetic variation, multiple

weak gene traits are selected and accumulated in the survivors, leading to the gradual evolution

of ‘polygenic’ herbicide resistance (Neve, 2005; Powles 2010; Neve 2014). At low rates, the

selection for herbicide resistance is not confined by population size as the multiple, weak

herbicide resistance endowing gene traits required for resistance pre-exist within the population

(Matthews 1996; Délye et al. 2013). What is especially threatening and unpredictable about the

evolution of polygenic resistance is the evolution of ‘generalist’ resistance biotypes that exhibit

cross resistance to herbicides yet to be in contact with the population or yet to be developed

(Hall et al. 2001; Powles and Yu 2010; Beckie and Tardif 2012).

PAGE | 5


Raphanus raphanistrum, A RESILIENT WEED IN AUSTRALIAN CROPPING SYSTEMS

Wild radish (Raphanus raphanistrum L.) is a member of the Brassicaceae family, originating from

the Mediterranean region before spreading north during the pre-Christian era through the trade

of produce and livestock (Chater 1964; Donaldson 1986; Holm 1997). Today, wild radish is a

global weed, colonising every continent except Antarctica (Figure 1.4A) (Bendtham 1954). In

Australia, wild radish is thought to have been accidently introduced multiple times (Woolls 1867)

via grain and hay imports in the middle of the 19th century (Vogel 1926). With the movement of

produce, machinery and livestock, wild radish has colonised every state of Australia (Figure 1.4B)

thriving in acid to neutral sands and loams (Woolls 1867; Gardner 1930). Wild radish is listed in

the top 180 worst agricultural weeds globally (Snow 2005) causing management problems for

45 different crop species in over 65 countries (Snow 2005). In Australia, wild radish was first

listed as a declared weed in 1909 (Gardner 1930) and is now considered the most problematic

dicot weed species in the Western Australian (WA) grainbelt (Alemseged et al. 2001; Borger et

al. 2012).

A. B.

FIGURE 1.4. A: WORLDWIDE DISTRIBUTION OF WILD RADISH [REPRODUCED FROM HOLM (1997)]. B: THE

DISTRIBUTION OF WILD RADISH WITHIN AUSTRALIA [REPRODUCED FROM PARSONS AND CUTHBERTSON (1992)].

Wild radish is problematic as it has a competitive advantage over grain crops, reducing yields

even at relatively at low densities (Code and Donaldson 1996; Cousens et al. 2001; Blackshaw et

al. 2002; Eslami et al. 2006). Wild radish affects wheat yield as a result of its early vigour

advantage (Cousens et al. 2001). An initial wild radish density of 200 plants m–2 reduced wheat

(Triticim aestivum L.) yield by 50% (Cheam 1995), with densities as low as 7 plants m–2 reducing

wheat yield by 10% (Code and Donaldson 1996). Wild radish plants that emerge with or shortly

after the crop cause the largest wheat yield reductions (Cheam 1995). Wild radish has an even

greater impact on canola (Brassica napus L.) yield with wild radish densities of 4 and 64 plants

PAGE | 6


m–2 reducing canola yield by 11% and 91%, respectively (Blackshaw et al. 2002). Due to the

impact wild radish has on yield, Australian grain growers spend an estimated 40 million dollars

annually on herbicidal products that specifically target wild radish (ABARE 2013).

BIOLOGY OF WILD RADISH

The biology of wild radish makes it an adaptive species that thrives despite modern weed control

systems (Borger et al. 2012; Webster and Nichols 2012). Wild radish is highly fecund, cross-

pollinated and genetically variable species. Wild radish is often present at high densities. An

isolated wild radish individual can produce up to 45,000 seeds m–2 when in competition with a

wheat crop with a population of 52 plants m–2 produced 17,000 seeds m–2 (Reeves et al. 1981).

In wheat crops, wild radish individuals can produce more than 1000 seeds m–2 (Cousens et al.

2008). The time of emergence significantly affects wild radish seed production, with early

emerging plants producing more seed (Mekenian and Willemsen 1975; Cheam 1986).

As well as being highly productive, wild radish exhibits a protracted seed germination pattern

necessitating multiple herbicide applications (Mekenian and Willemsen 1975; Cheam 1986;

Young 2001). On average it takes 4–5 days from seed imbibition to emergence, with wild radish

readily germinating over a wide temperature range (5–35°C) (Mekenian and Willemsen 1975;

Cheam 1995). Seed germination is at its greatest in southern Australia in autumn (Cheam 1986)

due to fluctuating temperatures (10–30°C) (Mekenian and Willemsen 1975; Piggin et al. 1978;

Reeves et al. 1981). After this, wild radish emergence slows depending on the innate dormancy

of the population (Piggin et al. 1978; Reeves et al. 1981; Panetta et al. 1988). The presence of

the seed pod inhibits wild radish germination with pod-extracted seed emerging 19 days after

burial compared to 58 days for pod-enclosed seed (Mekenian and Willemsen 1975; Cousens et

al. 2010). As well as protracted germination, which ensures that wild radish populations are not

easily controlled, wild radish seed can remain dormant in the soil for up to 7 years (Young 2001;

Hashem 2004; Newman 2007) with as much as 90% of freshly shed, pod-enclosed seed not

germinating the following season (Cheam 1986). The staggered germination and protracted

dormancy of wild radish allows unselected survivors to persist in cropping systems, maintaining

the genetic diversity of populations while under intense selection (Orr and Coyne 1992).

Dormancy in wild radish seed is thought to be due to the combined effects of a leachable

chemical inhibitor (Cheam 1984), which is gradually broken down after ripening, and the

presence of the seed pod that physically restricts moisture movement and emergence of the

seed radical (Young 2001).

PAGE | 7


GENETIC DIVERSITY

A striking feature of wild radish that contributes to its success as a weed is the significant genetic

variation exhibited both within and between wild radish populations (Mazer 1992; Williams

2001; Conner et al. 2003; Bhatti 2004; Madhou et al. 2005). Wild radish is a diploid species

(2n=18 chromosomes) (Cheam 1995) that is atypical of most colonising species as it is self-

incompatible and insect pollinated (Chater 1964). Outcrossing aids the significant genetic

diversity observed in wild radish populations (Conner and Via 1993), as every individual is

genetically different due to the segregation and recombination of genes from different parental

plants (David et al. 1993; Allard 1999). Due to the weight of wild radish pollen grains, wild radish

pollen is not considered readily moved by the wind (Cheam 1995), requiring an insect pollinator

for fertilisation (Chater 1964; Conner 1995). However foraging honey bees have been measured

to make singular movements greater than 300 m (Osborne et al. 1999) with a mean flight path

distance between flower visits of 0.38 m (Keasar et al. 1996), ensuring that wild radish plants

can readily share their genetic attributes over a wide area.

EFFECT OF HERBICIDAL SELECTIONS ON WILD RADISH

Herbicide dependency in Australia has resulted in the evolution of herbicide resistance in wild

radish. In Western Australia, six surveys have been conducted to identify the status of herbicide

resistance in wild radish populations (Hashem et al. 2001; Walsh et al. 2001; Hashem and

Dhammu 2002; Grima and Blake 2005; Walsh et al. 2007; Owen 2014). Of these, three surveys

were random in nature, designed to estimate the frequency and distribution of herbicide

resistance in the WA grainbelt (Walsh et al. 2001; Walsh et al. 2007; Owen 2014).

In 1999, 133 fields across the WA grainbelt were randomly sampled, resulting in the

identification of resistance to the acetolactate synthase (ALS) inhibiting herbicide, chlorsulfuron,

in 21% of populations (Table 1.1) (Walsh et al. 2001).

In 2003, 500 fields were randomly visited resulting in the collection of 90 wild radish populations

with ALS herbicide resistance identified in 32% of wild radish populations. This survey also

identified resistance to the phytoene desaturase (PDS) inhibitor, diflufenican (1%); and the

photosynthetic electron transport (PSII) inhibitor, atrazine (3%). Resistance to the synthetic

auxin 2,4-D was also identified in 3% (>20% survival) of populations, however 57% of populations

surveyed were developing auxinic herbicide resistance (<20% survival) (Table 1.1) (Walsh et al.

2007). By 2003, wild radish populations also exhibited multi-resistance, with 35% of populations

developing resistance to two modes of action, 18% to three modes of action and 7% to four

PAGE | 8


separate herbicidal modes of action (ALS, PDS, PSII inhibitors and synthetic auxins) (Walsh et al.

2004; Walsh et al. 2007).

In the latest random survey in 2010, 466 fields in the WA grainbelt were surveyed resulting in

the collection of 155 wild radish populations (Owen 2014). This survey clearly demonstrated that

the vast majority of wild radish populations in the Western Australian grainbelt were resistant

to the ALS inhibitor, chlorsulfuron (73%) (Table 1.1). Resistance to synthetic auxins (2,4-D) also

increased to 38% of populations with the majority of 2,4-D resistant populations collected from

the northern and central grainbelt. However, even though the incidence of auxinic herbicide

resistance was high, resistance was not strong (plants affected at field applied rates) with all

populations severely affected by auxinic herbicide treatment (Walsh et al. 2007; Owen 2014).

Therefore, even through 2,4-D resistance is widespread, 2,4-D is still considered effective on

most WA wild radish populations if combined with crop competition (Walsh et al. 2009). By

2010, resistance to the PDS inhibitor diflufenican was still low (5%). However a large number of

populations were developing resistance (44%) indicating that with further use resistance to PDS

inhibitors is likely to increase (Owen 2014). Even though resistance to the PSII inhibitor, atrazine,

was first identified in wild radish by Hashem et al. (2001), PSII resistance was still rare in 2010

with only one resistant population identified (Table 1.1) (Owen 2014). The reason for this lack

of PSII resistance evolution is still unknown. In 2010, the combined HPPD (4-

hydroxyphenylpyruvate dioxygenase) and PSII inhibitor Velocity (pyrasulfotole/bromoxynil) and

the EPSPS (5-enolpyruvylshikimate-3-phosphate synthase) inhibitor, glyphosate, were first

tested with no resistant wild radish populations identified (Owen 2014). However, prior to the

introduction of GR canola, glyphosate application was confined to non-selective pre-seeding

use; subsequent herbicide applications with different modes of action may have controlled any

glyphosate survivors. There have been no surveys specifically conducted following the sole use

of glyphosate.

Sampling of 130 wild radish populations in 2012 from problematic fields in Western Australia by

the agribusiness firm Landmark identified 88% of samples resistant to ALS-inhibiting herbicides,

46% resistant to PDS inhibitors, 16% resistant to synthetic auxins and 11% resistant to PSII

inhibitors. Although these results are not considered frequencies of resistance as they were

collected from fields which growers considered to have resistance, they clearly highlight that

herbicide resistance in wild radish is widespread, limiting management options for growers

(Broster 2014).

Herbicide resistance in wild radish is less evident in other states of Australia, with random field

surveys of southern New South Wales in 2011 and 2012 only identifying ALS resistance in wild

PAGE | 9


radish populations. In 2011 and 2012, 30 wild radish populations were collected from 178 fields.

Of these populations, ALS herbicide resistance was only found in two wild radish populations

(>20% survival) with an additional two populations confirmed as developing resistance (<20%

survival). In this study, no resistance to PDS inhibitors, PSII inhibitors, synthetic auxins or

glyphosate was identified (Broster et al. 2014). However, suspected resistant wild radish

populations from South Australia, New South Wales and Victoria, submitted to a commercial

testing survice have been confirmed to be resistant to PDS inhibitors and synthetic auxins

(Boutsalis 2013). Of the 61 wild radish samples submitted for herbicide resistance testing

between 2009 and 2013, 31% were resistant to the ALS-inhibiting herbicides chlorsulfuron,

metsulfuron or imazamox/imazapyr. Resistance to the synthetic auxins (2,4-D) was evident in

38% of populations, with 15% of populations resistant to the PDS inhibitor diflufenican. No

resistance to the EPSPS inhibitor glyphosate or the HPPD/PSII inhibitor mix was found (Boutsalis

2013). Herbicide resistance in Tasmanian wild radish populations is also rare with only two wild

radish populations identified as resistant to the ALS inhibitor, chlorsulfuron in 2012 (Broster

2013).

TABLE 1.1. CHANGE IN HERBICIDE RESISTANCE STATUS FOR WILD RADISH BETWEEN 2003 AND 2010 FOR WILD

RADISH POPULATIONS FROM THE WESTERN AUSTRALIAN GRAINBELT. R: RESISTANT POPULATION (SURVIVAL

GREATER THAN 20%), D: DEVELOPING RESISTANCE (SURVIVAL LESS THAN 1-19%), S: SUSCEPTIBLE POPULATION

(CONTROL IS EQUIVALENT TO A SUSCEPTIBLE CONTROL BIOTYPE).

Herbicide class 1999

(Walsh et al. 2001)

2003

(Walsh et al. 2007)

2010

(Owen 2014)

R R D S R D S

ALS inhibitors 21 32 22 46 73 11 16

PDS inhibitors – 1 45 54 5 44 51

Auxins – 3 57 40 38 38 24

PSII inhibitors – 3 12 85 1 0 99

EPSPS inhibitors – – – – 0 0 100

“THE ARMS RACE AGAINST HERBICIDE RESISTANCE”

With the rapid escalation of herbicide resistance biotypes globally (Figure 1.3A) (Heap 2014) and

little likelihood of new herbicidal modes of action becoming available in the near future (Duke

2012), the agrochemical industry has had to find alternative uses for the remaining effective

herbicides. One alternative is the development of transgenic (GM) glyphosate-resistant crops,

PAGE | 10


using recombinant DNA techniques (Jones 2011). In 1996, this technology revolutionised weed

control when Monsanto™ introduced transgenic (GM), glyphosate-resistant (Roundup Ready™)

crops (Padgette et al. 1996). In Roundup Ready™ crops, weed control was simplified as the highly

effective non-selective herbicide glyphosate could be used throughout a prolonged window of

crop development to efficiently and economically control weed species (Duke and Powles 2008;

Green 2014). As a result of the simplicity and flexibility afforded by transgenic crops, global

adoption of GM crops increased 100-fold, from 1.7 million ha in 1996 to 175.2 million ha by 2013

(James 2013). By 2013, 11 different GM crop species were planted in 27 countries (James 2013),

often completely changing weed control selections at a landscape level (Duke and Powles 2008;

Green 2014). The efficacy and flexibility of GM crops eliminated the need to use a range of

chemically-diverse selective herbicides, replacing them with the exclusive use of glyphosate

throughout the crop rotation. The elimination of weed control diversity produced the “perfect

evolutionary storm” (Duke and Powles 2008) where large areas of genetically-diverse species

such as Amaranthus sp. (Culpepper et al. 2006; Norsworthy et al. 2008; Steckel et al. 2008),

Ambrosia sp. (Westhoven et al. 2008; Brewer and Oliver 2009; Norsworthy et al. 2010) and

Lolium sp.(Powles et al. 1998; Perez and Kogan 2003; Perez-Jones et al. 2005) were recurrently

glyphosate-selected, resulting in the evolution of glyphosate resistance (Sammons et al. 2007;

Powles 2008; Green 2014).

In response to the rapid evolution of glyphosate resistance, the agrochemical industry

responded with an “arms race against herbicide-resistant weeds” (Duke and Powles 2009),

developing additional GM herbicide-tolerant traits including ALS inhibitor resistance, synthetic

auxin resistance (2,4-D [Enlist] and dicamba [Xtend]) and glufosinate resistance (Behrens et al.

2007; Green et al. 2008; Egan et al. 2011; Peng 2011; Green 2014). It is widely cautioned

however that these technologies need to be judiciously managed to not repeat the extreme

over-reliance and lack of diversity that lead to the widespread evolution of glyphosate resistance

(Beckie et al. 2011; Egan et al. 2011; Mortensen et al. 2012; Green 2014).

In 2009 the first Roundup Ready™ canola varieties were released in Western Australia (James

2011). Adoption of GR canola occurred with 639 tonnes of GM canola seed sold in Western

Australia in 2014 (Blight 2014). At a recommended sowing rate of 2.5 kg ha–1, this equates to an

estimated GM canola planting of 255,000 ha in 2014 (AOF 2014), a 53% increase in GM planting

since 2012 (AOF 2013). Given the increase in GR canola adoption, the threat of glyphosate

resistance is significant. Therefore a detailed understanding of the glyphosate susceptibility of

major weed species in Australia is needed.

PAGE | 11


HARVEST WEED SEED CONTROL

In response to the widespread evolution of herbicide resistance in Australia, new non-herbicidal

weed control techniques were developed (Walsh and Powles 2014), which use the advanced

sorting capacities of new combine harvester designs to capture weed seeds at harvest before

they re-enter the soil seed bank (Fogelfors 1982; Shirtliffe and Entz 2005; Walsh and Powles

2014). Controlling weed seeds at harvest through management of the chaff fraction is

collectively termed 'harvest weed seed control' (HWSC) which aims to destroy weed seeds by

burning via narrow windrows or chaff heaps or by impacting weed seed with the Harrington

Seed Destructor™ (Walsh et al. 2013). These techniques are highly effective, achieving more

than 99% wild radish control (Walsh and Newman 2007; Walsh and Powles 2007; Walsh et al.

2012). However over-reliance on these tactics has also been cautioned as, just like herbicides,

the efficiency of HWSC applies an intense selection for the evolution of HWSC evading attributes

in weed species (Walsh et al. 2013). One such attribute could be the evolution of early-maturing

biotypes which could shed mature seed prior to harvest.

Please note that a more comprehensive review of literature relevant to each chapter of this thesis

is included in the Introduction and Discussion sections of each experimental chapter to follow.

CONCLUSION

Fifty eight years have passed since Harper first predicted the evolution of herbicide resistance

in weeds (Harper 1956). Since then, herbicide over-reliance has resulted in the widespread

evolution of herbicide-resistant weeds globally (Heap 2014), now considered one of the greatest

agronomic problems in agriculture (OECD/FAO 2012). Recently, herbicide resistance was termed

“Evolution in Action” (Powles and Yu 2010) and as an evolutionary process, “the key to managing

herbicide resistance is to anticipate which weeds will survive a selection and provide sufficient

management diversity for their control” (Powles and Holtum 1994). It is therefore vital that the

selection processes leading to resistance evolution be well understood in order to design

effective and sustainable weed control strategies (Neve et al. 2009).

Despite the worldwide importance of wild radish as a weed in agriculture, there are several key

gaps in knowledge that need to be addressed including:

1. The current status of glyphosate resistance among wild radish populations.

2. The effects of low herbicide use rates on wild radish populations.

3. The evolutionary consequences of highly effective HWSC on wild radish populations.

PAGE | 12


RESEARCH OBJECTIVES

The overall objective of this study was to explore the potential of wild radish populations to

evolve resistance to herbicides or to HWSC. To address this, the specific aims of this thesis

included:

1. Ascertaining the current glyphosate resistance status of wild radish populations in the

WA grainbelt (Chapter 1).

2. Ascertaining the effectiveness of the sole use of glyphosate on weed species (Chapter

2).

3. Exploring the evolutionary response of a susceptible wild radish population to recurrent

selection with low rates of glyphosate and 2,4-D amine (Chapters 3 and 4).

4. Exploring the evolutionary capacity of wild radish to change flowering times in response

to selection (Chapter 5).

THESIS OUTLINE

This thesis is in agreement with the Postgraduate and Research Scholarship Regulation

1.3.1.33(1) of The University of Western Australia (Crawley, Western Australia) and is presented

as a series of five scientific papers. One manuscript has been published and is included as a PDF,

with the remaining four as manuscripts in preparation for submission. All information presented

resulted from work conducted towards this thesis. There are seven main chapters: (1) general

introduction, (2–6) experimental chapters as manuscripts and (7) general discussion. The

general introduction covers the broad background for the work presented in this thesis in order

to justify the research objectives presented above. A more focused review of literature is

presented in the introduction and discussion of each experimental chapter. The five

experimental chapters are presented in the format of scientific papers that can be read

individually or as a part of the whole thesis. Each experimental chapter includes the following

sections: Abstract, Introduction, Materials and Methods, Results, Discussion and References.

This thesis-as-a-series-of-papers results in some unavoidable repetition, especially in the

materials and methods of Chapters 4 and 5.

Chapter 1 is a general introduction describing the theoretical and problem-specific context of

this study. This includes literature relating to the evolution of herbicide resistance and the

biology of wild radish. This review largely focuses on literature pertaining to wild radish and the

herbicide traits likely to be introduced into Australian transgenic herbicide-tolerant cropping

systems. The gaps in knowledge and research objectives are identified. Chapter 2 documents

the first worldwide case of glyphosate resistance in wild radish. Chapter 3 presents the first

PAGE | 13


survey of the WA grainbelt coinciding with the initial release of transgenic GR canola. These

surveys highlighted that large wild radish populations were being treated, resulting in large

numbers of wild radish surviving glyphosate. This work highlighted the need to understand the

evolutionary consequence of high wild radish survival, which led to the first low-dose glyphosate

selection study on a dicot species (wild radish), which is described in Chapter 4. Crops resistant

to the synthetic auxin herbicide 2,4-D have recently been developed to provide an effective

alternative to control the increasing number of GR dicot weed species. Auxinic herbicides are

robust in terms of resistance with only 31 species classified as auxinic resistant, following more

than 50 years of widespread and intensive use. Nonetheless, the low-dose 2,4-D studies in

Chapter 5 highlight that the long-term sustainability of synthetic auxins is contingent on using

effective rates which achieve high levels of control. In order to control widespread herbicide

resistance in wild radish populations in Australia, non-herbicidal weed control tactics such as

HWSC are being incorporated into weed control strategies. However, Chapter 6 demonstrated

that wild radish has the capacity to alter flowering times in response to recurrent selection, such

as that imposed by HWSC. The findings from all chapters are summarised and the implications

for wild radish control is discussed in Chapter 7.

REFERENCES

ABARE (2013). AgSurf interactive farm survey data. Canberra, Australian Bureau of Agriculture

and Resource Economics.

ABARES (2013). Australian farm survey results 2010–11 to 2012–13. Canberra, Australian Bureau

of Agricultural and Resource Economics and Sciences (ABARES).

ABARES (2014). Agricultural commodities report. Canberra, Australian Bureau of Agricultural

and Resource Economics and Sciences (ABARES).

Abel, A. L. (1954). The rotation of weedkillers. British Weed Control Conference 1954 249–255.

Alemseged, Y., et al. (2001). A farmers survey of weed management and herbicide problems on

winter crops in Australia. Plant Protection Quarterly 16: 21–25.

Allard, R. W. (1999). Principles of Plant Breeding. New York, John Wiley and Sons.

AOF (2013). Annual crop report; 2013, Australian Oilseeds Foundation.

AOF (2014). Annual crop report; 2014 Australian Oilseeds Federation.

Beckie, H. J., et al. (2011). Risk assessment of glyphosate resistance in Western Canada. Weed

Technology 25(1): 159–164.

Beckie, H. J. and F. J. Tardif (2012). Herbicide cross resistance in weeds. Crop Protection 35: 15–

28.

PAGE | 14


Behrens, M. R., et al. (2007). Dicamba resistance: enlarging and preserving biotechnology-based

weed management strategies. Science 316(5828): 1185–1188.

Bendtham, G. H. and J. D. Hooker (1954). Handbook of the British Flora, L Reeve and Co Ltd.

Bhatti, M. (2004). Genetic variation in naturalised wild radish (Raphanus raphanistrum)

populations in the mediterranean climate of south-western Australia. PhD thesis,

University of Western Australia.

Blackman, G. E. (1950). Selective toxicity and the development of selective weedkillers. Journal

of the Royal Society of Arts 98(4820): 499–517.

Blackshaw, R. E., et al. (2002). Influence of wild radish on yield and quality of canola. Weed

Science 50(3): 344–349.

Blight, A. (2014). Western Australian Roundup-Ready canola sales in 2014. Monsanto, M.

Ashworth.

Borger, C. P. D., et al. (2012). Linking field and farmer surveys to determine the most important

changes to weed incidence. Weed Research 52(6): 564–574.

Borlaug, N. (2002). Feeding a world of 10 billion people: The miracle ahead. In Vitro Cellular &

Developmental Biology - Plant 38(2): 221–228.

Boutsalis, P. (2013). Herbicide resistant weeds. from www.plantscienceconsulting.com.

Brewer, C. E. and L. R. Oliver (2009). Confirmation and resistance mechanisms in glyphosate-

resistant common ragweed (Ambrosia artemisiifolia) in Arkansas. Weed Science 57(6):

567–573.

Broster, J. (2013). Charles Sturt University herbicide resistance testing service report. Wagga

Wagga, NSW, Charles Sturt University.

Broster, J. (2014). Charles Sturt University herbicide resistance testing service annual report.

Wagga Wagga, NSW, Charles Sturt University.

Broster, J., et al. (2014). Extent of herbicide resistance in three broadleaf weed species in

southern New South Wales. 19th Australasian Weeds Conference. C. Preston. Hotel Grand

Chancellor, Hobart, Tasmania, Australian Weed Science Society.

BRS. (2013). Land use of Australia data 2013. from adl.brs.gov.au/landuse.

Carroll, L. (1960). The garden of live flowers. Alice's adventures in wonderland and looking

through the looking glass. New York, New American Library: 345.

Chater, A. O. (1964). Cruciferae, Raphanus L., Cambridge University Press.

Cheam, A. H. (1984). Coat imposed dormancy controlling germination in wild radish and fiddle

dock seeds. Proceedings of the 7th Australian Weeds Conference: 184–189.

PAGE | 15

http://www.plantscienceconsulting.com/


Cheam, A. H. (1986). Seed production and seed dormancy in wild radish (Raphanus

raphanistrum) and some possibilities for improving control. Weed Research 26(6): 405–

413.

Cheam, A. H. and G. R. Code (1995). The biology of Australian weeds: 24 Raphanus raphanistrum

L. Plant Protection Quarterly 10(1): 2–13.

Code, G. R. and T. W. Donaldson (1996). Effect of cultivation, sowing methods and herbicides on

wild radish populations in wheat crops. Australian Journal of Experimental Agriculture

36(4): 437–442.

Conner, J. and S. Via (1993). Patterns of phenotypic and genetic correlations among

morphological and life history traits in wild radish, Raphanus raphanistrum. Evolution 47:

704–711.

Conner, J. D. and R. Rush (1995). The effect of wild radish floral morphology on pollination

efficiency by four taxa pollinators. Oecologia 104: 234–245.

Conner, J. K., et al. (2003). Expression of additive genetic variances and covariances for wild

radish floral traits: Comparison between field and greenhouse environments. Evolution

57(3): 487–495.

Cousens, R., et al. (2008). Small-scale spatial structure within patterns of seed dispersal.

Oecologia 158(3): 437–448.

Cousens, R. D., et al. (2001). Early growth and development of wild radish (Raphanus

raphanistrum L.) in relation to wheat. Australian Journal of Agricultural Research 52(7):

755–769.

Cousens, R. D., et al. (2010). The role of the persistent fruit wall in seed water regulation in

Raphanus raphanistrum (Brassicaceae). Annals of Botany 105(1): 101–108.

Culpepper, A. S., et al. (2006). Glyphosate-resistant Palmer amaranth (Amaranthus palmeri)

confirmed in Georgia. Weed Science 54(4): 620–626.

Darwin, C. (1859). On the origin of species by means of natural selection or the preservation of

favoured races in the struggle for life. London, John Murray.

Darwin, C. and A. R. Wallace (1858). On the tendency of species to form varieties and on the

perpetuation of varieties and species by natural means of selection. Journal of the Linnean

Society of London 2: 45–63.

Darwin, C. R. (1859). Notebook B: Transmutation of species (1837–1838).

David, J. L., et al. (1993). Outcrossing and selfing evolution in populations under directional

selection. Heredity 71(6): 642–651.

Dawkins, R. and J. R. Krebs (1979). Arms races between and within species. Proceedings of the

Royal Society of London. Series B, Biological Sciences 205(1161): 489–511.

PAGE | 16


Delp, C. J. (1980). Coping with resistance to plant disease. Plant Disease 64(7): 652–657.

Délye, C., et al. (2013). Deciphering the evolution of herbicide resistance in weeds. Trends in

Genetics 29(11): 649–658.

Donaldson, T. W. (1986). Wild radish (Raphanus raphanistrum L): a review of research on its

biology in Victoria, 1976–1982. Plant Protection Quarterly 1(4): 160–162.

Duke, S. and S. B. Powles (2009). Glyphosate-resistant crops and weeds: now and in the future.

AgBioforum 12(3): 346–357.

Duke, S. O. (2012). Why have no new herbicide modes of action appeared in recent years? Pest


Duke, S. O. and S. B. Powles (2008). Glyphosate-resistant weeds and crops. Pest Management

Science 64(4): 317–318.

Duke, S. O. and S. B. Powles (2008). Glyphosate: a once-in-a-century herbicide. Pest


Egan, J. F., et al. (2011). 2,4-Dichlorophenoxyacetic acid (2,4-D) resistant crops and the potential

for evolution of 2,4-D resistant weeds. Proceedings of the National Academy of Sciences

108(11): E37.

Eslami, S. V., et al. (2006). Wild radish (Raphanus raphanistrum) interference in wheat. Weed

Science 54(4): 749–756.

Fairbanks, D. J. and B. Rytting (2001). Mendelian controversies: a botanical and historical review.

American Journal of Botany 88(5): 737–752.

Fisher, R. A. (1936). Has Mendel's work been rediscovered? Annals of Science 1(2): 115–137.

Fogelfors, H. (1982). Collection of chaff, awns and straw when combining and its influence on

the seed bank and its composition of the weed flora. 23rd Swedish Weeds Conference.

Uppsala, Sweden, Swedish University of Agricultural Science, Department of Plant

husbandry and Research Information Centre: 339–345.

Gardner, C. (1930). A systematic census of plants occurring in Western Australia Perth,

Government of Western Australia.

Godfray, H. C. J., et al. (2010). Food security: The challenge of feeding 9 billion people. Science

327(5967): 812–818.

Green, J. M. (2014). Current state of herbicides in herbicide-resistant crops. Pest Management

Science 70(9): 1351–1357.

Green, J. M., et al. (2008). New multiple-herbicide crop resistance and formulation technology

to augment the utility of glyphosate. Pest Management Science 64(4): 332–339.

PAGE | 17


Grima, R. and A. Blake (2005). The incidence and severity of wild radish resistance in the NAR

results from an in situ survey. Crop Updates. Perth, Western Australia, Department of

Agriculture and Food. Weeds: 43–45.

Hall, L. M., et al. (1994). Mechanisms responsible for cross resistance and multiple resistance.

Herbicide resistance and world grains. S. Powles and D. L. Shaner. Boca Raton/Ann

Arbor/London/Tokyo, Lewis Publishers: 243–261.

Harper, J. C. (1956). The evolution of weeds in relation to herbicides. Proceedings of British

Weed Control Conference. 3: 179–188.

Hashem, A., et al. (2001). Resistance of wild radish (Raphanus raphanistrum) to acetolactate

synthase-inhibiting herbicides in the Western Australia wheat-belt. Weed Technology

15(1): 68–74.

Hashem, A. and H. S. Dhammu (2002). Cross-resistance to imidazolinone herbicides in

chlorsulfuron-resistant Raphanus raphanistrum. Pest Management Science 58(9): 917–

919.

Hashem, A., et al. (2001). Triazine resistance in a biotype of wild radish (Raphanus raphanistrum)

in Australia. Weed Technology 15(4): 636–641.

Hashem, A. B. and C. Wilkins (2004). How long is the seed bank life of wild radish? Agribusiness

Crop Updates. J. Peirce. Western Australia, Western Australian Department of Agriculture

and Food 17–18.

Heap, I. (2014). Global perspective of herbicide-resistant weeds. Pest Management Science

70(9): 1306–1315.

Heap, I. (2014). The international survey of herbicide resistant weeds. Retrieved 20 May 2014,

from www.weedscience.com.

