Invasion Ecology of Thrips in Relation to Seedling Cotton

Invasion ecology of thrips in relation to seedling cotton

Ranawelle Liyanage Rehan Silva Bsc. (Hons) The University of Queensland

A thesis submitted for the degree of Doctor of Philosophy at The University of Queensland in 2015 School of Biological Sciences

Abstract Cotton Gossypium hirsutum L. (Malvales: Malvaceae) crops are attacked by a variety of invertebrate pests that include the larvae of Lepidoptera, spider mites, aphids, locust, beetles and thrips. In Australia three species of thrips are regarded as cotton pests, Thrips tabaci Lindeman, Frankliniella schultzei Trybom and F. occidentalis Pergrande (Thysanoptera: Thripidae). Thrips damage can result in delayed development, the early production of fruiting bodies and increased susceptibility to disease. Despite this, the ecology of thrips in cotton producing regions is poorly understood; in particular host-plant relationships and the sources of thrips populations that invade cotton crops require further investigation. The physical presence of other arthropods or the presence of herbivory can influence the attractiveness of plants to colonizing arthropods. The chemical changes that occur in plants following arthropod herbivory are referred to as herbivore-induced plant responses and they can affect host plant attractiveness and acceptance and hence colonization patterns. Laboratory olfactometer experiments determined how the three species of thrips responded to pairwise combinations of cotton seedlings damaged by different arthropod herbivores. Responses of thrips to herbivore induced cotton plants varied dramatically. Frankliniella schultzei was more attracted to cotton seedlings damaged by foliar-feeding arthropods (cotton bollworm (Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae)) and root-feeding mealworms (Tenebrio molitor L. (Coleoptera: Tenebrionidae)) than to intact seedlings. Conversely, F. occidentalis was more attracted to intact seedlings than herbivore damaged seedlings. The responses of T. tabaci were different to those of both F. occidentalis and F. schultzei; T. tabaci did not discriminate between intact plants and those damaged by foliar or root feeding arthropods. Thus the responses of T. tabaci to herbivore damaged cotton plants were intermediate to those of F. schultzei and F. occidentalis. However, all three species of thrips were more attracted to plants damaged by two-spotted spider mites (Tetranychus urticae (Koch) (Trombidiforms: Tetranychidae)) than to intact plants. These results highlight the complexity and variability of the responses of thrips to herbivore induced cotton host plants. In the field, herbivore complexes can simultaneously attack cotton seedlings. Laboratory olfactometer experiments investigated the responses of different thrips species to cotton seedlings simultaneously damaged by more than one species of herbivore. The responses of F. occidentalis, F. schultzei and T. tabaci to plants dually infested by different combinations of herbivores were predictable based on their responses to plants damaged by single species alone. Investigation of temporal changes in the responses of thrips to herbivore damaged cotton seedlings showed that cotton seedling responses to herbivore damage were not instantaneous and that the period over which thrips behaviour modifying changes in the plants persisted was determined by both the type and extent of feeding damage. Both F. schultzei and F. occidentalis were attracted to plants ii damaged by two-spotted spider mites for much longer than they were to plants damaged by cotton bollworm larvae or bean spider mites. Similarly, F. schultzei was attracted to seedlings for much longer when higher densities of two-spotted spider mites damaged plants. The population sources of thrips that invade cotton crops and the abundance and species composition of thrips on cotton were investigated in field studies in the Namoi valley, New South Wales. Host plant utilization by thrips was quantified through a structured sampling program that estimated the presence and abundance of adult and larval thrips across various cultivated and uncultivated plant species. The floral and vegetative parts of different plant species were sampled regularly over two years, during which 65 different plant species were sampled. The seasonal composition of thrips on cotton changed from predominately T. tabaci on seedling cotton to F. schultzei and F. occidentalis on mature flowering cotton later in the season. High T. tabaci abundance on early season cotton was attributed to the high numbers of T. tabaci on the surrounding weed species, as hosts on which it was recorded were plentiful. In contrast, the patterns of F. occidentalis and F. schultzei abundance on cotton were not correlated with their abundances on the weeds but rather the emergence of cotton flowers The genetic relationships of thrips on cotton relative to those on the other host plant species sampled was investigated by analysis of mitochondrial CO1 gene sequences. No evidence was found for cryptic species of T. tabaci or F. occidentalis in the populations sampled from weeds and cotton, indicating that they were not from different breeding populations. It is inferred that T. tabaci or F. occidentalis were moving from the weeds and other crops onto cotton. Weeds clearly play an important role in the population biology of thrips and thus they must be considered in in the potential pest management strategies. Results of this study indicate that it could be possible to predict large out breaks of thrips (particularly T. tabaci) on cotton based on the relative abundance of different weed species and weather conditions that favour their growth.

iii

Declaration by author

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my research higher degree candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis.

Publications during candidature

Peer-reviewed papers

Silva R, Furlong MJ, Wilson LJ & Walter GH (2013) How predictable are the behavioral responses of insects to herbivore induced changes in plants? Responses of two congeneric thrips to induced cotton plants. PLoS ONE 8: e63611.

Silva R, Walter GH, Wilson LJ & Furlong MJ (2014) Responses of Thrips tabaci to odours of herbivore-induced cotton seedlings. Entomologia Experimentalis et Applicata151: 239-246.

Publications included in this thesis

Chapter 2

Contributor Statement of contribution Silva, R Designed experiments (70%) Wrote the paper (85%) Furlong, MJ Designed experiments (15%) Wrote and edited paper (5%) Walter, GH Designed experiments (15%) Wrote and edited paper (5%) Wilson, LJ Wrote and edited paper (5%)

Chapter 3

Silva R, Walter GH, Wilson & Furlong MJ (2014) Responses of Thrips tabaci to odours of herbivore-induced cotton seedlings. Entomologia Experimentalis et Applicata151: 239-246.

Contributions by others to the thesis Dr Michael Furlong, Professor Gimme Walter and Dr Lewis Wilson made significant contributions to the conception and design of the project, and provided advice and guidance regarding analyses throughout the research.

Statement of parts of the thesis submitted to qualify for the award of another degree

None

Acknowledgements

It has been a long, tough and challenging journey and I am so grateful that I have finally finished my PhD. However, this journey wouldn’t have been possible without the help, support and encouragement of so many people around me to whom I would like to take this opportunity to express my gratitude and appreciation.

I am grateful and would like to express my sincere thanks to my principle supervisors, Dr Michael Furlong and Professor Gimme Walter for their patience and supervision of the research. I greatly appreciate and value the help they provided me in developing logical thinking and teaching me on how to propose good questions and finding ways to get the answers.

I also would like to thank my co-supervisors Dr Lewis Wilson and Dr James Hereward. I am eternally gratefully for Dr Lewis Wilson for driving me to the field sites and helping me with the field work, I couldn’t have done it without his kind support and help.

A special thanks to Jason Callander, Gurion Ang, Alana Dane, Rini Murtiningsih and Praise Tadle for their continued friendship and consistent support throughout all the highs and lows of this thesis. I am especially thankful to both Jason Callander and Gurion Ang for their critical comments on certain chapters of this thesis.

I would also like acknowledge Desley Tree from the Queensland Department of Agriculture Fisheries and Forestry for teaching me all about thrips identification.

I also sincerely appreciate and thank the Crusier Research and Development Fund for their financial support of this research.

Finally I want to thank my family, in particular my father Mr Mohan Silva, for his keen interest and encouragement in my education and supporting throughout my studies, I also thank my sister Shamindrie Silva for her love and support. I am also eternally grateful to Matthew Trump for his support, he always cheered me up when I was down or facing tough times during the course of the PhD.

I dedicate this thesis to the memory of my late mother, Kumari Silva, she was an inspiration to me and I will never forget her encouragement and support she gave me throughout my education. vii

Keywords Thrips tabaci, Frankliniella schultzei, Frankliniella occidentalis, Gossypium hirsutum, induced resistance, induced susceptibility, olfactometer, attraction, weeds

Australian and New Zealand Standard Research Classifications (ANZSRC) ANZSRC code: 070308, Crop and Pasture Protection (Pests, Diseases and Weeds), 45% ANZSRC code: 060201, Behavioural Ecology, 45% ANZSRC code: 060411, Population, Ecological and Evolutionary Genetics, 10%

Fields of Research (FoR) Classification FoR code: 0703, Crop and Pasture production, 45% FoR code: 0602, Ecology, 45% FoR code: 0604, Genetics, 10%

viii

Fronticepiece. Centre: Frankliniella schultzei (black colour), Frankliniella occidentalis. Clockwise from top left; Rapistrum rugosum covered landscape, Brassica napus and Triticum aestivum field, Thrips tabaci, fieldwork, Ammi majus, cotton field and olfactometer setup.

Table of Contents Abstract ...... ii

Acknowledgements...... iv

Fronticepiece...... ix

Chapter 1 ...... 1

1.1 General introduction ...... 1

1.2 Thrips and cotton ...... 5

1.3 Thrips biology ...... 7

1.4 Biological attributes of invasive thrips ...... 9

1.5 Polyphagy: impact on pest status ...... 12

1.6 Thrips, non-cultivated and cultivated crops ...... 12

1.7 Host plant selection within a crop field ...... 14

1.8 Herbivore-induced plant responses ...... 14

1.9 Study species ...... 16

1.10 Structure of the thesis ...... 17

Chapter 2 ...... 24

2.1 Introduction ...... 26

2.2 Materials and methods ...... 28

2.2.1 Plants ...... 28

2.2.2 Insects ...... 29

2.2.3 Standard olfactometer tests ...... 30

2.2.4 Standard method for setting up arthropod enclosures ...... 31

2.2.5 Feeding styles, cotton responses and thrips behaviour ...... 32

2.2.6 Degree of foliage damage and thrips attraction ...... 35

2.2.7 Mite damage vs thrips damage on thrips attractancy ...... 35

2.2.8 Are thrips attracted to mites alone (no plants) or a combination of mites and damaged plants? ...... 36

2.2.9 Response of each Frankliniella species to cotton seedlings damaged by the other Frankliniella species...... 36

2.2.10 Statistical analysis...... 36

2.3 Results ...... 36

2.3.1 Feeding styles, cotton responses and thrips behaviour ...... 36

2.3.2 Degree of foliage damage and thrips attraction ...... 37

2.3.3 Mite damage vs thrips damage on thrips attractancy ...... 37

2.3.4 Are thrips attracted to mites alone (no plants) or a combination of mites and damaged plants? ...... 40

2.4 Discussion ...... 41

Chapter 3 ...... 44

3.1 Introduction ...... 46

3.2 Materials and methods ...... 47

3.2.1 Plants ...... 47

3.2.2 Arthropods ...... 47

2.2.3 Standard olfactometer tests ...... 49

2.2.4 Feeding mode, cotton responses and T. tabaci ...... 50

2.2.5 Degree of foliage damage and T. tabaci response ...... 51

2.2.6 Two spotted spider mites damage vs T. tabaci damage and the response of T. tabaci ..... 51

2.2.7 Statistical analysis...... 52

3.3 Results ...... 52

3.3.1 Feeding mode, cotton responses and T. tabaci ...... 52

3.3.2 Degree of foliage damage and T. tabaci response ...... 52

3.3.3 Two spotted spider mites damage vs T. tabaci damage and the response of T. tabaci ..... 52

3.4 Discussion ...... 54

Chapter 4 ...... 58

4.1 Introduction ...... 61

4.2 Materials and methods ...... 64

4.2.1 Literature search ...... 64 xi

4.2.2 Plants ...... 64

4.2.3 Arthropods ...... 64

4.2.4 Standard olfactometer tests ...... 67

4.2.5 Standard method for damaging cotton seedlings ...... 68

4.2.6 Responses of thrips to T. urticae and H. armigera damage ...... 68

4.2.7 Responses of thrips to Aphis gossypii and T. urticae damage ...... 69

4.2.8 Responses of thrips to Thrips and H. armigera damage ...... 70

4.2.9 Responses of thrips to Whitefly and H. armigera damage ...... 72

4.2.10 Responses of thrips to Tenebrio molitor and T. urticae damage ...... 73

4.2.11 Responses of thrips to single vs. dual herbivore damage ...... 73

4.2.12 Statistical analysis...... 73

4.3 Results ...... 74

4.3.1 Responses of thrips to T. urticae and H. armigera damage ...... 74

4.3.2 Responses of thrips to Aphis gossypii and T. urticae damage ...... 76

4.3.3 Responses of thrips to Thrips and H. armigera damage ...... 78

4.3.4 Responses of thrips to Whitefly and H. armigera damage ...... 78

4.3.5 Responses of thrips to Tenebrio molitor and T. urticae damage ...... 80

4.3.5 Responses of thrips to single vs. dual herbivore damage ...... 83

4.4 Discussion ...... 83

4.4.1 Predictability of responses-a combination of additivity and amount of damage ...... 83

4.4.2 Additivity of damage ...... 84

4.4.3 Effect of degree of damage ...... 85

4.4.4 Responses to arthropod feeding habits ...... 86

4.4.4 Thrips as atypical herbivores ...... 87

Chapter 5 ...... 88

5.1 Introduction ...... 90

5.2 Materials and methods ...... 91

5.2.1 Plants ...... 91

xii

5.2.2 Arthropods ...... 91

5.2.3 Standard olfactometer bioassays ...... 93

5.2.4 Standard method for damaging cotton seedlings ...... 94

5.2.5 Effects of post-feeding interval on responses of thrips to damaged cotton seedlings ...... 94

5.2.6 Effect of degree of damage and post-feeding interval on responses of thrips to damaged cotton seedlings ...... 95

5.2.7 Effect of dual herbivory and post-feeding interval on responses of thrips to damaged cotton seedlings ...... 95

5.2.8 Statistical analyses ...... 97

5.3 Results ...... 97

5.3.1 Effects of post-feeding interval on responses of thrips to damaged cotton seedlings ...... 97

5.3.2 Effect of degree of damage and post-feeding interval on responses of thrips to damaged cotton seedlings ...... 100

5.3.3 Effect of dual herbivory and post-feeding interval on responses of thrips to damaged cotton seedlings ...... 103

5.4 Discussion ...... 106

5.4.1 Timing of response ...... 106

5.4.2 Duration of response ...... 107

5.4.3 Comparison in responses between different sources of herbivory ...... 107

5.4.4 Comparison of responses from the same mode of herbivory ...... 108

5.4.5 Effect of damage intensity of duration of response ...... 108

5.4.6 Responses to dual herbivory ...... 109

Chapter 6 ...... 111

6.1 Introduction ...... 114

6.2 Materials and methods ...... 118

6.2.1 Sampling sites and methods ...... 118

6.2.2 Thrips and plant identification ...... 118

6.2.3 Seasonal abundance of potential host plants ...... 126

6.2.4 Effect of rainfall on abundance of thrips’ host plants ...... 126

xiii

6.2.5 Seasonal abundance of thrips on non-cultivated hosts ...... 126

6.2.6 Seasonal use of non-cultivated hosts by thrips ...... 126

6.2.7 Seasonal abundance of thrips on cotton and other crops ...... 128

6.2.8 DNA extraction and COI sequencing: Molecular analysis of Thrips tabaci and Frankliniella occidentalis ...... 128

6.2.9 Statistical analysis ...... 128

6.3 Results ...... 129

6.3.1 Seasonal abundance of potential host plants ...... 129

6.3.2 Effect of rainfall on abundance of thrips’ host plants ...... 129

6.3.3 Seasonal abundance of thrips on non-cultivated hosts ...... 140

6.3.4 Seasonal use of non-cultivated hosts by thrips ...... 152

6.3.5 Seasonal abundance of thrips on cotton and other crops ...... 164

6.3.6 DNA extraction and COI sequencing: Molecular analysis of Thrips tabaci and Frankliniella occidentalis ...... 170

6.4 Discussion ...... 176

6.4.1 Seasonal abundance of weed species and thrips ...... 176

6.4.2 Host plant use by thrips ...... 177

6.4.3 The importance of flowers in determining abundance of thrips ...... 178

6.4.4 Thrips abundance and diversity on cotton and other crops ...... 178

6.4.5 Sources from which thrips invade cotton and management tactics ...... 180

Chapter 7 ...... 182

7.1 Responses of thrips to herbivore-induced responses in cotton ...... 182

7.2 The sources of the three thrips species that invaded cotton ...... 184

7.3 Implications for management of thrips in relation to cotton ...... 185

7.4 Future research questions ...... 186

References ...... 187

xiv

List of figures

Figure 1.1 Plants are attacked and fed upon by a diversity of organisms with different methods of feeding. The most diverse taxon of attackers are the arthropods. (A) Cycad plant (Cycas revoluta) with Theclinesthes onycha (Hewitson) (Lepidoptera: Lycaenidae) larvae feeding on cycad leaves. (B) Australian crow (Euploea core Cramer (Lepidoptera: Nymphalidae)) larva feeding on Nerium oleander. (C) Hibiscus harlequin bug (Tectocoris diophthalmus Thunberg (Hemiptera: Scutelleridae)) feeding on Hibiscus leaves. (D) Aphid spp feeding on Mock orange leaves (Philadelphus coronarius). (E) Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae) larvae feeding on Pigeon pea (Cajanus cajan) flowers. (F) Mammal herbivory, red-necked wallaby (Macropus rufogriseus) feeding on grass and cattle spp feeding on lucerne leaves (Medicago sativa).