Hilton, H. W. (1957). Herbicide tolerant strain of weeds, Hawaiian Sugar Planters Association

Annual Report: 69–71.

Holm, L. D., et al. (1997). World weeds: Natural histories and distribution, John Wiley and Sons.

Holt, J. S., et al. (1993). Mechanisms and agronomic aspects of herbicide resistance. Annual

Review of Plant Physiology and Plant Molecular Biology 44(1): 203–229.

Jabeen, H. (2014). World Bank, World Development Report 2014: Risk and opportunities.

Population and Development Review 40(2): 373–374.

James, C. (2011). Global status of commercialised biotech/GM crops: 2011. ISAAA Brief. Ithaca,

NY., International Service for the Acquisition of Agri-biotech Applications Brief 43.

James, C. (2013). Global status of commercialised biotech/ GM crops: 2013. ISAAA Brief. Ithaca,

NY., International Service for the Acquisition of Agri-Biotech Applications Brief 44.

PAGE | 18

http://www.weedscience.com/


Jasieniuk, M., et al. (1996). The evolution and genetics of herbicide resistance in weeds. Weed

Science 44(1): 176–193.

Jones, J. (2011). Why genetically modified crops? Philosophical Transactions of the Royal Society

369(1942): 1807–1816.

Keasar, T., et al. (1996). Innate movement rules in foraging bees: Flight distances are affected by

recent rewards and are correlated with choice of flower type. Behavioral Ecology and

Sociobiology 39(6): 381–388.

Kutschera, U. and K. Niklas (2004). The modern theory of biological evolution: an expanded

synthesis. Naturwissenschaften 91(6): 255–276.

Lamarck, J. B. (1809). Philosophie zoologique: ou exposition des considérations relative à

l'histoire naturelle des animaux. Paris, Dentu et L'Auteur.

Ma, Z. and T. J. Michailides (2005). Advances in understanding molecular mechanisms of

fungicide resistance and molecular detection of resistant genotypes in phytopathogenic

fungi. Crop Protection 24(10): 853–863.

Macnair, M. R. (1991). Why the evolution of resistance to anthropogenic toxins normally

involves major gene changes: the limits to natural selection. Genetica 84(3): 213–219.

Madhou, P., et al. (2005). Genetic variation in populations of Western Australian wild radish.

Australian Journal of Agricultural Research 56: 1079–1087.

Matthews, J. M. (1996). Aspects of the population genetics and ecology of herbicide resistant

annual ryegrass. Doctor of Philosophy, University of Adelaide.

Maxwell, B. D., Mortimer, A.M. (1994). Selection of herbicide resistance. Resistance in plants:

Biology and biochemistry. S. B. Powles and Holtum, J. A. M., Lewis Publishers: 1–26.

Mazer, S. J., Wolfe, L.M. (1992). Planting density influences the expression of genetic variation

in seed mass in wild radish (Raphanus sativus L.). American Journal of Botany 79: 1185–

1193.

Mekenian, M. R. and R. W. Willemsen (1975). Germination characteristics of Raphanus

raphanistrum. I. Laboratory studies. Bulletin of the Torrey Botanical Club 102(5): 243–252.

Melander, A. L. (1914). Can insects become resistant to sprays. Journal of Economic Entomology

7(2): 167–173.

Mendel, G. (1866). Verhandlungen des naturforschenden. Vereines in Brünn,. Abhandlungen.

3–47.

Mortensen, D. A., et al. (2012). Navigating a critical juncture for sustainable weed management.

BioScience 62(1): 75–84.

PAGE | 19


Neve, P. (2013). The rate debate: do low or reduced rates exacerbate the evolution of

resistance? Global Herbicide Resistance Challenge, Fremantle, Western Australia,

Australian Herbicide Resistance Initiative.

Neve, P., et al. (2014). Expanding the eco-evolutionary context of herbicide resistance research.

Pest Management Science 70(9): 1385–1393.

Neve, P. and S. Powles (2005). Recurrent selection with reduced herbicide rates results in the

rapid evolution of herbicide resistance in Lolium rigidum. Theoretical and Applied

Genetics 110(6): 1154–1166.

Neve, P., et al. (2009). Evolutionary-thinking in agricultural weed management. New Phytologist

184(4): 783–793.

Newman, P., et al. (2007). Decimate a wild radish population in five years. Agribusiness Crop

Updates 2007. A. Douglas. Burswood Casino, Perth, Department of Agriculture and Food

Western Australia: 1–4.

Norsworthy, J. K., et al. (2008). Confirmation and control of glyphosate-resistant Palmer

amaranth (Amaranthus palmeri) in Arkansas. Weed Technology 22(1): 108–113.

Norsworthy, J. K., et al. (2010). Confirmation and control of glyphosate-resistant giant ragweed

(Ambrosia trifida) in Tennessee. Weed Technology 24(1): 64–70.

OECD/FAO (2012). OECD-FAO Agricultural Outlook 2012–2021. Paris, France.

Orr, H. A. and J. A. Coyne (1992). The genetics of adaptation: A reassessment. The American

Naturalist 140(5): 725–742.

Osborne, J. L., et al. (1999). A landscape-scale study of bumble bee foraging range and constancy,

using harmonic radar. Journal of Applied Ecology 36(4): 519–533.

Owen, M. J. (2014). Multiple herbicide-resistant Raphanus raphanistrum (wild radish) now

dominate across the Western Australian grainbelt, Unpublished, University of Western

Australia.

Padgette, S. R., et al. (1996). New weed control opportunities: Development of soybeans with a

Roundup Ready gene. Herbicide-Resistant Crops. S. O. Duke. Boca Raton, CRC Press: 53–

84.

Panetta, F. D., et al. (1988). Survival and fecundity of wild radish (Raphanus raphanistrum L.)

plants in relation to cropping, time of emergence and chemical control. Australian Journal

of Agricultural Research 39(3): 385–397.

Parsons, W. T. and E. G. Cuthbertson (1992). Noxious weeds of Australia. Melbourne, Inkarta

Press.

Peng, W. (2011). GM crop cultivation surges, but novel traits languish. Nature Biotechnology

29(4): 302.

PAGE | 20


Perez-Jones, A., et al. (2005). Identification of glyphosate-resistant Italian ryegrass (Lolium

multiflorum) in Oregon. Weed Science 53(6): 775–779.

Perez, A. and M. Kogan (2003). Glyphosate-resistant Lolium multiflorum in Chilean orchards.

Weed Research 43(1): 12–19.

Piggin, C. M., et al. (1978). Germination of wild radish (Raphanus raphanistrum). Proceedings of

the 1st Conference of Australian Weed Science Society: 233–240.

Powles, S, B. and J. Holtum (1994). Herbicide resistance in plants: biology and biochemistry. Boca

Raton, Lewis Publishers.

Powles, S, B., et al. (1998). Evolved resistance to glyphosate in rigid ryegrass (Lolium rigidum) in

Australia. Weed Science 46: 604–607.

Powles, S, B. and Q. Yu (2010). Evolution in action: plants resistant to herbicides. Annual Review

of Plant Biology 61: 317–347.

Powles, S, B. (2008). Evolved glyphosate-resistant weeds around the world: lessons to be learnt.


Powles, S, B., et al. (1997). Herbicide resistance: Impact and management. Advances in

Agronomy 58: 57–93.

Reeves, T. G., et al. (1981). Seed production and longevity, seasonal emergence and phenology

of wild radish (Raphanus raphanistrum). Australian Journal of Experimental Agriculture

21(112): 524–530.

Rosenberg, M. (1990). The mother of invention: Evolutionary theory, territoriality, and the

origins of agriculture. American Anthropologist 92(2): 399–415.

Roush, R. T. and J. A. McKenzie (1987). Ecological genetics of insecticide and acaricide resistance.

Annual Review of Entomology 32(1): 361–380.

Sakschewski, B., et al. (2014). Feeding 10 billion people under climate change: How large is the

production gap of current agricultural systems? Ecological Modelling 288: 103–111.

Sammons, R. D., et al. (2007). Sustainability and stewardship of glyphosate and glyphosate-

resistant crops. Weed Technology 21(2): 347–354.

Shirtliffe, S. J. and M. H. Entz (2005). Chaff collection reduces seed dispersal of wild oat (Avena

fatua) by a combine harvester. Weed Science 53(4): 465–470.

Snow, A. and L. Campbell (2005). Can feral radishes become weeds. Crop Ferality and

Volunteerism. J. Gressel. Boca Raton, FL., CRC Press: 193–207.

Steckel, L. E., et al. (2008). Palmer amaranth (Amaranthus palmeri) in Tennessee has low level

glyphosate resistance. Weed Technology 22(1): 119–123.

Switzer, C. M. (1957). The existence of 2,4-D resistant strains of wild carrot. Proceedings of the

Northeastern Weed Control Conference. 11: 315–318.

PAGE | 21


Vencill, W. K., et al. (2012). Herbicide resistance: Toward an understanding of resistance

development and the impact of herbicide-resistant crops. Weed Science 60(Special Issue

1): 2–30.

Vogel, F. (1926). Contribution to knowledge of the growing place requirements of wild radish

and Sinapis arvensis. Bovarian Farming Year Book. Frankfurt, Germany.

Wallace, A. R. (1855). On the laws which has regulated the introduction of species. Annals of

Natural History 16(2).

Walsh, M. and P. Newman (2007). Burning narrow windrows for weed seed destruction. Field

Crops Research 104(1–3): 24–30.

Walsh, M., et al. (2013). Targeting weed seeds in-crop: A new weed control paradigm for global

agriculture. Weed Technology 27(3): 431–436.

Walsh, M. J., et al. (2001). High frequency of chlorsulfuron-resistant wild radish (Raphanus

raphanistrum) populations across the Western Australian wheatbelt. Weed Technology

15(2): 199–203.

Walsh, M. J., et al. (2012). Harrington seed destructor: A new non-chemical weed control tool

for global grain crops. Crop Science 52(3): 1343–1347.

Walsh, M. J., et al. (2009). Combined effects of wheat competition and 2,4-D amine on phenoxy

herbicide resistant Raphanus raphanistrum populations. Weed Research 49(3): 316–325.

Walsh, M. J., et al. (2007). Frequency and distribution of herbicide resistance in Raphanus

raphanistrum populations randomly collected across the Western Australian wheatbelt.


Walsh, M. J. and S. B. Powles (2007). Management strategies for herbicide-resistant weed

populations in Australian dryland crop production systems. Weed Technology 21(2): 332–

338.

Walsh, M. J. and S. B. Powles (2014). Management of herbicide resistance in wheat cropping

systems: learning from the Australian experience. Pest Management Science 70(9): 1324–

1328.

Walsh, M. J., et al. (2004). Multiple-herbicide resistance across four modes of action in wild

radish (Raphanus raphanistrum). Weed Science 52(1): 8–13.

Webster, T. M. and R. L. Nichols (2012). Changes in the prevalence of weed species in the major

agronomic crops of the southern United States: 1994/1995 to 2008/2009. Weed Science

60(2): 145–157.

Weintraub, R. L. (1953). 2,4-D mechanisms of action. Journal of Agricultural and Food Chemistry

1(3): 250–254.

PAGE | 22


Westhoven, A. M., et al. (2008). Field presence of glyphosate-resistant Horseweed (Conyza

canadensis), common lambsquarters (Chenopodium album), and giant ragweed

(Ambrosia trifida) biotypes with elevated tolerance to glyphosate. Weed Technology

22(3): 544–548.

Whitehead, C. W. and C. M. Switzer (1963). The differing response of strains of wild carrot to

2,4-D and related herbicides. Canadian Journal of Plant Science 43(3): 255–262.

Williams, J. C. (2001). Sources of phenotypic variation in floral traits in wild radish (Raphanus

raphanistrum). American Journal of Botany 88(9): 1577–1581.

Woolls, W. (1867). Plant introduced accidently (1866). A contribution to the flora of Australia.

Sydney, F White: 136–152.

Wright, S. (1984). Evolution and the genetics of populations, Genetic and biometric foundations.

The University of Chicago Press.

WSSA (2010). Common and chemical names of herbicides approved by the Weed Science Society

of America. Weed Science 58(4): 511–518.

Young, K. (2001). Germination and emergence of wild radish (Raphanus raphanistrum L.). PhD

thesis, University of Melbourne.

PAGE | 23


PAGE | 24

CHAPTER 2. FIRST GLYPHOSATE-RESISTANT WILD RADISH (Raphanus raphanistrum L.) POPULATIONS

CHAPTER 2.

IDENTIFICATION OF THE FIRST GLYPHOSATE-RESISTANT WILD RADISH (Raphanus raphanistrum L.) POPULATIONS

Michael B. AshworthAB, Michael J. WalshAB, Ken C. FlowerB, Stephen B. PowlesAB

Published: Pest Management Science 70(9): 1432–1436

25 April 2014

Keywords: glyphosate, glyphosate resistance, wild radish, evolution, herbicide resistance

A Australian Herbicide Resistance Initiative, School of Plant Biology, The University of Western

Australia, Crawley, WA 6009, Australia B School of Plant Biology, The University of Western Australia, Crawley, WA 6009, Australia

PAGE | 25

CHAPTER 2. FIRST GLYPHOSATE-RESISTANT WILD RADISH (Raphanus raphanistrum L.) POPULATIONS

PAGE | 26

CHAPTER 2. FIRST GLYPHOSATE-RESISTANT WILD RADISH POPULATIONS

PAGE | 27


PAGE | 28


PAGE | 29


PAGE | 30


PAGE | 31


SUPPORTING INFORMATION

PAGE | 32

CHAPTER 3. GLYPHOSATE-RESISTANT SPECIES IN THE GRAINBELT OF WESTERN AUSTRALIA

CHAPTER 3.

THE IDENTIFICATION OF GLYPHOSATE-RESISTANT Lolium rigidum AND Raphanus raphanistrum POPULATIONS FROM THE FIRST WESTERN

AUSTRALIAN PLANTINGS OF GLYPHOSATE-RESISTANT CANOLA

ABSTRACT

Transgenic glyphosate-resistant (GR) canola was first commercially grown in Western Australia

in 2010. This provided the opportunity to obtain baseline data regarding the control of wild

radish by glyphosate as used as an in-crop post-emergent herbicide. Two targeted surveys (2010

and 2011) were conducted across the 14 million ha grainbelt of Western Australia. The 2010

survey was conducted at the late-flowering stage of GR canola, while the 2011 survey was

conducted at an earlier growth stage (6–8 leaves), approximately 2–3 weeks after the second

in-crop glyphosate application. During both surveys, a total of 239 fields were visited,

representing an estimated combined survey area of 24,000 ha. The 2011 survey alone

represented a 23% sub-sample of the total GR canola planting in the WA grainbelt for that

season. This study identified glyphosate resistance in one population of wild radish (Raphanus

raphanistrum L.) and in eight annual ryegrass (Lolium rigidum L.) populations. No glyphosate

resistance was identified in barley grass species (Hordeum sp.), brome grass species (Bromus

sp.), wild oat species (Avena sp.), capeweed (Arctotheca calendula L.) or small-flowered mallow

(Malvia parviflora L.). Thus, despite a long history of use in Western Australia, glyphosate still

provides excellent weed control of most species however some resistance is evident.

INTRODUCTION

Over the past 50 years, weed control has been efficiently and economically achieved through

the use of herbicides. Of all herbicides available, glyphosate is the world's largest selling and

most important herbicide (Baylis 2000). Glyphosate (N-(phosphonomethyl)glycine) acts on the

enzyme EPSPS (5-enolpyruvylshikimate-3-phosphate synthase) involved in the biosynthesis of

aromatic amino acids required for the production of anthocyanins, lignin, growth promoters,

growth inhibitors, phenolics and proteins (Amrhein et al. 1980; Steinrücken and Amrhein 1980).

However, the efficacy of glyphosate is threatened by persistent over-reliance and the lack of

weed control diversity surrounding its use. In North and South America, the widespread

adoption of glyphosate-resistant (GR) crops has applied a continuous and intense selection

PAGE | 33


pressure with glyphosate, resulting in the evolution of GR in 14 species and species shifts to

weeds that are only partially controlled by glyphosate (Powles 2008; Vila-Aiub 2008; Heap 2014).

Western Australian (WA) dryland cropping systems are following this trend towards glyphosate

over-reliance, with the commercial release of genetically-modified glyphosate-resistant (GR)

canola in 2010. Glyphosate provides effective and cheap control of a wide range of problematic

WA weed species (Walsh et al. 2007; Owen and Powles 2009; Michael et al. 2010; Owen et al.

2014). As a result of glyphosate over-reliance in WA, glyphosate resistance has been identified

in multiple resistant populations of annual ryegrass (Lolium rigidum Gaud.) (Neve et al. 2004;

Owen and Powles 2010), wild radish (Raphanus raphanistrum L.) (Ashworth et al. 2014) and

jungle rice (Echinochloa colona Link.) (Gaines et al. 2012). Random surveys of the WA grainbelt

at harvest have not identified glyphosate resistance in other problematic weed species including

wild oat sp. (Owen and Powles 2009), barley grass sp. (Owen 2014), brome grass sp. (Owen

2014) or fleabane sp. (Owen et al. 2009).

In 2010, genetically-modified glyphosate-resistant (GR) canola was commercially grown for the

first time in the grainbelt of WA. This provided a unique opportunity to establish baseline data

concerning the prevalence of GR weed populations following the sole use of glyphosate as an

in-crop post-emergent herbicide. Here, we report the frequency and distribution of GR weed

species in the WA grainbelt prior to the long-term adoption of transgenic GR crops.

MATERIALS AND METHODS

POPULATION COLLECTION

In 2010 and 2011, GR canola fields were inspected for surviving weed species following

glyphosate treatment. GR canola crop field locations were identified from regional agronomists

and consultants with each field confirmed to have received two glyphosate applications at more

than the recommended field rate for wild radish control in Australia (540 g ha–1) (Figure 3.1).

Fields were sampled by walking three inverted V transects, 100 m into each field and sampling

seed from surviving weed species.

In 2010, 73 GR canola fields were surveyed (Figure 3.1) between 4 and 22 October 2010, at the

early flowering stage of canola (GS 67) (Lancashire et al. 1991), sampling siliques from surviving

wild radish plants. In total, 24 wild radish populations were collected. Siliques were stored in a

non-airconditioned glasshouse over summer to after ripen, before the seed was extracted using

a modified 'grist mill'.

In 2011, 166 GR canola fields were surveyed (Figure 3.1) by M Ashworth over a four-week period

between 17 May and 24 June 2011, at the 6–8 true-leaf stage of canola (GS 18 (Lancashire et al.

PAGE | 34


1991)), 2–3 weeks after the second recommended glyphosate application. Surviving weeds were

carefully removed with soil and intact root systems. The shoots were trimmed to reduce

transplant shock, wrapped in moistened paper towel and express posted to The University of

Western Australia (UWA) using an overnight courier service. On arrival, all samples were

immediately planted into 180 mm plastic pots containing potting mixture (25% peat moss, 25%

sand and 50% mulched pine bark), and allowed to re-grow. During the 2011 survey, 27 annual

ryegrass, 19 wild radish and four capeweed whole plant populations were collected.

FIGURE 3.1. LOCATIONS OF SURVEYED GR CANOLA FIELDS ACROSS THE WA GRAINBELT IN 2010 AND 2011.

AVERAGE RAINFALL ISOHYETS SHOWN, HIGH: 450–750 MM, MEDIUM: 325–450 MM AND LOW: <325 MM.

RESISTANCE TESTING

For all experiments in this study, plants were grown outdoors during their normal growing

season from May to November at UWA. Plants were watered as required and fertilised weekly

with 2 g Scotts PolyFeed™ soluble fertiliser (N 19% [urea 15%, ammonium 1.9%, nitrate 2.1%],

P 8%, K 16%, Mg 1.2%, S 3.8%, Fe 400 mg kg–1, Mn 200 mg kg–1, Zn 200 mg kg–1, Cu 100 mg kg–1,

HIGH

LOW

MEDIUM

PAGE | 35


B 10 mg kg–1, Mo 10 mg kg–1). Herbicide treatments were applied using a twin nozzle laboratory

sprayer fitted with 110° 01 flat fan spray jets (Tee jet™) delivering herbicide in 100 L ha–1 of

water at 210 kPa, travelling at a speed of 3.6 km h–1.

2010 collected population resistance testing

In April 2011, approximately 250 seeds from each of the 24, 2010 field-collected wild radish

populations were germinated on solidified water agar (0.6% w/v) containing 1 µM karrikinolide

(KAR1) in darkness for 2 days. Karrikinolide is a derived compound from smoke that has been

shown to break wild radish seed dormancy (Long 2011). Pre-germinated seeds were planted into

four replicate polystyrene foam trays (400 mm wide x 500 mm long x 150 mm deep) containing

potting mixture (25% peat moss, 25% sand and 50% mulched pine bark). The known susceptible

wild radish population WARR7 (referred to hereafter as S1) (Walsh et al. 2004) and the

transgenic GR canola line (Brassica napus cv. Roundup Ready™ Cobbler) were included as

glyphosate susceptible and resistant controls respectively. At the two true-leaf stage, seedlings

were treated with glyphosate (Roundup Ready Herbicide™ 690 g kg–1, Nufarm, Laverton North,

Vic., Australia) at 540 g ha–1 as previously described. Plant survival was assessed 42 days after

treatment (DAT) by inspecting the rosette of each seedling. If new green growth was evident,

the plant was deemed a survivor. All surviving plants were cut back to the last newly-emerging

leaf to apply additional survival stress and allowed to re-grow for 14 days. Surviving populations

were then re-treated with glyphosate (540 g ha–1) with survival re-assessed 42 DAT. Using these

procedures, of the 24 wild radish populations collected in 2010, 18 were killed by glyphosate

treatments, while six populations exhibited >20% survival and therefore judged worthy of

further study. In 2012, the six wild radish populations that survived initial glyphosate screening

were screened in a full dose–response study to quantify glyphosate resistance. The wild radish

population (S1) was included as a susceptible control. Populations were planted (20 per pot) in

180 mm plastic pots filled with potting mixture and maintained outside as previously described.

At the two true-leaf stage all populations were treated with glyphosate at 0, 270, 540, 810, 1080

and 2160 g ha–1. Plant survival was assessed 42 DAT. One population (collected from Wongan

Hills, WA [30˚ 54’S 116˚ 43’E] referred to hereafter as R) was deemed to be two-fold glyphosate

resistant when compared to the S1 population. Thirty surviving plants of this R population were

re-potted into 305 mm plastic pots and isolated to prevent the ingress of foreign pollen. Once

all individuals were flowering, plants were manually crossed using the 'Beestick method' in

which flowers are crossed using a bee carcase as outlined by (Williams 1980), ensuring a random

pattern of cross pollination (panmixia). At maturity, all siliques were collected and threshed as

previously described. The extracted seed represented the glyphosate-selected progeny subset

PAGE | 36


of the R population (referred hereafter as R1). To confirm the heritability of glyphosate

resistance in this wild radish population, a further dose response study was conducted in 2013

with the R1 population and three known herbicide-susceptible wild radish populations—S1,

WARR 33 (referred to hereafter as S2) (Walsh et al. 2007) and WARR 36 (referred to hereafter

as S3) (Table 3.3) were planted (20 per pot) into 180 mm plastic pots and maintained as

previously described. At the two true-leaf stage, plants were sprayed with glyphosate at 0, 150,

300, 450, 600, 750, 900 and 1800 g ha–1 for the S1–S3 populations and 0, 450, 750, 1050, 1500,

3000, 4500, 6000 and 7000 g ha–1 for the R1 population. All populations were assessed for

survival 42 DAT. Aboveground shoot biomass was harvested, dried at 65°C for 7 days before

weighing.

2011 collected population resistance testing

In the 2011 field survey of GR canola crops, all fields were visited weeks after in-crop glyphosate

use, enabling visual assessment of weed survival. From these glyphosate treated GR canola fields

survivors of 27 annual ryegrass, 19 wild radish and four capeweed populations were collected

and transplanted into pots. Once ryegrass had approximately 25 mm of new leaf growth and

dicot species had two new leaves, all populations were tested using the Syngenta Quick-Test™

(RISQ) as previously described by Boutsalis (2001) and Walsh et al. (2001). The known

susceptible wild radish population S1 and the GR transgenic canola line (Brassica napus cv.

Roundup Ready™ Cobbler) were included as controls. Glyphosate was applied at the

recommended field rate of 540 g ha–1 as previously described. Plant survival was assessed 42

DAT by inspecting the new growth of each treated plant. If new growth was evident, the plant

was deemed a survivor. Surviving plants were cut back to the last emerging leaf to apply

additional survival stress. Plants that re-grew new leaves were re-treated with glyphosate at 540

g ha–1 and re-assessed 42 DAT.

Of the 27 ryegrass populations collected as survivors in GR crops, 19 did not survive glyphosate

treatment and were therefore judged to be glyphosate susceptible. The eight annual ryegrass

populations collected as survivors from GR canola fields and subsequently survived the

glyphosate RISQ test were re-potted into 250 mm plastic pots containing standard potting

mixture to be watered and fertilised as previously described. Prior to flowering, each of the eight

populations were isolated to prevent the ingress of foreign pollen and allowed to randomly cross

with each other. At maturity, all seed heads were collected and threshed, with the extracted

seed representing the glyphosate-selected progeny subset of each field-collected ryegrass

population. This glyphosate-selected progeny was tested for resistance by planting 200 seeds

into four replicate 200 x 250 x 50 mm trays containing potting mixture. The annual ryegrass

PAGE | 37


populations VLR1 (Christopher et al. 1991) and NLR70 (Powles et al. 1998) were included as

known susceptible and known GR controls, respectively. At the two-leaf stage, seedlings were

treated with single (450 g ha–1) and double (900 g ha–1) the recommended glyphosate rate for

annual ryegrass. Plant survival was assessed 28 DAT. Surviving plants were severely trimmed to

apply additional survival stress. Following 20 mm of regrowth, all plants were re-treated with

glyphosate (900 g ha–1) and re-assessed for survival 28 DAT.

WEED PREVALENCE IN 2011 GR CANOLA FIELDS

While inspecting GR canola fields in 2011, visual estimates of the diversity and density of weeds

species present were recorded by rating ten replicate 250 mm x 250 mm randomly-placed

quadrats per site. It was observed that plants surviving glyphosate had either plant injury with

fresh regrowth or were of advanced growth stage, in the vicinity of plants showing glyphosate

injury symptoms. Any weeds that emerged after glyphosate application were not considered

survivors, but were included in the assessment. These late-emerging weeds had no more than

2–3 leaves for grasses or 2 true leaves for dicots. Species diversity at each site was ascertained

by identifying the species present (Hussey et al. 2007). Plant density was rated based on

photographic comparisons representing a scale of plant densities, photographed one metre

above the ground. Due to the damaged state of all glyphosate-affected plants, barley grass,

brome grass and wild oat species could not be visually differentiated; therefore they were

grouped for initial plant density ratings. All late-emerging species could be readily differentiated,

therefore were reported individually.

DATA ANALYSIS

Survival data was analysed using non-linear regression analyses with the DRC package in R 3.0.0

(R Development Core Team 2011; http://www.R-project.org). Data was fitted to a three

parameter log-logistic model [1] where the upper limit was fixed at 1.0 (Streibig et al. 1993; Price

et al. 2012):

Y= c + {1 – c / (1 + exp [b (log x – log e)]} [1]

where Y denotes plant survival expressed as a percentage of the untreated control in response

to herbicide dose x, c is the lower asymptotic value of Y, and b is the slope of the curve around

e, which is the dose causing 50% mortality (LD50) or biomass (GR50). LD50 or GR50 parameters were

compared using the selectivity indices (SI) function (R 3.0.0) to determine if the ratios between

these values were significantly different (P<0.05). A Lack-of-Fit test was also applied to each

fitted curve to ascertain the appropriateness of the model [1]. A ratio of recurrently selected

and basal populations (R/S) at the estimated LD50 level was used to highlight the change in

PAGE | 38

http://www.r-project.org/


population susceptibility to glyphosate. Data was plotted using SigmaPlot v.12 (Systat Software

Inc., 2011).

RESULTS AND DISCUSSION

WILD RADISH

In the 2010 GR canola crop survey, wild radish populations were collected from 24 fields (32%)

of the 73 GR canola fields surveyed. The subsequent screening of these populations collected

from the northern grainbelt of WA in 2010 identified one possible GR population from Wongan

Hills (R) (Table 3.1.) which exhibited 76% survival when screened at 540 g ha–1. At this glyphosate

rate, the S1 population was killed while the Roundup Ready™ canola control was unaffected.

Dose response screening quantified this population [R] to be two-fold (LD50) glyphosate resistant

when compared to S1 (Table 3.3). Dose response screening of the glyphosate-selected progeny

(R1) identified inherited resistance, increasing the resistance level to 3.5-fold (LD50) (Table 3.3)

when compared to the pooled values of three known susceptible populations (S1–S3). Mean

survival of the R1 population was 84% and 59% at the recommended (540 g ha–1) and double

glyphosate rate (1080 g ha–1), respectively. Even though this population survived glyphosate, its

biomass was dramatically reduced, with the R1 population exhibiting a GR50 of 350 g ha–1,

compared to the pooled GR50 rate of 207 g ha–1 for the S1–S3 populations. This GR wild radish

population is in addition to the two populations identified in Chapter 2, which were found in

glyphosate maintained fallow.

Surveying 166 GR canola fields across the WA grainbelt in 2011 did not identify glyphosate

resistance in wild radish using the RISQ methodology. During this survey, wild radish survivors

(whole plants) were collected from 39 GR canola fields, representing 23% of GR fields surveyed.

When glyphosate was applied (540 g ha–1), 24 of the populations (62%) were killed. Following

trimming and re-treatment (540 g ha–1) all remaining populations died. These treatments killed

the susceptible control (S1) following a single glyphosate application (540 g ha–1) while the

resistant control population (Brassica napus cv. Roundup Ready™ Cobbler) survived.

During the 2010 and 2011 GR canola field surveys, it was evident that the first commercial GR

canola plantings in WA (2010) were established on fields with high wild radish densities. This is

a dangerous practice, posing significant risk for glyphosate resistance evolution as large

populations were treated. In the 2011 survey, wild radish was the second most-prevalent

species, infesting 66% (110) of GR canola fields (Table 3.1) with 86% (95) of these fields

containing economically-damaging wild radish densities (>10 plants m–2) (Streibig et al. 1989;

Boz 2005) (Table 3.2). Alarmingly, 15% (22) of these fields contained extremely high wild radish

PAGE | 39


densities of more than 100 plants m–2. The sole reliance on glyphosate as an in-crop selective

herbicide in GR canola did not effectively control these high densities with 33% (37) of fields

containing wild radish survivors following two glyphosate applications. Of these, most of the

fields (83%) contained more than 1 plant m–2. Of concern were the 13 fields with surviving

densities of more than 10 plants m–2 following glyphosate treatment. Resistance testing of these

populations however only identified one GR population as previously described (Table 3.1). The

high survival rates in this study and low frequency of resistance reported is clear evidence that

sole reliance on glyphosate cannot effectively control large wild radish densities.

Wild radish has a protracted emergence pattern during the growing season (Mekenian and

Willemsen 1975; Cheam 1986). Glyphosate does not have soil residual activity, due to rapid

inactivation (Sprankle et al. 1975), therefore late-emerging wild radish are not controlled in GR

canola crops, as evident in the high proportion of surveyed fields with late-emerging populations

(42%) (Table 3.2). Even though late-emerging wild radish has less effect on canola yield

(Blackshaw et al. 2002), the high frequency of seed-bearing wild radish found in the 2010 GR

canola field survey highlights that late-emerging individuals provide significant contributions to

the soil seed bank. In order to control these late-emerging cohorts, herbicide-tolerant canola

crops have historically been treated with residual PSII or ALS herbicide chemistries (Kirkegaard

et al. 2008). However, these PSII or ALS herbicides cannot be used in GR canola. The application

of glyphosate in GR canola is also restricted to before the six-leaf stage and with no residual

control from glyphosate, crop competition was solely relied on to suppress the growth of any

late-emerging populations. However as 32% of the GR canola fields in 2010 contained wild radish

at flowering, it was clear that crop competition alone was ineffective at providing wild radish

control.

Prior to the introduction of GR canola, glyphosate could only be used as a pre-seeding

knockdown herbicide and was rarely used alone, owing to insufficient wild radish control

(Panetta et al. 1988; Peltzer et al. 2009). This study clearly demonstrates that sole reliance on

glyphosate as an in-crop selective in GR canola does not provide effective control of large wild

radish populations. If widely used, the lack of control in GR canola phases has also been shown

to favour proliferation of uncontrolled species, leading to shifts in the weed flora (Reddy 2004;

Culpepper 2006; Owen 2008). Any proliferation would directly place other wild radish control

options in subsequent cropping phases at increased risk of resistance evolution. Consequently,

incorporation of additional weed control diversity into GR canola systems is strongly

recommended.

PAGE | 40


ANNUAL RYEGRASS

The results of the 2011 survey of the 166 GR canola fields concurs with previous in-crop ryegrass

surveys conducted before harvest (Owen et al. 2014). Both studies show that glyphosate

resistance in annual ryegrass is present in the southern and eastern, moderate to high rainfall

zone (450–700 mm yr–1) of the WA grainbelt. Despite a long history of glyphosate use in the

central and northern grainbelt of WA, no additional GR annual ryegrass populations were found.

In the 2011, GR canola crop survey, annual ryegrass was the most common weed species

infesting GR canola, found in 91% of the 166 fields surveyed (Table 3.1). Plant densities were

generally high with 74% (112 fields out of 166) of fields having more than 50 plants m–2 and 15%

(22 fields out of 166) containing more than 100 plants m–2 (Table 3.1). Following glyphosate

treatment, 30% (45 fields out of 166) of the GR canola fields surveyed had survivors, of which

18% (8 fields out of 166) had >20 plants per m–2 (Table 3.2). From the 27 ryegrass populations

collected, eight populations originating from the southern and eastern high rainfall zone (450–

700 mm yr–1) were found to be glyphosate resistant (Table 3.1). Glyphosate resistance in these

populations is inherited, as the progeny of all eight populations showed high levels of survival

(>94%) at double the recommended rate (900 g ha–1). At this rate, the GR control population

(NLR70) was unaffected while the susceptible population (VLR1) was killed. Re-treating all

surviving populations (900 g ha–1) did not further decrease survival or biomass, confirming that

all eight populations are resistant to glyphosate. As a result, this survey highlights that 5% of the

GR canola fields in this survey contained GR annual ryegrass, concurring with the results of

random surveys of ryegrass in WA crop fields conducted by Owen (2014). The identification of

these GR annual ryegrass populations highlights that additional weed control tactics are

necessary in GR canola systems, including effective pre-emergent herbicides (such as trifluralin)

and harvest weed seed control.

OTHER WEED SPECIES

No GR barley grass, brome grass or wild oat populations were found in this study of GR canola

fields in WA (Owen et al. 2014, Walsh et al. 2007). However, glyphosate resistance has been

recently documented in five brome grass (Bromus diandrus L.) populations in South Australia [3]

and Victoria [2] (Preston 2014). In the WA GR canola surveys, these grasses were found in 54%

of GR canola fields with 38% of fields containing high densities (>50 plants m–2). Late-emerging

populations were common with 29% (49 fields out of 166) of fields contained late-emerging

barley grass species, 11% (18 fields out of 166) brome grass species and 13% (23 fields out of

166) wild oat species. While no resistance was found (Table 3.1), the widespread distribution

PAGE | 41


and densities of these grasses are of concern; therefore these species should be carefully

monitored.

Capeweed (Arctotheca calendula L.) is a weed species increasing in prevalence in WA cropping

fields (Borger et al. 2012). In this study, capeweed was found in 28% (46 fields out of 166) of GR

canola fields with four fields containing survivors. Screening these four surviving populations

found them to be glyphosate susceptible (Table 3.1). The 2011 survey also identified two fields

containing small-flowered mallow (Malva parviflora L.) however these populations were also

found to be susceptible to glyphosate (Table 3.1).

CONCLUSION

GR crops offer an additional tool to control the multiple herbicide resistance that is prevalent in

annual ryegrass and wild radish populations in the WA grainbelt. This survey is unique in that it

was conducted in the first year of commercial GR canola plantings in WA. Firstly, this survey

established that when wild radish and ryegrass populations were large, glyphosate alone could

not provide season-long weed control, presumably because staggered emergence of weeds

meant that late-emerging plants were not exposed to glyphosate (non-residual). Secondly, this

survey revealed one glyphosate-resistant wild radish population and eight glyphosate-resistant

ryegrass populations. This and Chapter 2 are the first reports of glyphosate resistance in wild

radish, whereas glyphosate-resistant ryegrass is known. In light of this and similar surveys

(Boutsalis et al 2012; Broster et al. 2011; Broster et al. 2012; Broster 2013; Owen et al. 2014), it

is important that weed control is diversified around the use of glyphosate to minimise the risks

of resistance evolution (Sammons et al. 2007; Powles and Yu 2010). When weed control diversity

is incorporated, GR cropping systems have been found to be sustainable, with low levels of

glyphosate resistance evolution (Beckie et al. 2006; Beckie 2011). In Australia, this diversity may

involve the use of diverse pre-seeding herbicide strategies such as the double knock (glyphosate

followed by paraquat/diquat) (Borger and Hashem 2007) and the incorporation of residual

herbicides (Beckie et al. 2011). Increased crop competition (Beckie et al. 2008) and harvest weed

seed control (Walsh et al. 2012) are also highly-effective tools that increase diversity.

PAGE | 42


TABLE 3.1. FREQUENCY AND DENSITY OF WEED SPECIES PRESENT IN 166 GR CANOLA FIELDS IN 2011. FIGURES DENOTE MEAN PLANT DENSITY OF TEN REPLICATE (250 MM X 250

MM) QUADRANT RATINGS. PLANT DENSITY PRIOR TO GLYPHOSATE TREATMENT DENOTED BY SURVIVING AND DAMAGED PLANTS AT ADVANCED GROWTH STAGE (>3 LEAF FOR

GRASSES, >2 LEAF FOR DICOTS). PARENTHESES INDICATE SPECIES PLANT DENSITY AS A PERCENTAGE OF THE TOTAL NUMBER OF FIELDS REGISTERING THAT SPECIES.

Fields

surveyed

Fields before

glyphosate

application

Fields after

glyphosate

application

Number of fields with weed species and densities treated with

glyphosate

Samples

collected

Confirmed

resistant

Low

(0–10 m–2)

Medium

(10–50 m–2)

High

(50–100+ m–2)

Very high

(>100+ m–2)

Annual ryegrass 166 151 45 10 (7%) 29 (19%) 90 (59%) 22 (15%) 27 8

Wild radish 166 110 37 15 (14%) 61 (55%) 34 (31%) 0 19 1

Barley grass sp., Brome

grass sp., Wild oat sp.

166 90 0 10 (11%) 46 (51%) 28 (31%) 6 (7%) 0 0

Capeweed 166 46 4 8 (18%) 25 (54%) 13 (28%) 0 3 0

Malva 166 2 0 2 (100%) 0 0 0 0 0

PAGE | 43


TABLE 3.2. FREQUENCY AND DENSITY OF WEED SPECIES PRESENT FOLLOWING GLYPHOSATE TREATMENT IN 166 GR CANOLA FIELDS IN 2011. FIGURES DENOTE MEAN PLANT DENSITY

OF TEN REPLICATE (250 MM X 250 MM) QUADRANT RATINGS. PARENTHESES INDICATE SPECIFIC PLANT DENSITIES AS A PERCENTAGE OF THE TOTAL NUMBER OF FIELDS REGISTERING

THAT SPECIES. PLANT DENSITY SURVIVING GLYPHOSATE APPLICATION DENOTED BY SURVIVING PLANTS AT ADVANCED GROWTH STAGE (>2 LEAF FOR GRASSES, >1 LEAF FOR DICOTS).

LATE-EMERGING SPECIES (<3 LEAF FOR GRASSES, <2 LEAF FOR DICOTS).

Total fields

after

application

Number of fields (% in brackets) with surviving plants following

glyphosate application

(at densities ranging from <1 plant m–2 to 20 plant m–2)

Proportion of fields

with late-emerging

populations (%)

<1 plant m–2 1 plant m–2 10 plants m–2 20 plants m–2

Annual ryegrass 45 9 (20%) 22 (49%) 6 (13%) 8 (18%) 25

Wild radish 37 6 (16%) 18 (48%) 7 (19%) 6 (16%) 42

Barley grass sp. 0 0 0 0 0 30

Brome grass sp. 0 0 0 0 0 11

Wild oat sp. 0 0 0 0 0 14

Capeweed 4 3 (75%) 0 1 (25%) 0 0

Malva 0 0 0 0 0 0

PAGE | 44


TABLE 3.3. PARAMETER ESTIMATES FROM THREE PARAMETER LOG-LOGISTIC MODEL [1] USED TO CALCULATE LD50 AND R/S VALUES FOR SUSCEPTIBLE WILD RADISH POPULATIONS

(S1–S3) AND FIELD COLLECTED (R) AND FIELD-COLLECTED PROGENY (R1) POPULATIONS. STANDARD ERRORS IN PARENTHESES.

Biotype Designation Population Glyphosate

resistance status

Location

(WA)

Year Co-ordinates b c e (LD50)

(g ha–1)

P valueA R/S

Field

collected

S1 WARR7 Susceptible Yuna 1999 –28.34̊S 115.01̊E 3.15 (0.26) –1.63 (1.78) 384 (18) – – R1 – Resistant Wongan Hills 2010 –30.54S 116. 43E 4.03 (0.42) 5.06 (2.77) 752 (22) <0.05 2.0a

S1 WARR7 Susceptible Yuna 1999 –28.34̊S 115.01̊E 3.89 (0.32) –6.24 (2.72 441 (28) S2 WARR33 Susceptible Belmunging 2004 –29.11̊S 116.26̊E 1.88 (0.23) –9.02 (3.84) 379(45) S3 WARR36 Susceptible Carnac Island 2013 –32.12S 115.66E 2.05 (0.22) –8.92 (3.29) 421 (40)

Pooled

WARR7;33;36 Susceptible – – – 2.22 (0.15) –8.41 (1.91) 422 (21) – – Progeny R1 – Resistant Wongan Hills 2010 –30.54S 116. 43E 1.47 (0.12) –6.62 (3.72) 1470 (102) <0.05 3.5b

A LD50 P value comparing the difference between field collected and susceptible control populations using the SI function in the DRC package in R. v3.0.0.

Lack-of-fit P value for the appropriateness of the three parameter model [1] for field-collected dose–response: 0.84.

Lack-of-fit P value for the appropriateness of the three parameter model [1] for progeny dose–response: 0.99.

PAGE | 45


ACKNOWLEDGEMENTS

This research was funded by the Grains Research Development Corporation of Australia,

through a Grains Research Scholarship and the Australian Government through an Australian

Post Graduate Award to M. Ashworth. We thank AHRI and plant growth facility staff at The

University of WA who provided assistance. We also thank the Grower Group Alliance of WA and

Industry agronomists for assistance in providing field locations and grower contacts.

REFERENCES

Amrhein, N., et al. (1980). The site of inhibition of the shikimate pathway of glyphosate. II.

Interference of glyphosate with chorismate formation in vivo and in vitro. Plant Physiology

66: 830–834.

Ashworth, M. B., et al. (2014). Identification of the first glyphosate-resistant wild radish

(Raphanus raphanistrum L.) populations. Pest Management Science 70(9): 1432–1436.

Baylis, A. (2000). Why glyphosate is a global herbicide: strengths, weaknesses and prospects.

Pest Management Science 56: 299–308.

Beckie, H. J. (2011). Herbicide-resistant weed management: focus on glyphosate. Pest


Beckie, H. J., et al. (2011). Risk assessment of glyphosate resistance in Western Canada. Weed

Technology 25(1): 159–164.

Beckie, H. J., et al. (2006). A decade of herbicide-resistant crops in Canada. Canadian Journal of

Plant Science 86(4): 1243–1264.

Beckie, H. J., et al. (2008). Weed suppression by canola and mustard cultivars. Weed Technology

22(1): 182–185.


Science 50(3): 344–349.

Borger, C. P. and A. Hashem (2007). Evaluating the double knockdown technique: sequence,

application interval, and annual ryegrass growth stage. Australian Journal of Agricultural

Research 58: 265–271.



Boutsalis, P. (2001). Syngenta Quick-Test: A rapid whole-plant test for herbicide resistance.

Weed Technology 15(2): 257–263.

Boz, Ö. (2005). Economic threshold for wild radish (Raphanus raphanistrum L.) control in wheat

fields. Turkish Journal of Agriculture & Forestry 29(3): 173–177.

PAGE | 46


Broster, J. (2013). Victorian and New South Wales wild radish survey results 2012–2013 Wagga

Wagga, NSW, School of Agricultural & Wine Sciences, Charles Sturt University.

Broster, J., et al. (2012). Herbicide resistance frequencies in annual ryegrass (Lolium sp.) and

other grass species in Tasmania. Plant Protection Quarterly 27.

Broster, J., et al. (2011). Herbicide resistance levels in annual ryegrass (Lolium rigidum. Gaud.) in

southern New South Wales. Plant Protection Quarterly 26.



413.

Christopher, J. T., et al. (1991). Cross resistance to herbicides in annual ruegrass (Lolium

rigidum). 2. Chlorsulfuron resistance involved wheat like detoxification system. Plant

Physiology 95: 1036–1043.

Culpepper, S. A. (2006). Glyphosate-induced weed shifts 1. Weed Technology 20(2): 277–281.

Gaines, T. A., et al. (2012). Evolved resistance to glyphosate in junglerice (Echinochloa colona)

from the tropical ord river region in Australia. Weed Technology 26(3): 480–484.



Hussey, B. M. J., et al. (2007). Western weeds: A guide to weeds of Western Australia, Weeds

Society of Western Australia.

Kirkegaard, J., et al. (2008). Break crop benefits in temperate wheat production. Field Crops

Research 107(3): 185–195.

Lancashire, P. D., et al. (1991). A uniform decimal code for growth stages of crops and weeds.

Annals of Applied Biology 119(3): 561–601.

Long, R. L, et al. (2011). Seeds of Brassicaceae weeds have an inherent or inducible response to

the germination stimulant karrikinolide. Annals of Botany 108(5): 933–944.



Michael, P. J., et al. (2010). Herbicide-resistant weed seeds contaminate grain sown in the

Western Australian grainbelt. Weed Science 58(4): 466–472.

Neve, P., et al. (2004). Multiple herbicide resistance in a glyphosate-resistant rigid ryegrass

(Lolium rigidum) population. Weed Science 52(6): 920–928.

Owen, M. and S. Powles (2009). Distribution and frequency of herbicide-resistant wild oat

(Avena spp.) across the Western Australian grain belt. Crop and Pasture Science 60: 25–

31.

PAGE | 47



Owen, M. D. K. (2008). Weed species shifts in glyphosate-resistant crops. Pest Management

Science 64(4): 377–387.

Owen, M. J. (2014). Herbicide resistance in Hordeum, Bromus and Avena species collected across

the Western Australian grainbelt, University of Western Australia.

Owen, M. J., et al. (2014). Multiple herbicide-resistant Lolium rigidum (annual ryegrass) now

dominates across the Western Australian grainbelt. Weed Research 54(3): 314–324.

Owen, M. J., et al. (2009). A survey in the southern grain belt of Western Australia did not find

Conyza sp. resistant to glyphosate. Weed Technology 23(3): 492–494.

Owen, M. J. and S. B. Powles (2009). Distribution and frequency of herbicide-resistant wild oat

(Avena spp.) across the Western Australian grainbelt. Crop and Pasture Science 60(1): 25–

31.

Owen, M. J. and S. B. Powles (2010). Glyphosate-resistant rigid ryegrass (Lolium rigidum)

populations in the Western Australian grain belt. Weed Technology 24(1): 44–49.

Panetta, F. D., et al. (1988). Survival and fecundity of wild radish (Raphanus raphanistrum L)



Peltzer, S. C., et al. (2009). Weed management in wide-row cropping systems: a review of current

practices and risks for Australian farming systems. Crop and Pasture Science 60(5): 395–

406.

Powles, S., et al. (1998). Evolved resistance to glyphosate in rigid ryegrass (Lolium rigidum) in

Australia. Weed Science 46: 604–607.

Powles, S. and Q. Yu (2010). Evolution in action: plants resistant to herbicides. Annual Review of

Plant Biology 61: 317–347.

Powles, S. B. (2008). Evolved glyphosate-resistant weeds around the world: lessons to be learnt.


Preston, C. (2014). Australian glyphosate resistance register. Retrieved May 2014, from

http://www.weedscrc.org.au/glyphosate.

Price, W. J., et al. (2012). Estimation of dose-response models for discrete and continuous data

in weed science. Weed Technology 26(3): 587–601.

Reddy, K. N. (2004). Weed control and species shift in bromoxynil and glyphosate-resistant

cotton (Gossypium hirsutum) rotation systems. Weed Technology 18(1): 131–139.



Sprankle, P., et al. (1975). Rapid inactivation of glyphosate in the soil. Weed Science 23(3): 224–

228.

PAGE | 48

http://www.weedscrc.org.au/glyphosate


Steinrücken, H. C. and N. Amrhein (1980). The herbicide glyphosate is a potent inhibitor of 5-

enolpyruvylshikimic acid-3-phosphate synthase. Biochemical and Biophysical Research

Communications 94(4): 1207–1212.

Streibig, J. C., et al. (1989). Estimation of thresholds for weed control in Australian cereals. Weed

Research 29(2): 117–126.

Streibig, J. C., et al. (1993). Dose-response curves and statistical models. Herbicide Bioassays. P.

Kudsk and J. C. Streibig. Boca Raton, CRC Press: 29–55.

Vila-Aiub, M. et al. (2008). Glyphosate-resistant weeds of South American cropping systems: an

overview. Pest Management Science 64(4): 366–371.



15(2): 199–203.






Williams, P. H. (1980). Bee-sticks, an aid in pollinating Cruciferae. Hortscience 15(6): 802–803.

PAGE | 49


PAGE | 50

CHAPTER 4. LOW DOSE GLYPHOSATE SELECTION IN WILD RADISH (Raphanus raphanistrum L.)

CHAPTER 4.

WILD RADISH (Raphanus raphanistrum L.) RECURRENT SELECTION WITH GLYPHOSATE OVER FOUR GENERATIONS RESULTS IN LOW-LEVEL

GLYPHOSATE AND ALS HERBICIDE RESISTANCE

ABSTRACT

Wild radish (Raphanus raphanistrum L.) is a highly problematic, global, dicot species which

causes economic loss by aggressively competing for water, nutrients and light. Of all herbicide

classes, glyphosate has been one of the slowest to evolve resistance, with glyphosate resistance

only reported in 25 weed species worldwide following more than 40 years of intensive use. This

study, tests the potential for glyphosate resistance to evolve in wild radish as a result of

recurrent selection at low glyphosate doses. Following four generations of low glyphosate dose

selection, the progenies of an initially herbicide-susceptible wild radish population (WARR7)

displayed a modest level of glyphosate resistance, with a resistant to susceptible (R/S) ratio of

2.4 (LD50). Along with this modest level of glyphosate resistance, concomitant weak cross

resistance was evident to the acetolactate synthase (ALS)-inhibiting herbicides imazamox (4.7-

fold) and metosulam (3.7-fold). This report is the first to confirm the evolution of weak

glyphosate resistance in a dicot species as a result of the recurrent use of sublethal glyphosate

rates. This study serves as a reminder of the importance of using robust glyphosate rates in order

to preserve the effectiveness of glyphosate.

INTRODUCTION

Wild radish is a cosmopolitain weed species (Snow 2005) which causes economic loss in crops

by aggressively competing for water, nutrients and light (Reeves et al. 1981; Blackshaw et al.

2002; Eslami et al. 2006). Wild radish is a genetically-diverse species (Kercher and Conner 1996;

Madhou et al. 2005) that has evolved multiple herbicide resistance, to inhibitors of acetolactate

synthase (ALS) (Hashem et al. 2001), phytoene desaturase (PDS) (Walsh et al. 2004),

photosynthetic electron transport (Hashem et al. 2001) and synthetic auxins (Walsh et al. 2004).

Resistance to glyphosate was also recently reported in Western Australian wild radish

populations (Ashworth et al. 2014). As an outcrossing species (Cheam 1995), wild radish is

PAGE | 51


considered to be at high risk of resistance evolution as it can share and accumulate

advantageous alleles among surviving individuals within a population.

Glyphosate (N-(phosphonomethyl) glycine) is the world's largest selling and most important

herbicide (Baylis 2000). Glyphosate is unique in acting on the enzyme EPSPS (5-

enolpyruvylshikimate-3-phosphate synthase) which is involved in the biosynthesis of aromatic

amino acids required for the production of anthocyanins, lignin, growth promoters, growth

inhibitors, phenolics and proteins (Amrhein et al. 1980; Steinrücken and Amrhein 1980).

Since 1974, glyphosate has been the world's most widely used herbicide because it is efficacious,

economical and environmentally benign (Duke and Powles 2008). Consequently glyphosate is

persistently relied upon for weed control prior to crop establishment (Llewellyn et al. 2012) and

in fallows (Farooq et al. 2011). When reviewed in early 1990’s (following 20 years of glyphosate

use), no cases of field-evolved glyphosate resistance had been identified (Holt et al. 1993; Dyer

1994), giving rise to the view that glyphosate resistance evolution was very unlikely (Jasieniuk

and Maxwell 1994; Bradshaw et al. 1997). This assumption was based upon the lack of evolved

resistance and that plants do not readily metabolise glyphosate (Bradshaw et al. 1997; Duke

2011). It was only after 24 years of intensive use that glyphosate resistance was first identified

(Powles et al. 1998). Since 1996 however, with the widespread adoption of transgenic

glyphosate-resistant (GR) crops in the Americas, glyphosate is now widely used as an in-crop

herbicide to exclusively treat large areas (Duke and Powles 2008; Green 2011). This intense

selection (outside the normal range of the population will survive) resulted in the evolution of

glyphosate resistance in several important and economically-damaging weed species (Powles

2008; Heap 2014), through the selection of extremely rare (predicted 1 x 10–8) (Jasieniuk et al.

1996; Neve et al. 2003), single major resistance gene traits (Lorraine-Colwill et al. 2001; Ng et al.

2004; Zelaya et al. 2004).

In the field, however, plants are often exposed to herbicide dosages that are lower than those

recommended. This is due a variety of human factors (such as growers' cutting rates),

environmental factors (such as herbicide drift and variability of herbicide interception within

weed canopies) and biological factors (such as plants greater than the optimal size for a given

rate) (Vila-Aiub and Ghersa 2005). When selected at intermediate doses, individuals that possess

both weak and strong gene traits can contribute to plant survival (Powles and Yu 2010; Yu and

Powles 2014). It was first suggested by Gressel (1995) that reducing the applied dose of a

herbicide could lead to ‘creeping’ resistance as a result of the accumulation of multiple

mutations of smaller effect. Studies have demonstrated that low-rate selection on the

outcrossing species annual ryegrass with the ACCase-inhibiting herbicide diclofop-methyl (Neve

PAGE | 52


and Powles 2005) [which can be metabolised in wheat (Christopher et al. 1992)] resulted in rapid

evolution of strong resistance to diclofop-methyl in controlled environments (Neve and Powles

2005; Neve and Powles 2005; Busi et al. 2012) and in the field (Manalil et al. 2011). This

resistance resulted from increased diclofop-methyl metabolism as a result of a metabolism-

based mechanism mediated by the selection and accumulation of multiple P450 enzymes (Busi

et al. 2013; Yu et al. 2013; Gaines et al. 2014). Unlike diclofop-methyl, glyphosate is not readily

metabolised in higher plants (Franz et al. 1997) and is known to inhibit some wheat cytochrome

P450 enzymes (Lamb et al. 1998; Xiang et al. 2005). However species still exhibit some

phenotypic variation in response to glyphosate (Boerboom et al. 1991; Busi and Powles 2009;

Ashworth et al. 2014), with recurrent low-dose selection with glyphosate in annual ryegrass

resulting in a two-fold increase in glyphosate resistance (Busi and Powles 2009).

This study investigates the potential of a known herbicide-susceptible wild radish population to

evolve glyphosate resistance when recurrently selected with low glyphosate rates over four

generations. The evolution of cross resistance to other herbicide modes of action was also

assessed.


PLANT MATERIAL

This selection study was conducted with the known glyphosate-susceptible wild radish biotype

WARR 7 (referred hereafter as G0), originally collected in 1999 from Yuna, Western Australia

(latitude –28.34 S, longitude 115.01 E) (Walsh et al. 2004). Since collection, seed stocks of this

population have been maintained and multiplied in the absence of herbicide selection while

preventing ingression of any herbicide-resistance genes. Using this population, four successive

generations of recurrent glyphosate selection was applied commencing in June 2011 (Selection

1; referred hereafter as S1), April 2012 (S2), September 2012 (S3) and January 2013 (S4) using

the general procedure below.

GENERAL PROCEDURE FOR POPULATION SELECTION

This pot study was conducted over four consecutive generations by selecting survivors from

dose–response studies involving six glyphosate rates. Four hundred seeds per rate were planted

into four replicate polystyrene foam trays measuring 400 mm wide x 500 mm long x 150 mm

deep, containing standard potting mixture (25% peat moss, 25% sand and 50% mulched pine

bark). After planting, seedlings were grown in the outdoors (April–October) facility at The

University of Western Australia (Perth, Western Australia) where they were watered as required

and fertilised weekly with 2 g Scotts PolyFeed™ soluble fertiliser (N 19% [urea 15%, ammonium

PAGE | 53


1.9%, nitrate 2.1%], P 8%, K 16%, Mg 1.2%, S 3.8%, Fe 400 mg kg–1, Mn 200 mg kg–1, Zn 200 mg

kg–1, Cu 100 mg kg–1, B 10 mg kg–1, Mo 10 mg kg–1). At the two true-leaf stage, seedlings were

counted prior to being treated with glyphosate (Roundup Ready Herbicide™ 690 g kg–1, Nufarm,

Laverton North, Vic., Australia) at rates described in Table 4.1. Herbicide treatments were

applied using a twin nozzle laboratory sprayer fitted with 110° 01 flat fan spray jets (Tee jet™)

delivering herbicide in 118 L ha–1 of water at 210 kPa, travelling at a speed of 3.6 km h–1. After

treatment, plants were returned to the outdoor area. Survival was assessed 42 days after

treatment (DAT) with treated plants with new growth considered alive. The glyphosate dose at

which 20% of the population survived was chosen as the low-dose selection. From these

survivors, 20 plants were selected, based on maximum regrowth (S1–S4, Table 4.1). Prior to

flowering, selected survivors were isolated to ensure cross pollination only among survivors and

to prevent ingress of foreign pollen. Once all plants were actively flowering, the population was

manually crossed using the 'Beestick method' outlined by Williams (1980) ensuring a random

pattern of cross pollination (panmixia). At maturity all siliques were collected with the seed

threshed using a modified 'grist mill'. The separated seed progeny represented the next

generation for glyphosate selection. During each selection a separate population of 20

randomly-chosen untreated plants was isolated and crossed using the general procedure to

produce four generations of unselected controls (C1–C4) (Table 4.1).

FINAL DOSE–RESPONSE EVALUATIONS OF RECURRENTLY-SELECTED WILD RADISH

POPULATIONS

In April 2013, the commencing population (G0), selected (S1–S4) and unselected progenies (C1–

C4) were outdoors grown during the normal growing season and evaluated within a final

glyphosate dose–response bioassay. Twenty seeds were planted into four replicate 180 mm

plastic pots containing standard potting mixture. At the two true-leaf stage, all populations were

sprayed with glyphosate at 0, 150, 300, 450, 600, 750, 1800, 2700 g ha–1 (field recommended

rate 550 g ha–1) as previously described (Figures 4.1 & 4.2). Following treatment, all pots were

returned to outdoor conditions. Plant survival was assessed 42 DAT with the aboveground shoot

biomass harvested and dried at 65°C for 7 days before weighing.

CROSS-RESISTANCE PROFILES OF COMMENCING SUSCEPTIBLE (G0) AND FOURTH GENERATION

GLYPHOSATE-SELECTED PROGENY (S4)

To determine the extent of any cross resistance to dissimilar herbicide modes of action, the G0

and S4 populations were treated at the recommended rate with a range of herbicides with

known activity against wild radish (Table 4.4). In April 2013, 20 seeds of G0 and S4 were each

sown into four replicate 180 mm plastic pots containing standard potting mixture. At the two-

PAGE | 54


leaf stage the recommended rate of each herbicide (Table 4.4) was applied as previously

described. Plant survival was assessed 35 DAT with the exception of the phenoxyacetic acid

herbicides (MCPA, 2,4-D amine) which were assessed 42 DAT. Survivors from each herbicide

treatment were isolated and crossed as previously described to produce progeny (Table 4.4).

This progeny was screened as previously described to confirm heritability of cross resistance

traits.

To quantify the response to ALS-inhibiting herbicides, a dose–response study was conducted as

previously described with chlorsulfuron (750 g kg–1, Nufarm, Laverton North, Vic., Australia) at

0, 3.75, 7.5, 15, 30 and 60 g ha–1; metsosulam (100 g kg–1, Bayer Cropscience, Hawthorn East,

Vic., Australia) at 0, 1.2, 2.5, 5, 10 and 20 g ha–1; and imazamox (700 g kg–1, Crop Care, Murarrie,

Qld, Australia) at 0, 8, 16, 32, 64 and 128 g ha–1 (all corresponding to 0.25 ×, 0.5 ×, 1 ×, 2 × and 4

× the recommended rate, respectively). Plant survival was assessed 35 DAT.

DATA ANALYSIS

Survival and biomass data were analysed using non-linear regression analyses with the DRC

package in R 3.0.0 (R Development Core Team 2011; http://www.R-project.org). Data was fitted

to a three parameter log-logistic model [1], where the upper limit was fixed to 100 (Streibig et

al. 1993; Price et al. 2012):



to herbicide dose x, c is the lower asymptotic value of Y and b is the slope of the curve around

e, which is the dose causing 50% mortality (LD50) or biomass reduction (GR50). LD50 and GR50

parameters of the selected (S1–S4) and unselected progeny (C1–C4) were compared to the

commencing (G0) population using the selectivity indices (SI) function (R 3.0.0) to determine if

the ratios between these values differed significantly (P<0.05). A Lack-of-Fit test was also applied

to each fitted curve to ascertain the appropriateness of the model [1]. A ratio of recurrently

selected and commencing populations (R/S) at the estimated LD50 level was used to highlight

the change in population susceptibility to glyphosate. Data was plotted using SigmaPlot v.12

(Systat Software Inc., 2011).

PAGE | 55



TABLE 4.1. SEED PROGENY COLLECTED FROM BASAL WILD RADISH POPULATION WARR 7 [G0]. SELECTED PLANTS

(N) AT A SPECIFIC SUBLETHAL DOSE OF GLYPHOSATE (g ha–1). GLYPHOSATE RECURRENT SELECTED GENERATIONS

S1–S4. RECURRENTLY UNSELECTED GENERATIONS C1–C4.