Figure 1.2 Stages of cotton maturity. In Australia cotton is planted in October and begins to mature in March. (A) Seedling cotton. (B) Mature cotton, bolls not open. (C) Chemically defoliated when 60% of bolls have opened, usually in mid- to late March.

Figure 1.3 Cotton is attacked by a variety of different arthropod herbivores, including the following. (A) Two-spotted spider mites (Tetranychus urticae Koch (Trombidiforms: Tetranychidae)). (B) Cotton aphid (Aphis gossypii Glover (Hempitera: Aphididae)). (C) Cowpea aphid (Aphis craccivora Koch ((Hempitera: Aphididae)). (D) Cottonseed bug (Oxycarenus luctuosus Montrouzier (Hemiptera: Lygaeidae)). (E) Cotton harlequin bug (Tectocoris diophthalmus Thunberg (Hemiptera: Scutelleridae)). (F) Tomato thrips (Frankliniella schultzei Trybom (Thysanoptera: Thripidae)). (G) Cotton bollworm larva (H. armigera). (H) Adult silverleaf whitefly (Bemisia tabaci Gennadius (Hemiptera: Alegrodidae)). (I) Western flower thrips (Frankliniella occidentalis Pergrande (Tysanoptera: Thripidae)).

Figure 1.4 Thrips damage to cotton seedlings. (A) Undamaged cotton seedling (B) Mild leaf distortion, usually caused by 1 to 5 thrips per plant. (C) Severe leaf distortion – leaves appear malformed and crinkle, usually caused by 5 to 20 thrips per plant. (D) Thrips feeding on undersides of leaves causing ‘silvery’ damaged areas, thrips frass is also visible.

Figure 1.5 Thysanoptera are divided into two suborders. (A) Terebrantia. (B) Tubulifera.

Figure 1.6 Female Frankliniella schultzei adult feeding on Malvaviscus arboreus pollen. (B) F. schultzei second instar larvae feeding.

Figure 1.7 Examples of thrips species. (A) Thrips imaginis (Thripidae). (B) Frankliniella occidentalis (Thripidae). (C) Pseudanaphothrips achaetus (Thripidae). (D) Thrips tabaci (Thripidae). (E) Frankliniella schultzei (Thripidae). (F) Tenothrips frici (Thripidae) on Sow thistle. (G) Tubulifera spp.

Figure 1.8 Frankliniella occidentalis life cycle. (A) First instar larvae. (B) Second instar larvae. (C) Pro-pupa. (D) Pupa. (E) Female adult.

Figure 1.9 Weeds and other cultivated crops allow population buildup of pest arthropods, ultimately allowing pest arthropods like thrips to move to cultivated crops.

Figure 1.10 A stylized schematic representation of the effects of herbivore-induced plant responses, on subsequent arthropod-plant interactions.

Figure 2.1 Cotton seedlings were raised in pots containing organic potting mix in a ventilated greenhouse. All seedlings used in experiments were at the two-leaf stage.

Figure 2.2 Thrips species investigated. (A) Frankliniella occidentalis. (B) F. schultzei.

Figure 2.3 Thrips were reared in glass jars lined with paper towels; green beans were used as a food and oviposition substrate. For both adult and larval thrips, pollen collected from Hibiscus spp flowers was provided as an additional food supplement. To prevent thrips from escaping glass jars were covered by two layers of nylon mesh.

Figure 2.4 Olfactometer set up. (A) The two side arms of the Y-tube are connected to two glass chambers containing test plants. The stem of Y-tube is connected to a vacuum pump which in turn is connected to an airflow regulator. (B) To determine whether a thrips is attracted to a particular odour source, a choice was considered made once a thrips breached the half way (4.5cm) mark up of a arm of the Y-tube and reminded there for more than 20s.

Figure 2.5 Inverted plastic cups were used to construct cages to enclose arthropods used to damage the potted cotton seedlings. Cups contained two ventilation windows covered with nylon mesh. Once arthropods were placed on seedlings, the rim of the cup was pushed onto the soil surface to prevent escape.

Figure 2.6 Arthropod herbivores used to damage cotton seedlings. (A) Adult two-spotted spider mites (Tetranychus urticae). (B) Second instar Helicoverpa armigera larvae. (C) Root-feeding mealworm larva (Tenebrio molitor).

Figure 2.7 Root herbivory. (A) Undamaged root mass. (B) Mealworms burrowed into the soil and damaged roots by feeding or dislodging them. (C) Root damage was measured by exposing seedlings to slow running water in a plastic box. The root mass was removed from the stem and all broken and dislodged pieces of root collected.

Figure 2.8 Responses of two congeneric thrips species, F. schultzei and F. occidentalis, to various forms of herbivory inflicted on cotton seedlings. NS – not significant, * P<0.05, ** P<0.01, *** P< 0.001.

Figure 2.9 Quantitative relationship between the number of individual thrips or mites damaging cotton seedlings and responses of F. schultzei and F. occidentalis. NS – not significant, * P<0.05, ** P<0.01, *** P< 0.001.

Figure 2.10 Responses of F. schultzei and F. occidentalis when exposed to volatiles emitted by mite damaged and conspecific thrips damaged cotton seedlings. NS – not significant, * P<0.05, ** P<0.01, *** P< 0.001.

Figure 2.11 Response of thrips to mites alone (no plants) or a combination of mites and damaged plants. (A) Response of F. schultzei and F. occidentalis to mites alone (B) Response of F. schultzei and F. occidentalis to mite damaged plants from which mites were removed and on plants with mites in situ. NS – not significant.

Figure 3.1 Adult female Thrips tabaci.

Figure 3.2 Arthropod herbivore species used to damage cotton seedlings. (A) Second instar Helicoverpa armigera larvae. (B) Adult two-spotted spider mites (Tetranychus urticae). (C) Root-

xvi feeding mealworm larvae (Tenebrio molitor). (D) Frankliniella occidentalis. (E) Frankliniella schultzei.

Figure 3.3 Olfactometer responses of female T. tabaci to cotton seedlings that had either been damaged mechanically, by H. armigera larvae, female adult thrips (F. occidentalis, F. schultzei or T. tabaci), female adult two spotted spider mites (T. urticae), or had their roots damaged by mealworm larvae (T. molitor) (n = 60 T. tabaci per treatment in all cases). NS – not significant, * P<0.05, ** P<0.01, *** P< 0.001, χ² test.

Figure 3.4 Effect of damaging herbivore density on the subsequent attraction of adult female T. tabaci to damaged seedlings in a Y-tube olfactometer (n = 60 T. tabaci per treatment). NS – not significant, * P<0.05, ** P<0.01, *** P< 0.001, χ² test.

Figure 3.5 Responses of adult female T. tabaci to cotton seedlings damaged by adult female two spotted spider mites (T. urticae) or conspecifics in a Y-tube olfactometer (n = 60 T. tabaci per treatment). NS – not significant, * P<0.05, ** P<0.01, *** P< 0.001, χ² test.

Figure 4.1 Studies that have investigated the effects of herbivory on the behavioural responses (ovipositional and attractancy) of herbivorous arthropods. From a total of 147 published studies abstracted in Thomson Reuters Web of Science® between 1990 and 2013, only 10 investigated the effects of multiple herbivory on plants. Of these 10 studies, only 3 investigated the effects of plant responses on herbivore behaviour. The 10 studies that looked at the effects of multiple herbivory, herbivores with different modes of feeding were used to damage plants: herbivore chewing on foliage (CF), herbivore sucking on foliage (SF) and herbivore chewing on roots (CR).

Figure 4.2 Thrips species investigated. (A) Frankliniella occidentalis. (B) Thrips tabaci. (C) F. schultzei.

Figure 4.3 Arthropods used to damage cotton seedlings. (A) Two-spotted spider mites (Tetranychus urticae). (B) Cotton aphid (Aphis gossypii). (C) Mealworm (Tenebrio molitor). (D) Cotton bollworm larvae (Helicoverpa armigera).

Figure 4.4 Whitefly species used in experiments. (A) Adult silverleaf whitefly (Bemisia tabaci B- Biotype), note the absence of hairs on the silverleaf whitefly nymphs and adults have a split between wings (B) Adult greenhouse whitefly (Trialeurodes vaporariorum), hairs present on nymphs and wings of adults overlap.

Figure 4.5 Schematic representation of the herbivore complexes used to damage cotton seedlings. Cotton seedlings were damaged by different densities of Tetranychus urticae (n = 10, 25 or 50 per plant) and a single density of Helicoverpa armigera (n = 10 per plant).

Figure 4.6 Schematic representation of the herbivore complexes used to damage cotton seedlings. Cotton seedlings were damaged by same densities (10, 25 or 50 of each species of arthropod per plant) of Tetranychus urticae and A. gossypii.

Figure 4.7 Schematic representation of the herbivore complexes used to damage cotton seedlings. Cotton seedlings were damaged by different densities of thrips species (T. tabaci, F. occidentalis or F. schultzei; n = 10, 25 or 50 per plant) and but a single density of Helicoverpa armigera (n=10 per plant).

Figure 4.8 Schematic representation of the herbivore complexes used to damage cotton seedlings. Cotton seedlings were damaged by different densities of each whitefly species (B. tabaci or T. xvii vaporariorum; n = 10, 25 or 50 per plant) and a single density of Helicoverpa armigera (n=10 per plant).

Figure 4.9 Olfactometer responses of Frankliniella occidentalis, Thrips tabaci and F. schultzei to undamaged cotton seedlings and seedlings damaged by Tetranychus urticae alone or seedlings simultaneously damaged by T. urticae and Helicoverpa armigera (n = 60 thrips per treatment in all cases). χ2 tests ns, not significant, ***P<0.001.

Figure 4.10 Olfactometer responses of Frankliniella occidentalis, Thrips tabaci and F. schultzei to undamaged cotton seedlings and seedlings damaged by Aphis gossypii alone or seedlings simultaneously damaged by A. gossypii and Tetranychus urticae (n = 60 thrips per treatment in all cases). χ² tests: ns, not significant, ***P<0.001.

Figure 4.11 Olfactometer responses of Frankliniella occidentalis, Thrips tabaci and F. schultzei to undamaged cotton seedlings and seedlings damaged by selected thrips species alone or seedlings simultaneously damaged by selected thrips and Helicoverpa armigera (n = 60 thrips per treatment in all cases). χ2 tests ns, not significant, ***P<0.001.

Figure 4.12 Olfactometer responses of F. occidentalis, T. tabaci and F. schultzei to undamaged cotton seedlings and seedlings damaged by either B. tabaci or T. vaporarium and cotton seedlings simultaneously damaged by B. tabaci and H. armigera or T. vaporariorum and H. armigera (n = 60 thrips per treatment in all cases). χ² tests: ns, not significant, ***P<0.001.

Figure 4.13 Olfactometer responses of Frankliniella occidentalis, Thrips tabaci and F. schultzei to undamaged cotton seedlings and seedlings damaged by either Tenebrio molitor alone or seedlings simultaneously damaged by T. molitor and Tetranychus urticae (n = 60 thrips per treatment in all cases). χ² tests: ns, not significant, ***P<0.001.

Figure 4.14 Olfactometer responses of Frankliniella occidentalis, Thrips tabaci and F. schultzei to cotton seedlings damaged by Tetranychus urticae and cotton seedlings simultaneously damaged by both T. urticae and Helicoverpa armigera larvae (n = 60 thrips per treatment in all cases). χ² tests: ns, not significant, ***P<0.001.

Figure 5.1 Thrips species investigated. (A) Frankliniella occidentalis. (B) Thrips tabaci. (C) Frankliniella schultzei.

Figure 5.2 Herbivore species used to damage cotton seedlings. (A) Two-spotted spider mites (T. urticae). (B) Bean spider mites (T. ludeni). (C) Second instar H. armigera larvae.

Figure 5.3 Experimental time line showing the times (h) following herbivore feeding that the responses of thrips to herbivore damaged plants were measured. (A) Arthropods were allowed to feed on plants for 5 h after which herbivores were removed and thrips responses were measured immediately in standard olfactometer experiments (B) Arthropods were allowed to feed on plants for 24 h, after which herbivores were removed and thrips responses were measured at immediately and thereafter 24 h intervals until each species of thrips showed no preference for either damaged or undamaged plants in standard olfactometer experiments.

Figure 5.4 Responses of female thrips (Frankliniella occidentalis and F. schultzei) in olfactometer assays when presented with undamaged and damaged cotton seedlings 5 h after the onset of feeding by 10 Helicoverpa armigera larvae, and also (in a separate test) daily following 24 h of feeding damage to the cotton plants by 10 larvae. Assays were conducted every day until a given species of thrips stopped discriminating between previously damaged and undamaged plants. Data are xviii presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness- of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test.

Figure 5.5 Responses of female thrips (Frankliniella occidentalis, F. schultzei and Thrips tabaci); in olfactometer assays when presented with undamaged and damaged cotton seedlings 5 h after the onset of feeding by 50 two-spotted spider mites (Tetranychus urticae), and also (in a separate test) daily following 24 h of feeding on the cotton plants by 50 mites. Assays were conducted every day until a given species of thrips stopped discriminating between previously damaged and undamaged plants. Data are presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness-of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test.

Figure 5.6 Responses of female thrips (Frankliniella occidentalis, F. schultzei and Thrips tabaci) in olfactometer assays when presented with undamaged and damaged cotton seedlings 5 h after the onset of feeding by 50 bean spider mites (Tetranychus ludeni) and in a separate test, daily following 24 h feeding by spider mites. Assays were conducted every day until a given species of thrips stopped discriminating between previously damaged and undamaged plants. Data are presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness-of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test.

Figure 5.7 Responses of female Frankliniella schultzei thrips in olfactometer assays when presented with undamaged and damaged cotton seedlings 5 h after the onset of feeding by 100 two- spotted spider mites (Tetranychus urticae) and (in a separate test) daily following 24 h feeding by two-spotted spider mites. Assays were conducted every day until F. schultzei stopped discriminating between previously damaged and undamaged plants. Data are presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness-of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test.

Figure 5.8 Responses of female thrips (Frankliniella occidentalis and F. schultzei) in olfactometer assays when presented with undamaged and damaged cotton seedlings 5h after the onset of feeding by 50 two-spotted spider mites (Tetranychus urticae) plus 10 Helicoverpa armigera larvae, and (in a separate test) daily following 24 h feeding. Assays were conducted every day until a given species of thrips stopped discriminating between previously damaged and undamaged plants. Data are presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness- of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test.