Glyphosate dose

(g ha–1)

Population

size

Herbicide

efficiency (%

control) A

Plants selected

(N)

Progeny

0 20 0 20 C1

67.5 382 9 0

135 391 10 0

270 372 15 0

378 369 26 0

540 396 54 20 S1

0 20 0 20 C2

270 361 56 0

540 392 88 20 S2

810 383 92 0

1080 379 97 0

0 20 0 20 C3

270 362 12 0

540 374 58 0

810 387 83 20 S3

1080 378 98 0

1620 391 100 0

0 20 0 20 C4

270 351 16 0

540 357 32 0

810 376 64 0

1080 379 87 20 S4

1620 368 100 0

A Calculated as 100 – % plant survival

PAGE | 56


RESULTS

EFFECT OF SUBLETHAL GLYPHOSATE ON G0 POPULATION

With the known herbicide-susceptible wild radish population WARR7 (G0), four successive

generations of recurrent glyphosate selection were conducted (S1–S4). Concomitantly, four

concurrently-grown generations were maintained as controls to be treated similarly but without

glyphosate selection (C1–C4). Finally, all populations were evaluated in a full glyphosate dose–

response study to ascertain the population response to glyphosate. As expected, glyphosate

dose–response studies confirmed that the G0 population, although susceptible to glyphosate

displays sufficient intra-population diversity at low glyphosate doses in which low-dose

glyphosate selection could be conducted. Comparing the commencing G0 population with the

subsequent glyphosate-selected populations (S1–S4) found that recurrent glyphosate selection

at sublethal doses for four generations marginally increased glyphosate resistance (Table 4.2,

Figure 4.1, Plate 4.1). After three generations of glyphosate selection, a negligible shift in

glyphosate susceptibility was observed, with the LD50 only 1.4-fold higher, being 518 g ha–1 to

719 g ha–1 (S3) (Table 4.2). However, in the fourth selected generation (S4), the LD50 increased

to 1222 g ha–1 (2.4-fold) resulting in 78% survival in the S4 population at the above

recommended rate of 750 g ha–1 (Table 4.2). At this rate (750 g ha–1), the G0 population was

killed. Despite the 2.4-fold modest level of glyphosate resistance in the fourth generation (S4),

plant biomass was glyphosate affected. A small increase in plant biomass was only recorded in

the fourth generation of selection, increasing the rate required to reduce plant biomass from 60

g ha–1 (GR50; G0) to 204 g ha–1 (GR50; S4) (Table 4.2, Figure 4.1, Plate 4.1).

Recurrently-grown generations in the absence of glyphosate selection (C1–C4) did not change

the population's susceptibility to glyphosate (Figure 4.2; Table 4.2) however some differences in

the generations biomas response to glyphosate was evident when compared to the G0

population (Figure 4.2, Table 4.3). This is expected to be due to the significant genetic variability

within wild radish populations. This study demonstrates that the modest progressive shift

towards glyphosate resistance in the selected population (S1–S4) resulted from glyphosate

selection and not from other environmental factors.

CROSS-RESISTANCE PATTERN

Evidence of weak cross resistance to ALS-inhibiting herbicides was identified in the recurrently

glyphosate-selected wild radish populations. Screening of the G0 and S4 populations with a

range of herbicides with known activity against wild radish identified a heritable resistance to

imazamox and metosulam (Table 4.5, Figure 4.3). Dose response studies quantified a 4.7-fold

PAGE | 57


increase in imazamox survival (LD50) and a 3.6-fold increase in metosulam survival (Table 4.5,

Figure 4.3) when compared with the G0 population. However there was only a 2.6-fold increase

in plant biomass for imazamox (GR50) with no increase in biomass reported for metosulam (GR50)

when compared with the G0 population (Table 4.6). No cross resistance was evident to the ALS-

inhibiting herbicide chlorsulfuron (Tables 4.5 & 4.6).

PAGE | 58


A. Glyphosate (g ha-1)

0 200 400 600 800 1000 1200 1400 1600 1800

Pla

nt S

urvi

val (

%)

0

20

40

60

80

100

B. Glyphosate (g ha-1)

0 200 400 600 800 1000 1200 1400 1600 1800

Sho

ot b

iom

ass

(% u

ntre

ated

con

trol)

0

20

40

60

80

100

120

FIGURE 4.1. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE LOW-DOSE GLYPHOSATE-

SELECTED GENERATIONS S1 (…□…), S2 (…∆…), S3 (…◊…) AND S4 (…○…). EACH SYMBOL REPRESENTS THE MEAN OF EIGHT RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED

SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG-LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 STANDARD ERROR OF THE MEAN.

PAGE | 59


A. Glyphosate (g ha-1)

0 200 400 600 800 1000 1200 1400 1600 1800

Pla

nt S

urvi

val (

%)

0

20

40

60

80

100

B. Glyphosate (g ha-1)

0 200 400 600 800 1000 1200 1400 1600 1800

Sho

ot b

iom

ass

(% u

ntre

ated

con

trol)

0

20

40

60

80

100

120

FIGURE 4.2. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE UNSELECTED CONTROL

GENERATIONS C1 (…□…), C2 (…∆…), C3 (…◊…) AND C4 (…○…), DERIVED FROM CROSSING 20 RANDOMLY-SELECTED PLANTS IN THE ABSENCE OF SELECTION. EACH SYMBOL

REPRESENTS THE MEAN OF EIGHT RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG-LOGISTIC MODEL

[1]. VERTICAL BARS REPRESENT 1 STANDARD ERROR OF THE MEAN.

PAGE | 60


TABLE 4.2. PARAMETER ESTIMATES FOR SURVIVAL FROM THE THREE PARAMETER LOG-LOGISTIC MODEL [1] USED TO ESTIMATE LD50 PARAMETERS FOR PERCENT SURVIVAL BY NON-

LINEAR REGRESSION ANALYSIS. STANDARD ERRORS FOR EACH PARAMETER ESTIMATE ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF LD50 FOR THE

COMMENCING (G0) AND RESPECTIVE GLYPHOSATE SELECTED (S1–S4) OR UNSELECTED CONTROL (C1–C4) POPULATIONS.

Progeny Selection rate

(g ha–1)

Population

size

Survival

(%)

Plants

selected

b c e LD50

(g ha–1)

P

valueA

R/S

G0 – – – – 5.35 (0.56) –2.17 (2.04) 518 (12) – –

Glyphosate

selected

S1 540 396 54 20 7.21 (0.77) –0.89 (1.86) 562 (10) <0.05 1.1

S2 540 392 88 20 4.43 (0.58) 0.50 (2.91) 636 (20) <0.05 1.2

S3 810 387 83 20 6.81 (1.04) 7.66 (2.91) 719 (18) <0.05 1.4

S4 1080 379 87 20 2.35 (0.25) –3.88 (6.85)` 1222 (96) <0.05 2.4

Unselected C1 0 20 – 20 5.54 (0.45) –1.93 (1.51) 505 (9) 0.29 1.0

C2 0 20 – 20 6.55 (0.54) –1.26 (1.43) 513 (8) 0.87 1.0

C3 0 20 – 20 5.61 (0.47) –2.18 (1.57) 507 (9) 0.49 1.0

C4 0 20 – 20 7.02 (0.71) –0.78 (1.65) 513 (9) 0.67 1.0

A LD50 P value comparing the difference between the selected and commencing populations assessed by the SI function in the DRC package in R. v2.14.1.

Lack-of-fit P value for appropriateness of the three parameter logistic model [1] – 0.84.

PAGE | 61


TABLE 4.3. PARAMETER ESTIMATES FOR BIOMASS FOR EACH GLYPHOSATE SELECTED AND UNSELECTED POPULATION USING THE THREE PARAMETER LOG-LOGISTIC MODEL [1] TO

ESTIMATE GR50 PARAMETERS. STANDARD ERRORS FOR EACH PARAMETER ESTIMATE ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF GR50 FOR THE

COMMENCING (G0) AND THE SELECTED (S1–S4) OR CONTROL (C1–C4) POPULATIONS.

Population b c e GR50

(g ha–1) P valueA R/S

G0 0.88 (0.56) –4.80 (10.14) 60 (33) – –

Glyphosate

selected

S1 0.94 (0.27) –5.48 (5.28) 78 (16) 0.66 1.3

S2 0.41 (0.34) –18.98 (31.44) 40 (27) 0.60 0.7

S3 0.86 (0.42) –6.02 (11.04) 91 (29) 0.51 1.5

S4 0.69 (0.24) –15.36 (17.24) 204 (65) <0.05 3.4

Unselected

C1 1.18 (0.33) –3.76 (4.52) 98 (18) 0.43 1.6

C2 1.66 (0.26) –1.49 (2.57) 141 (11) 0.21 2.4

C3 1.80 (0.52) –2.67 (5.47) 180 (28) 0.18 3.0

C4 2.23 (0.47) –0.15 (2.90) 157 (14) 0.17 2.6

A LD50 P value comparing the difference between the selected and commencing populations assessed by the SI function in the DRC package in R. v2.14.1.


PAGE | 62


TABLE 4.4. POPULATION SURVIVAL FOR THE COMMENCING WILD RADISH POPULATION (G0), GLYPHOSATE-SELECTED GENERATION (S4) AND HERBICIDE-SELECTED (S4) PROGENY

WHEN SPRAYED AT THE RECOMMENDED RATE OF A RANGE OF HERBICIDES WITH KNOWN ACTIVITY AGAINST WILD RADISH. EACH RESULT IS THE MEAN OF FOUR REPLICATES. STANDARD

ERRORS OF THE MEANS ARE IN PARENTHESES.

Herbicide

mode of action

Herbicide active Rate

(g ha–1)

Adjuvant Mean plant survival (%)

Unselected (G0) 4th generation

selected (S4)

Unselected (G0) 4th generation

selected (S4)

Cross-resistant progeny

ALS inhibitor Chlorsulfuron 15 0.1% BS1000 3 (2) 7 (1) 1.5 (1.5) 11 (4)

ALS inhibitor Metosulam 5 0.1% BS1000 0 25 (4) 0 35 (5)

ALS inhibitor Sulfometuron-methyl 7.5 0.1% BS1000 0 0 – –

ALS inhibitor Imazamox 32 0.1% BS1000 0 52 (1) 0 61 (4)

Synthetic auxin 2,4-D amine 500 – 0 0 – –

Synthetic auxin MCPA amine 1000 – 4 (2) 6 (2) 5 (2) 15 (2)

PDS inhibitor Diflufenican 100 – 5 (1) 9 (5) 1 (1) 19 (2)

PSII Bromoxynal 400 – 0 0 – –

PSII Diuron 1000 – 0 0 – –

PSII Metribuzin 280 – 0 0 – –

PSII Atrazine 1000 2% Hasten 0 0 – –

PSI Diquat 100 – 0 0 – –

PAGE | 63


A. Imazamox (g ha-1)

0 10 20 30 40 50 60

Mea

n %

Sur

viva

l

0

20

40

60

80

100

B. Metosulam (g ha-1)

0 5 10 15 20

Mea

n %

Sur

viva

l

0

20

40

60

80

100

FIGURE 4.3. DOSE–RESPONSE SURVIVAL CURVES FOR IMAZAMOX (A) AND METOSULAM (B) FOR THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE GLYPHOSATE-

SELECTED GENERATION S4 (…○…). EACH SYMBOL REPRESENTS THE MEAN OF SIX RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED SURVIVAL CURVES USING A THREE

PARAMETER LOG-LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 STANDARD ERROR OF THE MEAN.

PAGE | 64


TABLE 4.5. PARAMETER ESTIMATES FOR SURVIVAL FOR IMAZAMOX, METOSULAM AND CHLORSULFURON USING THE THREE PARAMETER LOG-LOGISTIC MODEL [1] TO ESTIMATE LD50

PARAMETERS FOR THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE GLYPHOSATE-SELECTED GENERATION (S4) (…○…). STANDARD ERRORS FOR EACH

PARAMETER ESTIMATE ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF LD50 FOR THE COMMENCING POPULATION (G0) AND THE GLYPHOSATE-SELECTED

GENERATION (S4).

Biotype b c

e LD50

(g ha–1) P valueA R/S P valueB

Imazamox G0 1.94 (0.15) 0.8 (0.92) 4.65 (0.26) – – 0.87

S4 1.26 (0.11) 1.3 (1.24) 22.16 (0.71) <0.05 4.7

Metosulam G0 1.62 (0.24) 1.3 (1.31) 0.45 (0.08) – – 0.11

S4 1.31 (0.20) 3.2 (3.52) 1.69 (0.13) <0.05 3.7

Chlorsulfuron G0 5.76 (0.45) –0.1 (0.68) 4.24 (0.05) – – 0.15

S4 3.89 (0.24) 2.1 (1.49) 4.68 (0.10) 0.16 1.1

A LD50 P value comparing the difference between selected and commencing populations using the SI function in the DRC package in R. v2.14.1. B Lack-of-fit P value for appropriateness of the three parameter logistic model [1] for each herbicide treatment.

PAGE | 65


TABLE 4.6. PARAMETER ESTIMATES FOR BIOMASS FOR IMAZAMOX, METOSULAM AND CHLORSULFURON USING THE THREE PARAMETER LOG-LOGISTIC MODEL [1] TO ESTIMATE GR50

PARAMETERS FOR THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE GLYPHOSATE-SELECTED GENERATION (S4) (…○…). STANDARD ERRORS FOR EACH

PARAMETER ESTIMATE ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF GR50 FOR THE COMMENCING POPULATION (G0) AND THE GLYPHOSATE-SELECTED

GENERATION (S4).

Biotype b c

e GR50

(g ha–1) P valueA R/S P value B

Imazamox G0 2.45 (0.83) –1.1 (1.32) 6.5 (0.89) – – 0.99

S4 0.63 (0.12) 0.2 (0.61) 17.60 (4.44) >0.05 2.7

Metosulam G0 2.59 (0.45) 1.2 (1.45) 6.4 (0.44) – – 0.83

S4 1.43 (0.18) 0.4 (0.84) 6.6 (0.63) 0.80 1.0

Chlorsulfuron G0 2.59 (0.47) 0.5 (1.00) 1.0 (0.07) – – 0.61

S4 1.70 (0.27) –2.2 (2.41) 1.0 (0.10) 0.68 1.0

A LD50 P value comparing the difference between the selected and commencing populations using the SI function in the DRC package in R. v2.14.1. B Lack-of-fit P value for appropriateness of the three parameter logistic model [1] for each herbicide treatment.

PAGE | 66


DISCUSSION

EVOLUTIONARY RESPONSE OF WILD RADISH TO LOW-DOSE GLYPHOSATE SELECTION

Herbicide resistance is the evolved and inherited capacity of a plant population to withstand

the normal recommended herbicide dose that had been previously effective in controlling the

same population (Vencill et al. 2012). Commencing with a glyphosate-susceptible wild radish

population, this study reports that four successive generations of low rate, recurrent

glyphosate selection resulted in only low-level glyphosate resistance (Table 4.2, Figure 4.1,

Plate 4.1), with glyphosate dose response studies quantifying a maximum 2.4-fold shift

towards glyphosate resistance in the fourth generation of selection. This result concurs with

similar experiments conducted with ryegrass (Busi and Powles 2009) and indeed in glyphosate

recurrent selection work conducted with Lolium perenne (Johnston and Faulkner 1991),

birdsfoot trefoil (Lotus corniculatus L.) (Boerboom et al. 1991) and fescue (Festuca Iongifolia

and rubra L.) (Johnston et al. 1989) which exhibited a maximum 2–5-fold increase in

glyphosate resistance following glyphosate selection.

The genetic basis for the modest level of glyphosate resistance obtained in this study remains

to be identified, however it has been observed that the progressive resistance shifts in this study

are characteristic of the accumulation and enrichment of minor gene trait/s (Busi and Powles

2009). Outcrossing appears to be an essential biological attribute enabling this resistance to

evolve as the closely-related self-pollinated member of the Brassicaceae, Arabidopsis thaliana

L., failed to evolve glyphosate resistance following seven generations of glyphosate selection

(Brotherton et al. 2007). Similarly, low-dose glyphosate selection did not select glyphosate

resistance in the self-pollinating grass species wild oat (Avena fatua L.) (Busi and Powles;

unpublished). However, unlike previous selection studies (Johnston et al. 1989; Boerboom et al.

1991; Busi and Powles 2009), all selections in this study (S1–S4) were performed at or above the

field recommended rate for wild radish control in Australia (540 g ha–1). This report should

therefore act as a clear warning regarding the evolutionary consequences of high wild radish

survivorship following glyphosate use, as any decrease in population susceptibility to glyphosate

may lead to population increases, placing glyphosate and other wild radish control options at

risk of resistance evolution (Jasieniuk et al. 1996; Neve et al. 2003).

Recurrent selection for four generations with the known herbicide-susceptible wild radish

biotype (G0) also resulted in unexpected and relatively weak cross-resistance to the ALS-

inhibiting herbicides imazamox (4.7-fold) and metosulam (3.6-fold). The genetic and mechanistic

basis of this cross resistance remains to be determined; however it appeared to be herbicide

PAGE | 67


specific with no decrease in susceptibility to the ALS-inhibiting herbicide chlorsulfuron. The lack

of cross resistance to chlorsulfuron was unexpected as chlorsulfuron cross resistance in common

amongst annual ryegrass populations (Christopher 1991). As expected, the lack of resistance at

the field rate for the ALS-inhibiting herbicide sulfometuron-methyl (Table 4.4)(which cannot be

metabolised in wheat (Triticum aestivum L.) or annual ryegrass (Anderson and Swain 1992))

indicates that the cross resistance in the S4 population was not the result of a pre-existing ALS

gene mutation (Yu et al. 2003), which is also considered unlikely to pre-exist in a small wild radish

population of just 396 plants.

CONCLUSION

This study concurs with the previous study in grasses by Busi and Powles (2009) that recurrent

low-rate glyphosate selection in an outcrossing dicot species (wild radish) over four consecutive

generations only resulted in the evolution of modest glyphosate resistance. Along with modest

glyphosate resistance, this study demonstrated a concomitant weak cross resistance to the ALS-

inhibiting herbicides metosulam and imaxamox. The genetic and biochemical basis of this weak

glyphosate resistance and cross resistance remains unexplained and requires further

investigation. There are several human, environmental and biological factors that can reduce

the efficiency of glyphosate in the field and therefore trigger the small shifts in glyphosate

resistance demonstrated in this study (Duke 1988). As a consequence glyphosate use should be

monitored to ensure that only highly-effective rates are applied and its use is combined with

weed control tactics that maximise weed control diversity to limit selection of GR weed species

(Powles 2008; Powles and Yu 2010).

ACKNOWLEDGEMENTS

This research was funded by the Grains Research Development Corporation of Australia (GRDC)

and the Australian Government through an Australian Post Graduate Award to M Ashworth.

Thank you to AHRI and plant growth facilities staff at The University of Western Australia who

provided technical assistance.

PAGE | 68


PLATE 4.1. PHOTOGRAPHIC REPRESENTATION OF THE RESPONSE OF THE GLYPHOSATE-SELECTED GENERATION TO GLYPHOSATE. G0 REPRESENTS THE UNSELECTED, COMMENCING

WILD RADISH POPULATION. S1–S4 REPRESENT FOUR GLYPHOSATE-SELECTED GENERATIONS (S1–S4).

PAGE | 69


REFERENCES

Amrhein, N., et al. (1980). The site of inhibition of the shikimate pathway of glyphosate. II. Interference of glyphosate with chorismate formation in vivo and in vitro. Plant Physiology 66: 830–834.

Anderson, J. J. and R. S. Swain (1992). Metabolism of sulfometuron-methyl in wheat and its possible role in wheat intolerance. Journal of Agricultural and Food Chemistry 40(11): 2279–2283.

Ashworth, M. B., et al. (2014). Identification of the first glyphosate-resistant wild radish (Raphanus raphanistrum L.) populations. Pest Management Science 70(9): 1432–1436.

Baylis, A. (2000). Why glyphosate is a global herbicide: strengths, weaknesses and prospects. Pest Management Science 56: 299–308.

Blackshaw, R. E., et al. (2002). Influence of wild radish on yield and quality of canola. Weed Science 50(3): 344–349.

Boerboom, C. M., et al. (1991). Recurrent selection for glyphosate tolerance in Birdsfoot-trefoil. Crop Science 31(5): 1124–1129.

Bradshaw, L. D., et al. (1997). Perspectives on glyphosate resistance. Weed Technology 11(1): 189–198.

Brotherton, J. E., et al. (2007). Identification of Arabidopsis thaliana variants with differential glyphosate responses. Journal of Plant Physiology 164: 1337–1345.

Busi, R., et al. (2012). Understanding the potential for resistance evolution to the new herbicide pyroxasulfone: field selection at high doses versus recurrent selection at low doses. Weed Research 52(6): 489–499.

Busi, R., et al. (2013). Evolved polygenic herbicide resistance in Lolium rigidum by low-dose herbicide selection within standing genetic variation. Evolutionary Applications 6: 231–242.

Busi, R. and S. B. Powles (2009). Evolution of glyphosate resistance in a Lolium rigidum population by glyphosate selection at sublethal doses. Heredity 103(4): 318–325.

Cheam, A. H., Code, G.R. (1995). The biology of Australian weeds: 24 Raphanus raphanistrum L. Plant Protection Quarterly 10(1): 2–13.

Christopher, J. T., et al. (1992). Resistance to acetolactate synthase-inhibiting herbicides in annual ryegrass (Lolium rigidum) involves at least two mechanisms. Plant Physiology 100(4): 1909–1913.

Duke, S. (1988). Glyphosate herbicides – Chemistry, degradation and mode of action. P. C. Kearney, Kaufman, D. D. New York, Dekker. 3: 1–70.

Duke, S. O. (2011). Glyphosate degradation in glyphosate-resistant and -susceptible crops and weeds. Journal of Agricultural and Food Chemistry 59(11): 5835–5841.

Duke, S. O. and S. B. Powles (2008). Glyphosate: a once-in-a-century herbicide. Pest Management Science 64(4): 319–325.

Dyer, W. E. (1994). Resistance to glyphosate. Herbicide resistance in plants: Biology and biochemistry. S. B. Powles and J. A. M. Holtum. New York, NY, CRC Lewis Publishers: 229–241.

Eslami, S. V., et al. (2006). Wild radish (Raphanus raphanistrum) interference in wheat. Weed Science 54(4): 749–756.

PAGE | 70


Farooq, M., et al. (2011). Crop yield and weed management in rainfed conservation agriculture. Soil and Tillage Research 117(0): 172–183.

Franz, J. E., et al. (1997). Glyphosate: A unique global herbicide. Washington, DC., Americal Chemical Society.

Gaines, T. A., et al. (2014). RNA-Seq transcriptome analysis to identify genes involved in metabolism-based diclofop resistance in Lolium rigidum. The Plant Journal: 865–876.

Green, J. M. (2011). Outlook on weed management in herbicide-resistant crops: Need for diversification. Outlooks on Pest Management 22(3): 100–104.

Gressel, J. (1995). Creeping resistances: The outcome of using marginally effective or reduced rates of herbicides. . Brighton Crop Protection Conference – Weeds, Brighton, British Crop Protection Council Publications.

Hashem, A., et al. (2001). Resistance of wild radish (Raphanus raphanistrum) to acetolactate synthase-inhibiting herbicides in the Western Australia wheat-belt. Weed Technology 15(1): 68–74.

Hashem, A., et al. (2001). Triazine resistance in a biotype of wild radish (Raphanus raphanistrum) in Australia. Weed Technology 15(4): 636–641.

Heap, I. (2014). The international survey of herbicide resistant weeds. Retrieved 20 May 2014, from www.weedscience.com.

Holt, J. S., et al. (1993). Mechanisms and agronomic aspects of herbicide resistance. Annual Review of Plant Physiology and Plant Molecular Biology 44(1): 203–229.

Jasieniuk, M., et al. (1996). The evolution and genetics of herbicide resistance in weeds. Weed Science 44(1): 176–193.

Jasieniuk, M. and B. Maxwell (1994). Populations genetics and the evolution of herbicide resistance in weeds. Phytoprotection 75(4): 25–35.

Johnston, D. T. and J. S. Faulkner (1991). Herbicide resistance in the Graminaceae a plant breeder's view. Herbicide resistance in weeds and crops. J. C. Caseley, G. W. W Cussans and R. T. Atkin. Oxford, Butterworth-Heinemann: 319–330.

Johnston, D. T., et al. (1989). Selection for tolerance to glyphosate in fine-leaved Festuca species. 6th International Turfgrass Research Conference, Tokyo.

Kercher, S. and J. K. Conner (1996). Patterns of genetic variability within and among populations of wild radish, Raphanus raphanistrum (Brassicaceae). American Journal of Botany 83(11): 1416–1421.

Lamb, D. C., et al. (1998). Glyphosate is an inhibitor of plant cytochrome P450: Functional expression of thlaspiarvensae cytochrome P45071B1/reductase fusion protein in Escherichia coli. Biochemical and Biophysical Research Communications 244(1): 110–114.

Llewellyn, R. S., et al. (2012). Extensive use of no-tillage in grain growing regions of Australia. Field Crops Research 132(0): 204–212.

Lorraine-Colwill, D. F., et al. (2001). Inheritance of evolved glyphosate resistance in Lolium rigidum (Gaud.). TAG Theoretical and Applied Genetics 102(4): 545–550.

Madhou, P., et al. (2005). Genetic variation in populations of Western Australian wild radish. Australian Journal of Agricultural Research 56: 1079–1087.

Manalil, S., et al. (2011). Rapid evolution of herbicide resistance by low herbicide dosages. Weed Science 59(2): 210–217.

PAGE | 71



Neve, P., et al. (2003). Simulating evolution of glyphosate resistance in Lolium rigidum I: population biology of a rare resistance trait. Weed Research 43(6): 404–417.

Neve, P. and S. Powles (2005). High survival frequencies at low herbicide use rates in populations of Lolium rigidum result in rapid evolution of herbicide resistance. Heredity 95(6): 485–492.

Neve, P. and S. Powles (2005). Recurrent selection with reduced herbicide rates results in the rapid evolution of herbicide resistance in Lolium rigidum. Theoretical and Applied Genetics 110(6): 1154–1166.

Ng, C.-H., et al. (2004). Inheritance of glyphosate resistance in Goosegrass (Eleusine indica). Weed Science 52(4): 564–570.

Powles, S., et al. (1998). Evolved resistance to glyphosate in rigid ryegrass (Lolium rigidum) in Australia. Weed Science 46: 604–607.

Powles, S. and Q. Yu (2010). Evolution in action: plants resistant to herbicides. Annual Review of Plant Biology 61: 317–347.

Powles, S. B. (2008). Evolved glyphosate-resistant weeds around the world: Lessons to be learnt. Pest Management Science 64(4): 360–365.

Price, W. J., et al. (2012). Estimation of dose–response models for discrete and continuous data in weed science. Weed Technology 26(3): 587–601.

Reeves, T. G., et al. (1981). Seed production and longevity, seasonal emergence and phenology of wild radish (Rahanus raphanistrum). Australian Journal of Experimental Agriculture 21(112): 524–530.

Snow, A., Campbell, L. (2005). Can ferral radishes become weeds. Crop Ferality and Volunteerism. J. Gressel. Boca Raton, FL., CRC Press: 193–207.

Steinrücken, H. C. and N. Amrhein (1980). The herbicide glyphosate is a potent inhibitor of 5-enolpyruvylshikimic acid-3-phosphate synthase. Biochemical and Biophysical Research Communications 94(4): 1207–1212.

Streibig, J. C., et al. (1993). Dose-response curves and statistical models. Herbicide Bioassays. P. Kudsk and J. C. Streibig. Boca Raton, CRC Press: 29–55.

Vencill, W. K., et al. (2012). Herbicide resistance: Toward an understanding of resistance development and the impact of herbicide-resistant crops. Weed Science 60(sp1): 2–30.

Vila-Aiub, M. M. and C. M. Ghersa (2005). Building up resistance by recurrently exposing target plants to sublethal doses of herbicide. European Journal of Agronomy 22(2): 195–207.

Walsh, M. J., et al. (2004). Multiple-herbicide resistance across four modes of action in wild radish (Raphanus raphanistrum). Weed Science 52(1): 8–13.


Xiang, W.-S., et al. (2005). Expression of a wheat cytochrome P450 monooxygenase in yeast and its inhibition by glyphosate. Pest Management Science 61(4): 402–406.

Yu, Q., et al. (2013). Enhanced rates of herbicide metabolism in low herbicide-dose selected resistant Lolium rigidum. . Plant Cell and Environment 36: 818–827.

Yu, Q. and S. Powles (2014). Metabolism-based herbicide resistance and cross-resistance in crop weeds: A threat to herbicide sustainability and global crop production. Plant Physiology In Press.

PAGE | 72


Yu, Q., et al. (2003). ALS gene proline (197) mutations confer ALS herbicide resistance in eight separated wild radish (Raphanus raphanistrum) populations. Weed Science 51(6): 831–838.

Zelaya, I. A., et al. (2004). Inheritance of evolved glyphosate resistance in Conyza canadensis L Theoretical and Applied Genetics 110(1): 58–70.

PAGE | 73


PAGE | 74

CHAPTER 5. LOW DOSE 2,4-D AMINE SELECTION IN WILD RADISH (Raphanus raphanistrum L.)

CHAPTER 5.

RECURRENT SELECTION WITH REDUCED 2,4-D AMINE RATES RESULTS IN THE RAPID EVOLUTION OF 2,4-D AND ALS HERBICIDE RESISTANCE IN

WILD RADISH (Raphanus raphanistrum L.)

ABSTRACT

When used at effective rates, resistance to auxinic herbicides has been slow to evolve when

compared to other modes of action. Here we report the evolutionary response of a herbicide-

susceptible population of wild radish (Raphanus raphanistrum L.) and confirm that sublethal

doses of 2,4-D amine can lead to the rapid evolution of 2,4-D resistance. Following four cycles

of 2,4-D selection, the progeny of a susceptible wild radish population evolved 2,4-D resistance,

increasing the LD50 9.6-fold from 14 g ha–1 to 138 g ha–1. This resulted in 11% survival at the field

recommended rate of 500 g ha–1. Further increases in resistance were considered likely if

selected past the fourth generation. Along with 2,4-D resistance, cross resistance was also

reported to the acetolactate synthase (ALS)-inhibiting herbicides metosulam (3.5-fold), and

chlorsulfuron (2.3-fold). Pre-treatment of the 2,4-D selected population (S4) with the potent

inhibitor of cytochrome P450 monooxygenases, malathion, restored chlorsulfuron to full

efficacy, indicating that mechanism providing cross resistance to chlorsulfuron was likely

metabolism based mediated by increased P450 activity. The mechanism/s and genetic control

of the auxinic and ALS cross resistance in this study remain to be investigated.

INTRODUCTION

Synthetic auxin herbicides such as 2,4-dichlorophenoxyacetic acid (2,4-D), 2-methyl-4-

chlorophenoxyacetic acid (MCPA) and 3,6-dichloro-2-methoxybenzoic acid (dicamba) are

effective tools for controlling dicot weeds in a variety of crops and other situations. These auxinic

herbicides induce a sustained increase in the plant hormone indole-3-acetic acid (IAA) which

leads to unregulated growth (Sterling and Hall 1997) and the accumulation of ethylene, abscisic

acid and reactive oxygen species, which lead to necrosis and death (Grossmann 2000;

Grossmann 2010). Synthetic auxins are a highly-effective herbicide class, used annually on an

estimated 200 million ha globally (Green 2012). With transgenic auxin-herbicide-resistant crops

soon to be commercialised, it is expected that this global use with further increase (Behrens et

al. 2007; Green et al. 2008; Wright et al. 2010). However, unlike other modes of action such as

PAGE | 75


ALS-inhibiting herbicides (Powles and Yu 2010), the evolution of auxin resistance has been slow

with resistance only reported in 31 species (Heap 2013) despite over 50 years of use (Weintraub

1953).

When auxin herbicides are used at recommended field rates, inheritance studies have indicated

that resistance is conferred by single major genes (with just a few cases two major genes)

(Coupland 1994; Preston and Mallory-Smith 2001; Mithila et al. 2011). However, it is known that

weedy plant populations contain continuous and heritable variation in herbicide susceptibility

(Holliday and Putwain 1980; Patzoldt et al. 2002). When outcrossing species are selected at low

herbicide rates (within the populations standing genetic variability for herbicide response) both

weak and strong gene traits are selected and accumulated in survivors until resistance is evident

(Powles and Yu 2010; Délye et al. 2013). The evolution of polygenic herbicide resistance by

selection at low herbicide doses was first demonstrated in the grass species annual ryegrass

(Lolium rigidum) (Neve and Powles 2005), resulting in the rapid evolution of diclofop-methyl

resistance through the accumulation of multiple genes including cytochrome P450

monooxygenase (P450) enzymes (Busi et al 2011; Yu et al 2013; Gaines et al 2014). Other than

in Chapter 4, no such low-herbicide-dose recurrent evolutionary studies have been conducted

in dicot species. Wild radish (Raphanus raphanistrum L.) is a good candidate for such studies.

Wild radish is a highly problematic global weed (Snow 2005) causing economic loss in crops by

aggressively competing for water, nutrients and light (Reeves et al. 1981; Blackshaw et al. 2002;

Eslami et al. 2006). Since the 1970s, wild radish control has been dominated by herbicides

resulting in the evolution of resistance to multiple modes of action, including acetolactate

synthase inhibiting (ALS) (Hashem et al. 2001), phytoene desaturase (PDS) (Walsh et al. 2004),

photosynthetic electron transport (Hashem et al. 2001), 5-enolpyruvylshikimate-3-phosphate

synthase (EPSPS) (Ashworth et al. 2014) and synthetic auxin herbicides (Walsh et al. 2004). Due

to its high genetic variability and obligate cross-pollination (Cheam 1995; Kercher and Conner

1996; Madhou et al. 2005), wild radish was evaluated for its ability to evolve resistance to low-

dose selection with an auxinic herbicide. It is hypothesised that minor genes with additive effect

will be selected at low 2,4-D rates in wild radish.

This study investigated the potential of a known herbicide-susceptible wild radish population to

evolve 2,4-D resistance when recurrently selected at low 2,4-D amine doses. The evolution of

cross resistance to chemically dissimilar herbicides was also assessed.

PAGE | 76



PLANT MATERIAL

This selection study was conducted using the known herbicide-susceptible wild radish

(Raphanus raphanistrum L.) biotype WARR7 (referred hereafter as G0) originally collected in

1999 from Yuna, Western Australia (28.34° S, 115.01° E) (Walsh et al. 2004). Since collection,

seed stocks of this population have been maintained and multiplied in isolation without

herbicide selection preventing the ingression of any herbicide resistance genes. Using this

known 2,4-D susceptible population, four successive generations were recurrently selected with

low rates of 2,4-D amine in June 2011 (S1), April 2012 (S2), September 2012 (S3) and January

2013 (S4) using the general procedure detailed below (Table 5.1).

GENERAL PROCEDURE FOR POPULATION SELECTION

This pot study was conducted over four consecutive generations by selecting survivors from

dose–response studies involving six 2,4-D amine rates. Seeds (400 per rate) were planted

approximately 1 cm deep into four replicate polystyrene foam trays (400 mm wide x 500 mm

long x 150 mm deep) containing standard potting mixture (25% peat moss, 25% sand and 50%

mulched pine bark). After planting , seedlings were grown in the outdoor growth facility at The

University of Western Australia, being watered as required and fertilised weekly with 2 g Scotts

PolyFeed™ soluble fertiliser (N 19% [urea 15%, ammonium 1.9%, nitrate 2.1%), P 8%, K 16%, Mg

1.2%, S 3.8%, Fe 400 mg kg–1, Mn 200 mg kg–1,Zn 200 mg kg–1, Cu 100 mg kg–1, B 10 mg kg–1, Mo

10 mg kg–1). At the two true-leaf stage, seedlings were counted before being treated with

Amicide 625 (625 g 2,4-D Amine L–1, Nufarm, Laverton North, Vic., Australia) (Table 5.1).

Herbicide treatments were applied using a twin nozzle laboratory sprayer fitted with 110° 01

flat fan spray jets (Tee jet™) delivering herbicide in 118 L ha–1 of water at 210 kPa, travelling at

a speed of 3.6 km h–1. After treatment, plants were returned to the outdoor area. Survival was

assessed 42 days after treatment (DAT). Plants with new green growth were considered

survivors. The 2,4-D dose used for each selection was based upon 20% survival. From this

selected rate, 20 plants were sub-sampled based on maximum regrowth (S1–S4, Table 5.1). Prior

to flowering, the selected survivors were isolated to ensure cross pollination only among

survivors and to prevent ingress of foreign pollen. Once all plants were actively flowering, plants

were crossed manually using the 'Beestick method' as outlined by Williams (1980), ensuring a

random pattern of cross pollination (panmixia). At maturity all siliques were collected with the

seed threshed using a modified 'grist mill'. The separated seed progeny represented the next

generation for subsequent 2,4-D selection. During each selection, a separate population of 20

PAGE | 77


randomly-chosen untreated plants was isolated and crossed using the general procedure to

produce four successive generations of unselected controls (C1–C4) (Table 5.1).

TABLE 5.1. SEED PROGENY COLLECTED FROM BASAL WILD RADISH POPULATION WARR 7 [G0]. SELECTED PLANTS

(N) AT A SPECIFIC SUBLETHAL DOSE OF 2,4-D AMINE (g ha–1). 2,4-D AMINE RECURRENT SELECTED GENERATIONS

S1–S4. RECURRENTLY UNSELECTED GENERATIONS C1–C4.

2,4-D Amine

(g ha–1)

Population

size

Herbicide efficiency

(% control)A

Plants selected

(N)

Progeny

0 20 0 20 C1

31 361 7 0 –

62.5 374 17 0 –

125 382 71 20 S1

250 373 88 0 –

500 381 100 0 –

0 20 0 20 C2

125 384 14 0 –

250 396 88 20 S2

500 391 97 0 –

750 372 98 0 –

1000 100 0 –

0 20 0 20 C3

125 374 46 0 –

250 393 77 20 S3

500 386 96 0 –

750 371 99 0 –

1000 381 100 0 –

0 20 0 20 C4

125 391 6 0 –

250 362 24 0 –

500 372 48 0 –

750 379 76 20 S4

1000 362 100 0 – A Calculated as 100 – (% plant survival)

PAGE | 78


DOSE–RESPONSE ANALYSIS

In May 2013, the commencing (G0), selected (S1–S4) and control (C1–C4) progenies were

evaluated in a final 2,4-D dose–response study conducted under field conditions. For each 2,4-

D dose rate, 20 seeds per pot were planted into four replicate 180 mm diameter plastic pots

containing standard potting mixture. At the two true-leaf stage, the commencing (G0) and

control populations (C1–C4) were sprayed with 2,4-D amine at 0, 45, 90, 125, 250, 500, 750 and

1000 g ha–1 (field rate 500 g ha–1) with the selected lines (S1–S4) treated at 0, 90, 180, 250, 500,

750, 1000, 1500 and 2000 g ha–1 as previously described. Plant survival was assessed 42 DAT

with aboveground shoot biomass harvested and dried at 65°C for 7 days before weighing.

CROSS-RESISTANCE PROFILES OF COMMENCING SUSCEPTIBLE (G0) AND FOURTH GENERATION

2,4-D AMINE-SELECTED PROGENY (S4)

To determine the extent of any cross resistance to dissimilar herbicide modes of action, the G0

and S4 populations were treated at the recommended rate with a range of herbicides with

known activity against wild radish (Table 5.4). In April 2013, 20 seeds of each population were

established in four replicate 180 mm plastic pots containing standard potting mixture. At the

two-leaf stage, the recommended rate of each herbicide (Table 5.4) was applied as previously

described. Plant survival was assessed 35 DAT, with the exception of phenoxyacetic acid

herbicide treatments (MCPA, 2,4-D amine) which were assessed 42 DAT. Survivors from the

chlorsulfuron, metosulam, MCPA and 2,4-D ester treatments were isolated and crossed as

previously described to produce progeny (Table 5.4). This progeny was treated as previously

described and assessed to confirm the heritability of the cross resistance traits.

To quantify the strength of the cross resistance to the ALS-inhibiting herbicides, dose–response

studies were conducted as previously described using chlorsulfuron (750 g kg–1, Nufarm,

Laverton North, Vic., Australia) at 0, 3.75, 7.5, 15,30, 60 g ha–1; metsosulam (100 g kg–1, Bayer

Cropscience, Hawthorn East, Vic., Australia) at 0, 1.2, 2.5, 5, 10, 20 g ha–1, imazamox (700 g kg–

1, Crop Care, Murarrie, Qld, Australia) at 0, 8, 16, 32, 64, 128 g ha–1 and sulfometuron-methyl

(750 g kg–1, Nufarm, Laverton North, Vic., Australia) at 0, 1.5, 3, 6, 12, 24 g ha–1 (all corresponding

to 0.25 ×, 0.5 ×, 1 ×, 2 ×, 4 × the recommended dose). Plant survival and biomass was assessed

35 DAT.

A separate dose–response study was conducted to indicate the possible involvement of P450

enzyme metabolism as the basis for the chlorsulfuron cross resistance in the S4 population. The

commencing (G0) and fourth generation selected (S4) populations were established and

maintained as previously described before being treated with chlorsulfuron, with and without

PAGE | 79


pre-treatment with the known P450 inhibitor malathion (1000 g ha–1) (Table 5.6). At the two

true-leaf stage, 1000 g ha–1 of malathion (Maldison™ 500 g kg–1, Nufarm, Laverton North, Vic.,

Australia) was applied to each population as previously described. Thirty minutes later,

chlorsulfuron was applied at 0, 25, 50, 100, 200 g ha–1. Survival and biomass was assessed 35

DAT.

DATA ANALYSIS

Survival and biomass data was analysed using non-linear regression analyses with the DRC

package in R 3.0.0 (R Development Core Team 2011; http://www.R-project.org). Data was fitted

to a three parameter log-logistic model [1] with the upper limit fixed to 100 (Streibig et al. 1993;

Price et al. 2012):



to herbicide dose x, c is the lower asymptotic value of Y, and b is the slope of the curve around

e, which is the dose causing 50% mortality (LD50) or biomass reduction (GR50). LD50 and GR50

parameters of the selected (S1–S4) and unselected progeny (C1–C4) were compared to the

commencing (G0) population using the selectivity indices (SI) function (R 3.0.0) to determine if

the ratios between these values were significantly different (P<0.05). A Lack-of-Fit test was also

applied to each fitted curve to ascertain the appropriateness of the model [1]. A ratio of

recurrently selected and commencing populations (R/S) at the estimated LD50 level was used to

highlight the change in population susceptibility to 2,4-D. Data was plotted using SigmaPlot v.12

(Systat Software Inc., 2011).

RESULTS

EFFECT OF SUBLETHAL 2,4-D AMINE SELECTIONS ON THE COMMENCING POPULATION (G0)

Commencing with the known 2,4-D susceptible wild radish population (G0), four successive

generations of recurrent low dose 2,4-D amine selection was applied to produce four selected

generations (S1–S4) which were compared with four concurrently-grown generations treated in

the same way for four generations but without 2,4-D selection (C1–C4). In the final study, all

populations were grown outdoors and subjected to a full 2,4-D amine dose–response study to

ascertain the population response to 2,4-D amine.

The G0 population was confirmed to be susceptible to 2,4-D amine, with an LD50 of 14 g ha–1

(Table 5.2). As expected, at low 2,4-D doses, significant intra-population variability in 2,4-D

response was evident with 100% mortality achieved at 250 g ha–1 (LD99; 263 g ha–1). At the 2,4-

PAGE | 80



D rate (low) that caused 80% mortality, 20 surviving individuals were sub-sampled, crossed and

maintained for seed production, to serve as the S1 generation for the next cycle of 2,4-D

selection. Following three generations of low dose 2,4-D amine selection, heritable increases in

survival were evident, increasing the LD50 rate from 14 g ha–1 (G0) to 22 g ha–1 (S1), 54 g ha–1 (S2)

and 63 g ha–1 (S3) (Table 5.2). The fourth selection (S4) displayed clear evidence of 2,4-D amine

resistance, with the LD50 rate being 138 g ha–1 (Table 5.2: Figure 5.1: Plate 5.1). Comparing all

2,4-D selected progenies (S1–S4) with the starting G0 population, it is evident that four

generations of recurrent low dose 2,4-D amine selection resulted in the evolution of 2,4-D amine

resistance in this initially 2,4-D susceptible wild radish population. After four generations of low

dose 2,4-D selection the S4 generation was 9.6-fold resistant to 2,4-D, when compared to the

commencing G0 population, demonstrating that 2,4-D resistance evolved in this wild radish

population following four generations of low dose 2,4-D selection. At the field applied 2,4-D

amine rate, 11% of the S4 population survived, while the G0 populations was completely

controlled. Along with increased survival, plant growth was progressively less affected with

successive 2,4-D amine selections, increasing the rate required to reduce plant biomass by 50%

(GR50) from 60 g ha–1 (G0) to 111 g ha–1 (S1 and S2) and 171 g ha–1 (S3). Curve fitting in the fourth

generation (S4) did not effectively estimate the GR50 parameter due to a high standard error

(Table 5.3); therefore the GR50 rate was estimated from the raw data to be greater than 250 g

ha–1, resulting in a >4.2-fold increase in biomass when compared to the G0 population. For the

four control generations grown in the absence of 2,4-D selection, no shifts in either survival or

biomass were recorded (C1–C4), indicating that environmental factors did not contribute to the

2,4-D resistance shifts in the 2,4-D selected populations (S1–S4) (Tables 5.2, 5.3; Figure 5.2).

The mechanistic and genetic basis of the 2,4-D amine resistance in this study has not yet been

investigated, however increased survival was also recorded for the other phenoxyacetic acid

herbicide formulations: 2,4-D ester (11% survival at 540 g ha–1) and MCPA (7% survival at 1000

g ha–1). Both phenoxyacetic acid formulations controlled the G0 population (Table 5.4).

CROSS RESISTANCE TO CHLORSULFURON AND METASULAM

In addition to the evolution of 2,4-D amine resistance from four generations of low 2,4-D rate

recurrent selection (Figure 5.1), there was evidence of heritable cross resistance to the ALS-

inhibiting herbicides chlorsulfuron and metosulam. Initial screening found 58% of the S4

population survived chlorsulfuron at the recommended field rate (15 g ha–1) with 30% survival

at the recommended rate for metosulam (5 g ha–1) (Table 5.4). At these rates, the G0 population

was killed. Screening of the progeny from cross-resistance screening found these traits to be

heritable with 80% survival recorded for chlorsulfuron (15 g ha–1) and 59% for metosulam (5 g

PAGE | 81


ha–1) at the field recommended rate (Table 5.4). Full dose–response screening of the G0 and S4

populations quantified a modest 2.3-fold increase in chlorsulfuron (Figure 5.3, Table 5.5) and a

3.5-fold increase in metosulam resistance (Figure 5.4; Table 5.5) following four generations of

2,4-D amine selection. The genetic or biochemical basis of this chlorsulfuron and metosulam

cross resistance is yet to be investigated, however the mechanism/s appear to be strong as plant

biomass was not greatly affected by chlorsulfuron (>9.7-fold; GR50) (Figure 5.3; Table 5.5) or

metosulam application (7.1-fold;GR50) (Figure 5.4; Table 5.5). At the recommended field rate (15

g ha–1), chlorsulfuron treatment reduced biomass by 31% (S4) while metosulam (5 g ha–1)

reduced biomass by 43% when compared to the untreated equivalent control (S4). At these

rates, the G0 population was killed (nil biomass). No cross resistance was evident to the ALS-

inhibiting herbicides imazamox or sulfometuron-methyl (Table 5.5).

An indicator of the mechanistic basis for chlorsulfuron cross resistance in the 2,4-D selected

plants is that pre-treatment with the P450 inhibitor malathion (1000 g ha–1) completely reversed

chlorsulfuron cross resistance in the S4 population. In the absence of chlorsulfuron, malathion

pre-treatment had no effect on survival or biomass in either population (S4 or G0). However,

the application of malathion followed by chlorsulfuron reduced chlorsulfuron cross resistance

in the S4 population to full susceptibility (Figure 5.5, Table 5.6, Plate 5.2), decreasing the LD50

rate in the malathion-treated S4 population to 10% of the non-malathion treated (S4) population

(Table 5.6). Following malathion treatment, the G0 population became increasingly susceptible

to chlorsulfuron halving the LD50 rate (Table 5.6). Malathion pre-treatment also reduced the rate

required for 50% biomass reduction in the selected population (S4) by 87% from 31 g ha–1 to 4

g ha–1 (GR50) (Figure 5.5, Table 5.6). At the recommended chlorsulfuron rate (15 g ha–1), plant

biomass in the S4 population decreased from 51 g plant–1 to 6 g plant–1 following malathion pre-

treatment (Figure 5.5; Table 5.6).

PAGE | 82


A. 2,4-D Amine (g ha-1)

0 100 200 300 400 500

Pla

nt s

urvi

val (

%)

0

20

40

60

80

100

B. 2,4-D Amine (g ha-1)

0 100 200 300 400 500

Sho

ot b

iom

ass

(% u

ntre

ated

con

trol)

0

20

40

60

80

100

120

FIGURE 5.1. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE LOW-DOSE 2,4-D AMINE-

SELECTED GENERATIONS S1 (…□…), S2 (…∆…), S3 (…◊… ) AND S4 (…○…). EACH SYMBOL REPRESENTS THE MEAN OF NINE RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED

SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG-LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT STANDARD ERROR OF THE MEAN.

PAGE | 83


A. 2,4-D Amine (g ha-1)

0 100 200 300 400 500

Pla

nt S

urvi

val (

%)

0

20

40

60

80

100

B. 2,4-D Amine (g ha-1)

0 100 200 300 400 500

Sho

ot b

iom

ass

(% u

ntre

ated

con

trol)

0

20

40

60

80

100

120

FIGURE 5.2. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION G0 (__•__) AND THE UNSELECTED CONTROL

GENERATIONS C1 (…□…), C2 (…∆…), C3 (…◊…) AND C4 (…○…), DERIVED FROM CROSSING 20 RANDOMLY-SELECTED PROGENY. EACH SYMBOL REPRESENTS THE MEAN OF EIGHT

RATE TREATMENTS. THE PLOTTED LINES ARE PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG-LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT

1 STANDARD ERROR OF THE MEAN.

PAGE | 84


TABLE 5.2. PARAMETER ESTIMATES FROM THE THREE PARAMETER LOG-LOGISTIC MODEL [1] USED TO ESTIMATE LD50 PARAMETERS FOR PERCENT SURVIVAL BY NON-LINEAR

REGRESSION ANALYSIS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF LD50 FOR THE COMMENCING

POPULATION (G0) AND RESPECTIVE 2,4-D AMINE-SELECTED (S1–S4) OR CONTROL (C1–C4) POPULATIONS.

Progeny Selection rate

(g ha–1)

Population

size

Survival

(%)

Plants

selected

b c e (LD50)

(g ha–1)

P valueA R/S

G0 – – – – 1.27 (0.42) –0.87 (1.88) 14 (6) –

2,4-D

selected

S1 125 382 71 20 1.88 (0.28) –0.44 (0.54) 22 (2) <0.05 1.5

S2 250 396 88 20 2.13 (0.33) –0.49 (0.79) 54 (5) <0.05 3.7

S3 250 393 77 20 1.83 (0.36) 0.69 (1.57) 63 (7) <0.05 4.4

S4 750 379 76 20 1.52 (0.14) –0.30 (2.04) 138 (8) <0.05 9.6

Unselected C1 0 20 – 20 1.75 (0.22) –0.49 0.71) 39 (5) 0.14 2.7

C2 0 20 – 20 1.54 (0.37) –0.76 (1.21) 18 (4) 0.51 1.2

C3 0 20 – 20 1.63 (0.27) –0.65 (0.88) 20 (3) 0.31 1.4

C4 0 20 – 20 `1.16 (0.37) –0.85 (1.39) 10 (5) 0.64 0.7

A LD50 P value comparing the difference between selected and commencing populations assessed by the SI function in the DRC package in R. v2.14.1.


PAGE | 85


TABLE 5.3. PARAMETER ESTIMATES FROM THE THREE PARAMETER LOG-LOGISTIC MODEL [1] USED TO ESTIMATE GR50 PARAMETERS FOR BIOMASS IN THE FORM OF PERCENT MEAN

CONTROL CALCULATED BY NON-LINEAR REGRESSION ANALYSIS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO

OF GR50 FOR THE COMMENCING POPULATION (G0) AND RESPECTIVE 2,4-D AMINE-SELECTED (S1–S4) OR CONTROL (C1–C4) POPULATIONS.

Population b c e GR50 (g ha–1) P valueA R/S

G0 4.66 (1.17) 0.86 (3.42) 60 (6) –

2,4-D selected S1 3.99 (0.96) –0.45 (2.91) 112 (10) <0.05 1.8

S2 3.70 (0.67) –0.54 (2.26) 111 (8) <0.05 1.8

S3 1.77 (0.35) –2.60 (4.20) 171 (19) <0.05 2.8

S4 0.66 (0.20) –51.42 (50.65) >250* – >4.2

Unselected C1 2.41 (0.51) –1.06 (3.52) 65 (7) 0.59 1.1

C2 1.73 (0.53) –1.88 (4.18) 44 (7) 0.14 0.8

C3 3.52 (0.78) –0.56 (3.29) 71 (7) 0.28 1.2

C4 3.39 (1.31) 0.26 (1.13) 60 (10) 0.96 1.0

A LD50 P value comparing the difference between selected and commencing populations assessed by the SI function in the DRC package in R. v2.14.1.

Lack-of-fit P value for appropriateness of the three parameter logistic model [1] (G0;S1–S3;C1–C4) – 0.99.

Lack-of-fit P value for appropriateness of the three parameter logistic model [1] (S4) – 0.24. * Estimated GR50 from raw data.

PAGE | 86


TABLE 5.4. POPULATION SURVIVAL FOR THE COMMENCING WILD RADISH POPULATION (G0) AND THE FOURTH LOW DOSE 2, 4-D AMINE-SELECTED GENERATION (S4) WHEN SPRAYED

AT THE RECOMMENDED RATE WITH A RANGE OF HERBICIDES WITH KNOWN ACTIVITY AGAINST WILD RADISH. EACH RESULT IS THE MEAN OF FOUR REPLICATES. STANDARD ERRORS OF

THE MEANS ARE IN PARENTHESES.

Herbicide mode

of action

Herbicide active Rate

(g ha–1)

Adjuvant Mean plant survival (%)

Commencing

population (G0)

4th generation

Selected (S4)

Commencing

population (G0)

(Cross-resistant

progeny control)

4th generation

Selected (S4)

(Cross-resistant

progeny)

ALS inhibitor Chlorsulfuron 15 0.1% BS1000 0 58 (2) 0 80 (3)

ALS inhibitor Metosulam 5 0.1% BS1000 0 30 (8) 0 59 (2)

ALS inhibitor Sulfometuron-methyl 7.5 0.1% BS1000 0 0

ALS inhibitor Imazimox 32 0.1% BS1000 0 5 (3)

Synthetic auxin MCPA amine 1000 – 0 7 (4) 0 16 (1)

Synthetic auxin 2,4-D ester 540 – 0 11 (5) 2 (1) 18 (2)

PDS Inhibitor Diflufenican 100 – 3 (1) 1 (1)

PSII Bromoxynal 400 – 0 0

PSII Diuron 1000 – 0 0

PSII Metribuzin 280 – 0 0

PSII Atrazine 1000 2% Hasten 0 0

EPSPS inhibitor Glyphosate 540 – 0 0

PSI Diquat 100 – 0 0

PAGE | 87


A. Chlorsulfuron (g ha-1)

0 10 20 30 40 50 60

Pla

nt s

urvi

val (

%)

0

20

40

60

80

100

B. Chlorsulfuron (g ha-1)

0 10 20 30 40 50 60

Sho

ot b

iom

ass

(% u

ntre

ated

con

trol)

0

20

40

60

80

100

120

FIGURE 5.3. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION (G0) (__•__) AND THE FOURTH LOW DOSE 2,4-D

AMINE SELECTED GENERATION (S4) (…∆…) FOLLOWING CHLORSULFURON TREATMENT. EACH SYMBOL REPRESENTS THE MEAN OF FOUR REPLICATES. THE PLOTTED LINES ARE

PREDICTED SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG-LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 STANDARD ERROR OF THE MEAN.

PAGE | 88


A. Metosulam (g ha-1)

0 5 10 15 20

Pla

nt s

urvi

val (

%)

0

20

40

60

80

100

B. Metosulam (g ha-1)

0 5 10 15 20

Sho

ot b

iom

ass

(% u

ntre

ated

con

trol)

0

20

40

60

80

100

120

FIGURE 5.4. DOSE–RESPONSE CURVES FOR SURVIVAL (A) AND BIOMASS (B) OF THE COMMENCING WILD RADISH POPULATION (G0) (__•__) AND THE FOURTH LOW DOSE 2,4-D

AMINE SELECTED GENERATION (S4) (…∆…) FOLLOWING METOSULAM TREATMENT. EACH SYMBOL REPRESENTS THE MEAN OF FOUR REPLICATES. THE PLOTTED LINES ARE PREDICTED

SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG-LOGISTIC MODEL [1]. VERTICAL BARS REPRESENT 1 STANDARD ERROR OF THE MEAN.

PAGE | 89


TABLE 5.5. PARAMETER ESTIMATES FOR CHLORSULFURON, METOSULAM, IMAZAMOX AND SULFOMETURON-METHYL CALCULATED BY NON-LINEAR REGRESSION ANALYSIS USING A

THREE PARAMETER LOG-LOGISTIC MODEL [1] TO ESTIMATE LD50 PARAMETERS FOR SURVIVAL AND GR50 PARAMETERS FOR BIOMASS (% OF MEAN CONTROL). STANDARD ERRORS

FOR PARAMETER ESTIMATES ARE IN PARENTHESES. R/S VALUES WERE CALCULATED AS A RATIO OF LD50 OR GR50 BETWEEN THE COMMENCING POPULATION (G0) AND THE FOURTH

LOW DOSE 2,4-D AMINE-SELECTED GENERATION (S4).

A. Survival

Biotype B c e (LD50) (g ha–1) P valueA R/S P valueB Chlorsulfuron G0 6.03 (0.46) –1.29 (1.53) 4 (0.1) – 0.34

S4 1.23 (0.13) 13.22 (3.91) 19 (1.5) <0.05 4.4 Metosulam G0 1.95 (0.24) 0.02 (0.68) 1 (0.1) – 0.36

S4 1.36 (0.11) 28.75 (1.70) 4 (0.2) <0.05 4.8 Imazamox G0 2.15 (0.21) –0.45 (0.80) 5 (0.2) – 0.96

S4 2.34 (0.31) –0.08 (0.97) 5 (0.3) 0.83 1.0 Sulfometuron-

methyl

G0 2.01 (0.24) 1.34 (1.72)

2 (0.1) – 0.77 S4 3.26 (0.35) 2.43 (1.80) 2 (0.1) 0.72 1.0

B. Biomass

Biotype B c e (GR50) (g ha–1) P valueA R/S P valueB Chlorsulfuron G0 2.59 (0.58) 0.64 (2.55) 3 (0.2) – 0.99

S4 0.58 (0.06) 12.71 (49.72) 58 (6.9) <0.05 >9.7 Metosulam G0 1.33 (0.34) –8.71 (11.47) 1 (0.2) – 0.81

S4 0.58 (0.13) 0.33 (5.47) 7 (1.6) <0.05 7.1 Imazamox G0 1.33 (0.27) –4.84 (5.07) 5 (0.8) – 0.11

S4 1.27 (0.16) –5.28 (4.49) 6 (0.6) 0.53 1.1 Sulfometuron-

methyl

G0 1.95 (0.43) 10.34 (8.14) 7 (0.8) – 0.10 S4 4.23 (2.16) 2.85 (7.53) 8 (0.4) 0.87 1.1

A LD50 P value comparing the difference between selected and commencing populations using the SI function in the DRC package in R. v2.14.1. B Lack-of-fit P value for appropriateness of the three parameter logistic model [1] for each herbicide treatment.

PAGE | 90


A. Chlorsulfuron (g ha-1)

0 5 10 15 20 25 30

Pla

nt s

urvi

val (

%)

0

20

40

60

80

100

G0S4 G0 plus malathionS4 plus malathion

B. Chlorsulfuron (g ha-1)

0 5 10 15 20 25 30

Sho

ot b

iom

ass

(% u

ntre

ated

con

trol)

0

20

40

60

80

100

120

S4 plus malathion S4 G0 plus malathion G0

FIGURE 5.5. DOSE–RESPONSE CURVES OF SURVIVAL (A) AND BIOMASS (B) FOR MALATHION PRE-TREATMENT PRIOR TO CHLORSULFURON. THE PLOTTED LINES ARE PREDICTED

SURVIVAL (A) AND BIOMASS (B) CURVES USING A THREE PARAMETER LOG-LOGISTIC MODEL [1] . DASHED LINES AND CIRCULAR SYMBOLS (- - - - -) DENOTE THE COMMENCING WILD

RADISH POPULATION (G0); SOLID LINES AND TRIANGULAR SYMBOLS (_____) DENOTE THE FOURTH LOW DOSE 2,4-D AMINE SELECTED GENERATION (S4). SOLID SYMBOLS ( ▲, •)

DENOTE NO MALATHION PRE-TREATMENT. HOLLOW SYMBOLS (∆,◦) DENOTE MALATHION PRE-TREATMENT. VALUES ARE THE MEAN OF FOUR REPLICATES WITH VERTICAL BARS

REPRESENTING 1 STANDARD ERROR OF THE MEAN.

PAGE | 91


TABLE 5.6. PARAMETER ESTIMATES FOR MALATHION PRE-TREATMENT PRIOR TO CHLORSULFURON, CALCULATED BY NON-LINEAR REGRESSION ANALYSIS USING A THREE PARAMETER

LOG-LOGISTIC MODEL [1] TO ESTIMATE LD50 PARAMETERS FOR SURVIVAL (A) AND GR50 PARAMETERS FOR PLANT BIOMASS (% UNTREATED CONTROL) (B). STANDARD ERRORS FOR

PARAMETER ESTIMATES ARE IN PARENTHESES. PRE-TREATMENT RATIOS WERE CALCULATED AS THE RATIO OF LD50 OR GR50 BETWEEN THE MALATHION PRE-TREATED AND NON-PRE-

TREATED PLANTS WITHIN BOTH THE COMMENCING POPULATION (G0) OR THE FOURTH LOW DOSE 2,4-D AMINE SELECTED GENERATION (S4).

A. Survival

Biotype Malathion pre-treatment

(g ha–1)

b C e (LD50)

(g ha–1)

P value A Pre-treatment

ratio

G0 0 2.36 (0.65) 0.33 (1.15) 4 (0.8) – – G0 1000 2.46 (0.63) –4.86 (8.34) 2 (0.7) <0.05 0.5 S4 0 1.04 (0.16) 35.44 (2.16) 23 (2) – – S4 1000 1.42 (0.32) –1.02 (3.32) 2 (0.7) <0.05 0.1

B. Biomass

Biotype Malathion pre-treatment

(g ha–1)

b C e (GR50)

(g ha–1)

P value A Pre-treatment

ratio

G0 0 1.69 (0.63) –1.58 (3.81) 3 (1) – – G0 1000 2.45 (1.21) –2.48 (3.82) 2 (1) <0.05 0.7 S4 0 0.51 (0.18) 16.87 (118.27) 31 (8) – – S4 1000 1.27 (0.40) –18.25 (45.41) 4 (1) <0.05 0.1

A LD50 and GR50 P value comparing the difference between selected and commencing populations assessed by the SI function in the DRC package in R. v2.14.1.