Figure 5.9 Responses of female Frankliniella schultzei in olfactometer assays when presented with undamaged and damaged cotton seedlings 5 h after the onset of feeding by both 50 two-spotted spider mites (Tetranychus urticae) and two Helicoverpa armigera larvae, and (in a separate test) daily following 24 h feeding. Assays were conducted every day until these thrips stopped discriminating between previously damaged and undamaged plants. Data are presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness-of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test.

Figure 6.1 In Australia cotton is planted in October, seedling cotton in the Namoi valley, New South Wales, 27 October 2011.

Figure 6.2 Cotton begins to mature in March (A) Mature cotton, 23rd February 2012 (B) Chemically defoliated plants when 60% of bolls have opened, 8th April 2013.

Figure 6.3 Location of the sampling areas in the Namoi valley. A total of 12 sites were chosen, each site contained either non-crop plants, crop plants or a combination of both.

xix

Figure 6.4 Pictures of individual sampling sites. Sites were different from one another in terms of the availability of host plants and topography of the land.

Figure 6.5 Each site contained variety of different flowering plants. (A) Site containing two species of flowering plants. (B) Site containing three species of flowering plants. (C) Site containing a single flowering plant species. (D) Site containing a single crop species.

Figure 6.6 At each site plants were sampled by haphazardly collecting 10 inflorescences and 10 leaves of each species, they were then placed in 100% ethanol in specimen containers. The figure shows, Sonchus oleraceus, note the presence of thrips (Tenothrips frici) on the inflorescences.

Figure 6.7 Sampling different plant species for thrips. From top left; sampling canola (Brassica napus), sampling wheat (Triticum aestivum), canola field, sampling cotton.

Figure 6.8 Individual adult and larval thrips were slide mounted in a drop of Hoyers medium and viewed under a compound light microscope .

Figure 6.9 Abundance of plant species on each sample date and at each site was rated as: 6, present over >60%; 5, present over 40-60%; 4, present over 20-40%; 3, present over 10-20%; 2, present over 5-10%; 1, present but less than 5%.

Figure 6.10 Mean plant abundance (% groundcover). (A) Mean plant abundance across all sites. Sp=Spring, Su=Summer, A=Autumn, W=Winter (B) Mean plant abundance at each site.

Figure 6.11 Images of site 1 showing how the vegetation cover changed with season. During spring (October) weed host plants abundance and diversity was high as compared to summer (November, December and February) when abundance and diversity was low.

Figure 6.12 Images of site 6 showing how the vegetation cover changed with season. During spring (October) weed host plants abundance and diversity was high as compared to summer (December and February) and winter (August) when abundance and diversity was low.

Figure 6.13 Mean number of plant species. (A) Number of plant species across all sites (B) Number of plant species at each site.

Figure 6.14 Relationship between mean plant abundance and monthly rainfall, but the rainfall was offset by 1 month rainfall, y = 2.6506x, R² = 0.3274.

Figure 6.15 Mean number of adult thrips per host plant species across all sites, Namoi valley, 2011- 2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei. Sp=Spring, Su=Summer, A=Autumn, W=Winter.

Figure 6.16 Mean number of adult thrips per host plant species at each site, Namoi valley, 2011- 2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei.

Figure 6.17 Mean number of larval thrips per host plant species across all sites, Namoi valley, 2011-2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei. Sp=Spring, Su=Summer, A=Autumn, W=Winter.

Figure 6.18 Mean number of larval thrips per host plant species at each site, Namoi valley, 2011- 2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei.

Figure 6.19 Mean number of plants species from which adult thrips were recorded across all sites, Namoi valley, 2011-2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei. Sp=Spring, Su=Summer, A=Autumn, W=Winter.

Figure 6.20 Mean number of plants species from which adult thrips were recorded at each site, Namoi valley, 2011-2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei.

Figure 6.21 Relationship between mean number of hosts on which thrips were found and number of hosts available (A) Thrips tabaci, y = 0.1358x – 1.5152, R² = 0.3692 (B) Frankliniella occidentalis, y = 0.0766x – 1.0299, R² = 0.4917 (C) F. schultzei, y = 0.0086x – 0.0638, R² = 0.1839.

Figure 6.22 Mean number of plants species from which larval thrips were recorded across all sites, Namoi valley, 2011-2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei. Sp=Spring, Su=Summer, A=Autumn, W=Winter.

Figure 6.23 Mean number of plants species from which larval thrips were recorded at each site, Namoi valley, 2011-2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei.

Figure 6.24 Mean (± SE) numbers of adult thrips collected from flowers and leaves across host plant species from which most thrips were collected. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei.

Figure 6.25 Mean (± SE) numbers of larval thrips collected from flowers and leaves across host plant species from which most thrips were collected. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei.

Figure 6.26 Mean number of adult thrips recorded on host plant species across all sites. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei. Sp=Spring, Su=Summer, A=Autumn, W=Winter.

Figure 6.27 Relative host use by Thrips tabaci adults based on the number of times they were recorded on that host and the mean numbers recorded on that host. Hosts in the upper right of each season plot are likely to be highly significant.

Figure 6.28 Relative host use by Thrips tabaci larvae based on the number of times they were recorded on that host and the mean numbers recorded on that host. Hosts in the upper right of each season plot are likely to be highly significant.

Figure 6.29 Relative host use by Frankliniella occidentalis adults based on the number of times they were recorded on that host and the mean numbers recorded on that host. Hosts in the upper right of each season plot are likely to be highly significant.

Figure 6.30 Relative host use by Frankliniella occidentalis larvae based on the number of times they were recorded on that host and the mean numbers recorded on that host. Hosts in the upper right of each season plot are likely to be highly significant.

Figure 6.31 Relative host use by Frankliniella schultzei adults based on the number of times they were recorded on that host and the mean numbers recorded on that host. Hosts in the upper right of each season plot are likely to be highly significant.

xxi

Sp Su W Sp Su A W Sp 35 Sp Su W Sp Su A W Sp 35 30 Sp Su W Sp Su A W Sp 35 30 25 30 25 20 25 20 sample 15 20 sample 15

10 sample SE) number of thrips recorded recorded per of number thrips SE) 15 10

5 recorded per of number thrips SE)

10 SE) number of thrips recorded recorded per of number thrips SE)

Mean Mean ( Figure 6.32 Relative host use by Frankliniella schultzei larvae based on the number of times they 5 0 were recorded on that host and the mean numbers recordedFrankliniella on that schultzei host. Hosts in the upper right of Oct-11 Nov-11 Mean ( Feb-12 Aug-12 Oct-12 Nov-125 Dec-12 Feb-13each seasonApr-13 plotMay-13 are likelyJul-13 to beSep-13 highlyNov-13 significant. 0 Figure 6.33 Mean numbers of adult thrips per sampleFrankliniella across alloccidentalis cottonFrankliniella crop sites schultzei, Namoi valley, Oct-11 Nov-11 Feb-12 Mean ( Aug-12sampleOct-12 eventNov-12 Dec-12 Feb-13 Apr-13 May-13 Jul-13 Sep-13 Nov-13 0 2011-2013. (A) adult thrips (B) larval thrips. Thrips tabaci tabaci Frankliniella occidentalis occidentalis F.Frankliniella schultzei Oct-11 Nov-11sampleFeb-12 eventAug-12 Oct-12 Nov-12 Dec-12 Feb-13 Apr-13 May-13 Jul-13 Sep-13 Nov-13 schultzei. Sp=Spring, Su=Summer, A=Autumn, W=Winter. Thrips tabaci Frankliniella occidentalis sample event Figure 6.34 Comparisons of thrips density on cotton flowers and cotton leaves. (A) Thrips tabaciThrips tabaci (B) Frankliniella occidentalis (C) F. schultzei.

Figure 6.35 Mean number of adult thrips recorded on crop host plants across all sites. (A) Triticum aestivum (B) Brassica x napus (C) Vicia faba (D) Cicer arietinum. Black line indicates when crops were available for sampling. Sp=Spring, Su=Summer, A=Autumn, W=Winter.

Figure 6.36 Comparisons of thrips density on Triticum aestivum flowers and leaves. (A) Thrips tabaci (B) Frankliniella occidentalis.

Figure 6.37 Four Thrips tabaci haplotypes, with most individuals represented by two haplotypes separated by a single mutation. Insects from four targeted host plants were represented by these two haplotypes.

Figure 6.38 Most Frankliniella occidentalis individuals were represented by two haplotypes, with individuals from all ten host plants represented in these two haplotypes.

Figure 6.39 Global phylogeny of Thrips tabaci showing the three main lineages samples from current study are shown in blue (continued on following page).

Figure 6.40 Global phylogeny of Frankliniella occidentalis showing the two main lineages samples from current study are shown in blue (continued on following page).

xxii

List of Tables

Table 1.1 Ovipositional and attractancy responses of various arthropod herbivores to host plants that have been “induced” in various ways. Modes of induction were; herbivore chewing on foliage (HC-F), herbivore chewing on roots (HC-R), herbivore sucking on foliage (HS-F), mechanical damage to foliage (MD-F), mechanical damage to roots (MD-R) and the application of the biochemical elicitor jasmonic acid (JA) and methyl jasmonate (MeJA).

Table 6.1 Total numbers of adult and larval Thrips tabaci, Frankliniella occidentalis and F. schultzei, collected from host plant species, summed across all 12 sample sites (Figure 6.4) in the Namoi valley, 2011-2013.

Table 6.2 Total numbers of other thrips species collected from the plants species listed in table 6.1. The only plant species listed here that does not also appear in Table 6.1 is Centaurea melitensis.

Table 6.3 Plant species from which no thrips species were collected during the surveys detailed in Tables 6.1 and 6.2.

Table 6.4 Total numbers of adult and larval Thrips tabaci, Frankliniella occidentalis and F. schultzei, collected from crop host plant species, summed across all 4 sample sites (Figure 6.4) in the Namoi valley, 2011-2013.

xxiii

List of Abbrevations used in the thesis

COI Cytochrome Oxidase Subunit One d.f. degrees of freedom CSIRO Commonwealth Scientific and Industrial Research Organisation DNA Deoxyribonucleic acid PCR Polymerase chain reaction TI Thrips imaginis PAA Pseudanaphothrips achaetus PPA Pseudanaphothrips parvus TF Tenothrips frici MA Microcephalothrips abdominalis FO Frankliniella occidentalis FS Frankliniella schultzei TT Thrips tabaci HIPV Herbivore induced plant volatiles

xxiv

Chapter 1

1.1 General introduction In natural and crop ecosystems, plants are subject to attack from a diverse suit organisms that ranges from pathogens to herbivorous mammals and includes a wide array of herbivorous arthropods (Figure 1.1) (Dicke, 2009; Johnson, 2011; Wilson, 2001). Consequences of such attacks are various forms of damage including the loss of leaf tissue, damage to apical meristems and reproductive structures and the reduced ability to photosynthesise or assimilate (Crawley, 1983). Cotton (Gossypium hirsutum L. (Malvales: Malvaceae)) (Figure 1.2) is an important field crop worldwide, it comprises 40% of the world’s fiber market and is grown in over 100 countries (ICAC, 2014). In Australia, the cotton industry generates in excess of US$ 1 billion per year in export revenue and is one of Australia’s largest rural export earners (ICAC, 2014). However in Australia and throughout the world, cotton is attacked by a large number of invertebrate pests (Oiver et al., 2001; Matthews & Tunstall, 1994; Williams et al., 2011). Such pests include the larvae of Lepidoptera, spider mites, aphids, locusts, beetles and thrips (Figure 1.3). In many cotton- growing regions around the world, thrips are common pests (Attique & Ahmad, 1990; Wilson & Bauer, 1993; Atakan et al., 1998; Deligeorgidis et al., 2002) and most crops receive some protection against thrips by seed treatment and insecticide application at planting (Lei & Wilson, 2004). Damage caused by thrips may not be the biggest economic problem but once populations of thrips establish they can build up rapidly in the earlier periods of the growing season (Wilson & Bauer, 1993). These infestations damage leaves and also render cotton plants vulnerable to other problems (Lei & Wilson, 2004). The ecology of thrips in cotton regions is poorly understood. In particular, their host-plant ranges and the sources of thrips populations that invade cotton crops have not been thoroughly investigated. With respect to cotton in Australia, Milne & Walter (2000) demonstrated the presence of high densities of Thrips tabaci Lindeman (Thysanoptera: Thripidae) on inflorescences and vegetative parts of wheat and five common weed species near cotton fields; but there is no direct evidence to demonstrate that it is these thrips that invade cotton.

(A) (C)

(B)

(D)

(E) (F)

(A)

(B)

(C)

(A) (B) (C)

(D) (E) (F) (F)

(G) (H) (I)

Figure 1.3 Cotton in Australia is attacked by a variety of different arthropod herbivores, including the following. (A) Two-spotted spider mites (Tetranychus urticae Koch (Trombidiforms: Tetranychidae)). (B) Cotton aphid (Aphis gossypii Glover (Hempitera: Aphididae)). (C) Cowpea aphid (Aphis craccivora Koch ((Hempitera: Aphididae)). (D) Cottonseed bug (Oxycarenus luctuosus Montrouzier (Hemiptera: Lygaeidae)). (E) Cotton harlequin bug (Tectocoris diophthalmus Thunberg (Hemiptera: Scutelleridae)). (F) Tomato thrips (Frankliniella schultzei Trybom (Thysanoptera: Thripidae)). (G) Cotton bollworm larva (H. armigera). (H) Adult silverleaf whitefly (Bemisia tabaci Gennadius (Hemiptera: Alegrodidae)). (I) Western flower thrips (Frankliniella occidentalis Pergrande (Tysanoptera: Thripidae)). 4

1.1 Thrips and cotton

Thrips are regarded as early season pests of cotton in Australia (Forrester & Wilson, 1988) and they decline in abundance and pest status after this stage of the crop. Adult and larval thrips that infest cotton seedlings feed on the contents of plant epidermal cells and damaged plant tissue often appears silvery (Figure 1.4) (Cook et al., 2011). The resulting damage also leads to the deformation of plant parts, causing leaves to twist and become distorted (Figure 1.4) (Cook et al., 2011). Thrips damage to cotton seedlings may also result in the reduced plant height and leaf area and delay the production of fruiting bodies (Hawkins et al., 1966; Forrester & Wilson, 1988; Williams et al., 2011). The consequences of such injury to plant growth parameters result in yield loss or delayed maturity, though often plants are able to compensate (Wilson et al., 1994). Thrips are also regarded as beneficial insects in some situations as they consume the eggs of spider mites, which are also serious pests of cotton (Wilson et al., 1996) and in doing so, they can delay mite outbreaks (Wilson et al., 1996). Management of thrips relies heavily on the use of prophylactic neonicotinoid seed treatments, despite the fact that in many cases thrips populations are unlikely to cause economic damage. This widespread use of neonicotinoid seed treatments has selected for resistance to this group of insecticides in co-incident cotton aphid populations, so more targeted use would be beneficial in reducing ongoing selection pressures.

A) B) C) D)

The economic justification for the suppression of thrips in early season cotton remains controversial. Generally, the recommended action is not to control thrips (Brecke et al., 1996) as it 5 is now accepted that early season pest damage on cotton is predominantly cosmetic with little effect on crop yield or maturity (Sadras & Wilson, 1998; Wilson et al., 2003). Furthermore, controlling thrips with insecticides for cosmetic reasons could have negative consequences as beneficial arthropods could be reduced early in the season and the thrips themselves perform some useful functions in that they are facultative predators of two spotted spider mites (Wilson et al., 1996). Despite the irrational decision making, farmers still actively control thrips populations in cotton and this study was designed to better understand the ecology of thrips in cotton. My general aim was to investigate aspects of the invasion ecology of thrips species. In particular, I was interested in why thrips invade early season cotton, what they do there and where they come from. Three species of thrips are regarded as early season pests of cotton in Australia; Thrips tabaci Lindeman, Frankliniella schultzei Trybom (Forrester and Wilson, 1988) and the western flower thrips, F. occidentalis Pergrande (Thysanoptera: Thripidae) (Williams et al., 2011). All of these thrips species are regarded as polyphagous (Milne & Walter, 1998b; Milne et al., 2007; Wimmer et al., 2008). Previous research has shown that thrips prefer to colonize cotton seedlings that also harbor mites (Wilson et al., 1996). This raises the possibility that mites either produce odours that are attractive to thrips or that they induce changes in the plants that are attractive to thrips. If different types of plant damage (e.g. different herbivores, biotic vs abiotic damage) cause different responses in plants, then different thrips species may vary in their responses to these changes. In the laboratory, I investigated the olfactory responses of these three species of thrips to various pairwise combinations of cotton seedlings damaged by various arthropod herbivores. This investigation is covered in chapters 2 to 5. The other issue investigated was that of where the three species of thrips that invade cotton come from. I quantified seasonal patterns of host plant use by thrips in the field in a study centered around a cotton growing region in the Namoi valley, New South Wales. Wilson and Bauer (1993) showed that thrips populations on seedling cotton were dominated by Thrips tabaci, while after flowering F. schultzei dominated. The cause of this change is unknown. Since that work the western flower thrips, F. occidentalis, has been recorded in cotton regions in Australia (Williams et al., 2011). Further, the sources of the different thrips species colonizing cotton crops and changes in abundance of different thrips species within and between cotton growing seasons has received only limited attention. Wheat crops are anecdotally considered to be a major source of thrips populations that move into seedling cotton (Wilson & Bauer, 1993), but this has never been confirmed, nor the species involved identified. It is also known that thrips species use a range of other crops as well as weed and grass hosts. Identifying seasonal dynamics of the different thrips species on both cultivated and non-cultivated sources is crucial to understanding when higher thrips populations might be expected in cotton and this will inform when greater value from seed treatments might be 6 expected. Surveys of non-crop weed hosts for thrips in cotton regions were conducted to identify the thrips species present, and this forms the second component of this study. To provide background for the data chapters that follow, I detail in the rest of this chapter what is known about thrips biology and introduce the three species of thrips that I studied. Finally I, present the general structure of my thesis.

1.2 Thrips biology

Thrips are generally between one to four millimeters in length, yellow, brown or black in colour and constitute a single order of insects, the Thysanoptera (Mound, 2005; Morse & Hoddle, 2006). They are further divided into two suborders: Terebrantia and Tubulifera Roughly 60% of the species are in Tubulifera (Figure 1.5), which contains a single family, the Phlaeothripidae (Reitz et al., 2011). There are 13 families in the Terebrantia but five of these are known only from fossils (Mound, 2011). The remaining eight families are: Melanthripidae, Fauriellidae, Stenurothripidae, Heterothripidae, Uzelothripidae, Merothripide, Aeolothripidae, and Thripidae (see examples of Thripidae species in Figure 1.7) (Mound & Morris, 2007; Mound, 2011). The adults and larvae of both suborders share a unique structural feature; only the left of the two mandibles is fully developed (Moritz et al., 2004). Beyond that, the differences between the two suborders include the fact that the tenth abdominal segment of adults in the Tubulifera suborder are tubular in shape with a terminal anus and genital opening at the base of the tube in both sexes, the females of Terebrantia have a saw-like ovipositor and wing structures also differ (Reitz et al., 2011). The life history also differs, in contrast to species of Terebrantia, life history of members of the Tubulifera involves three pupal stadia following the two larval instars (Reitz et al., 2011). Roughly 90% of thrips considered to be pests are from the Terebrantian family Thripidae (Moritz et al., 2004). The life cycle of terebrantian thrips are similar to one another and consists of an egg stage, two active larval feeding stages and two mobile non-feeding pro-pupal and pupal stages before the adult stage (Figure 1.8) (Reitz et al., 2011). Most terebrantian female thrips use their saw-like ovipositor to make an incision into plant tissue and then the egg is embedded into the incision (Lewis, 1973). Eggs are usually laid within non-woody aboveground plant tissue, including stems, foliage, flowers and fruits (Reitz et al., 2011). Both adult and larval terebrantian thrips exhibit highly thigmotactic behaviour where by adult and larvae prefer concealed, protected habitats on plants, such as flowers, buds, or where leaves are appressed against one another (Lewis, 1973). The non-feeding pupal stages are mainly found off plants either in the leaf litter or in the top-most layer of soil (Reitz et al., 2011).

(A) (B)

Figure 1.5 Thysanoptera are divided into two suborders. (A) Terebrantia. (B) Tubulifera.

Thrips species are known to have broad and mixed diets, a number of species are known to feed on pollen (Figure 1.6) (Kirk, 1984), and some species that are considered to be phytophagous are also predaceous (Kirk, 1997). Adult thrips and larvae feed by sucking contents from individual plant cells (Figure 1.6). Feeding is achieved by the actions of a feeding tube, comprised of the maxillae fitting together laterally with a tongue and groove system. Finger-like processes hold together the apices of the stylets with a subterminal feeding aperture (Lewis, 1997). Feeding takes place when thrips puncture leaf tissue and suck plant cell contents along the feeding tube, while a salivary pump injects saliva into the plant (Mound, 2005).

(A)

(B)

Figure 1.6 Female Frankliniella schultzei adult feeding on Malvaviscus arboreus pollen. (B) F. schultzei second instar larvae feeding.

1.3 Biological attributes of invasive thrips

Certain attributes associated with thrips make them ideal invasive species, and many species of thrips are preadapted to an invasive lifestyle (Morse & Hoddle, 2006). Thrips exhibit high fecundity, short generation times, they have a broad range of lifestyles and have the ability to readily adopt and utilize a variety of resources in a variety of habitats (Morse & Hoddle, 2006). The last characteristic is of particular importance as insect herbivores can perform well in invaded areas due to their ability to exploit alternative resources (Bernays & Chapman, 1994). One attribute that makes this possible is the fact that most pest species of thrips are highly polyphagous, adults of certain species are known to feed on a number of plant species (Reitz et al., 2011). These polyphagous species are described as phytophagous opportunists that take advantage of available hosts (Mound & Teulon, 1995). As a consequence, this feeding strategy has important implications for the understanding of thrips ecology and research has therefore tended to focus on how this functional and mechanistic relationship between thrips and host plants affects the invasive aspect of its ecology.

(A) (B)

(F)

(E)

(G)

(A) (B) (C) (D) (E)

Two active larval feeding stages Two mobile non-feeding pro-pupal and pupal stages

Figure 1.8 Frankliniella occidentalis life cycle. (A) First instar larvae. (B) Second instar larvae. (C) Pro-pupa. (D) Pupa. (E) Female adult.

1.4 Polyphagy: impact on pest status

Polyphagous insect species utilize several host plant species comprising different families to survive (Bernays & Chapman, 1994) and it has been claimed that when principal cultivated or wild host plants are not available or are unattractive, development occurs on alternative cultivated or wild host plants (Titmarsh et al., 1990). So although F. schultzei maybe specialized on only a few species of host plants, the polyphagy component of its host relationships represents a survival mechanism allowing survival through periods of stressful conditions (Milne & Walter, 2000) by utilizing alternative (albeit less suitable) host plants. Studies on other insects have also shown a similar pattern. The pentatomid bug (Nezara viridula L. (Hemiptera: Pentatomidae)) moves from one host plant species to another as the season progresses, (Kiritani et al., 1965). This host-switching behaviour is said to be an adaptive response to the limited food sources available, thereby enabling insects to persist when conditions are not suitable for their primary host plants (Velasco & Walter, 1993). This ability to use an array of host plants brings to point an important consideration; when investigating the ecology of any pest species, it is important to investigate its ecology in relation to more than just a single crop species (Milne & Walter, 1998b). Investigations of host plant use should be carried out across all the plant species with which the pest species is habitually associated, so that, a more comprehensive picture of the spatio-temporal dynamics of the species emerges (Milne & Walter, 1998b).

1.5 Thrips, non-cultivated and cultivated crops

Many pest thrips species have wide host ranges that include both non-cultivated and cultivated host plants and host plant selection is important during periods of dispersal that may be induced by plant senescence or destruction (Doederlein & Sites, 1993). Movement of individuals of any particular species from one locality to another is brought about by unfavorable conditions in the habitat of origin or when a species responds to certain environmental stimuli (Walter, 2003; Loxdale & Lushai, 1999; Dingle & Drake, 2007). For example, in a classic example Jacobsen (1945) showed that stinkbug populations that developed on Salsola iberica adjacent to wheat crops subsequently migrated to the wheat crops. Thus if weeds grow external to crops they can provide pest insects with alternative host plants and thereby serve as a source population which can then move into adjacent crops (Figure 1.9) (Norris & Kogan, 2000). Milne & Walter (2000) reported that the presence of high densities of T. tabaci on inflorescences and vegetative parts of wheat and five common weed species near cotton fields. These results are consistent with the idea that T. tabaci migrates into cotton from nearby hosts, at least in the Namoi area of New South Wales (Wilson &

Bauer, 1993), however, no direct evidence exists to demonstrate that it is these thrips that invade cotton. Thrips populations build up and develop on weeds or other cultivated crops and migrate to other often nearby crop plants (e.g onion fields) when weed hosts mature or when cultivated crops are harvested (Doederlein & Sites, 1993). North & Shelton (1986) demonstrated that T. tabaci leave cereal and forage crops when they have become senescent or are harvested and that they then colonize nearby cabbage crops. Accordingly, feeding activity caused by F. occidentalis on chrysanthemum inflorescences resulted in senescent of inflorescences causing F. occidentalis to colonize healthy plants (Rhainds & Shipp, 2003). Katayamia (2006) showed that F. occidentalis, a pest of ornamentals, vegetables and fruit crops in Japan, inhabited numerous weed species around fields and that their reproductive host changed as the months changed but most importantly results suggested that F. occidentalis ultilized numerous weed species in the spring. Other studies have also shown similar patterns whereby due to the polyphagous nature of most thrips, reservoirs are not just considered to be crops but rather also weeds growing around the fields (Pearsall & Myers, 2000; Atakan & Uygur, 2005; Aliakbarpour & Salmah, 2011; Chellemi et.al., 1994). Thus factors affecting the observed movement patterns include the phenology of other crops and weeds in the region, physiological changes in the cotton plant with age which typically result in a decline in thrips numbers and environmental factors such as climatic variables (Wilson & Bauer, 1993). All of these studies suggest that thrips populations build up and persist on nearby weed or cultivated crops and then moves to subsequent crops; however no direct evidence exists to suggest this is the case.

Cotton Weed host plants Cultivated crops

Figure 1.9 Weeds and other cultivated crops allow population buildup of pest arthropods, ultimately allowing pest arthropods like thrips to move to cultivated crops.

1.6 Host plant selection within a crop field

Once in the vicinity of the crop habitat, arthropod herbivores (i.e. thrips) must find their host plants. Choosing a host plant is governed by many factors (Bernays, 2001). For example, in phytophagous insects, the behavourial processes leading to host location and acceptance for oviposition and feeding may include orientated movement from a distance in response to colour, shape, odor and/or nutritional quality (Wise & Weinberg, 2002). These are all important plant cues used by herbivores to orient to their host plants (Wise & Weinberg, 2002; Anderson et al, 2011). Another important factor that may influence plant attractiveness and subsequent colonization of a plant by the invading arthropod is the physical presence of other arthropods (which may be conspecifics) or the presence of herbivory (Wise & Weinberg, 2002; Maddox & Root, 1990; Fritz, 1992). Furthermore, given that a single plant species can host more than one species of herbivore, there is also the possibility that the invading arthropod will encounter plants that have previously been exposed to a range of different herbivores (Maddox & Root, 1990; Fritz, 1992). This may alter the attractiveness of plants to invading arthropod herbivores and affect host-plant acceptance by the invading herbivore species (Shiojiri et al., 2002; Kessler et al., 2004; Rodriguez-Saona et al., 2005; Poelman et al., 2008).

1.7 Herbivore-induced plant responses Regardless of the quantity of damage or the species doing that damage, the chemical changes that occur in plants following arthropod herbivory are referred to as herbivore-induced plant responses, and such responses include the synthesis and emission of herbivore-induced plant volatiles (HIPVs) (Karban & Baldwin, 1997). These HIPVs can ultimately alter the interaction of the plant and its environment by modifying its interaction with arthropod herbivores that may be present after the initial bouts of herbivory (Dicke & van Loon, 2000; Hilker & Meiners, 2010; Karban, 2011; Kessler & Heil, 2011). Research into herbivore-induced responses in plants was pionerred more than 30 years ago (Haukioja & Hanhimӓki, 1985; Haukioja & Niemelӓ, 1977; Niemelӓ et al., 1979; Haukioja, 1980). More recent work has shown that the consequences of these responses include the suppression of further herbivore feeding (Anderson & Agrell, 2005; Miyazaki et al., 2013), reduced preference (i.e. oviposition or attraction) of certain herbivores for plants induced by herbivore feeding (Karban, 1989; Karban & Baldwin, 1997; Agrawal & Sherriffs, 2001) and the attraction of carnivorous natural enemies of herbivores (Figure 1.9) (Dicke et al., 1990; Turlings et al., 1990; Shiojiri et al., 2001; Vos et al., 2001; Dicke & Hilker, 2003). Ecologists have highlighted the potential role induced-plant responses play in regulating herbivore populations (Karban, 1993; Soler et al., 2009). For example, ovipositing arthropods distinguish among individual plants within a population based on prior herbivory (Wold & Marquis, 1997; Agrawal, 1999). Such studies demonstrate how herbivore-induced plant responses can influence the spatial distribution and 14 population dynamics of arthropod herbivores. However, the ecological function and significance of these chemical changes measured in plants remains to be proven and remains a topic of discussion and disagreement (Hare, 2011; Takabayashi & Dicke, 1996). Arthropod-plant interactions are complex; some studies have reported that different herbivore species induce similar plant responses whereas others have highlighted specificity in responses (Kessler & Halitschke, 2007). For example, Raphanus sativus plants damaged by Pieris rapae L. (Lepidoptera: Pieridae) larvae early in the season become resistant to a number of subsequent herbivores from different feeding guilds (Agrawal & Sherriffs, 2001). However, other studies have shown specificity, for example Plutella xylostella L. (Lepidoptera: Plutellidae) preferred to oviposit on previously damaged leaves of cabbage (Brassica olerceae) while the Mammestra brassicae L. (Lepidoptera: Noctuidae) preferred to oviposit on undamaged cabbage leaves (Poelman et al., 2010). Arthropod-plant interactions can be further complicarted by the effects of subterranean pests on foliar feeding arthropod herbivores. Studies have shown that root- and foliar-feeding herbivores can interact via the same host plant (Anderson et al., 2011; Masters & Brown, 1992; Moran & Whitham, 1990). It is now evident that such interactions have the potential to influence community structure as, the population dynamics of either above- or below-ground herbivores can be influbnced by the presence of the other (Masters et al., 1993; van der Putten et al., 2001; Soler et al., 2008; White & Andow, 2006; Andersen, 1987). Herbivore-induced plant responses should not be viewed as only having a defensive role, this is in part due to the fact that herbivore feeding can also result in induced susceptibility to additional attack and may even have no effect (Agrawal, 2000; Agrawal & Sherriffs, 2001). There is no general pattern; both types of responses (induced resistance and induced susceptibility) have been reported, so there is variability (Table 1.1) that remains to be understood. Table 1.1 provides an overview of published studies of arthropod-plant interactions showing that the effect of host plant damage on subsequent arthropod attraction and oviposition is often not predictable. To foraging herbivores, HIPVs represent complex information and it is difficult to predict whether herbivores will be attracted or repelled or even whether the induced responses are likely to have an impact on herbivores that subsequently feed on the plant (Dicke & van Loon, 2000). Accordingly, Colorado potato beetles were more attracted to herbivore damaged potato plants compared to undamaged plants (Landolt et al., 1999) even though the damage caused by Colorado potato beetles to potato plants stimulate the increased production of proteinase inhibitors that suppress Colorado potato beetle feeding (Green & Ryan, 1972). Similarly, herbivore-damaged cabbage (Brassica oleracea) and canola (B. napus) plants were the preferred choice for oviposition over undamaged plants for the diamondback moth whereas, diamondback moth oviposition was greater on 15 undamaged Chinese cabbage (B. rapa) over herbivore damaged Chinese cabbage plants (Silva & Furlong, 2012). Thus host plant responses to herbivory can be complex and measurement of one response by a given insect to herbivore induced changes in the plant is unlikely to define the insect-plant relationship. Most studies, however, only consider the effects of damage caused by a single species of herbivore (Table 1.1). In the field plants are often simultaneously fed upon and damaged by multiple herbivore species (Strauss, 1991; Vos et al., 2001). Plant responses to feeding by multiple herbivore species is poorly understood and thus warrants further investigation. In this thesis I investigated patterns of host plant use by three species of thrips in a cotton growing region to determine the sources of infestation of cotton crops by a structured and quantitative sampling regime in the field. I also investigated the olfactory responses of three species of thrips to cotton seedlings damaged by various arthropod herbivores. As background to detailing the specific research questions I addressed, I now outline the current understanding of the three thrips species I investigated.