Lack-of-fit P value for appropriateness of the three parameter logistic model [1] minus malathion, Survival: 0.47, Biomass: 0.99.

Lack-of-fit P value for appropriateness of the three parameter logistic model [1] plus malathion, Survival: 0.96, Biomass: 0.98.

PAGE | 92


DISCUSSION

WILD RADISH RESPONSE TO RECURRENT LOW DOSE 2,4-D AMINE SELECTION

This study clearly demonstrates that in a small unselected wild radish population (G0), there is

sufficient genetic variability in 2,4-D amine response to evolve resistance to 2,4-D amine when

recurrently selected at low 2,4-D doses (Table 5.2, Figures 5.1, Plate 5.1). Following four

consecutive generations of low 2,4-D dose selection, the fourth (S4) generation displayed clear

evidence of 2,4-D amine resistance up to 9.6-fold higher (LD50) than the starting G0 population.

As 2,4-D selection consistently increased both the survival and biomass index’s in this study, 2,4-

D selection past the fourth generation is considered likely to further increase resistance. Along

with resistance to 2,4-D amine, increased survival following 2,4-D ester and MCPA amine

treatment was observed in the S4 generation, highlighting that 2,4-D amine selection can lead

to decreased susceptibility across multiple phenoxyacetic acid formulations within the synthetic

auxin herbicide group. This study is the first to demonstrate the consequences of low-dose 2,4-

D selection on a previously-susceptible dicot species. Like wild radish studied here, heritable

phenotypic variability in 2,4-D amine response has been measured in populations of chickweed

(Stellaria media) and cleavers (Galium aparine) (Lutman and Snow 1987; Barnwell et al. 1989).

The evolutionary shifts in this study as a result of four generations of herbicide selection concurs

with previous reports of low-dose selection with the grass species Lolium rigidum L. using the

acetyl-coenzyme carboxylase (ACCase)-inhibiting herbicide diclofop-methyl (Neve and Powles

2005) which, like 2,4-D amine, is metabolised in wheat (Zimmerlin and Durst 1992). Low-dose

diclofop-methyl selections of Lolium rigidum L. (Neve and Powles 2005) resulted in polygenic

accumulation of multiple herbicide-degrading enzymes including cytochrome P450

monooxygenases, leading to increased diclofop-methyl metabolism (Busi et al. 2013; Yu et al.

2013; Gaines et al. 2014). The mode of action of auxinic herbicides plus the mechanisms of

resistance in field-evolved weed species is not well understood (Grossmann 2010; Mithila et al.

2011); therefore the genetic basis of the 2,4-D resistance in this study has not been determined.

However it is unlikely that a rare, highly resistant, monogenic gene trait would pre-exist in the

small (382 plants) unselected population used in this study (McKenzie and Batterham 1994). The

progressive increases in resistance observed in the selected populations (S1–S4) (Figure 5.1) are

characteristic of the progressive accumulation and enrichment of minor gene traits (Neve and

Powles 2005). Since 2,4-D amine is metabolised in wheat through the activity of cytochrome

P450 monooxygenases (Bristol et al. 1977; Zimmerlin and Durst 1992; Hamburg et al. 2000;

Siminszky 2006) and 2,4-D and MCPA conjugates (hypothesised to be the result of P450

enzymes) have been identified in several dicot species such as peas (Pisum sativum L.); (Andreae

PAGE | 93


and Good 1957; Collins and Gaunt 1970), soybean (Feung et al. 1973), tobacco (Nicotiana

tobacum L.; (Feung et al. 1975) and sunflowers (Helianthus annum L.; (Feung et al. 1975). It is

plausible that an outcrossing dicot species such as wild radish could evolve resistance to 2,4-D

through low-rate selection and accumulation of minor genes encoding the regulation of P450

metabolism. However as 2,4-D conjugates retain some herbicidal activity and can be hydrolysed

back into active 2,4-D (Feung et al. 1977), low-dose selected 2,4-D resistant individuals are likely

to still be significantly affected by 2,4-D treatment.

Since the first report of auxinic herbicide-resistant weed populations (Fireweed (Senecio

madagascariensis) in Hawaii (Hilton 1957)), the number of field-evolved auxinic herbicide-

resistant weeds has slowly increased to 31 species worldwide (Coupland 1994; Heap 2014).

However characterisation of the mechanisms of auxinic herbicide resistance has only been

investigated in a few species (Mithila et al. 2011) to include reduced foliar and root uptake

(yellow starthistle (Centaurea solstitialis) (Fuerst et al. 1996)), reduced translocation to the roots

(common hempnettle (Weinberg et al. 2006)) and decreased auxinic herbicide binding to auxin

receptors (Arabidopsis (Marchant et al. 1999), wild mustard (Brassica rapa) (Zheng and Hall

2001) and false cleavers (Van Eerd et al. 2005)). However there is little difference in auxinic

herbicide metabolism between resistant and susceptible biotypes in false cleavers (Van Eerd et

al. 2005), yellow starthistle (Fuerst et al. 1996), prickly lettuce (Burke et al. 2009), kochia

(Cranston et al. 2001) and wild mustard (Peniuk et al. 1993). Differences in auxinic herbicide

metabolism however has been identified in the roots of resistant common hempnettle

(Weinberg et al. 2006) and common chickweed (Stellaria media) (Coupland et al. 1990). A recent

study on one field-evolved 2,4-D resistant population of wild radish [WARR20 (Walsh et al.

2009)] identified that field-evolved 2,4-D resistance is conferred by a reduced translocation

mechanism out of the treated leaf, with no differences in 2,4-D metabolism between resistant

and susceptible wild radish populations (D.Goggin unpublished).

CROSS RESISTANCE

Alarmingly, the results in this study show that along with 2,4-D resistance, there was

concomitant evolution of cross resistance to other P450-metabolisable herbicides with different

modes of action. In addition to resistance to 2,4-D amine, a level of cross resistance occurred to

the ALS-inhibiting herbicides chlorsulfuron (4-fold; LD50) (Figure 5.3, Table 5.5) and metosulam

(5-fold; LD50) (Figure 5.4, Table 5.5). The mechanistic basis providing this cross resistance was

strong with surviving plants actively growing following herbicide application (chlorsulfuron: 10-

fold (GR50); metosulam: 7-fold (GR50) (Table 5.5)). The strength of these cross-resistant

individuals is of concern, as ALS-inhibiting herbicides are widely used to selectively control wild

PAGE | 94


radish from within monocot crops such as wheat (Triticum aestivum) (Hashem et al. 2001) and

have been developed for use in transgenic ALS-resistant crops (Green 2014).

The lack of resistance to the ALS-inhibiting herbicide sulfometuron-methyl (which cannot be

metabolised by wheat (Anderson and Swain 1992)) indicates that the evolved cross resistance

in the S4 generation was unlikely to be the result of an altered target site mechanism (Yu et al.

2003). Furthermore, malathion pre-treatment implicates the involvement of P450 enzymes in

chlorsulfuron cross-resistance. Malathion is an organophosphate insecticide which effectively

synergises chlorsulfuron activity to control maize (Zea mays) (Kreuz and Fonné-Pfister 1992) and

annual ryegrass (Lolium rigidum) (Christopher et al. 1994; Preston et al. 1996) by inactivating

certain P450 enzymes (Correia and Ortiz de Montellano 2005). In this study, malathion pre-

treatment (1000 g ha–1) restored chlorsulfuron to full efficacy in the 2,4-D amine resistant

population (S4) (Figure 5.5). Since chlorsulfuron can be metabolised in wheat through the

activity of cytochrome P450 monooxygenases (Christopher et al. 1991; Gerwick et al. 1994) and

P450 enzymes CYP 81A12 and CYP 81A21 are responsible for ALS resistance in Echliochloa

phyllopogon L. (Iwakami et al. 2014), it is considered plausible that the cross resistance to

chlorsulfuron in this study involves P450 enzymes.

Previous studies have highlighted that the activity of P450 enzymes CYP71A11, CYP81B2,

CYP81C1, and CYP81C2 in 2,4-D treated cultured tobacco cells can be induced by 2,4-D (Yamada

et al. 2000). In addition, the activity of P450 enzymes CYP71R4, CYP72A, CYP81B1, CYP81A and

CYP92A in annual ryegrass also increased following 2,4-D application (Duhoux and Délye 2013).

The induction of P450 enzymes as a result of 2,4-D treatment has been demonstrated to lead to

a broad spectrum, non-heritable, protective mechanism inducing diclofop-methyl resistance in

annual ryegrass (Han et al. 2013). However, the 2,4-D induction of P450 enzymes does not

appear to be the cause of the ALS cross-resistance in this study, as all cross-resistance traits were

heritable (Table 5.4).

IMPLICATION OF AUXINIC HERBICIDE RATE ON RESISTANCE EVOLUTION

When herbicides are applied at high doses, the high level of mortality results in the selection of

rare resistance genes that impart resistance traits outside the population's natural genetic

variability for herbicide response (Darmency 1994; Jasieniuk et al. 1996; Neve et al 2014). When

used at recommended (high) rates, the evolution of 2,4-D resistance has been extremely slow,

leading to only 31 resistant species evolving following more than 70 years of commercial use

(Coupland 1994; Heap 2013). This has favoured selection of monogenic gene traits (in a few

cases involving two major genes) in the limited number of inheritance studies conducted

(reviewed by Grossman (2010) and Mithila et al (2011)). However unlike high doses, lower

PAGE | 95


herbicide rates (doses that kill most plants but many survive) selects for all possible resistance-

endowing genes, both weak and strong (Powles and Yu 2010; Délye et al. 2013), which are

predicted to occur at far higher frequencies (1 × 102) in a population (Lande 1983). As a result,

this study demonstrates that low 2,4-D rate selection can lead to the rapid evolution of 2,4-D

resistance and some cross resistance to dissimilar but metabolisable ALS herbicides. As cross

resistance can be affected by a range of biotic and abiotic stresses (Marrs 1996; Schuler and

Werck-Reichhart 2003), these cross-resistance patterns can be unpredictable (Petit et al. 2010),

confounding the effectiveness of resistance management strategies such as rotation or the use

of herbicide mixtures (Délye 2013; Lagator et al. 2013).

CONCLUSION

The evolution of non-target site, metabolism-based herbicide resistance poses a significant

threat to the management of herbicide resistance worldwide, as resistance can rapidly evolve

from small populations conferring an unpredictable array of cross-resistance traits (Yu and

Powles 2014). This report provides a stark evolutionary warning regarding the consequences of

low 2,4-D amine doses on genetically diverse, outcrossing species such as wild radish. Highly-

effective rates combined with weed control tactics that continually maximise diversity is

necessary to preserve the effectiveness of these finite herbicide resources (Powles and Yu 2010;

Duke 2012). With transgenic, synthetic auxin-resistant crops under development for the control

of the rapidly increasing glyphosate-resistance threat in the Americas (Wright et al. 2010; Green

2011), it is important that these new technologies are accompanied by appropriate stewardship

practices which continually promote the use of high, infrequently applied herbicide doses

(Gressel and Segel 1990; Neve and Powles 2005; Gressel 2009).

ACKNOWLEDGEMENTS


and the Australian Government through an Australian Post Graduate Award to M Ashworth. We

thank AHRI and plant growth facilities staff at The University of Western Australia who provided

technical assistance.

PAGE | 96


PLATE 5.1. PHOTOGRAPHIC REPRESENTATION OF THE 2,4-D AMINE DOSE–RESPONSE QUANTIFIED IN FIGURE 5.1. G0 REPRESENTS THE UNSELECTED, SUSCEPTIBLE BASE

POPULATION. GENERATIONS S1–S4 REPRESENT THE POPULATION RESPONSE TO FOUR GENERATIONS OF 2,4-D AMINE SELECTION.

PAGE | 97


PLATE 5.2. PHOTOGRAPHIC REPRESENTATION OF THE CHLORSULFURON CROSS-RESISTANCE RESPONSE TO MALATHION PRE-TREATMENT IN FIGURE 5.5. G0 REPRESENTS THE

UNSELECTED, SUSCEPTIBLE BASE POPULATION. S4 REPRESENTS THE FOURTH GENERATION OF 2,4-D SELECTION. MALATHION TREATMENT WAS APPLIED 30MIN BEFORE

CHLORSULFURON.

PAGE | 98


REFERENCES

Anderson, J. J. and R. S. Swain (1992). Metabolism of sulfometuron-methyl in wheat and its

possible role in wheat intolerance. Journal of Agricultural and Food Chemistry 40(11):

2279–2283.

Andreae, W. A. and N. E. Good (1957). Studies on 3-indoleacetic acid metabolism. IV.

Conjugation with aspartic acid and ammonia as processes in the metabolism of carboxylic

acids. Plant Physiology 32(6): 566–572.

Ashworth, M., et al. (2014). Identification of the first glyphosate resistant wild radish (Raphanus

rapanistrum L.) populations. Pest Management Science 70(9): 1432–1436.

Barnwell, P., et al. (1989). Differential auxin sensitivity and mecoprop resistance in chickweed

(Stellaria media L.) populations. Brighton Crop Protection Conference. Weeds: 427–431.

Behrens, M. R., et al. (2007). Dicamba resistance: enlarging and preserving biotechnology-based

weed management strategies. Science 316(5828): 1185–1188.


Science 50(3): 344–349.

Bristol, D. W., et al. (1977). Metabolism of 2,4-dichlorophenoxyacetic acid by wheat cell

suspension cultures. Journal of Agricultural and Food Chemistry 25(6): 1308–1314.

Burke, I. C., et al. (2009). Resistance of a prickly lettuce (Lactuca serriola) biotype to 2,4-D. Weed

Technology 23(4): 586–591.

Busi, R., et al. (2013). Evolved polygenic herbicide resistance in Lolium rigidum by low-dose

herbicide selection within standing genetic variation. Evolutionary Applications 6: 231–

242.

Busi, R., et al. (2011). Genetic control of a cytochrome P450 metabolism-based herbicide

resistance mechanism in Lolium rigidum. Heredity 106(5): 817–824.

Cheam, A. H. and G. R. Code (1995). The biology of Australian weeds: 24 Raphanus raphanistrum

L. Plant Protection Quarterly 10(1): 2–13.

Christopher, J. T., et al. (1991). Cross resistance to herbicides in annual ryegrass (Lolium rigidum).

2. Chlorsulfuron resistance involved wheat like detoxification system. Plant Physiology 95:

1036–1043.

Christopher, J. T., et al. (1994). Malathion antagonizes metabolism based chlorsulfuron

resistance in Lolium rigidum. Pesticide Biochemistry and Physiology 49: 172–182.

Collins, D. J. and J. K. Gaunt (1970). The metabolic fate of 4-chloro-2-methyl-phenoxyacetic acid

in peas. Biochemistry Journal 118(3): 54.

PAGE | 99


Correia, M. and P. Ortiz de Montellano (2005). Inhibition of cytochrome P450 enzymes.

Cytochrome P450. P. Ortiz de Montellano, Springer US: 247–322.

Coupland, D. (1994). Resistance to auxin analog herbicides. Herbicide resistance in plants:

Biology and biochemistry. S. B. Powles and J. A. M. Holtum. Boca Roton, Lewis Publishers,

CRC Press: 171–214.

Coupland, D., et al. (1990). Uptake, translocation, and metabolism of mecoprop in a sensitive

and a resistant biotype of Stellaria media. Pesticide Biochemistry and Physiology 36(1):

61–67.

Cranston, H., et al. (2001). Dicamba resistance in kochia. Weed Science. 49:164–170.

Darmency, H. (1994). Genetics and herbicide resistance in weeds and crops. Herbicide resistance

in plants: Biology and biochemistry. S. Powles and J. A. M. Holtum. Boca Raton, Lewis

Publishers, CRC Press: 263–297.

Délye, C. (2013). Unravelling the genetic bases of non-target-site-based resistance (NTSR) to

herbicides: a major challenge for weed science in the forthcoming decade. Pest


Délye, C., et al. (2013). Deciphering the evolution of herbicide resistance in weeds. Trends in

Genetics 29(11): 649–658.

Duhoux, A. and C. Délye (2013). Reference genes to study herbicide stress response in Lolium

sp: Up-regulation of P450 genes in plants resistant to acetolactate-synthase inhibitors.

PLoS ONE 8(5): e63576.



Eslami, S. V., et al. (2006). Wild radish (Raphanus raphanistrum) interference in wheat. Weed

Science 54(4): 749–756.

Feung, C. S., et al. (1973). Metabolism of 2,4-dichlorophenoxyacetic acid. V. Identification of

metabolites in soybean callus tissue cultures. Journal of Agricultural and Food Chemistry

21(4): 637–640.

Feung, C. S., et al. (1975). Metabolism of 2,4-dichlorophenoxyacetic acid. VII. Comparison of

metabolites from five species of plant callus tissue cultures. Journal of Agricultural and

Food Chemistry 23(3): 373–376.

Feung, C. S., et al. (1977). Metabolism of 2,4-dichlorophenoxyacetic acid. 11. Herbicidal

properties of amino acid conjugates. Journal of Agricultural and Food Chemistry 25(4):

898–900.

Fuerst, E. P., et al. (1996). Physiological characterization of picloram resistance in yellow

starthistle. Pesticide Biochemistry and Physiology 56(2): 149–161.

PAGE | 100


Gaines, T. A., et al. (2014). RNA-Seq transcriptome analysis to identify genes involved in

metabolism-based diclofop resistance in Lolium rigidum. The Plant Journal 78(5): 865–

876.

Gerwick, B. C., et al. (1994). Mechanisms of tolerance to triazolopyrimidine sulfonanilide

herbicides. Herbicides inhibiting branched-chain amino acid biosynthesis. J. Stetter,

Springer Berlin Heidelberg. 10: 145–160.

Green, J. M. (2011). Outlook on weed management in herbicide-resistant crops: Need for

diversification. Outlooks on Pest Management 22(3): 100–104.

Green, J. M. (2012). The benefits of herbicide-resistant crops. Pest Management Science 68(10):

1323–1331.


Science 70(9): 1351–1357.

Green, J. M., et al. (2008). New multiple-herbicide crop resistance and formulation technology

to augment the utility of glyphosate. Pest Management Science 64(4): 332–339.

Gressel, J. (2009). Evolving understanding of the evolution of herbicide resistance. Pest


Gressel, J. and L. Segel (1990). Herbicide rotations and mixes: effective stategies to delay

resistance. Managing resistance to agrochemicals. M. Green, M. LeBaron and W. Molberg.

Washington, DC, Americal Chemical Society: 430–458.

Grossmann, K. (2000). Mode of action of auxin herbicides: a new ending to a long, drawn out

story. Trends in Plant Science 5(12): 506–508.

Grossmann, K. (2010). Auxin herbicides: current status of mechanism and mode of action. Pest


Hamburg, A., et al. (2000). Comparative degradation of [14C]-2,4-dichlorophenoxyacetic acid in

wheat and potato after foliar application and in wheat, radish, lettuce, and apple after soil

application. Journal of Agricultural and Food Chemistry 49(1): 146–155.

Han, H., et al. (2013). Enhanced herbicide metabolism induced by 2,4-D in herbicide susceptible

Lolium rigidum provides protection against diclofop-methyl. Pest Management Science

69(9): 996–1000.


synthase-inhibiting herbicides in the Western Australia wheat-belt. Weed Technology

15(1): 68–74.



PAGE | 101


Heap, I. (2013). The international survey of herbicide resistant weeds. Retrieved 17 December

2013, from www.weedscience.com.



Hilton, H. W. (1957). Herbicide tolerant strain of weeds, Hawaiian Augar Planters Association

Annual Report: 69–71.

Holliday, R. J. and P. D. Putwain (1980). Evolution of herbicide resistance in Senecio vulgaris:

Variation in susceptibility to simazine between and within populations. Journal of Applied

Ecology 17(3): 779–791.

Iwakami, S., et al. (2014). Cytochrome P450 CYP81A12 and CYP81A21 are associated with

resistance to two acetolactate synthase inhibitors in Echinochloa phyllopogon. Plant

Physiology.

Jasieniuk, M., et al. (1996). The evolution and genetics of herbicide resistance in weeds. Weed

Science 44(1): 176–193.

Kercher, S. and J. K. Conner (1996). Patterns of genetic variability within and among populations

of wild radish, Raphanus raphanistrum (Brassicaceae). American Journal of Botany 83(11):

1416–1421.

Kreuz, K. and R. Fonné-Pfister (1992). Herbicide-insecticide interaction in maize: Malathion

inhibits cytochrome P450-dependent primisulfuron metabolism. Pesticide Biochemistry

and Physiology 43(3): 232–240.

Lagator, M., et al. (2013). Herbicide mixtures at high doses slow the evolution of resistance in

experimentally evolving populations of Chlamydomonas reinhardtii. New Phytologist

198(3): 938–945.

Lande, R. (1983). The response to selection on major and minor mutations affecting a metrical

trait. Heredity 50(1): 47–65.

Lutman, P. J. W. and H. S. Snow (1987). Variations in the tolerance of Galium aparine (Cleavers)

and Stellaria media (Chickweed) to Mecoprop. Brighton Crop Protection Conference.

Farnham, United Kingdom, British Crop Protection Council.: 411.



Marchant, A., et al. (1999). AUX1 regulates root gravitropism in Arabidopsis by facilitating auxin

uptake within root apical tissues. EMBO J 18(8): 2066–2073.

Marrs, K. A. (1996). The functional genomics of glutathione-s-transferases in plants. Annual

Review of Plant Physiology and Plant Molecular Biology 47(1): 127–158.

PAGE | 102




McKenzie, J. A. and P. Batterham (1994). The genetic, molecular and phenotypic consequences

of selection for insecticide resistance. Trends in Ecology & Evolution 9(5): 166–169.

Mithila, J., et al. (2011). Evolution of resistance to auxinic herbicides: Historical perspectives,

mechanisms of resistance, and implications for broadleaf weed management in

agronomic crops. Weed Science 59(4): 445–457.

Neve, P. and S. Powles (2005). High survival frequencies at low herbicide use rates in populations

of Lolium rigidum result in rapid evolution of herbicide resistance. Heredity 95(6): 485–

492.

Neve, P. and S. Powles (2005). Recurrent selection with reduced herbicide rates results in the

rapid evolution of herbicide resistance in Lolium rigidum. Theoretical and Applied

Genetics 110(6): 1154–1166.



Patzoldt, W. L., et al. (2002). Variable herbicide responses among Illinois waterhemp

(Amaranthus rudis and A. tuberculatus) populations. Crop Protection 21(9): 707–712.

Peniuk, M. G., et al. (1993). Physiological investigations into the resistance of a wild mustard

(Sinapis arvensis L.) biotype to auxinic herbicides. Weed Research 33(6): 431–440.

Petit, C., et al. (2010). Complex genetic control of non-target-site-based resistance to herbicides

inhibiting acetyl-coenzyme A carboxylase and acetolactate-synthase in Alopecurus

myosuroides. Plant Science 178(6): 501–509.



Preston, C. and C. Mallory-Smith (2001). Biochemical mechanisms, inheritance, and molecular

genetics of herbicide resistance in weeds. Herbicide resistance in world grains. S. Powles

and D. Shaner. Boca Raton, FL, CRC Press: 23–60.

Preston, C., et al. (1996). Multiple resistance to dissimilar herbicide chemistries in a biotype of

Lolium rigidum due to enhanced activity of several herbicide degrading enzymes.

Pesticide Biochemistry and Physiology 54(2): 123–134.

Price, W. J., et al. (2012). Estimation of dose-response models for discrete and continuous data



of wild radish (Rahanus raphanistrum). Australian Journal of Experimental Agriculture

21(112): 524–530.

Schuler, M. A. and D. Werck-Reichhart (2003). Functional genomics of P450's. Annual Review of

Plant Biology 54(1): 629.

PAGE | 103


Siminszky, B. (2006). Plant cytochrome P450-mediated herbicide metabolism. Phytochemistry

Reviews 5(2): 445–458.

Snow, A. and L. Campbell (2005). Can ferral radishes become weeds. Crop ferality and

volunteerism. J. Gressel. Boca Raton, FL., CRC Press: 193–207.

Sterling, T. M. and J. C. Hall (1997). Mechanism of action of natural auxins and auxinic herbicides

Herbicide activity: Toxicology, biochemistry and molecular biology. R. M. Roe, J. D. Burton

and R. J. Kuhr. Amsterdam, IOS Press: 111–141.

Streibig, J. C., et al. (1993). Dose-response curves and statistical models. Herbicide bioassays. P.


Van Eerd, L. L., et al. (2005). Physiological and biochemical characterization of quinclorac

resistance in a false cleavers (Galium spurium L.) bBiotype. Journal of Agricultural and

Food Chemistry 53(4): 1144–1151.

Walsh, M. J., et al. (2009). Combined effects of wheat competition and 2,4-D amine on phenoxy

herbicide resistant Raphanus raphanistrum populations. Weed Research 49(3): 316–325.



Weinberg, T., et al. (2006). MCPA (4-chloro-2-ethylphenoxyacetate) resistance in hemp-nettle

(Galeopsis tetrahit L.). Journal of Agricultural and Food Chemistry 54(24): 9126–9134.

Weintraub, R. L. (1953). 2,4-D mechanisms of action. Journal of Agricultural and Food Chemistry

1(3): 250–254.


Wright, T. R., et al. (2010). Robust crop resistance to broadleaf and grass herbicides provided by

aryloxyalkanoate dioxygenase transgenes. Proceedings of the National Academy of

Sciences 107(47): 20240–20245.

Yamada, T., et al. (2000). Molecular cloning of novel P450 species induced by chemical

treatments in cultures tobacco cells. Pesticide Biochemistry and Physiology 68: 11–25.

Yu, Q., et al. (2013). Enhanced rates of herbicide metabolism in low herbicide-dose selected

resistant Lolium rigidum. Plant Cell and Environment 36: 818–827.

Yu, Q. and S. Powles (2014). Metabolism-based herbicide resistance and cross-resistance in crop

weeds: A threat to herbicide sustainability and global crop production. Plant Physiology

doi: http://dx.doi.org/10.1104pp114.242750.

Yu, Q., et al. (2003). ALS gene proline (197) mutations confer ALS herbicide resistance in eight

separated wild radish (Raphanus raphanistrum) populations. Weed Science 51(6): 831–

838.

PAGE | 104


Zheng, H. -G. and J. C. Hall (2001). Understanding auxinic herbicide resistance in wild mustard:

physiological, biochemical, and molecular genetic approaches. Weed Science 49(2): 276–

281.

Zimmerlin, A. and F. Durst (1992). Aryl hydroxylation of the herbicide diclofop by a wheat

cytochrome P450 monooxygenase: Substrate specificity and physiological activity. Plant

Physiology 100(2): 874–881.

PAGE | 105


PAGE | 106

CHAPTER 6. RECURRENT SELECTION ON WILD RADISH (Raphanus raphanistrum L.) FLOWERING TIME

CHAPTER 6.

RECURRENT SELECTION ON WILD RADISH (Raphanus raphanistrum L.) FLOWERING TIME LEADS TO RAPID CHANGES IN GENOTYPE

ABSTRACT

Harvest weed seed control (HWSC) is an emerging tool that offers much needed diversity from

the sole use of herbicides to control the growing number of multiple herbicide-resistant weed

species worldwide. However the efficiency of HWSC is contingent on effective collection of weed

seeds at harvest, with the evolution of earlier flowering biotypes considered likely to lead to

HWSC evasion through early weed seed shedding prior to harvest. This study demonstrates that

an unselected population of wild radish (Raphanus raphanistrum L.) contains the phenotypic

diversity to halve its flowering time (FT) from 60 days after emergence (DAE) (FD50) in the

commencing population (G0) to 29 DAE (FD50) following five generations of early FT selection.

Three generations of late FT selection almost doubled FT to 114 DAE (FD50). As a result of

divergent FT selection, the range in FT increased to 85 days at the population level (EF5–LF1;

FD50). Even though early FT selection halved FT (FD50), the initiation of flowering only decreased

by 11 days (EF5) indicating that shifts in the initiation of flowering was unlikely to directly result

in HWSC evasion. However early FT selection consistently reduced the range over which the

early FT selected populations flowered, resulting in 77% of the EF5 generation flowering before

the initiation of flowering in the G0 population. Plant biomass and height was also greatly

affected by FT selection, with earlier flowering biotypes likely to be less competitive. However

earlier FT biotypes also flowered at a lower height and grew more prostrate, reducing the

effective cutting height for wild radish interception. This study demonstrates the adaptive ability

of wild radish populations and highlights that management precautions may be needed in order

to preserve the effectiveness of HWSC.

INTRODUCTION

In modern agriculture, herbicides are the dominant technology for controlling weeds. However,

despite their effectiveness, herbicides have not been able to eradicate weeds from fields.

Rather, evolutionary forces have acted on the genetic variation within weed populations,

resulting in survivors that possess a diverse range of herbicide resistance alleles (Powles and Yu

2010; Duke 2012). With no new herbicide modes of action being introduced (Duke 2012), the

PAGE | 107


widespread evolution of herbicide resistance in over 238 species (Heap 2014) has necessitated

the development of alternative, non-herbicidal weed control strategies that manipulate niches

in the biology of species for control (Murphy 1998; Madafiglio et al. 2006; Walsh and Powles

2007; Walsh et al. 2013). Harvest weed seed control (HWSC) is a novel, non-herbicidal technique

for weed management which aims to intercept, concentrate and destroy weed seeds retained

at grain harvest, before they re-enter the soil seedbank (Walsh et al. 2013). HWSC describes a

range of techniques developed in Australia which involve a diversity of chaff handling techniques

such as chaff carts, narrow windrow burning, tramlines and the Harrington Seed Destructor™ to

destroy weed seed thus causing a reduction in seedbank weed seed return, leading to rapid

decline in the soil seedbank of annual weed species (Walsh and Newman 2007; Walsh et al.

2012; Walsh et al. 2013).

One weed species that is biologically suited for HWSC is wild radish (Raphanus raphanistrum L.).

Wild radish is ranked in the top 180 worst weeds globally, causing management difficulties in 45

different crop species in over 65 countries (Snow 2005). In Australia, wild radish is considered

the most problematic dicot weed species (Alemseged et al. 2001; Borger et al. 2012), causing

significant yield losses in commercial dryland and horticultural crops (Code and Donaldson 1996;

Blackshaw et al. 2002). Despite a long history of herbicide application, wild radish populations

have flourished, evolving resistance to multiple herbicide groups such as inhibitors of

acetolactate synthase (ALS) (Hashem et al. 2001), phytoene desaturase (PDS) (Walsh et al.

2004), photosynthetic electron transport (Hashem et al. 2001), synthetic auxin herbicides

(Walsh et al. 2004) and inhibitors of EPSPS (5-enolpyruvylshikimate-3-phosphate synthase)

(Ashworth et al. 2014).

The capacity of wild radish to adapt is attributed to its significant genetic diversity (Kercher and

Conner 1996; Madhou et al. 2005) and its ability to produce large amounts of seed (Reeves et

al. 1981; Cousens et al. 2008). Even though wild radish produces much seed, an ecological

opportunity for control exists as seed of wild radish is non-shattering and retained within siliques

on the plants at the time of grain harvest with 95% of the siliques being retained above the

cutting height of modern harvesters at crop maturity (>150 mm above the soil surface) (Murphy

1998; Walsh et al. 2013; Walsh and Powles 2014). With increased threshing and sorting

capacities of modern harvesters, intercepted seed can now be efficiently concentrated and

diverted for subsequent impact or heat treatment, resulting in over 90% control (Walsh and

Newman 2007; Walsh and Powles 2007; Walsh et al. 2012).

HWSC techniques are now widely used in Western Australia and increasingly across the nation.

HWSC are in essence a new form of weed control being practiced across large areas. However,

PAGE | 108


as for any weed control tool, HWSC, if used persistently, is a selection pressure for any

mechanism allowing wild radish seed to survive or evade HWSC techniques. Currently wild

radish flowering time is synchronised with crops , however as the effectiveness of HWSC is

contingent on weed seeds remaining intact on plants above the cutting height at harvest (Walsh

et al. 2012), it has been suggested that HWSC may select for more prostrate (Ferris 2007) and/or

earlier seed shedding biotypes (Baker 1974), with earlier flowering considered likely to increase

the risk of fruit abscission prior to harvest (Panetsos and Baker 1967; Panetta et al. 1988). With

significant variability in flowering time (FT) evident both within and between wild radish

populations (Conner et al. 2003; Madhou et al. 2005; Ridley and Ellstrand 2010), there is a

potential for FT selection imposed by HWSC. FT selection could lead to heritable changes in the

FT in wild radish populations, as wild radish and canola (Brassica napus L.) populations have

been shown to adapt FT in response to moisture constraints (Conner et al. 2003; Franke et al.

2006; Franks et al. 2007; Ridley and Ellstrand 2010; Franks 2011).

This study investigates the potential for a wild radish population to adapt its FT in response to

recurrent FT selection. Changes in the height and biomass of this population in response to FT

selection were also assessed.


POPULATION

This selection study was conducted using the wild radish biotype WARR7 (referred hereafter as

G0), originally collected in 1999 from Yuna, Western Australia (28.34° S, 115.01° E), prior to the

development of HWSC practices (Walsh et al. 2004). Since collection, seed stocks of this

herbicide-susceptible population have been maintained and multiplied, while preventing the

ingression of external genes. This wild radish population is considered typical of wild radish

generally present throughout the WA grainbelt. Commencing with this population (G0), five

successive generations of recurrent early FT selection were conducted in October 2011 (EF1),

December 2011 (EF2), March 2012 (EF3), July 2012 (EF4) and December 2012 (EF5), with three

generations of late FT selection conducted in September 2012 (LF1), January 2013 (LF2) and April

2013 (LF3). During each selection, concurrently-grown control populations were maintained in

the same conditions except in the absence of FT selection (CE1–CE5; CL1–CL3). When possible,

the previous selection was re-grown to check the phenotypic advancement of each selection

and to produce progeny to ascertain the heritability of each FT selection (EF1C–EF4C; LF1C, LF2C)

(Figure 6.1). At all times, plants were well watered and with optimum fertilisation.

PAGE | 109


INITIAL FLOWERING DATE SELECTION PROCEDURE

The initial FT selection (EF1) was made from a commencing G0 population of 1300 plants. Wild

radish seeds (G0) greater than 2.2 mm in diameter were pre-germinated on agar solidified water

(0.6% w/v), at room temperature (20˚C), in darkness for 2 days. Seeds with >5 mm of emerged

radical were transplanted (5 seedlings per pot) to a depth of 10 mm into 260, 305 mm diameter

pots containing standard potting mixture (25% peat moss, 25% sand and 50% mulched pine

bark). Pots were maintained in the outdoor growth facility at The University of Western Australia

(Perth) during their normal growing season (June–October). All pots were watered regularly and

fertilised weekly with 2 g Scotts Cal-Mag grower plus™ soluble fertiliser (N 15% [urea 11.6%,

ammonium 1.4%, nitrate 2%), P 2.2%, K 12.4%, Ca 5%, Mg 1.8%, S 3.8%, Fe 120 mg kg–1, Mn 60

mg kg–1, Zn 15 mg kg–1, Cu 15 mg kg–1, B 20 mg kg–1, Mo 10 mg kg–1). The plants were monitored

daily to observe the first signs of anthesis. At this time of first flowering, 13 plants (1%) were

selected, based on the number of days from emergence to opening of the first flower (as marked

by the protrusion of the corolla beyond the calyx). These selected plants were isolated to ensure

cross pollination only among the selection and to prevent ingress of foreign pollen. Once all

selected plants were flowering, all earlier flowers were removed to minimise any unintended

drift in selection due to differences in female fitness among selected individuals (Conner et al.

1996; Sahli and Conner 2011). Newly-opened flowers were crossed using the 'Beestick method'

as outlined by Williams (1980), ensuring a random pattern of cross pollination (panmixia). At

maturity, the same number of siliques was harvested from each plant in the population and

bulked. Mature siliques were then processed using a modified 'grist mill' with seed progeny

representing the first selected generation (EF1; LF1) (Table 6.1). Concurrently, a random sample

of 13 plants was selected and maintained as described above to form the first generation of the

unselected control line (CE1; CL1) (Table 6.1).

Conversely to early FT, the initial FT selection for late flowering (LF1) was made from a

commencing population of 1300 plants using the previously described procedure. All plants

were monitored daily to observe anthesis. Only the last 13 plants to flower among the 1300

plants (1%) were selected. These were isolated and cross-pollinated among themselves as

previously described. At maturity, siliques were harvested and processed, with seed progeny

representing the first long flowering selected generation (LF1) (Table 6.1).

SUBSEQUENT SELECTIONS GENERAL PROCEDURE

Subsequent directional FT selections were conducted in a temperature-controlled glasshouse

(nominal setting 25°C day/15°C night) with sunlight. Large seed (>2.2 mm diameter) of the initial

selected populations (EF1 or LF1) and the initial control populations (CE1 or CL1) were

PAGE | 110


germinated on solidified water agar (0.6% w/v), at room temperature (20˚C) in darkness for 2

days. After germination, 250 pre-germinated seeds (>5 mm emerged radical) were seeded into

separate 220 mm diameter pots, watered twice daily to field capacity using an automated

irrigation system and fertilised weekly as previously described. The date of emergence was

noted for each pot. At first flowering, 20 plants were selected based on the number of days from

emergence to the opening of the first flower. These selected plants were isolated and crossed

as previously described to produce early-selected generations (EF2–EF5) and, conversely, the

late-selected generations (LF2; LF3) (Figure 6.1). Concurrently, 20 randomly-selected seeds from

each respective control line (CE1; CL1) were planted, maintained and crossed as previously

described to produce unselected early-control generations (CE2–CE5) and late-control

generations (CL2; CL3) (Figure 6.1). For the early-selected generations (EF1–EF4) and late-

selected generations (LF1–LF2), 20 randomly-selected seeds were germinated during the

following generation, maintained and crossed as previously described to produce unselected

progeny which were used to confirm the heritability of each FT selection (Figure 6.1).

PAGE | 111


FIGURE 6.1. HIERARCHY OF FLOWERING TIME SELECTION APPLIED TO THE COMMENCING WILD RADISH

POPULATION (G0).

PAGE | 112


TABLE 6.1. FLOWERING TIME ADVANCEMENT (DAYS TO FLOWERING AND CUMULATIVE GROWING DAY DEGREES (GDD)) OF EACH WILD RADISH ACCESSION DURING SELECTION.

Selected generations Unselected control generations Phenotypic advancement in first

flowering between selected and

control generations

Selected line

Selection

intensity

(1-selection

ratio)

Sowing date Selection dates

Days to

first

flowering

Cumulative

GDD (°C d)

Control

line

Days to

first

flowering

Cumulative

GDD (°C d)

Change in

days to first

flowering

Change in

cumulative GDD

(°C d)

Early

flowering

time

selection

Commencing G0 0.01 5 Sept 2011 5 Oct 2011 30 446 EF1 0.05 9 Dec 2011 30 Dec 2011 21 382 CE1 26 467 –5 –85 EF2 0.05 9 March 2012 31 March 2012 22 369 CE2 30 468 –8 –99 EF3 0.05 23 May 2012 4 July 2012 42 349 CE3 72 521 –30 –172 EF4 0.05 4 Dec 2012 24 Dec 2012 19 269 CE4 34 506 –15 –237 EF5 Final early selected CE5 Final early control – –

Late

flowering

time

selection

Commencing G0 0.01 4 June 2012 4 Sept 2012 68 (92)1 447 (936)2

LF1 0.05 8 Nov 2012 6 Jan 2013 25 (48 )1 512 (1024)2 CL1 21 482 4 30 LF2 0.05 11 Feb 2013 14 April 2013 28 (62)1 484 (2214)2 CL2 21 436 7 48

LF3 Final late selected CL3 Final late control – –

1 Bracketed data denotes days to last flowering selection (last 10% of the population) 2 Bracketed data denotes cumulative day degrees until final last flowering individuals were selected (last 10% of the population)

PAGE | 113


ANALYSIS OF SELECTION AND CROSSING LINES

The rate of FT progression was evaluated by growing the G0, selected (EF1–EF5; LF1–LF3),

control (CE1–CE5; CL1–CL5) and selected progeny (EF1C–EF4C; LF1C, LF2C) generations at the

same time within temperature-regulated glasshouse conditions during a period of stable to

gradually increasing day length (June onwards, 2013) (Plate 6.4). Large seed (>2.0 mm in

diameter) from each population was pre-germinated on agar (0.6% w/v) solidified water in

darkness for 2 days. Seventy five seeds from each population (with >5 mm emerged radical)

were seeded 10 mm deep into individual 220 mm diameter pots containing standard potting

mixture (25% peat moss, 25% sand and 50% mulched pine bark). All selected, control and

progeny generations were arranged within the glasshouse in a randomised block design (3

blocks of 25 plants per treatment) with the date of emergence noted for each pot. All pots were

watered to field capacity every 2 hours (during the day) using an automated irrigation system

(Plate 6.4), with 2 g Scotts Cal-Mag grower plus™ soluble fertiliser applied weekly, as previously

described. For the duration of the experiment, temperatures were maintained above the base

temperature for wild radish growth (4.5°C) (Mekenian and Willemsen 1975; Reeves et al. 1981)

to a nominal setting of 25°C day and 15°C night. Air temperature and photoperiod was recorded

every 15 minutes using a environment-controlling thermister and light photometer (Schneider

Electric; www.schneider-electric.com) located 1 m above the pots in the centre of the

glasshouse. For the duration of the experiment, the date of flowering and height of the first

flower were recorded daily for each pot. Aboveground biomass at the initiation of flowering was

harvested and dried at 65°C for 7 days before weighing.

DATA ANALYSIS

To compare the FT response of recurrently-selected wild radish populations, non-linear regression

analysis was performed using the DRC package in R 3.0.0 (R Development Core Team 2011;

http://www.R-project.org) (Streibig et al. 1993; Price et al. 2012). The observed population

flowering over time was fitted to a four-parameter logistic model [1]:

Y=c + [d – c/1 + exp {b (log x – log e)]} [1]

where Y denotes cumulative flowering as a percentage of the total population, e is the FD50

denoting the time or accumulated temperature to flowering response is half-way between the

upper limit, d (fixed to the total percentage of the population collected) and c is the lower

asymptotic value of Y (set to 0). The parameter b denotes the relative slope around e. FD50

parameters were compared between selected and unselected (G0) populations using the

selectivity indices (SI) function (R 3.0.0) to determine if the ratios between the FD50 values were

PAGE | 114

http://www.schneider-electric.com/


significantly different (P<0.05). A Lack-of-Fit test was also applied to each curve to ascertain the

appropriateness of the model [1] in R3.0.0.

The relative importance of accumulated thermal units (GDD) to flowering was compared

between each selection as described by Marcellos and Single (1971) using equation [2]:

GDD = ∑ (Tmax + Tmin)/2 – Tbase [2]

where Tmax is the daily maximum temperature, Tmin is the daily minimum temperature and Tbase

is the base temperature for WR (4.5 C) (Mekenian and Willemsen 1975; Reeves et al. 1981;

Norsworthy et al. 2010). The accumulated photoperiod (hours) prior to flowering was reported

as the accumulated photoperiod 7 d prior to flowering as described by Norsworthy et al (2010).

Height of the first flower and the aboveground biomass at flowering was checked for

homogeneity of variance, normality and independence of residuals as described by Onofri et al.

(2010) using a two-way analysis of variance (ANOVA) in Genstat version 6.1.0.200 (VSN

International, www.vsni.co.uk/genstat). Biomass at flowering was log transformed prior to a

two-way ANOVA. Means were calculated and separated using Tukey’s protected LSD at the 5%

level of significance. Biomass data was back transformed prior to plotting. All data was plotted

using SigmaPlot v.12 (Systat Software Inc., 2011).

RESULTS

EFFECT OF RECURRENT EARLY-FLOWERING TIME SELECTION

In one large final experiment, the G0 population and all successive selected (EF1–EF5; LF1–LF3),

unselected controls (CE1–CE5; CL1–CL3) and selected progeny generations (EF2C–EF5C; LF2C–

LF3C) were grown in a temperature-controlled glasshouse to evaluate the phenotypic

advancement of each FT selection. Following five generations of early FT selection, there was a

large change in FT. From plants selected over five successive generations for early flowering, FT

was halved at the population level (FD50) from 59 DAE (G0) to 29 DAE (EF5) (Figure 6.2, Table

6.2), reducing the thermal requirement prior to flowering (GDD) from 634°C d (G0) to 344°C d

(EF5) (Table 6.4). These reductions in FT were evident during selection with the thermal

requirement before flowering decreasing by 85, 99, 172 and 237°C d in the EF1 to EF4

generations, respectively, when compared to the concurrently grown but unselected controls

(CE1–CE4) (Table 6.4). These FT reductions were considered heritable with FT in all selected

population progenies (EF2C–EF5C) not appreciably shifting from the parental lines (EF1–EF4)

(Table 6.2). In the absence of selection, the concurrently-grown control generations (CE1–CE5)

changed FT negligibly compared to the G0 population, demonstrating that FT reductions in the

selected generations (EF1–EF5) were primarily due to the effects of FT selection (Table 6.2).

PAGE | 115


FT reductions at the population level (FD50) in the early FT selected generations were more

affected by the reduction in the distribution of FT rather than any decrease in time to the

initiation of flowering in the population. Following five generations of early FT selection, the

initiation of flowering decreased by 11 days (EF5) (Table 6.2), however the distribution of FT in

the population decreased four-fold, from an initial range of 52 days (G0) to 13 days (EF5). This

decrease in the distribution of FT resulted in 77% of the EF5 generation flowering prior to the

initiation of flowering in the unselected G0 population (Table 6.2; Figure 6.2).

As well as reductions in FT, pleiotropic effects on plant height and biomass at flowering were

evident. Five generations of early FT selection reduced mean height at flowering by 2.6-fold from

88 cm in the G0 population to 33 cm (EF5) (Figure 6.5). Concurrently, mean plant biomass at

flowering decreased 5.5-fold from 22 g (G0) to 4 g (EF5) (Figure 6.4). In the absence of selection

(CE1–CE5), there was no change in population biomass or flowering height from the G0

population.

EFFECT OF RECURRENT LATE-FLOWERING TIME SELECTION

Conversely to the progressive flowering date reductions observed through early FT selection,

late FT selection resulted in large stepwise increases in FT in both the first (LF1) and third (LF3)

generation (Table 6.3, Figure 6.3). Three generations of late FT selection increased the length of

the vegetative stage two-fold, from 59 DAE (G0; FD50) to 114 DAE (LF3; FD50) (Table 6.3). This

resulted in a 2.1-fold increase in the thermal requirement prior to flowering from 634°C d (G0)

to 1314°C d (LF3) (Table 6.5). However the initiation of flowering was only delayed by 23 days

following a single generation of selection (LF1). Additional selection did not further delay the

initiation of flowering (LF2; LF3) (Table 6.3). The distribution of flowering however progressively

increased from 52 days in the G0 population to 84 days following three generations of late FT

selection (Table 6.3, Figure 6.3). These FT increases were heritable with all late-selected

progenies (LF2C–LF3C) not appreciably shifting FT from their parental populations (LF1–LF2)

(Table 6.3). In the absence of selection, the concurrently-grown generations (CL1–CL3) were not

differentiable from the G0 population, again demonstrating that FT increases in the late FT

selected populations (LF1–LF3) were primarily due to the effects of late FT selection (Table 6.3).

Late FT selection also progressively increased plant biomass and height of the first flower.

Following late FT selection, height of the first flower progressively increased from a mean of 88

cm in the G0 population to 112 cm, 121 cm and 141 cm in the LF1, LF2 and LF3 generations,

respectively (Figure 6.5). Late FT selection also increased mean plant biomass at flowering from

21.6 g (G0) to 26 g, 35 g and 46 g per plant in the LF1, LF2 and LF3 generations, respectively

(Figure 6.4).

PAGE | 116


Emergence to flowering (days)

0 20 40 60 80 100

Cum

ulat

ive

fow

erin

g (%

)

0

20

40

60

80

100

FIGURE 6.2. THE OBSERVED POPULATION RESPONSE TO EARLY FLOWERING TIME SELECTION AGAINST THE UNSELECTED COMMENCING WILD RADISH POPULATION G0 (__•__).

EARLY FLOWERING TIME SELECTED GENERATIONS EF1 (…∆…), EF2 (…X…), EF3 (…□… ), EF4 (…◊…) AND EF5 (…○…). EACH SYMBOL REPRESENTS CUMULATIVE DATA POINTS OF

75 REPLICATE PLANTS. THE PLOTTED LINES ARE PREDICTED CUMULATIVE FLOWERING DATE CURVES FITTED TO A FOUR-PARAMETER LOGISTIC MODEL [1].

PAGE | 117


TABLE 6.2. PARAMETER ESTIMATES (DAYS TO FLOWERING) FOLLOWING EARLY FLOWERING TIME SELECTION USING THE FOUR-PARAMETER LOGISTIC MODEL [1] USED TO ESTIMATE

FD50 PARAMETERS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. SELECTION RATIOS WERE CALCULATED BASED ON FD50 VALUES FOR THE UNSELECTED

COMMENCING POPULATION (G0) AND RESPECTIVE SELECTED OR UNSELECTED CONTROL PROGENY.

Population d b e FD50 (DAE)C e FD50

selection ratio

P valueA Shift from G0

(days; FD50)

First flowering

individual

Flowering range

(days)

Base G0 100 –9.79 (0.26) 59 (0.2) efg – – 38 52

Selected EF1 100 –9.65 (0.26) 51 (0.1) d 0.8 <0.05 –8 36 35

EF2 100 –19.19 (0.57) 57 (0.1) e 0.9 <0.05 –2 35 31

EF3 100 –11.70 (0.31) 45 (0.1) c 0.7 <0.05 –14 35 19

EF4 100 –11.38 (0.38) 37 (0.1) b 0.6 <0.05 –22 31 19

EF5 100 –14.38 (0.42) 29 (0.1) a 0.5 <0.05 –30 27 13

Unselected CE1 100 –16.93 (0.57) 61 (0.1) fg 1.0 <0.05 2 42 34

CE2 100 –10.54 (0.29) 57 (0.1) e 1.0 <0.05 –2 42 32

CE3 100 –11.10 (0.30) 59 (0.1) efg 1.0 0.42 0 41 37

CE4 100 –9.96 (0.29) 58 (0.1) ef 1.0 0.37 –1 36 45

CE5 100 –12.63 (0.41) 62 (0.1) g 1.0 <0.05 3 41 41

EF1 progeny EF2C 100 –10.76 (0.29) 46 (0.1) c 0.9 <0.05 –5 B 37 27

EF2 progeny EF3C 100 –26.07 (1.05) 60 (0.1) efg 1.1 0.21 3 B 44 32

EF3 progeny EF4C 100 –11.82 (0.32) 48 (0.1) c 1.0 0.27 3 B 36 20

EF4 progeny EF5C 100 –12.09 (0.41) 38 (0.1) b 1.0 0.32 1 B 29 23

A LD50 P value comparing the difference between selected and commencing populations assessed by the SI function in the DRC package in R. v2.14.1. B Progeny shifts compared parental population (EF2C to EF1, EF3C to EF2, EF4C to EF3 and EF5C to EF4). C FD50 parameters separated using Tukey’s protected LSD at the 5% level of significance.

Lack-of-fit P value for appropriateness of the four-parameter logistic model [1] – 1.0.

PAGE | 118


Emergence to flowering (days)

0 20 40 60 80 100 120 140

Cum

ulat

ive

flow

erin

g (%

)

0

20

40

60

80

100

FIGURE 6.3. THE OBSERVED POPULATION RESPONSE TO LATE-FLOWERING TIME SELECTION AGAINST THE UNSELECTED COMMENCING WILD RADISH POPULATION G0 (__•__). LATE-

SELECTED GENERATIONS LF1 (…X…), LF2 (…∆…) AND LF3 (…□…). EACH SYMBOL REPRESENTS CUMULATIVE DATA POINTS OF 75 REPLICATE PLANTS (EXCEPT LF3 = 65 PLANTS

(MAX 86%)). THE PLOTTED LINES ARE PREDICTED CUMULATIVE FLOWERING DATE CURVES FITTED TO A FOUR-PARAMETER LOGISTIC MODEL [1].

PAGE | 119


TABLE 6.3. PARAMETER ESTIMATES FOR THE DAYS TO FLOWERING FOLLOWING LATE-FLOWERING SELECTION USING THE FOUR-PARAMETER LOGISTIC MODEL [1] USED TO ESTIMATE

FD50 PARAMETERS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. SELECTION RATIOS WERE CALCULATED BASED ON FD50 VALUES FOR THE UNSELECTED

COMMENCING POPULATION (G0) AND RESPECTIVE SELECTED OR UNSELECTED CONTROL PROGENY.

Population d b e FD50 (DAE)C e FD50

selection ratio

P

valueA

Shift from G0

(days; FD50)

First

flower

Flowering range

(days)

Base G0 100 –9.79 (0.26) 59 (0.2) a – – 38 52

Selected LF1 100 –13.56 (0.44) 81 (0.2) c 1.3 <0.05 22 61 41

LF2 100 –10.15 (0.30) 80 (0.2) c 1.3 <0.05 21 59 48

LF3 86 –9.98 (0.34) 114 (0.3) d 1.9 <0.05 55 52 84

Unselected CL1 100 –11.86 (0.34) 60 (0.1) a 1.0 0.26 1 39 42

CL2 100 –14.04 (0.44) 63 (0.1) a 1.1 <0.05 4 47 34

CL3 100 –13.98 (0.42) 60 (0.1) a 1.0 0.14 1 46 32

LF1 progeny LF2C 100 –8.91 (0.27) 74 (0.2) b 0.9 <0.05 –7 B 42 58

LF2 progeny LF3C 100 –10.48 (0.31) 83 (0.2) c 1.0 0.34 3 B 61 45

A LD50 P value comparing the difference between selected and commencing populations assessed by the SI function in the DRC package in R. v2.14.1. B Progeny shifts compared parental population (LF2C to LF1, LF3C to LF2). C FD50 parameters separated using Tukey’s protected LSD at the 5% level of significance.


PAGE | 120


TABLE 6.4. PARAMETER ESTIMATES OF THE CUMULATIVE DAY DEGREES (GDD) UNTIL FLOWERING FOLLOWING EARLY FLOWERING DATE SELECTION USING THE THREE PARAMETER

LOGISTIC MODEL [1] TO ESTIMATE FD50 GDD PARAMETERS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. SELECTION RATIOS WERE CALCULATED BASED

ON FD50 VALUES FOR THE UNSELECTED COMMENCING POPULATION (G0) AND RESPECTIVE SELECTED OR UNSELECTED CONTROL PROGENY.

Population d b e FD50 (GDD) P valueA R/S

Base G0 100 –8.25 (0.22) 634 (2.1) – –

Selected

EF1 100 –9.05 (0.25) 535 (1.6) <0.05 0.8

EF2 100 –15.94 (0.48) 598 (1.0) <0.05 0.9

EF3 100 –12.70 (0.35) 469 (1.0) <0.05 0.7

EF4 100 –12.03 (0.40) 399 (0.9) <0.05 0.6

EF5 100 –17.28 (0.52) 344 (0.6) <0.05 0.5

Unselected

CE1 100 –13.83 (0.46) 665 (1.4) <0.05 1.0

CE2 100 –9.01 (0.25) 598 (1.8) <0.05 0.9

CE3 100 –9.32 (0.25) 635 (1.9) 0.82 1.0

CE4 100 –8.51 (0.24) 617 (2.0) <0.05 1.0

CE5 100 –10.60 (0.34) 672 (1.8) <0.05 1.1

EF1 progeny EF2C 100 –11.32 (0.33) 486 (1.2) <0.05 0.9

EF2 progeny EF3C 100 –20.39 (0.80) 645 (0.9) <0.05 1.1

EF3 progeny EF4C 100 –12.14 (0.33) 494 (1.1) <0.05 1.1

EF4 progeny EF5C 100 –13.04 (0.45) 393 (0.9) <0.05 1.0

A LD50 P value comparing the difference between selected and commencing (G0) populations assessed by the SI function in the DRC package in R. v2.14.1.


PAGE | 121


TABLE 6.5. PARAMETER ESTIMATES OF THE CUMULATIVE DAY DEGREES (GDD) UNTIL FLOWERING FOLLOWING LATE-FLOWERING SELECTION USING THE FOUR-PARAMETER LOGISTIC

MODEL [1] USED TO ESTIMATE FD50 GDD PARAMETERS. STANDARD ERRORS FOR PARAMETER ESTIMATES ARE IN PARENTHESES. SELECTION RATIOS WERE CALCULATED BASED ON

FD50 VALUES FOR THE UNSELECTED COMMENCING POPULATION (G0) AND RESPECTIVE SELECTED OR UNSELECTED CONTROL PROGENY.

Population d b e FD50 (GDD) P valueA R/S

Base G0 100 –8.25 (0.22) 634.02 (2.14) – Selected LF1 100 –12.54 (0.41) 894.99 (2.08) <0.05 1.4

LF2 100 –9.34 (0.28) 879.50 (2.66) <0.05 1.4 LF3 86 –8.80 (0.30) 1314.09 (4.15) <0.05 2.1

Unselected CL1 100 –9.83 (0.27) 639.66 (1.86) 0.49 1.0 CL2 100 –11.91 (0.38) 693.83 (1.67) <0.05 1.1 CL3 100 –11.54 (0.34) 647.86 (1.58) 0.27 1.0

LF1 progeny LF2C 100 –8.06 (0.26) 777.16 (2.80) <0.05 0.9 LF2 progeny LF3C 100 –9.50 (0.29) 922.62 (2.85) 0.21 1.0

A LD50 P value comparing the difference between selected and commencing (G0) populations assessed by the SI function in the DRC package in R. v2.14.1.


PAGE | 122


FIGURE 6.4. POPULATION BIOMASS (PER PLANT) RESPONSE AT FLOWERING FOR THE COMMENCING (G0), EARLY SELECTED (EF1–EF5) AND LATE SELECTED (LF1–LF3)

GENERATIONS. EACH SYMBOL REPRESENTS THE POPULATION MEAN OF THE SELECTED POPULATION WITH THE VERTICAL AND HORIZONTAL BARS REPRESENTING 1 STANDARD ERROR

OF THE MEAN (N=75; LF3 N=64).

PAGE | 123


FIGURE 6.5. POPULATION HEIGHT RESPONSE AT FLOWERING FOR THE COMMENCING (G0), EARLY SELECTED (EF1–EF5) AND LATE SELECTED (LF1–LF3) GENERATIONS. EACH

SYMBOL REPRESENTS THE POPULATION MEAN WITH THE VERTICAL BARS REPRESENTING 1 STANDARD ERROR OF THE MEAN (N=75; LF3 N=64).

PAGE | 124


DISCUSSION

EARLY SELECTION

This study conducted over five consecutive generations clearly demonstrates that an initially

unselected wild population of 1300 wild radish individuals (G0) contains sufficient genetic

diversity to rapidly adapt its FT in response to FT selection. Five generations of early FT selection

halved FT at the population level (FD50), while three generations of late FT selection doubled FT

(FD50). This resulted in a total FT divergence of 83 days at the population level (FD50) following

five early and three late-flowering selections (EF5–LF3) from the commencing population (G0).

These rapid changes in FT obtained with wild radish concurs with previous bi-directional FT

selection studies on Chinese daikon radish (Raphanus sativus L.), where late FT selected gene

trait/s had an additive effect on environmental cues (Vahidy 1969). FT shifts in response to early

FT selection have also been observed in both field and glasshouse environments in wild mustard

(Brassica rapa L.) (Franke et al. 2006; Franks et al. 2007; Franks 2011).

ADAPTABILITY OF WILD RADISH POPULATIONS TO SELECTION

Wild radish is a highly adaptable species which has consistently thrived in a diverse range of

environments (Madhou et al. 2005; Snow 2005) and production systems (Lemerle et al. 1996;

Alemseged et al. 2001; Borger et al. 2012). This study demonstrates that an unselected wild

radish population (G0) of 1300 plants contains sufficient genetic variability to initiate flowering

at far lower thermal requirements than normally observed in field-collected populations.

Previous studies have indicated that wild radish can reach the reproductive stage in as little as

600°C d (Reeves et al. 1981; Cheam 1986; Malik et al. 2010). However in this study, five

generations of early FT selection reduced the thermal requirement prior to flowering to 344°C

d (FD50). At an individual level, a thermal requirement of just 281°C d was observed resulting in

individual wild radish plants flowering 22 DAE (EF5) with a biomass of just 2.4 g (Plate 6.1).

Identification of these early flowering individuals in the EF5 generation is significant as these

individuals had become less sensitive to photoperiod or temperature cues, flowering during a

short photoperiod of 9.5 h per day with an average daily temperature of 15°C. The significant

reductions in wild radish flowering time and thermal requirement evolved in this study highlight

the genetic potential of wild radish populations to evolve genotypes that could potentially evade

HWSC by flowering and producing seed early.

PAGE | 125


PLATE 6.1. PHOTOGRAPHIC EXAMPLES OF FT SELECTED POPULATION BIOMASS AND HEIGHT. LEFT: FIVE TIMES

EARLY FT SELECTED GENERATION EF5 (BIOMASS 2.4 G, FLOWERING HEIGHT 17 CM). RIGHT: THREE TIME LATE

FT SELECTED GENERATION LF3 PRIOR TO FLOWERING (BIOMASS 72 G, PLANT HEIGHT 146 CM)

This study however understates the full adaptive response of the commencing wild radish

population (G0) following three generations of late FT selection. This study was suspended 149

DAE (or after 1565°C d) with 14% (11 plants) from the final LF3 generation failing to initiate

flowering, with a vegetative biomass at 139% of mean biomass of the LF3 generation (Plate 6.1).

However a total FT divergence of 127 days from the first flowering individual to the suspension

of this study is another clear demonstration of the remarkable capacity wild radish populations

have to adapt FT in response to selection (Conner et al. 2003; Simard and Legere 2004; Madhou

et al. 2005).

The genetic basis for FT adaptation in this study has not been determined, however quantitative

trait loci (QTL) studies in wild mustard, oilseed canola (Brassica napus L.) and Aribidopsis thalina

L. have identified multiple loci in FT initiation (Osborn et al. 1997; Cai et al. 2008; Colautti and

Barrett 2010; Raman et al. 2013). Over 80 different genes have also been identified to affect FT

initiation in response to external and endogenous cues (Simpson and Dean 2002). The

progressive early FT shifts shown in this study are indicative of the polygenic accumulation of

minor gene traits. However the large stepwise increases in FT following late FT selection is

consistent with previous observations indicating that wild radish population responses to late-

PAGE | 126


flowering selection is indicative of the involvement major gene trait/s (Vahidy 1969; Conner

1993).

As the selection of flowering genes can be affected by environmental cues (Lempe et al. 2005),

the glasshouse-based flowering date selections in this study cannot be directly applied to wild

radish population responses in the field. However this study provides valuable insight into the

evolutionary capacity of wild radish populations to rapidly evolve in response to high intensity

selections.

Implications for harvest weed seed control resistance

While large FT shifts were observed at the population level (FD50), the time to the initiation of

flowering only decreased by 11 days following five generations of early FT selection (EF5).

Delaying HWSC by this period has resulted in negligible loss of WR siliques (Walsh and Powles

2014). However, early FT selection consistently reduced the distribution of FT in the population,

resulting in 77% of the selected population (EF5) flowering before the initiation of flowering in

the unselected population (G0). The rate of fruit abscission was not determined in this study,

however, as a consequence of early FT selection, the number of individuals in the population

carrying well-matured pods at harvest would rapidly increase. This increase in the proportion of

well-matured pods is expected to increase the probability of silique abscission prior to harvest,

especially during periods of water deficit, high temperature or wind (Taghizadeh et al. 2010;

Taghizadeh et al. 2012).