1.8 Study species

Frankliniella schultzei Trybom (Figure 1.6) has been recorded on a wide range of plant species and although it is found frequently on flowers it is not restricted to these structures (Vierbergen & Mantel, 1991; Milne & Walter, 2000). They are also found on seedling and terminal buds of cotton (Wilson & Bauer, 1993), leaves of Achyranthes aspera L. (Ananthakrishnan et al., 1982), leaves and fruits of tomato (Davidson & Bald, 1930) and foliage of lucerne (Kirk, 1987). This species is an important pest as it damages economically important plants by feeding directly on various plant parts (Vierbergen & Mantel, 1991) or indirectly through the transmission of tomato spotted wilt virus (Mound & Houston, 1987). Thrips tabaci (Figure 1.6) is generally not regarded as a flower thrips although it does sometimes invade flowers (Kirk, 1987). This thrips species also feeds on a wide variety of plants, mainly crops and weed species (Shelton et al., 1987; Sites et al., 1992). They are important pests on various crops, mainly cabbage, onion, tobacco, tomatoes and cotton (Shelton et al., 1987; Sites et al., 1992; Wilson & Bauer, 1993). This species also damages plants through the transmission of tobacco streak ilarvirus (Klose et al., 1996). Frankliniella occidentalis Pergrande (Figure 1.6) is a major pest of agricultural and horticultural crops worldwide (Kirk, 2002) and causes direct feeding damage to leaves, fruits and flowers (Childers, 1997). It is also a major vector of tospoviruses (Wijkamp et al., 1995). Its hosts include a wide range of flowering plant species (Bryan & Smith, 1956). The taxonomy of this species is problematic as several of the species synomymised with it were originally described from only a 16 few specimens (Bryan & Smith, 1956) so their status remains unclear. Three colour forms are present in this species, ranging from pale, intermediate to dark and their relative abundance is dependent on the season. Population studies in the San Francisco bay area have revealed that some seasons are more favorable than others (Bryan & Smith, 1956). This high level of variation in its biology has led some to suggest that the F. occidentalis as currently recognized may in fact represent a complex of cryptic species. Rugman-Jones et al. (2010) conducted a study whereby nuclear and mitochondrial DNA sequences were compared between F. occidentalis populations across California. Using the nuclear-mitochondrial barcoding technique, these authors revealed the existence of two distinctive entities from the D2 domain of 28S and cytochrome c oxidase I gene sequences, thus suggesting the existence of two cryptic species of F. occidentalisin California (Rugman-Jones et al., 2010). Little quantitative information is available on the host use by any of the three species in a cotton growing region, nor is there any information available on the behavioural responses of the three species to cotton damaged by co-occuring pests. I aim to redress this situation through the research presented in my thesis, the structure of which is outlined next.

1.9 Structure of the thesis

The thesis contains five experimental chapters which are arranged to reflect the logical development of my research. In my behavioural ecological investigations I first worked on understanding how co-occurring pests on cotton (Figure 1.3) affected the attractancy of thrips. Plant preferences for colonization between damaged and intact seedlings were examined in Y-tube olfactometer choice tests to determine the relative attraction of the plant odours emitted as a result of different types of damage (Chapters 2-5). Chapter 2 covers the olfactory responses of F. occidentalis and F. schultzei to cotton seedlings damaged by different arthropod herbivore species. Results demonstrated that F. occidentalis and F. schultzei were attracted to herbivore-damaged cotton seedlings in very different ways. Based on the results obtained in Chapter 2, the same methods were employed to determine the olfactory responses of T. tabaci to cotton seedlings induced by these arthropod herbivore species (Chapter 3). Results showed that the responses of T. tabaci differed from those of both F. occidentalis and F. schultzei and they were often intermediate to the responses shown by the two Frankliniella species. The studies conducted in Chapters 2 and 3 only considered the effects of damage caused by a single herbivore species. Chapter 4 extended this work and examined how the responses of the three species of thrips differed if complexes of co-occurring herbivores simultaneously damaged cotton seedlings. Results showed that responses of all three species of thrips to plants dually infested by different combinations of target herbivores was predictable based

17 on their responses to plants damaged by single species infestations (Chapter 4). Chapter 5 investigated how the responses of the study species of thrips changed temporally following selected herbivore damage to cotton seedlings. Chapter 6 investigated the patterns of host plant use by three thrips species associated with cotton in the cotton growing region of the Namoi valley, New South Wales. A structured sampling program was conducted, specifically the presence and abundance of adult thrips and larvae across various crop and non-crop plant species was quantitatively measured. At a more basic level, this part of my research relied on sound taxonomy, so morphological criteria formed the basis for structuring a preliminary investigation of molecular markers for the organisms. Results suggest that the surrounding weed species are indeed the sources from which thrips populations migrate onto nearby cotton fields in the Namoi area. COI sequence analysis has revealed that T. tabaci and F. occidentalis on weeds and cotton are from the same population. The main aims of this study were to investigate the source from which thrips invade seedling cotton and the role of cotton plants in mediating thrips movement within cotton crop systems, more specifically; I addressed how herbivore-induced plant responses affected such interactions. In the third section, made up only of Chapter 7, I conclude the thesis by interpreting how results obtained in chapters 2 to 6 help determine the ecology of thrips in the cotton producing region of the Namoi valley. I base my interpretation on my understanding of thrips ecology and behaviour generated by the investigations reported in the previous chapters. The most important general conclusion is that the interactions between thrips and herbivore-damaged cottons are highly complex.

(2) HIPVs may attract or repel herbivores and influence oviposition behaviour

(3) HIPVs attract carnivores and natural enemies of (1) Arthropod feeding on above- herbivores ground and below-ground plant parts causes the release of HIPVs (Herbivore-Induced Plant Volatiles)

Figure 1.10 A stylized schematic representation of the effects of herbivore-induced plant responses, on subsequent arthropod-plant interactions.

Behavioral Mode of Herbivore Host plant reference response induction order species Oviposition Increased HC-F Brassica oleracea Lepidoptera Plutella xylostella Silva & Furlong, 2012; Shiojiri et al., 2002; Poelman et al., 2008; Lu et al., 2004; Reddy et al., 2004 Lycopersicon esculentum Spodoptera exigua Rodriguez-Saona et al., 2005 Gossypium hirsutum Spodoptera littoralis Anderson & Alborn, 1999 Raphanus raphanistrum Pieris rapae Agrawal & Sherriffs, 2001 MD-F Brassica oleracea Plutella xylostella Silva & Furlong, 2012 JA Lu et al., 2004 MD-F, MD-R Brassica napus Silva & Furlong, 2012 HS-F Solanum tuberosum Diptera Episyrphus balteatus Harmel et al., 2007 Phaseolus vulgaris Thysanoptera Frankliniella occidentalis De Vries et al., 2006 Decreased HC-F Solanum carolinense Coleoptera Leptinotarsa juncta Wise & Weinberg, 2002 Solanum dulcamara Plagiometriona clavata Viswanathan et al., 2005 Mimosa pigra Coelocephalapion aculeatum Heard, 1995 Sasa nipponica Diptera Procystiphora uedai Tabuchi et al., 2010 Nicotiana attenuata Lepidoptera Manduca quinquemaculata Kessler & Baldwin, 2001

Nicotiana tabacum Heliothis virescens De Moraes et al., 2001 Lycopersicon esculentum Helicoverpa armigera Lin et al., 2008 Zea mays Ostrinia furnacalis Huang et al., 2009 MeJA Kessler & Baldwin, 2001 JA Capsicum annuum Diptera Liriomyza trifolii Tebayashi et al., 2007 Brassica oleracea Lepidoptera Pieris rapae Bruinsma et al., 2007 Pieris brassicae Bruinsma et al., 2007 Arabidopsis Thysanoptera Frankliniella occidentalis Abe et al., 2009 Brassica rapa Abe et al., 2009 HC-R Brassica nigra Lepidoptera Pieris brassicae Soler et al., 2010 Gossypium hirsutum Spodoptera littoralis Anderson et al., 2011 HC-F, MD-F Brassica rapa Coleoptera Phaedon cochleariae Rostás & Hilker, 2002 Lycopersicon esculentum Lepidoptera Spodoptera exigua Rodriguez-Saona, 2005 Brassica oleracea Mamestra brassicae Poelman et al., 2008 Nicotiana tabacum Thysanoptera Frankliniella occidentalis Delphia et al., 2007 MD-F Brassica rapa Lepidoptera Plutella xylostella Silva & Furlong, 2012 HS-F Solanum lycopersicum Diptera Liromyza trifolii Mayer et al., 2002 HC-F, JA Brassica campestris Plutella xylostella Lu et al., 2004 No effect HC-F Lythrum salicaria Coleoptera Hylobius transverovittatus Hunt-Joshi & Blossey, 2005 Brassica oleracea Lepidoptera Pieris rapae Shiojiri et al., 2002; Poelman et al., 2008 Solanum lycopersicum Hemiptera Bemisia argentifolli Mayer et al., 2002

Triticum aestivum Hymenoptera Cephus cinctus Buteler et al., 2009 HC-R Galerucella calmariensis Hunt-Joshi & Blossey, 2005 Brassica nigra Lepidoptera Pieris rapae Soler et al., 2010; Soler et al., 2009 Neuroptera Chrysoperala carnea Soler et al., 2009 MD-F Brassica rapa Coleoptera Phaedon cochleariae Rostás & Hilker, 2002 Raphanus raphanistrum Lepidoptera Pieris rapae Agrawal & Sherriffs, 2001 MD-R Brassica rapa Plutella xylostella Silva & Furlong, 2012 Attraction Increased HC-F Capsicum annuum Coleoptera Anthonomus eugenii Addesso et al., 2011 Malus domestica Lepidoptera Cydia pomonella Hern & Dorn, 2002 JA Brassica oleracea Plutella xylostella Lu et al., 2004 HS-F Gossypium hirsutum Thysanoptera Frankliniella occidentalis Agrawal & Colfer, 2000 HC-F, MD-F Brassica oleracea Lepidoptera Mamestra brassicae Rojas, 1999 Gossypium hirsutum Trichoplusia ni Landolt, 1993 Decreased MD-F, HC-F Zea mays Hemiptera Rhopalosiphum maidis Bernasconi et al., 1998 (winged) HC-F Nicotiana tabacum Lepidoptera Heliothis virescens De Moraes et al., 2001 Brassica oleracea Trichoplusia ni Landolt, 1993 Zea mays Homoptera Cicadulina storeyi Oluwafemi et al., 2011 MD-F Nicotiana tabacum Thysanoptera Frankliniella occidentalis Delphia et al., 2007 JA Brassica campestris Lepidoptera Plutella xylostella Lu et al., 2004

HS-F Gossypium hirsutum Thysanoptera Frankliniella occidentalis Agrawal & Colfer, 2000 No effect MD-F, HC-F Zea mays Hemiptera Rhopalosiphum maidis Bernasconi et al., 1998 (wingless) MD-F Malus domestica Lepidoptera Cydia pomonella Hern & Dorn, 2002 Brassica oleracea Trichoplusia ni Landolt, 1993 Solanum lycopersicum Mamestra brassicae Rojas, 1999 HC-F Brassica oleracea Rojas, 1999

Chapter 2

How predictable are the behavioral responses of insects to herbivore induced changes in plants? Responses of two congeneric thrips to induced cotton plants

This chapter has been published in PLoS ONE 8: e63611. doi:10.1371/journal.pone.0063611, 2013. The work has been formatted and edited slightly, so that it is consistent with the rest of the thesis. For example, figures are placed within the text. In addition, the bibliography to this manuscript has been omitted. All references cited are included in the thesis bibliography.

Abstract Changes in plants following insect attack are referred to as induced responses. These responses are widely viewed as a form of defence against further insect attack. In this chapter I explore whether it is possible to make generalizations about induced plant responses given the unpredictability and variability observed in insect-plant interactions. Experiments were conducted to test for consistency in the responses of two congeneric thrips, Frankliniella schultzei Trybom and Frankliniella occidentalis Pergrande (Thysanoptera: Thripidae) to cotton seedlings (Gossypium hirsutum Linneaus (Malvales: Malvaceae)) damaged by various insect herbivores. In dual-choice experiments that compared intact and damaged cotton seedlings in an olfactometer, F. schultzei was attracted to seedlings damaged by Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae), Tetranychus urticae (Koch) (Trombidiforms: Tetranychidae), Tenebrio molitor L. (Coleoptera: Tenebrionidae), F. schultzei and F. occidentalis but not to mechanically damaged seedlings. In similar tests, F. occidentalis was attracted to undamaged cotton seedlings when simultaneously exposed to seedlings damaged by H. armigera, T. molitor or F. occidentalis. However, when exposed to F. schultzei or T. urticae damaged plants, F. occidentalis was more attracted towards damaged plants. A quantitative relationship was also apparent, F. schultzei showed increased attraction to damaged seedlings as the density of T. urticae or F. schultzei increased. In contrast, although F. occidentalis demonstrated increased attraction to plants damaged by higher densities of T. urticae, there was a negative relationship between attraction and the density of damaging conspecifics. Both species showed greater attraction to T. urticae damaged seedlings than to seedlings damaged by conspecifics. Results demonstrate that the responses of both species of thrips were context dependent, making generalizations difficult to formulate.

Keywords: induced resistance, induced susceptibility, Frankliniella schultzei, Frankliniella occidentalis, attraction, olfactometer, host location behavior, omnivory, facultative predation

2.1 Introduction Insect herbivores are diverse and feed on plants in various ways. Plants respond biochemically to herbivore, and pathogen, attack (Karban & Myers, 1989; Johnson, 2011) and these induced responses are widely considered to constitute a form of defence (Karban & Baldwin, 1997; Agrawal & Sherriffs, 2001). The volatiles released by plants in response to insect herbivory alter the course of the insect-plant interaction (Underwood & Rausher, 2002; Ohgushi, 2008) and reduced rates of attraction of certain herbivores to damaged plants are often cited as evidence of a defensive role for these chemicals (Karban & Myers, 1989; Karban & Baldwin, 1997; Agrawal & Sherriffs, 2001). Herbivore-induced plant interactions become more intricate when herbivore natural enemies are also considered. The biochemical changes induced in plants have been interpreted as indirect defences as they can attract predators and parasitoids of the damaging herbivores and they are presumed to ultimately increase the fitness of the induced plant (Kessler & Heil, 2011). The importance of induced plant responses is no longer controversial (Karban, 2011) but the function of these responses remains a question for debate. Results from specific studies have frequently been used to support functional interpretation of induced responses including their ecological impact, their effect on the fitness of the impacted herbivores and even possible co-evolutionary relationships between plants and the natural enemies of their herbivores (Shiojiri et al., 2010). It thus begs the question: under what circumstances can we make generalizations about induced plant responses to insect herbivory? Simply put, can we extrapolate from specific studies to draw broad scale conclusions about induced plant responses, or are conclusions from specific studies only valid to the particular systems studied? It is important to acknowledge that the majority of examples of induced ‘resistance’ have been measured from the perspective of the herbivore and there is little evidence of increased plant fitness, and therefore that the response should be considered a defence (Karban, 2011; Kessler & Heil, 2011; Hare, 2011). For instance, herbivores that feed on induced plants may take longer to complete development and consume more tissue than insects feeding on non-induced plants; consequently they may be more damaging to plants than herbivores that grow more quickly but consume less plant material (Karban, 2011). Conversley, longer feeding times may also make herbivores more apparent to natural enemies (Edwards & Wratten, 1983) thereby increasing the chances that they will be attacked. Despite these possibilities, there is no proof that the recorded attraction of natural enemies actually leads to the suppression of attacking herbivore populations (Hare, 2011). Further, most parasitoids do not immediately kill their hosts, which means that plant damage continues for some time after herbivores have been attacked by the attracted parasitoid natural enemies (Kessler & Heil, 2011). Whether herbivores induce specific volatiles that attract specific natural enemies is also debatable; for example, Microplitis croceipes (Cresson) 26

(Hymenoptera: Braconidae), which attacks the corn earworm (Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae)), was equally attracted to plants treated with regurgitant of the beet armyworm, Spodoptera exigua (Hübner) (Lepidoptera: Noctuidae) (which it does not attack), as it was to plants treated with H. zea regurgitant (Hare, 2011). In short there are many factors that determine the outcomes of insect-plant interactions and the observed measurable phenomena in one system may not be evident in others. In general, induced responses tend to negatively affect the oviposition and attractancy responses of adult lepidopterans (Kessler & Baldwin, 2001; Rojas, 1999), yet there is considerable variability in the responses of insects to damaged plants more generally. In some cases induced responses have lead to an increase in insect oviposition and attractancy responses (Silva & Furlong, 2012; Shiojiri et al., 2002; Rodriguez-Saona et al., 2005), and the plants involved become more susceptible to additional attack by herbivores (Karban, 2011). For example, the diamondback moth (Plutella xylostella L. (Lepidoptera: Plutellidae)), showed varying oviposition responses to three species of damaged Brassica plants; on both cabbage and canola oviposition was greater on damaged plants than on undamaged plants, but on Chinese cabbage oviposition was greater on undamaged plants (Silva & Furlong, 2012). Thus host plant responses to herbivory can be unpredictable and the measurement of one response by a given insect to herbivore induced changes in the plant is unlikely to define the insect-plant relationship. In view of such variability it is worth considering just how much can be generalized about induced plant responses and their evolutionary ecology. In this study, the interactions between two congeneric thrips species, Frankliniella schultzei Trybom and F. occidentalis Pergande (Thysanoptera: Thripidae), and cotton (Gossypium hirsutum), a host plant common to both, were investigated for consistency. By feeding on the cotyledons and the developing terminals of young cotton seedlings (Matthews & Tunstall, 1994), these thrips cause damage that can sometimes result in yield loss or delayed maturity, though often plants are able to compensate (Wilson et al., 1994; Sadras & Wilson, 1998). Hence, given evidence of induced responses in cotton to other herbivore species (Anderson & Alborn, 1999), we might predict that both species of thrips would respond similarly. Both F. schultzei and F. occidentalis have an additional aspect to their interaction with plants. Neither fits neatly into the ‘herbivore’ category as both species also consume eggs of spider mites, and their feeding can delay the development of mite outbreaks (Wilson et al., 1996; Agrawal & Klein, 2000). Indeed, mite eggs provide a valuable dietary component in relation to cotton as both thrips species have improved reproductive output if they have access to mite eggs on cotton leaf tissue (Trichilo & Leigh, 1986; Milne & Walter, 1997). Predictions about behaviour need to take account of these intricacies. We might expect, for example, that both species of thrips should 27 be attracted more strongly to cotton seedlings that have been damaged by mites, assuming that the value of mite eggs in the diet outweigh any negative effects of induced responses in the plant tissue. This further assumes that thrips can detect differences between plants damaged by mites or by other herbivores. Mite damaged cotton is attractive to F. occidentalis (Agrawal & Klein, 2000), but no such information is yet available for F. schultzei. In this study I tested for consistency in the responses of these two congeneric Frankliniella thrips to mechanical damage, damage induced by various herbivores and damage caused by potential prey. The study aimed to determine specifically (1) whether damage inflicted by arthropods with different modes of feeding affected the responses of thrips in similar ways, (2) whether the degree of damage affected subsequent attractiveness to thrips, (3) the effect of mite damage as opposed to thrips damage on the attractiveness to thrips, (4) how F. schultzei responds to F. occidentalis damaged cotton seedlings and (5) how F. occidentalis responds to F. schultzei damaged cotton seedlings.

2.2 Materials and methods

2.2.1 Plants Cotton plants (G. hirsutum var. 71RRF) were raised from seed in seed trays containing organic potting mix supplemented with slow-release fertilizer [Osmocote (N:P:K, 16:35:10); Scotts Australia, Baulkham Hills, New South Wales]. Five days after germination, seedlings were transplanted individually into pots (11 cm diameter) (Figure 2.1) with the same mixture of organic potting mix and fertilizer; they were watered daily and grown under natural light and temperature conditions in a ventilated greenhouse. Plants were used in experiments when they were at the two- leaf stage (BBCH code 12 (Munger et al., 1998)) (Figure 2.1).

Two-leaf stage

Figure 2.1 Cotton seedlings were raised in pots containing organic potting mix in a ventilated greenhouse. All seedlings used in experiments were at the two-leaf stage. 28

2.2.2 Insects Frankliniella schultzei and F. occidentalis (Figure 2.2) used in all experiments were from a laboratory culture that originated from adults collected from the field. Frankliniella schultzei were collected from Malvaviscus arboreus Cav. (Malvales: Malvacea) flowers at the St Lucia campus of The University of Queensland, in Brisbane, Queensland, Australia (27°29'43.9"S 153°00'38.2"E). Frankliniella occidentalis were either collected from clover Trifolium repens L. (Fabales: Fabaceae), at the Gatton Research Station, Gatton, Queensland (27°32'31.6"S 152°19'45.9"E), or from cotton (G. hirsutum) flowers at the Australian Cotton Research Institute at Myall Vale near Narrabri, New South Wales. Insects were reared in glass jars (17 cm in height, 8 cm diameter) (Figure 2.3) held in an incubator (25 ± 2ºC, L12:D12). To begin each culture, fresh green beans (Phaseolus vulgaris) were soaked in water for 16 h and then washed in warm soapy water to remove any insecticide residue. They were rinsed in clean water and dried before being used as a rearing substrate for the thrips. Cleaned beans were placed in glass jars lined with paper towel (Figure 2.3) and female thrips were released to feed and oviposit. To prevent thrips from escaping, glass jars were sealed using two layers of nylon mesh (22 × 22 × 22 cm, ≈ 1 mm2) with a black cardboard layer held between them (Figure 2.3). The green beans were left in the glass jars for three days after which they were removed and replaced with new ones; old beans were placed in new glass jars lined with paper towel for larval rearing. For both adult and larval thrips, pollen collected from Hibiscus spp (Malvales: Malvaceae) flowers was provided as an additional food supplement (Figure 2.3). All experiments were conducted on lab reared thrips 3-5 generations after field collection.

(A) (B)

Figure 2.2 Thrips species investigated (A) Frankliniella occidentalis (B) F. schultzei.

Two nylon mesh

Thrips larvae on beans

Beans Glass jar

Pollen grains

2.2.3 Standard olfactometer tests The same method was used in all olfactometer experiments. A clean glass Y-tube olfactometer (stem 9 cm; arms 9 cm each at a 45º angle; internal diameter 0.9 cm) (Figure 2.4) was used to compare the responses of thrips in various dual-choice experiments. Air was drawn over an activated charcoal filter and then through the olfactometer at a rate of 1L min-1. Experiments were conducted in a temperature controlled room at 25ºC (± 2ºC), and a light source was placed directly above the Y junction of the olfactometer to avoid bias. Single female thrips were released at the stem entrance of the Y-tube and observed for 10 min. When a single thrips moved more than half way (4.5 cm) (Figure 2.4) into one of the arms and remained there for more than 20s, it was recorded as being attracted to the odour source associated with that arm. After testing five individuals, the chambers with plants were rotated and the Y- tubes cleaned with 100% ethanol and dried. After 10 thrips individuals had been screened, the test plants were changed. A minimum of 60 individual thrips was used in each test.

B) (A) Glass chambers (B)

Y - tube

Tube connecting to vacuum Airflow regulator

2.2.4 Standard method for setting up arthropod enclosures Cotton seedlings for use in olfactometer experiments were either damaged mechanically or by H. armigera larvae, adult thrips, two spotted spider mites (T. urticae) or mealworms (T. molitor) (see below for details). Cages were constructed to enclose the arthropods used to damage the potted cotton seedlings. All arthropods were enclosed on potted cotton seedlings within inverted plastic cups (9.7 cm in height, 8 cm diameter) containing two windows (3.5 × 4.5 cm) covered with nylon mesh (≈1 mm2) (Figure 2.5). Once the relevant arthropods had been placed inside the cup, the rim was pushed onto the soil surface to enclose them on the seedling and they were allowed to feed for 24 h.

Nylon mesh

Damaged leaves

Plastic cup Ventilation window

2.2.5 Feeding styles, cotton responses and thrips behaviour A series of dual-choice olfactometer experiments was conducted to examine thrips behaviour in response to cotton seedling damage. In each experiment the response of thrips to damaged cotton seedlings and undamaged cotton seedlings was compared. The following forms of damage were inflicted on cotton seedlings: Mechanical damage. Both leaves of the cotton seedlings were damaged mechanically in a manner designed to simulate the feeding damage of various caterpillars that attack cotton plants. Circular holes (0.5 cm diameter) were cut through leaf tissue using a cork borer. A total of six holes (three

32 on each leaf) were cut. Plants were damaged 24 h before the start of the standard olfactometer experiments. H. armigera damage. Cotton leaves were damaged by second instar H. armigera larvae (Figure 2.6B). Ten larvae were introduced to the adaxial leaf surface and were enclosed as previously described. Larvae were allowed to feed on the seedlings for 24 h. Standard olfactometer experiments were performed using plants with H. armigera larvae in situ and with plants from which H. armigera larvae were removed after 24 hrs feeding. Thrips damage. Cotton foliage was independently damaged by F. schultzei and F. occidentalis (Figure 2.2). Fifty thrips (densities of this magnitude have been recorded in the field (Williams et al., 2011)) were introduced onto each plant and were enclosed as described above. Thrips were allowed to feed on seedlings for 24 h. Standard olfactometer experiments were performed using plants with thrips in situ or with plants from which thrips were removed after 24 hrs feeding. Mite damage. Fifty mites (Figure 2.6A) (densities of this magnitude have been recorded in the field (Williams et al., 2011)) were introduced onto the leaves and were enclosed as described above. Mites were allowed to feed on cotton seedlings for 24 h. Standard olfactometer experiments were performed using plants with mites in situ and with plants from which mites were removed after 24 hrs feeding. Root damage. Roots of potted cotton seedlings were damaged by using mealworms (T. molitor) (Figure 2.6C). These cause damage similar to that of false wireworm species (also Tenebrionidae, and which are pests of cotton (Williams et al., 2011)). They damage roots by feeding or by dislodging them as they burrow. Ten mealworms were placed on the soil surface and allowed to burrow into the soil. The response of thrips to root-damaged and undamaged above ground foliage was measured in standard olfactometer experiments 48 h after release of the mealworms, whereupon damage began to be inflicted. After the olfactometer experiments the cotton seedlings were removed from the pots and the mealworms extracted. Each cotton seedling was placed in a plastic box (45 × 25 × 10 cm) and exposed to slow running water. The root mass was removed from the stem of each plant and all broken and dislodged pieces of root in the soil were collected. All root material was then dried in an oven at 60ºC for 24 h. The dried root mass was then weighed and the number of dislodged pieces of root recorded before this material was pooled and weighed independently of the root mass. Root herbviory reduced root biomass by an average of 43%.

(A)

(B)

(C)

(A) (B)

Mealworms feeding on seedling roots (C)

Undamaged root mass Broken and dislodged pieces of root collected

2.2.6 Degree of foliage damage and thrips attraction To determine if a quantitative relationship was detectable between the degree of damage and subsequent thrips attraction, experiments were conducted in which thrips preference was compared between plants that had been subjected to a high degree of herbivory and plants that had been subjected to a much lower degree of herbivory by the same herbivore. The arthropods used to damage cotton seedlings were mites and thrips. These arthropods were independently enclosed on cotton seedlings as previously described and allowed to feed for 24 h when they were removed and standard olfactometer experiments performed. For the thrips-damaged plant tests, plants damaged by 50 adult thrips and plants damaged by 10 adult thrips were compared. Similarly, for mite- damaged tests, plants damaged by 50 mites and plants damaged by 10 mites were also compared.

2.2.7 Mite damage vs thrips damage on thrips attractancy The relative attraction of thrips to mite-damaged plants and thrips-damaged plants was investigated by exposing cotton seedlings to either 50 mites or to 50 thrips in insect enclosures for 24 h as previously described. Standard olfactometer experiments were performed using plants with herbivores in situ and with plants from which herbivores were removed after 24 hrs feeding. 35

2.2.8 Are thrips attracted to mites alone (no plants) or a combination of mites and damaged plants? Experiments were conducted to determine whether thrips were attracted to mite odors or whether the mites had to be present with damaged plants for thrips to respond. To establish if thrips were attracted to mite odours alone 50 mites were placed in one olfactometer chamber and the other chamber was left empty. Separate standard olfactometer tests were then performed to investigate the attraction of F. schultzei and F. occidentalis to mites. In a second experiment cotton seedlings were exposed to 50 mites for 24 h in arthropod enclosures as previously described, mites were then removed from some plants but left in situ on others and standard olfactometer tests performed to independently test the attraction of F. schultzei and F. occidentalis to mite-infested and mite- damaged plants from which mites had been removed.

2.2.9 Response of each Frankliniella species to cotton seedlings damaged by the other Frankliniella species Cotton seedlings were exposed to either 50 F. schultzei adults or 50 F. occidentalis adults for 24 h in arthropod enclosures as previously described. The thrips were then removed and standard olfactometer tests performed to test the attraction of F. schultzei to plants damaged by F. occidentalis relative to undamaged plants and the attraction of F. occidentalis to plants damaged by F. schultzei relative to undamaged plants.

2.2.10 Statistical analysis The number of thrips attracted to the different treatments in paired olfactometer tests was compared by a statistical χ² test using StatView (SAS Institute, 1999). Responses were converted to percentages for presentation.