It was also observed that FT selection had pleiotropic effects on plant size (biomass and height),

concurring with similar responses in Arabidopsis thaliana (Tienderen et al. 1996). The height of

the first flower was consistently reduced by early FT selection. Any reduction in flowering height

will reduce the maximum crop cutting height for effective WR interception. Reductions in the

time to flowering also consistently reduced the biomass of early FT selected populations.

Reductions in wild radish biomass at flowering has directly affected its reproductive capacity

(Stanton 1985; Cheam 1986; Conner and Via 1993; Conner et al. 1996; Sharon et al. 1996) with

low biomass individuals producing fewer, smaller flowers that produce less seeds per silique

(Conner and Via 1993; Conner et al. 1996; Conner et al. 1996; Williams 2001). The presence of

a steep fitness gradient with early FT selection implies that FT in wild radish populations are

likely to rapidly moderate if selection is relaxed, as later flowering biotypes will have higher

reproductive capacity (Baker 1974; Conner et al. 1996). It therefore may be prudent to relax the

use of HWSC when weed seed banks are reduced to economically-acceptable levels, to allow for

the recovery of FT traits in order to maintain long-term efficacy of HWSC techniques.

PAGE | 127


It was also consistently observed that early FT selected populations lacked the ability to support

reproductive branches due to insufficient biomass accumulation prior to flowering (Plate 6.2).

As a consequence, these populations were more prostrate which is likely to limit silique

interception in wild radish, as unsupported stems and siliques are more likely to lie below the

harvest cutting height (Plate 6.3). However, these low biomass, prostrate-growing biotypes

would have reduced competitive ability against well-established field crops (Cousens et al. 2001;

Harker et al. 2003; Beckie et al. 2008). Consequently HWSC tactics should always be used in

conjunction with highly-competitive crops in order to limit the reproductive capacity of any

prostrate-growing forms.

CONCLUSION

Integrated weed management techniques such as HWSC are essential for the sustainability of

modern herbicide resistance management strategies, as herbicides can no longer be solely

guaranteed to control multiple herbicide-resistant species such as wild radish (Walsh et al. 2004;

Walsh et al. 2007). However, HWSC techniques are themselves a selection pressure and it is

considered plausible that the recurrent use of HWSC on large, genetically-diverse populations

may lead to selection of early-shedding wild radish through selection for early flowering

biotypes. As the use of HWSC continues to increase, this study highlights the possibility that the

evolution of phenotypic traits would enable plant species such as wild radish to evade HWSC.

Therefore strategies to diversify the use of HWSC should be implemented so as to disrupt

selection intensities. Populations should also be routinely monitored for increased abscission of

siliquas prior to harvest. This study is of particular importance as HWSC is likely to contribute

significantly to diversifying herbicide resistance weed management strategies worldwide

(Powles and Yu 2010; Walsh et al. 2013).

PAGE | 128


PLATE 6.2. VISUAL DIFFERENCES IN WILD RADISH REPRODUCTIVE CAPACITY (UNMEASURED) WITH EARLY FT

SELECTION. LEFT: FOUR GENERATIONS OF EARLY FT SELECTION. RIGHT: UNSELECTED COMMENCING POPULATION

G0.

PLATE 6.3. PROSTRATE GROWTH IN THE EF5 GENERATION DUE TO THE EFFECTS OF LOW BIOMASS ACCUMULATION

PRIOR TO FLOWERING.

PAGE | 129

CHAPTER 6. RECURRENT SELECTION ON WILD RADISH (R. RAPHANISTRUM L.) FLOWERING TIME

PLATE 6.4. GLASSHOUSE AND POT IRRIGATION LAYOUT DURING FINAL COMPARISON OF ALL SELECTED AND

CONTROL LINES.

PAGE | 130


ACKNOWLEDGEMENTS


and the Australian Government through an Australian Post Graduate Award to M Ashworth.

Thankyou to Dr Martin Vila-Aiub for advice and assistance in designing this study and AHRI and

plant growth facilities staff at The University of Western Australia who provided technical

assistance.

REFERENCES

Alemseged, Y., et al. (2001). A farmers' survey of weed management and herbicide problems on

winter crops in Australia. Plant Protection Quarterly 16: 21–25.

Ashworth, M. B., et al. (2014). Identification of the first glyphosate-resistant wild radish

(Raphanus raphanistrum L.) populations. Pest Management Science 70(9): 1432–1436.

Baker, H. G. (1974). The evolution of weeds. Annual Review of Ecology and Systematics 5(1): 1–

24.

Beckie, H. J., et al. (2008). Weed suppression by canola and mustard cultivars. Weed Technology

22(1): 182–185.


Science 50(3): 344–349.



Cai, C. C., et al. (2008). The genetic basis of flowering time and photoperiod sensitivity in

rapeseed Brassica napus L. Russian Journal of Genetics 44(3): 326–333.



413.

Code, G. R. and T. W. Donaldson (1996). Effect of cultivation, sowing methods and herbicides on

wild radish populations in wheat crops. Australian Journal of Experimental Agriculture

36(4): 437–442.

Colautti, R. I. and S. C. H. Barrett (2010). Natural selection and genetic constraints on flowering

phenology in an invasive plant. International Journal of Plant Sciences 171(9): 960–971.

Conner, J. (1993). Tests for major genes affecting quantitative traits in wild radish (Raphanus

raphanistrum L.). Genetica 90(1): 41–45.



704–711.

PAGE | 131


Conner, J. K., et al. (2003). Expression of additive genetic variances and covariances for wild

radish floral traits: Comparison between field and greenhouse environments. Evolution

57(3): 487–495.

Conner, J. K., et al. (1996). Measurements of natural selection on floral traits in wild radish

(Raphanus raphanistrum). I. Selection through lifetime female fitness. Evolution 50(3):

1127–1136.


(Raphanus raphanistrum). II. Selection through lifetime male and total fitness. Evolution

50(3): 1137–1146.

Cousens, R., et al. (2008). Small-scale spatial structure within patterns of seed dispersal.

Oecologia 158(3): 437–448.

Cousens, R. D., et al. (2001). Early growth and development of wild radish (Raphanus

raphanistrum L.) in relation to wheat. Australian Journal of Agricultural Research 52(7):

755–769.



Ferris, D. G. (2007). Evolutionary differentiation in Lolium L. (Ryegrass) in response to the

mediterranean type climate and changing farming systems of Western Australia. PhD,

University of Western Australia.

Franke, D., et al. (2006). A steep decline in flowering time for Brassica rapa in southern

California: Population level variation in the field and the greenhouse. International Journal

of Plant Sciences 167: 83–92.

Franks, S. J. (2011). Plasticity and evolution in drought avoidance and escape in the annual plant

Brassica rapa. New Phytologist 190(1): 249–257.

Franks, S. J., et al. (2007). Rapid evolution of flowering time by an annual plant in response to a

climate fluctuation. Proceedings of the National Academy of Sciences 104(4): 1278–1282.

Harker, K. N., et al. (2003). Seeding rate, herbicide timing and competitive hybrids contribute to

integrated weed management in canola (Brassica napus). Canadian Journal of Plant

Science 83(2): 433–440.


synthase-inhibiting herbicides in the Western Australian wheat-belt. Weed Technology

15(1): 68–74.



PAGE | 132




Kercher, S. and J. K. Conner (1996). Patterns of genetic variability within and among populations

of wild radish, Raphanus raphanistrum (Brassicaceae). American Journal of Botany 83(11):

1416–1421.

Lemerle, D., et al. (1996). Survey of weeds and diseases in cereal crops in the southern wheat

belt of New South Wales. Australian Journal of Experimental Agriculture 36(5): 545–554.

Lempe, J., et al. (2005). Diversity of flowering responses in wild Arabidopsis thaliana strains. PLoS

Genetics 1(1): 109–118.

Madafiglio, G. P., et al. (2006). Seed production of Raphanus raphanistrum following herbicide

application during reproduction and effects on wheat yield. Weed Research 46(1): 50–60.



Malik, M. S., et al. (2010). Temperature and light requirements for wild radish (Raphanus

raphanistrum) germination over a 12-month period following maturation. Weed Science

58(2): 136–140.

Marcellos, H. and W. Single (1971). Quantitative responses of wheat to photoperiod and

temperature in the field. Australian Journal of Agricultural Research 22(3): 343–357.



Murphy, C. L., D. Medd, R. (1998). The ecology of wild radish: the key to sustainable

management. Proceedings of the 9th Australian Agronomy Conference. Wagga Wagga,

Australia: 591–592.

Norsworthy, J. K., et al. (2010). Time of emergence affects survival and development of wild

radish (Raphanus raphanistrum) in South Carolina. Weed Science 58(4): 402–407.

Onofri, A., et al. (2010). Current statistical issues in weed research. Weed Research 50(1): 5–24.

Osborn, T. C., et al. (1997). Comparison of flowering time genes in Brassica rapa, B.napus and

Arabidopsis thaliana. Genetics 146(3): 1123–1129.

Panetsos, C. A. and H. G. Baker (1967). The origin of variation in Raphanus sativus (Cruciferae)

in California. Genetica 38(1): 243–274.

Panetta, F. D., et al. (1988). Survival and fecundity of wild radish (Raphanus raphanistrum L.)





PAGE | 133



Price, W. J., et al. (2012). Estimation of dose–response models for discrete and continuous data


Raman, H., et al. (2013). Genetic and physical mapping of flowering time loci in canola (Brassica

napus L.). Theoretical and Applied Genetics 126(1): 119–132.


of wild radish (Rahanus raphanistrum). Australian Journal of Experimental Agriculture

21(112): 524–530.

Ridley, C. E. and N. C. Ellstrand (2010). Rapid evolution of morphology and adaptive life history

in the invasive California wild radish (Raphanus sativus) and the implications for

management. Evolutionary Applications 3(1): 64–76.

Sahli, H. F. and J. K. Conner (2011). Testing for conflicting and nonadditive selection: Floral

adaptation to multiple pollinators through male and female fitness. Evolution 65(5):

1457–1473.

Sharon, Y. S., et al. (1996). Foliar herbivory affects floral characters and plant attractiveness to

pollinators: Implications for male and female plant fitness. The American Naturalist

147(6): 1098–1107.

Simard, M. J. and A. Legere (2004). Synchrony of flowering between canola and wild radish

(Raphanus raphanistrum). Weed Science 52(6): 905–912.

Simpson, G. and C. Dean (2002). Arabidopsis, the rosetta stone of flowering time? Science 296:

285–289.

Snow, A. and L. Campbell (2005). Can ferral radishes become weeds. Crop Ferality and

Volunteerism. J. Gressel. Boca Raton, FL., CRC Press: 193–207.

Stanton, M. L. (1985). Seed size and emergence time within a stand of wild radish (Raphanus

raphanistrum L.): The establishment of a fitness hierarchy. Oecologia 67(4): 524–531.

Streibig, J. C., et al. (1993). Dose-response curves and statistical models. Herbicide Bioassays. P.


Taghizadeh, M. S., et al. (2010). Water deficit changes the anatomy of the fruit abscission zone

in Raphanus raphanistrum (Brassicaceae). Australian Journal of Botany 57(8): 708–714.

Taghizadeh, M. S., et al. (2012). Effects of relative emergence time and water deficit on the

timing of fruit dispersal in Raphanus raphanistrum L. Crop and Pasture Science 63(10):

1018–1025.

Tienderen, P. H. V., et al. (1996). Pleiotropic effects of flowering time genes in the annual crucifer

Arabidopsis thaliana (Brassicaceae). American Journal of Botany 83(2): 169–174.

Vahidy, A. A. (1969). The genetics of flowering time in Raphanus sativus L. (Chinese Radish). PhD,

University of Hawaii.

PAGE | 134


Walsh, M. and P. Newman (2007). Burning narrow windrows for weed seed destruction. Field

Crops Research 104(1–3): 24–30.



Walsh, M., et al. (2004). Multiple herbicide resistance in wild radish (Raphanus raphanistrum)

populations in Western Australia. 4th International Weed Society Congress. Durban,

South Africa: 40.






Walsh, M. J. and S. Powles (2014). High seed retention at maturity of annual weeds infesting

crop fields highlights the potential for harvest weed seed control. Weed Technology doi:

http://dx.doi.org/10.1614/WT-D-13-00183.1.

Walsh, M. J. and S. B. Powles (2007). Management strategies for herbicide-resistant weed

populations in Australian dryland crop production systems. Weed Technology 21(2): 332–

338.



Williams, J. C., J. (2001). Sources of phenotypic variation in floral traits in wild radish (Raphanus



PAGE | 135


PAGE | 136

CHAPTER 7. GENERAL CONCLUSIONS

CHAPTER 7.

GENERAL CONCLUSIONS

Globally, an over-reliance upon herbicides to control weed species has resulted in the

widespread evolution of herbicide-resistant weed populations. As an ongoing process based on

evolutionary principles, herbicide over-reliance has resulted in the selection of resistant weed

populations in which the resistance is conferred by a diverse range of mechanisms (Powles and

Yu 2010). While many studies have detailed the biochemical and genetic basis of herbicide

resistance, few studies have focused on understanding the ecological and evolutionary

processes underlying the evolution of herbicide resistance (Neve et al. 2014). The research in

this thesis is aimed to add to this limited evolutionary knowledge by demonstrating the adaptive

capacity of wild radish populations to evolve resistance to weed control tactics that are likely to

be intensively used in global grain production in the future.

The general introduction in Chapter 1 reviewed the relevant herbicide resistance evolutionary

theory and the current status of herbicide resistance in Australian wild radish populations. This

review highlighted that the genetic diversity, competitive ability, fecundity, germination

patterns, multi-season dormancy and obligate cross-pollination of wild radish provides it with

the capacity to persist and adapt to herbicidal control measures. As a result of widespread

herbicide use, multiple herbicide resistance in Western Australian (WA) wild radish populations

is known across four separate herbicide modes of action (Walsh et al. 2001; Walsh et al. 2004;

Walsh et al. 2007; Owen 2014). However, prior to this study, glyphosate resistance in wild radish

had not been documented (Owen 2014). An over-reliance on glyphosate in Australian

agriculture is placing at risk the continued efficacy of glyphosate for weed control in many

situations. One response to herbicide resistance in Australian crop production is the use of novel

harvest weed seed control (HWSC) practices. However, HWSC practices are themselves a

selection for traits enabling survival, therefore opportunities for increased selection diversity is

required

The main research questions that arose from this review were:

1. What is the current status of glyphosate resistance among wild radish populations in

Western Australia? (Chapter 2)

PAGE | 137


2. How effective is the sole use of glyphosate at controlling WA weed species in transgenic

GR crops? (Chapter 3)

3. What are the implications of glyphosate and 2,4-D amine dose rates on the evolution of

herbicide resistance in wild radish? (Chapters 4 and 5)

4. What is the capacity of a wild radish population to adapt flowering time in response to

selection? (Chapter 6)

The following discussion includes a summary of the main findings of each research chapter and

their practical implications from the perspective of managing the evolution of herbicide

resistance in wild radish populations. This chapter concludes with future research priorities.

KEY OBSERVATIONS AND RESISTANCE MANAGEMENT IMPLICATIONS

Chapter 2: Identification of the first glyphosate-resistant wild radish populations

Key Findings:

1. Identification of the first two cases of field-evolved glyphosate resistance in wild radish

(WARR37; 38).

2. Both populations exhibit multiple resistance to four modes of action including inhibitors

of 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS), acetolactate synthase (ALS)

inhibitors, phytoene desaturase (PDS) inhibitors and synthetic auxins.

The almost universal adoption of transgenic glyphosate-resistant crops in USA (since 1996) has

resulted in over-reliance on glyphosate and insufficient diversity in weed control resulting in the

widespread evolution of glyphosate-resistant weeds (Green 2014). The evolution of glyphosate

resistance is considered one of the most challenging and topical weed control issues in

agriculture, as glyphosate is heavily relied on for weed control in transgenic GR crops and

conservation farming systems (Duke and Powles 2008; Shaner and Beckie 2014). Following more

than 40 years of glyphosate use in Western Australia, the first two cases of field-evolved

glyphosate resistance in wild radish (WARR37; 38) were identified (Chapter 2). Both of these

populations display multiple resistance to three other modes of action, namely inhibitors of ALS

(chlorsulfuron, metosulam, sulfometuron-methyl and imazamox), synthetic auxin (2,4-D) and

PDS (diflufenican). The evolution of glyphosate resistance has been linked to glyphosate over-

reliance without sufficient weed control diversity (Powles 2008). However the wild radish

populations collected in this study were from fields in the WA grainbelt where glyphosate use

was confined to pre-seeding use, heavily augmented by the use of other chemically-diverse

herbicide modes of action (D’Emden and Llewellyn 2006). This study reflects a previous report

identifying glyphosate resistance in a multi-resistant population of annual ryegrass (Lolium

PAGE | 138


rigidum L.) from the WA grainbelt (Neve et al. 2004). This study therefore highlights the capacity

of cross-pollinated, genetically-diverse wild radish populations to accumulate several resistance

gene traits and exhibit multiple resistances. Persistent use of glyphosate resulted in the addition

of glyphosate resistance to this multiple resistance profile.

Chapter 3: Surveys to determine the current frequency of glyphosate resistance in weed

species in the Western Australian grainbelt

Key Findings:

1. Eight glyphosate-resistant annual ryegrass populations were identified.

2. Glyphosate resistance was identified in one wild radish population.

3. Glyphosate remains effective on other major weed species in the WA grainbelt.

4. High wild radish and annual ryegrass densities in the first GR canola crops in Western

Australia constitutes a significant risk for selecting additional GR populations.

5. Wild radish survivorship following the sole use of glyphosate indicates that additional

weed control diversity is required to ensure effective wild radish control in GR canola

crops.

Two novel surveys (2010 and 2011) of the first commercial transgenic GR canola plantings in the

WA grainbelt were conducted to establish the glyphosate resistance status of problematic weed

species following the sole use of glyphosate. These surveys identified eight GR annual ryegrass

(Lolium rigidum Gaud.) populations and one GR wild radish population. This study also

demonstrated that glyphosate still provided excellent control of barley grass (Hordeum sp.),

brome grass (Bromus sp.) and wild oat (Avena sp.) species, capeweed (Arctotheca calendula L.)

and mallow (Malvia parviflora L.) in Western Australia. Even though glyphosate resistance in

wild radish is considered rare, the large wild radish densities treated in the first GR canola fields

surveyed are of concern and often resulted in ineffective control. The large number of wild

radish survivors following glyphosate treatment in this study warrants the need for additional

weed control diversity to be incorporated into GR canola programs to limit evolution of GR wild

radish populations (Sammons et al. 2007; Duke and Powles 2008).

Chapter 4: Effect of sublethal glyphosate doses on resistance evolution in wild radish

Key Findings:

1. This study with a glyphosate-susceptible wild radish population selected for four

generations at low glyphosate dose resulted in a modest level of glyphosate resistance.

2. Glyphosate selection resulted in weak, unexplained cross resistance to the ALS-inhibiting

herbicides imazamox and metosulam, requiring further investigation.

PAGE | 139


The lack of wild radish control following the sole use of glyphosate, as described in Chapter 3,

necessitated the study in Chapter 4 which investigated the consequences of low dose glyphosate

selection on the evolution of glyphosate resistance in wild radish. Previous studies on the

outcrossing monocot species, annual ryegrass, demonstrated that the recurrent use of low

glyphosate doses results in modest change towards glyphosate resistance (Busi and Powles

2009). This study with wild radish concurs with the Lolium rigidum L. study by Busi and Powles

(2009) with four generations of recurrent low dose selection only resulting in evolution of weak

glyphosate resistance (2.4-fold). The modest increase in glyphosate resistance in this study could

potentially lead to ineffective control of wild radish populations in a field situation (concurring

with Chapter 3). Along with modest increases in glyphosate resistance, weak cross-resistance to

the ALS-inhibiting herbicides imazamox (4.7-fold) and metosulam (3.7-fold) was also identified;

however the biochemical basis of the weak resistance in this study remains to be investigated.

This study clearly demonstrates that glyphosate must be used at rates that continually ensure

high control. Additional weed control diversity should be used to control any glyphosate-

selected survivors.

Chapter 5: Implications of 2,4-D rate on the evolution of herbicide resistance in wild radish

Key Findings:

1. The first study to confirm significant evolutionary shifts in 2,4-D amine resistance

following four generations of sublethal 2,4-D amine selection.

2. 2,4-D amine selection resulted in the evolution of cross resistance to the ALS-inhibiting

herbicides chlorsulfuron and metosulam.

3. Cross resistance to chlorsulfuron was eliminated by malathion pre-treatment indicating

that chlorsulfuron cross-resistance was likely metabolism based, mediated by increased

cytochrome P450 monooxygenase activity.

Like EPSPS inhibitors, auxinic herbicides are a robust herbicide class that have been slow to

evolve herbicide resistance despite prolonged and intensive use (Mithila et al. 2011; Heap 2014).

However, the research in Chapter 5 demonstrates that this slow rate of 2,4-D resistance

evolution is contingent on the use of effective doses that continually select outside a

population's standing genetic variability for 2,4-D response. Four generations of low dose 2,4-D

amine selection resulted in the evolution of 2,4-D amine resistance (9.6-fold LD50). Alarmingly,

as well as resistance to 2,4-D amine, cross resistance to the ALS-inhibiting herbicides metosulam

(3.5-fold) and chlorsulfuron (2.3-fold) was also identified. The biochemical basis and genetic

control of the auxinic resistance in this study remains to be investigated, however when the

inhibitor of cytochrome P450 monooxygenase, malathion (Christopher et al. 1994) was applied,

PAGE | 140


the efficacy of chlorsulfuron was restored indicating that the cross resistance to chlorsulfuron

was likely metabolism based, mediated by increased P450 activity. This study implies that 2,4-D

stewardship practices to minimise the risk of auxinic herbicide resistance evolution must

incorporate the use of highly-effective rates that consistently select outside the population's

standing genetic variability in conjunction with weed control diversity. This research is

particularly pertinent with the imminent introduction of auxinic herbicide-resistant (2,4-D and

dicamba) crops in the America’s to control the growing number of GR dicot species (Wright et

al. 2010; Egan et al. 2011; Green 2014).

Chapter 6: Evolutionary potential of wild radish to evade harvest weed seed control (HWSC)

Key Findings:

1. Five generations of early flowering time selection halved wild radish flowering time at a

population level, however this only resulted in small shifts in the initiation of flowering.

2. Three generations of late-flowering selection doubled wild radish flowering time at a

population level.

3. Five generations of early flowering time selection reduced the height of the first flower

and biomass at the initiation of flowering.

With the widespread escalation of multiple herbicide resistance among wild radish populations

in Australia (Boutsalis 2013; Owen 2014), alternative non-herbicidal weed control strategies that

exploit seed retention characteristics of wild radish have been developed. This strategy aims to

intercept and destroy wild radish seeds at harvest (HWSC) before they enter the soil seedbank.

However, the efficiency of HWSC is contingent on wild radish retaining seed until harvest and

HWSC itself is a strong selection for biotypes able to escape HWSC. It is plausible that early

flowering biotypes may increase silique abscission prior to harvest, leading to HWSC evasion

(Walsh et al. 2013; Walsh and Powles 2014). This study demonstrated that wild radish has a

remarkable capacity to adapt its flowering time (FT) when populations are preferentially

selected. Five generations of early FT selection halved wild radish flowering time from 60 days

after emergence (DAE) (FD50) in the commencing population (G0) to 29 DAE (EF5: FD50).

Conversely, three generations of late FT selection almost doubled FT to 114 DAE (FD50). Even

though wild radish exhibited the capacity to readily adapt flowering time in response to

selection, FT shifts in this study were unlikely to lead to HWSC evasion (Walsh and Powles 2014)

as the initiation of flowering was only reduced by 11 days (EF5). Early FT selection however

consistently reduced the range over which the wild radish population flowered, which is likely

to increase the risk of fruit abscission during periods of moisture stress, high temperatures or

wind (Taghizadeh et al. 2010; Taghizadeh et al. 2012). Fruit abscission was not measured in this

PAGE | 141


study, and therefore requires further research. Conversely, the reduction in the distribution of

flowering time is likely to increase the effectiveness of seed-set reduction practices such as

herbicidal crop topping (Madafiglio 2002; Walsh and Powles 2009) as a greater proportion of

flowers would likely be impacted by a single herbicide treatment.

The selection of early FT genotypes also had pleiotropic effects on wild radish biomass and

height. Following five generations of early FT selection, mean height of the first flower was more

than halved from 88 cm (G0) to 33 cm (EF5). However, as harvest cutting heights of 15cm are

routinely used to intercept other multi-resistant species in Australia such as annual ryegrass

(Walsh et al. 2013), the reductions in wild radish flowering height in this study is likely to be

inconsequential to the efficiency of wild radish interception.

The consistent reduction in biomass following early FT selection highlights a potential

mechanism for HWSC evasion. Following five generations of early FT selection, mean biomass

at the initiation of flowering was reduced 5.5-fold. It was observed that these early FT selected

genotypes had not accumulated sufficient biomass (<4 g plant–1) to support their reproductive

stems, resulting in a prostrate form. If these genotypes were not grown within a supportive crop

stand then the siliquas from these plants would likely be located under the recommended

cutting height of 15 cm, therefore evading HWSC interception. However these early maturing,

low biomass wild radish genotypes are also expected to produce less seed. The evolutionary

response of wild radish in this study therefore implies that HWSC techniques should be

conducted in the midst of a diverse range of weed control techniques, combining the use of

effective herbicides and highly-competitive crops, in order to limit the evolutionary

opportunities wild radish has to respond to these selections.

The rapid reduction in vegetative biomass at flowering with early FT selection also provides an

opportunity to manage wild radish FT traits. Wild radish plants that have low biomass at

initiation of flowering have reduced female fitness (Stanton 1985; Conner and Via 1993; Conner

et al. 1996; Sharon et al. 1996), producing fewer, smaller flowers that produce less seed per

silique (Conner and Via 1993; Conner et al. 1996; Williams 2001). Therefore, once wild radish

populations are reduced to economically-suitable levels, periodic relaxation of selection would

likely allow late FT genotypes to proliferate resulting in the recovery of mean FT back towards a

reproductive and climatic optimum. However this hypothesis requires further investigation.

FUTURE RESEARCH PRIORITIES

From the research in this thesis, a number of future research opportunities have been identified

and are summarised below:

PAGE | 142


1. The mechanistic and genetic basis of glyphosate resistance in the wild radish populations

identified in Chapter 2.

2. The mechanistic and genetic basis of 2,4-D and ALS cross-resistance identified in Chapter

5.

3. Is the mechanism imparting 2,4-D amine resistance in Chapter 5 also imparting resistance

across other synthetic auxin chemistries?

4. The implications of different wild radish flowering times in Chapter 6 on silique abscission

rates and crop competition in different field crops and environments.

5. The potential for using plant hormones to delay flowering initiation in the early FT

selected population in Chapter 6 in order to maintain efficiency of HWSC.

6. The capacity of wild radish to evolve early vigour genotypes in response to selection (in

response to highly-competitive crop environments).

FINAL REMARK

The research within this PhD study consistently highlighted the remarkable capacity of wild

radish populations to evolve under selection pressure. With this knowledge, it is possible to

promote stewardship practices to counter resistance evolution. Therefore, this thesis makes a

significant contribution to the development of sustainable weed control strategies. Following

the identification of glyphosate resistance, wild radish now joins a notorious group of

genetically-diverse weed species (Amaranthus, Ambrosia and Conyza sp.) that have evolved

resistance to five separate herbicidal modes of action (Heap 2014). Currently, wild radish is

considered the second-most problematic species in Australian grain growing regions. However

if the evolutionary understanding identified in this thesis is used to diversify weed control

strategies to minimise persistent, strong selection pressure from any one control tool then wild

radish can be successfully controlled in Australian agriculture.

REFERENCES

Boutsalis, P. (2013). Herbicide resistant weeds. from www.plantscienceconsulting.com.

Busi, R. and S. B. Powles (2009). Evolution of glyphosate resistance in a Lolium rigidum

population by glyphosate selection at sublethal doses. Heredity 103(4): 318–325.

Christopher, J. T., et al. (1994). Malathion antagonizes metabolism based chlorsulfuron

resistance in Lolium rigidum. Pesticide Biochemistry and Physiology 49: 172–182.



704–711.

PAGE | 143

http://www.plantscienceconsulting.com/



(Raphanus raphanistrum). I. Selection through lifetime female fitness. Evolution 50(3):

1127–1136.

D’Emden, F. H. and R. S. Llewellyn (2006). No-tillage adoption decisions in southern Australian

cropping and the role of weed management. Australian Journal of Experimental

Agriculture 46(4): 563–569.

Duke, S. O. and S. B. Powles (2008). Glyphosate-resistant weeds and crops. Pest Management

Science 64(4): 317–318.

Egan, J. F., et al. (2011). 2,4-Dichlorophenoxyacetic acid (2,4-D) resistant crops and the potential

for evolution of 2,4-D resistant weeds. Proceedings of the National Academy of Sciences

108(11): E37.


Science 70(9): 1351–1357.



Madafiglio, G. P. (2002). Population managment of Raphanus raphanistrum L. (wild radish) by

regulating seed production. PhD Thesis, University of Western Sydney.

Mithila, J., et al. (2011). Evolution of resistance to auxinic herbicides: Historical perspectives,

mechanisms of resistance, and implications for broadleaf weed management in

agronomic crops. Weed Science 59(4): 445–457.



Neve, P., et al. (2004). Multiple herbicide resistance in a glyphosate-resistant rigid ryegrass

(Lolium rigidum) population. Weed Science 52(6): 920–928.

Owen, M. J. (2014). Multiple herbicide-resistant Raphanus raphanistrum (wild radish) now

dominates across the Western Australian grainbelt, University of Western Australia.



Powles, S. B. (2008). Evolved glyphosate-resistant weeds around the world: lessons to be learnt.




Shaner, D. L. and H. J. Beckie (2014). The future for weed control and technology. Pest


PAGE | 144



Sharon, Y. S., et al. (1996). Foliar herbivory affects floral characters and plant attractiveness to

pollinators: Implications for male and female plant fitness. The American Naturalist

147(6): 1098–1107.

Stanton, M. L. (1985). Seed size and emergence time within a stand of wild radish (Raphanus

raphanistrum L.): The establishment of a fitness hierarchy. Oecologia 67(4): 524–531.

Taghizadeh, M. S., et al. (2010). Water deficit changes the anatomy of the fruit abscission zone

in Raphanus raphanistrum (Brassicaceae). Australian Journal of Botany 57(8): 708–714.

Taghizadeh, M. S., et al. (2012). Effects of relative emergence time and water deficit on the

timing of fruit dispersal in Raphanus raphanistrum L. Crop and Pasture Science 63(10):

1018–1025.





15(2): 199–203.




Walsh, M. J. and S. Powles (2014). High seed retention at maturity of annual weeds infesting

crop fields highlights the potential for harvest weed seed control. Weed Technology doi:

http://dx.doi.org/10.1614/WT-D-13-00183.1.

Walsh, M. J. and S. B. Powles (2009). Impact of crop-topping and swathing on the viable seed

production of wild radish (Raphanus raphanistrum). Crop & Pasture Science 60(7): 667–

674.



Williams, J. C. (2001). Sources of phenotypic variation in floral traits in wild radish (Raphanus


Wright, T. R., et al. (2010). Robust crop resistance to broadleaf and grass herbicides provided by

aryloxyalkanoate dioxygenase transgenes. Proceedings of the National Academy of

Sciences 107(47): 20240–20245.

PAGE | 145

the evolutionary response of wild radish ( raphanus ... · guidance in statistical analysis, in...

Documents