2.3 Results

2.3.1 Feeding styles, cotton responses and thrips behaviour In general, the two congeneric thrips species responded differently to the various forms of herbivory inflicted on cotton seedlings. Frankliniella schultzei did not show differential attraction between mechanically damaged and undamaged plants (χ² = 1.4, d.f. = 1, P = 0.23; Figure 2.8). However, F. schultzei was more attracted to herbivore-damaged plants compared to undamaged plants, regardless of whether herbivores were in situ or if they had been removed (H. armigera in situ: χ² = 38.4, d.f. = 1, P<0.001; H. armigera removed: χ² = 40.6, d.f. = 1, P<0.001; root damage: χ² = 26.6, d.f. = 1, P<0.001; thrips (F. schultzei ) in situ: χ² = 22.4, d.f. = 1, P<0.001; thrips (F. schultzei ) 36 removed: χ² = 39.8, d.f. = 1, P<0.001; thrips (F. occidentalis) removed: χ² = 32.2, d.f. = 1, P<0.001; mites in situ: χ² = 34.8, d.f. = 1, P<0.001; mites removed: χ² = 19.2, d.f. = 1, P<0.001; Figure 2.8). Conversely, F. occidentalis was more attracted to undamaged plants than to damaged plants, regardless of whether the damage was mechanical (χ² = 15.0, d.f. = 1, P = 0.001) or whether herbivores were in situ or had been removed (H. armigera in situ: χ² = 26.6, d.f. = 1, P = 0.001; H. armigera removed: χ² = 41.6, d.f. = 1, P<0.001; root damage: χ² = 17.0, d.f. = 1, P<0.001; thrips (F. occidentalis) in situ: χ² = 24.1, d.f. = 1, P<0.001; thrips (F. occidentalis) removed: χ² = 32.2, d.f. = 1, P<0.001; Figure 2.8). However, when exposed to F. schultzei (removed) damaged cotton seedlings, F. occidentalis was more attracted to damaged plants compared to undamaged plants (χ² = 21.6, d.f. = 1, P<0.001; Figure 2.8). With respect to responses to mite damaged plants, F. occidentalis behaved similarly to F. schultzei in that mite-damaged seedlings were more attractive than undamaged seedlings (mites in situ: χ² = 13.0, d.f. = 1, P = 0.003; mites removed: χ² = 17.0, d.f. = 1, P<0.001; Figure 2.8).

2.3.2 Degree of foliage damage and thrips attraction The degree of damage inflicted by insect herbivores significantly affected the level of attractiveness to both thrips species. Significantly more F. schultzei were attracted to plants damaged by 50 thrips (F. schultzei) than to plants damaged by 10 thrips (F. schultzei) (χ² = 25.0, d.f. = 1, P<0.001; Figure 2.9) and to plants damaged by 50 mites than to plants damaged by 10 mites (χ² = 33.0, d.f. = 1, P<0.001; Figure 2.9). However significantly more F. occidentalis were attracted to plants damaged by 10 thrips (F. occidentalis) than to plants damaged by 50 thrips (F. occidentalis) (χ² = 21.6, d.f. = 1, P<0.001; Figure 2.9) whereas significantly more F. occidentalis were attracted to plants damaged by 50 mites than to plants damaged by 10 mites (χ² = 6.66, d.f. = 1, P = 0.001; Figure 2.9).

2.3.3 Mite damage vs thrips damage on thrips attractancy Both thrips species behaved similarly when presented with mite-damaged and thrips-damaged plants simultaneously. Significantly more F. schultzei were attracted to mite damaged plants than thrips damaged plants regardless of whether the arthroponds were present or had been removed (all arthropods in situ: χ² = 18.5, d.f. = 1, P<0.001; all arthropods removed: χ² = 12.8, d.f. =1, P = 0.003; Figure 2.10). Similarly, significantly more F. occidentalis were attracted to mite damaged plants than to thrips damaged plants (all arthropods in situ: χ² = 8.06, d.f. = 1, P = 0.005; all arthropods removed: χ² = 9.60, d.f. = 1, P = 0.001; Figure 2.10).

Damaged Undamaged Frankliniella schultzei FoliarFoliar - - mechanicalmechanical damage damage NS n = 70

Helicoverpa in situ *** n = 60

HelicoverpaHelicoverpa removedremoved *** n = 70

RootRoot damageddamage *** n = 60

F. schultzei in situ *** n = 61

n = 68 F.F. schultzeischultzei removed removed ***

F.F. occidentalisoccidentalis removed removed *** n = 60

MitesMites inin situ situ *** n = 72

Mites removed *** n = 60 Mites removed

Frankliniella occidentalis n = 60 FoliarFoliar - - mechanicalmechanical damage damage ***

*** n = 60 Helicoverpa in situ Helicoverpa removed n = 60 Helicoverpa removed ***

Root damage n = 60 Root damage *** n = 60 Figure 2.8 Responses of two F. occidentalis in situ ***

F. occidentalis removed n = 60 congeneric thrips species, F. schultzei F. occidentalis removed *** and F. occidentalis, to various forms of F. schultzei removed n = 60 F. schultzei removed *** herbivory inflicted on cotton seedlings. Mites in situ *** n = 60 Mites in situ NS – not significant, * P<0.05, ** Mites removed *** n = 60 Mites removed P<0.01, *** P< 0.001. 100 80 60 40 20 0 20 40 60 80 100 % of thrips responding 38

Plants damaged by 50 individuals Plants damaged by 10 individuals Frankliniella schultzei Damaged by: F.F. schultzeischultzei *** n = 63

Mites *** n = 64

Frankliniella occidentalis Damaged by: F.F. occidentalisoccidentalis *** n = 60

n = 60 Mites ***

100 80 60 40 20 0 20 40 60 80 100 % of thrips responding

Mite damaged Thrips damaged Frankliniella schultzei

All arthropods in situ *** n = 70

All arthropods removed *** n = 70

Frankliniella occidentalis

All arthropods in situ *** n = 60

n = 60 All arthropods removed ***

100 80 60 40 20 0 20 40 60 80 100 % of thrips responding Figure 2.10 Responses of F. schultzei and F. occidentalis when exposed to volatiles emitted by mite damaged and conspecific thrips damaged cotton seedlings. NS – not significant, * P<0.05, ** P<0.01, *** P< 0.001.

2.3.4 Are thrips attracted to mites alone (no plants) or a combination of mites and damaged plants? Both species behaved similarly when presented with mites in the absence of plants and mite- damaged plants. Frankiniella schultzei was not attracted to mites when they were presented in the absence of plants (χ2 = 1.66, d.f. = 1, P = 0.19) (Figure 2.11A) and it did not discriminate between mite damaged plants from which mites had been removed and plants on which mites remained in situ (χ2 = 1.06, d.f. = 1, P = 0.30) (Figure 2.11B). Similarly F. occidentalis was not attracted to mites when they were presented in the absence of plants (χ2 = 3.26, d.f. = 1, P = 0.07) (Figure 2.11A) and it did not discriminate between mite damaged plants from which mites had been removed and plants on which mites remained in situ (χ2 = 0.60, d.f. = 1, P = 0.44) (Figure 2.11B).

Mites alone (no plants) No mites A Frankliniella schultzei

NS n = 60

Frankliniella occidentalis

NS n = 60

100 80 60 40 20 0 20 40 60 80 100 % of thrips responding

Mites in situ Mites removed B Frankliniella schultzei

NS n = 60

Frankliniella occidentalis

NS n = 60

100 80 60 40 20 0 20 40 60 80 100 % of thrips responding

2.4 Discussion It was expected that both thrips species would show similar responses to induced changes in cotton plants and it was predicted that they would be attracted to plants damaged by mites but avoid plants damaged by other herbivores. However, the species responded quite differently, but consistently, to damage inflicted on cotton seedlings. Frankliniella schultzei was consistently attracted to cotton seedlings damaged by all herbivores tested, including conspecifics, but not to plants damaged mechanically (Figure 2.8). In contrast, F. occidentalis was more strongly attracted to undamaged cotton seedlings than to mechanically damaged seedlings or to seedlings damaged by conspecifics, H. armigera or T. molitor but it was attracted to seedlings damaged by F. schultzei or mites (Figure 2.8). The prediction that thrips would be attracted to plants damaged by mites, was thus correct and similar for both species. For F. occidentalis this result contradicts expectations based on its responses to plants damaged by other herbivores, but is consistent with previous work investigating its responses to mite damaged cotton plants (Trichilo & Leigh, 1986). Hence, the responses of both species of thrips to damage induced in host plants is clearly context dependent but unpredictable. Understanding the molecular and biochemical changes in plants and their perception by thrips will be essential to fully understand these herbivore-host plant interactions. Two particular results further highlight the complexity of the interactions between cotton and these two species of thrips. One, when given a choice between undamaged seedlings and seedlings damaged by F. schultzei, F. occidentalis was much more strongly attracted to F. schultzei damaged seedlings (Figure 2.8). This is surprising as F. occidentalis would be expected to respond in the same manner as it did when seedlings were damaged by conspecifics (Figure 2.8). Two, in the quantitative test to determine the relationship between the degree of damage and thrips’ response, the responses of both species were related to density (Figure 2.9) but in opposite directions; increased damaged caused by higher densities of conspecifics increased attraction of F. schultzei whereas the opposite was observed for F. occidentalis. Thus insect herbivore-induced responses should not be solely viewed as plant defence strategies; by using the term ‘defences’, we automatically infer that plants are resisting various forms of herbivory. In cotton, responses induced by mites may be a defence against further mite attack yet the same responses are evidently used by thrips as a cue for the presence of mite prey. Further, cotton seedling responses to damage by other foliage or root feeders resulted in the avoidance of these plants by F. occidentalis and attraction to them by F. schultzei. Although induced responses may be interpreted as direct or indirect defence mechanisms, they are perhaps more parsimomously interpreted as biochemical responses to herbivory with the resultant volatiles being utilized in different ways by different species.

The intricacy and reciprocity of the interactions that were observed between the two thrips species and cotton seedlings highlights just how difficult it is to generalize about the way in which these particular organisms respond to induced plants, and the results presented here add to the variability of interactions that have been recorded between insects and plants (Table 1.1, Chapter 1). This leads to questioning the validity of broad scale conclusions drawn by extrapolation from specific studies. From the current study it could be argued that F. occidentalis was more attracted to undamaged plants over herbivore damaged plants as a result of induced resistance. However De Vries et al. (2006) showed that F. occidentalis damaged bean and cucumber plants were more attractive to conspecific females than undamaged plants. This provides further evidence for the context, or system, specificity of such interactions and highlights the danger of broader scale interpretation without specific testing. Further adding to the complexity, the current study shows that the responses of one host plant to herbivory had opposing effects on two congeneric insect herbivores. Other studies have shown that multiple host plants have opposing effects on a single insect herbivore species (Silva & Furlong, 2012; Lu et al., 2004). For instance, the diamondback moth (Plutella xylostella) laid more eggs on Jasmonic acid (JA) treated cabbage plants and fewer on JA treated Chinese cabbage, however Pieris rapae did not show such a differential response between JA treated and control cabbage and Chinese cabbage plants (Lu et al., 2004). Thus extrapolation from specific studies needs to be viewed with caution, and conclusions should be viewed in a context dependent manner. A challenge in trying to understand the responses recorded in this system is to determine what exactly the thrips are responding to and why they do so. The current study provides clear evidence that both study species respond to herbivore induced changes in the host plant rather than the presence of herbivores themselves (Figure 2.8 and 2.11). Previous studies on the interaction between F. occidentalis and cotton have also found that these thrips are attracted to undamaged cotton plants (Agrawal & Colfer, 2000; Spence et al., 2007). The current study is the first to investigate the interaction between F. schultzei and its cotton host plant. The only response that was consistent across the two thrips species was that they were attracted to mite-damaged seedlings (Figure 2.8 & 2.10). A possible reason for this is that F. schultzei and F. occidentalis are predators of mite eggs (Wilson et al., 1996). Feeding studies have shown that omnivory helps to meet the nutritional needs of thrips (Agrawal et al., 1999; Milne & Walter, 1997) and it has been argued that induced responses of host plants influence F. occidentalis to prey more on spider mite eggs and less on cotton leaves; essentially induced responses to herbivory indirectly cause omnivorous thrips to increase predation of mites (Agrawal et al., 1999; Spence et al., 2007). This argument follows the nomenclature that plant induced responses constitute a defence strategy. Induced responses to herbivory brought on by mites can both directly and indirectly reduce 42 mite populations causing omnivores to increase predation and enhance their role in biological control strategies (Agrawal et al., 1999). These generalizations must be taken with caution, as results of the current study show that F. schultzei, which is also an omnivore, was attracted to damaged cotton seedlings regardless of whether mites or other arthropods had caused the damage. Mite damage possibly induces different responses in the plant to the other insect species tested, hence both thrips species may be attracted to indications of the presence of mites (potential prey) but have different orientations to indications of damage by other herbivores. A mechanistic explanation is potentially available because plant transcriptional responses can be specific to the elicitor compounds released by particular insect (or mite) herbivore feeding (Frey et al., 2000; Voelckel & Baldwin, 2004). These herbivore specific plant responses to feeding could be caused by a combination of factors resulting from herbivore specific physical damage and the type and amount of elicitors released during feeding (Hilker & Meiners, 2010). The feeding modes or sites of H. armigera, T. urticae and T. molitor are very different from one another and could have induced the cotton plants in very different ways. So it is possible that thrips are not responding to mite damage alone, as F. schultzei individuals were attracted to other insect damaged plants (Figure 2.8). This study is also the first to demonstrate that below-ground herbivory affects the responses of thrips species to their host plants (Figure 2.8), with the two thrips species responding differently to root damage on cotton. Studies on other insects have shown similar results (Anderson et al., 2011; Abe et al., 2009; Silva & Furlong, 2012). Induced responses brought on by root herbivory have recently been shown to result in systemic responses in the shoots (van Dam et al., 2009). This study shows that one plant with root damage (cotton) had different effects on insect herbivores (thrips) but previous studies have shown that root herbivory on different plant species have opposing effects on a single insect herbivore (Silva & Furlong, 2012), further illustrating the complex nature of insect-plant interactions. Interactions between below-ground and above-ground insect herbivores further complicates matters and only adds to the challenge in predicting induced plant responses and insect behavioural responses. An important conclusion that emerges from the current study is that methodology has important consequences when making generalizations about induced plant responses and developing hypotheses that relate to this phenomenon. The methods used will influence the validity of the test and thus the quality of any derived generalization (Frey et al., 2000). The results show that tests should include multiple insect herbivores if a single plant is used or multiple host plant species if a single insect herbivore is tested. This will provide more meaningful tests (and answers) into how plants respond to insect damage and how insects respond to plant damage, and will avoid the problem of selecting study organisms and designing tests that are simply likely to verify earlier generalizations. 43

Chapter 3 Responses of Thrips tabaci to odours of herbivore-induced

cotton seedlings

This chapter has been published in Entomologia Experimentalis et Applicata 151: 239-246, 2014. The work has been reformatted and edited slightly, so that it is consistent with the rest of the thesis. For example, figures are now placed within the text, references are cited in a style consistent with the rest of the thesis and the bibliography has been incorporated into the thesis bibliography.

Abstract Herbivore induced changes in plants have been widely viewed as defensive responses against further insect attack. However changes in plants as a consequence of herbivore feeding can elicit various responses in herbivores; these are variable, context dependent and often unpredictable. In this laboratory study the responses of Thrips tabaci Lindeman (Thysanoptera: Thripidae) to volatiles emitted by intact and herbivore-damaged or mechanically-damaged cotton seedlings (Gossypium hirsutum L. (Malvales: Malvaceae)) were investigated in dual-choice olfactometer assays; the study builds on previous work on Frankliniella schultzei Trybom and Frankliniella occidentalis Pergrande (Thysanoptera: Thripidae). Thrips tabaci showed increased attraction to seedlings subject to foliar mechanical damage and those with foliar damage inflicted by conspecifics or Tetranychus urticae Koch (Trombidiforms: Tetranychidae), upon which it preys. However, T. tabaci did not discriminate between intact seedlings and those with foliar damage inflicted by Helicoverpa armigera Hübner (Lepidoptera: Noctuidae), two other species of thrips, F. schultzei and F. occidentalis, or those with root damage inflicted by Tenebrio molitor L. (Coleoptera: Tenebrionidae). Attraction of T. tabaci was also affected by herbivore density on damaged plants. That is, seedlings damaged by higher densities of T. urticae or T. tabaci were more attractive than seedlings damaged by lower densities of the corresponding arthropod. Although attracted to plants damaged by conspecifics or T. urticae, T. tabaci showed greater attraction to seedlings damaged by T. urticae than to seedlings damaged by conspecifics. Results are discussed in the context of the responses of F. schultzei and F. occidentalis to herbivore-induced cotton seedlings, highlighting the complexity, variability and unpredictability of the responses of even closely related species of insects to plants under herbivore attack.

Keywords: Thysanoptera, induced resistance, induced susceptibility, olfactometer

3.1 Introduction The biochemical responses of plants to insect herbivory can significantly affect subsequent arthropod-plant interactions (Hilker & Meiners, 2010; Kessler & Heil, 2011; Karban, 2011). In cotton plants (Gossypium hirsutum) herbivory can result in the suppression of further herbivore feeding (Miyazaki et al. 2013; Anderson & Agrell, 2005; Kranthi et al., 2003; Bezemer et al., 2003; Alborn et al., 1996), reduced herbivore attraction and oviposition (Landolt, 1993; Karban & Carey, 1984; Karban, 1986; Agrawal & Klein, 2000) and the attraction of herbivore natural enemies (Ngumbi et al., 2009). Such responses have been widely interpreted as a form of defence against herbivore attack (Dicke & Baldwin, 2010; Karban & Myers, 1989; Karban & Baldwin, 1997; Agrawal & Sherriffs, 2001) but in many systems the ecological function and significance of the chemical changes measured in plants remains to be proven (Hare, 2011). In chapter 2, I investigated the responses of two congeneric thrips species, Frankliniella schultzei Trybom and F. occidentalis Pergande (Thysanoptera: Thripidae), to herbivore-induced cotton. The responses differed between the two species and were context dependent. Whereas F. schultzei was more attracted to cotton seedlings damaged by various arthropod pests (cotton bollworm, two-spotted spider mites and root damaging mealworms) than to intact seedlings, F. occidentalis was more attracted to intact seedlings with the exception of seedlings damaged by two- spotted spider mites (for more detail see chapter 2). Other studies investigating herbivore-induced changes in cotton plants have found similar unpredictable interactions. For example, oviposition by Spodoptera littoralis Boisduval (Lepidoptera: Noctuidae), was higher on S. littoralis-damaged plants than on undamaged plants (Anderson & Alborn, 1999). However, when S. littoralis was offered a choice between cotton plants with roots damaged by Agriotes lineatus L. (Coleoptera: Elateridae) and undamaged plants, oviposition was higher on undamaged plants (Anderson et al., 2011). These results highlight the difficulties in generalizing about insect-plant interactions and emphasize that responses of herbivores to herbivore-induced plants can be unpredictable. Indeed, rather than becoming more resistant, many plant species become measurably more susceptible to further insect attack following herbivory (Karban & Baldwin, 1997; Silva & Furlong, 2012). Thrips tabaci Lindeman (Thysanoptera: Thripidae) is also a pest of cotton crops in Australia (Wilson and Bauer, 1993; Williams et al., 2011) and elsewhere in the world (Cook et al., 2003; Deligeorgidis et al., 2002). To expand the study and further investigate the responses of thrips to herbivore induced plants, interactions between this species and cotton were investigated in laboratory experiments. Although F. schultzei, F. occidentalis and T. tabaci are foliar feeding pests of cotton, none is entirely phytophagous. Rather, they are all facultative predators of two spotted spider mite, Tetranychus urticae (Koch) (Trombidiformes: Tetranychidae) (Trichilo & Leigh 1986; Wilson et al., 1996; Milne & Walter, 1998a) and in cotton crops they commonly occur on plant 46 parts that are infested with their T. urticae prey (Wilson et al., 1996; Milne & Walter, 1998b). In this study I tested for consistency in the responses of T. tabaci to cotton seedlings subjected to foliar mechanical damage, foliar damage inflicted by H. armigera, F. schultzei, F. occidentalis, T. tabaci or T. urticae (which is a potential prey item for T. tabaci) and root damage inflicted by T. molitor. Specifically I determined (1) how damage inflicted by arthropods with different modes of feeding affected the responses of T. tabaci to cotton seedlings and how this compared with the previously reported responses of F. schultzei and F. occidentalis (Chapter 2), and (2) whether the degree of feeding damage affected the intensity of any response by T. tabaci.

3.2 Materials and methods

3.2.1 Plants Potted cotton plants (G. hirsutum var. Sicott 71RRF) were raised from seed to seedling in individual pots (11 cm diameter) containing organic potting mix supplemented with slow release fertilizer [Osmocote (N:P:K, 16:35:10); Scots Australia, Baulkham Hills, New South Wales]. Seedlings were watered daily and grown under natural light (L:D, 13:11 h) and temperature conditions (mean minimum=17ºC; mean maximum=28.8ºC) in a ventilated greenhouse and used in experiments when they were at the two-leaf stage (BBCH code 12 (Munger et al., 1998)) (Figure 2.1, chapter 2), 20 days after germination.

3.2.2 Arthropods Thrips tabaci Thrips tabaci (Figure 3.1) were obtained from a laboratory culture that originated from adults collected from Ammi majus (Apiales: Apiaceae) flowers at the Australian Cotton Research Institute, Myall Vale, Narrabri, New South Wales, Australia, (30°12'20.1"S 149°35'44.2"E). Insects were reared at 25±2°C (L12:D12) on clean green beans (Phaseolus vulgaris). Before being used as an adult food source and oviposition substrate, green beans were soaked in water for 16 h and then washed in warm soapy water to remove any insecticide residue. Cleaned beans were held in glass jars (17 cm height, 8 cm diameter) that were lined with paper towel. Thrips were prevented from escaping by sealing glass jars with two layers of nylon mesh (22×22×22 cm, ≈ 1 mm²) with a black cardboard layer held between them (Figure 2.3, Chapter 2). The green beans were replenished every 3 days; material exposed to thrips was then transferred to clean glass jars for larval rearing. All experiments were conducted on laboratory-reared thrips 3-5 generations after field collection.

Figure 3.1 Adult female Thrips tabaci.

Two spotted spider mites (T. urticae) To provide mites for use in experiments a culture was maintained on cotton seedlings (two-leaf stage) that originated from adults collected from cotton seedlings at the Australian Cotton Research Institute, Myall Vale, Narrabri, New South Wales, Australia. Mites were reared at 26 ± 2ºC (L12:D12) on pots (20 cm diameter) containing 20 cotton seedlings.

Helicoverpa armigera Second instar H. armigera used in experiments were obtained from a laboratory culture that originated from eggs collected from various crops at Toowomba, Queensland, Australia. Helicoverpa armigera were reared on a soybean-based artificial diet (Teackle, 1991).

Mealworms (T. molitor) Mealworm larvae were purchased from a commercial supplier (Pisces Enterprises Pty Ltd., Brisbane, Australia). Larvae were held in plastic containers (6 cm height, 15 cm diameter) that contained vermiculite as a burrowing substrate. Larvae were supplied with small pieces of carrot as a food source but were starved for 24 h prior to use in experiments.

(A) (B)

(C)

(D) (E)

Figure 3.2 Arthropod herbivore species used to damage cotton seedlings. (A) Second instar Helicoverpa armigera larvae. (B) Adult two-spotted spider mites (Tetranychus urticae). (C) Root- feeding mealworm larvae (Tenebrio molitor). (D) Frankliniella occidentalis. (E) Frankliniella schultzei.

3.2.3 Standard olfactometer tests The same method was used in all olfactometer experiments. A glass Y-tube olfactometer (stem 9 cm, arms 9 cm [at a 45° angle], internal diameter 0.9 cm and with air drawn over an activated

49 charcoal filter and then through the olfactometer at a rate of 1L min-1) (Figure 2.4, Chapter 2) was used to compare the responses of T. tabaci to different cotton seedling treatments in various dual- choice experiments. A white light source (Osram L 15W) was placed directly above the Y junction of the olfactometer to avoid bias. Single T. tabaci females were released at the stem entrance of the Y-tube and observed for up to 10 min. When a thrips moved more than half way (4.5 cm) (Figure 2.4, Chapter 2) into one of the arms and remained there for more than 20s, it was recorded as having been attracted to the odour source associated with that arm. A minimum of 60 individual thrips was used in each test. Single T. tabaci were tested in a series of experimental runs that each consisted of 10-15 individual insects; a minimum of 60 T. tabaci was used in each test and data were pooled for analysis.

3.2.4 Feeding mode, cotton responses and T. tabaci behaviour Cotton seedlings were damaged mechanically, or by H. armigera larvae, female adult thrips (F. occidentalis, F. schultzei or T. tabaci), female adult two spotted spider mites or mealworm larvae. Cotton seedlings were damaged by enclosing the arthropod herbivores on seedlings in cages. Cages consisted of inverted plastic cups (9.7 cm in height, 8 cm diameter) that contained two ventilation windows (3.5×4.5 cm) covered with nylon mesh (≈1 mm²) (Figure 2.5, Chapter 2). Once herbivores had been introduced to the seedlings, the cup was placed over the plant and its rim pushed into the soil to enclose the seedling and relevant arthropods. Cups were also placed over undamaged control plants but no arthropods were introduced. Arthropods were allowed to feed for 24 h and then cups were removed from damaged and control plants before they were used in olfactometer experiments. Mechanical damage was inflicted 24 h prior to experiments. A series of dual-choice olfactometer experiments was conducted to examine T. tabaci behaviour in response to undamaged cotton seedlings and cotton seedlings damaged in one of the following ways:

Mechanical foliar damage. The leaves of cotton seedlings were mechanically damaged to simulate the feeding damage caused by various caterpillars. Three circular holes (0.5 cm diameter) were cut through each of the two leaves using a cork borer. Helicoverpa armigera damage. Ten second instar larvae (Figure 3.2A) were introduced onto the adaxial leaf surface of leaves, they were then enclosed as described above and allowed to feed on the seedlings for 24 h. Standard olfactometer experiments were performed using plants with H. armigera larvae in situ and with plants from which H. armigera larvae had been removed after 24 hrs feeding. Thrips damage. Thrips tabaci (Figure 3.1), F. schultzei and F. occidentalis (Figure 3.2E & D) were used independently to damage cotton foliage. Fifty adult female thrips were introduced and 50 enclosed onto each plant (one thrips species at a time) and were allowed to feed on seedlings for 24 h. Standard olfactometer experiments were performed using plants with thrips in situ or plants from which thrips had been removed after 24 hrs feeding. Two spotted spider mite damage. Fifty adult female mites (Figure 3.2B) were introduced and enclosed on cotton plants as described above. Mites were allowed to feed on plants for 24 hrs after which standard olfactometer experiments were performed using plants with mites in situ or using plants from which mites, any eggs laid and webbing produced had been removed using a fine paint brush. Root damage. Root damage was inflicted by late instar mealworm (T. molitor) larvae (Figure 3.2C). Ten larvae were placed on the soil surface and allowed to burrow into the soil and feed on the seedling roots. The response of T. tabaci to root damaged and undamaged above ground foliage was measured in standard olfactometer experiments 48 h after release of larvae onto the soil surface. The degree of root herbivory in cotton seedlings was measured upon the completion of the olfactometer experiments (Figure 2.8, Chapter 2). Undamaged and herbivore damaged cotton seedlings were removed from the pots and the mealworms extracted from the soil surrounding damaged roots. Individual plants were placed into plastic boxes (45×25×10 cm) and the soil washed from the roots using slow running water. The root mass was removed from the stem and all broken and dislodged pieces of root in the soil were collected separately and stored separately. Root material was then dried in an oven at 60ºC for 24h and root dry weights determined. Across all replicates herbivory reduced root biomass by an average of 45 (±5)% (n = 12).

3.2.5 Degree of foliage damage and T. tabaci response To quantify the intensity of T. tabaci responses to plants relative to the degree of cotton seedling damage, thrips were exposed to volatiles from cotton seedlings that had been subject to high (damaged by 50 individual arthropods) and low (damaged by 10 individual arthropods) levels of herbivory. Cotton seedlings were independently damaged by both densities of adult two spotted spider mites and both densities of adult T. tabaci as previously described (Section 3.2.4). Standard olfactometer experiments were then performed with herbivores removed, to determine the response of T. tabaci to cotton seedlings damaged by high and low densities of each herbivore.

3.2.6 Two spotted spider mites damage vs T. tabaci damage and the response of T. tabaci The relative attraction of T. tabaci to cotton seedlings damaged by 50 two spotted spider mite or 50 T. tabaci (damaged as previously described) was investigated in standard olfactometer tests using plants with herbivores in situ and with plants from which herbivores had been removed after 24 hrs feeding. 51

3.2.7 Statistical analysis The number of thrips attracted to the different treatments in paired olfactometer tests was compared by a χ² test using StatView (SAS Institute, 1999). Responses were converted to percentages for presentation.

3.3. Results

3.3.1 Feeding mode, cotton responses and T. tabaci behaviour Thrips tabaci was more attracted towards mechanically damaged plants than to undamaged plants (χ² = 26.6, d.f. = 1, P<0.001; Figure 3.3). Similarly, they were also more attracted to plants that had been damaged by conspecifics (T. tabaci in situ: χ² = 24.0, d.f. = 1. P<0.001; T. tabaci removed: χ² = 21.6, d.f. = 1, P<0.001; Figure 3.3) or by two spotted spider mites (mites in situ: χ² = 29.4, d.f. = 1, P<0.001; mites removed: χ² = 35.2, d.f. = 1, P<0.001) than to undamaged plants (Figure 3.3). However, they did not discriminate between undamaged plants and plants that had been damaged by other species of thrips (F. schultzei (removed): χ² = 0.26, d.f. = 1, P = 0.606; F. occidentalis (removed): χ² = 0.600, d.f. = 1, P = 0.439; Figure 3.3). Similarly, they did not discriminate between H. armigera damaged plants and undamaged plants, regardless of whether larvae were in situ (χ² = 0.600, d.f. = 1, P = 0.439) or had been removed (χ² = 1.067, d.f. = 1, P = 0.302), nor did they discriminate between undamaged plants and plants that had suffered root damage from mealworm feeding (χ² = 0.067, d.f. = 1, P = 0.796) (Figure 3.3).

3.3.2. Degree of foliage damage and T. tabaci response The degree of damage inflicted on cotton seedlings significantly affected the level of attractiveness to T. tabaci. Significantly more T. tabaci were attracted to plants damaged by 50 conspecifics than to plants damaged by 10 conspecifics (χ² = 24.0, d.f. = 1, P<0.001; Figure 3.4) and to plants damaged by 50 two spotted spider mites than to plants damaged by 10 two spotted spider mites (χ² = 21.6, d.f. = 1, P<0.001; Figure 3.4).

3.3.3. Two spotted spider mite damage vs T. tabaci damage and the response of T. tabaci Significantly more T. tabaci were attracted to two spotted spider mite damaged plants than to plants damaged by conspecifics, regardless of whether the arthropods were present or had been removed (all arthropods in situ: χ² = 24.0, d.f. = 1, P<0.001; all arthropods removed: χ² = 19.2, d.f. = 1, P<0.001; Figure 3.5).

Damaged Undamaged

Foliar - mechanical damage ***

HelicoverpaH. armigera inin situsitu ns

HelicoverpaH. armigera removedremoved ns

Root damage ns

Two-spotted spider Mitesmites inin situsitu ***

Two-spotted spiderMites mites removedremoved ***

T.T. tabaci tabaci in in situ situ *** T. tabaci removed T. tabaci removed *** F. schultzei removed ns F. schultzei removed F. occidentalis removed ns F. occidentalis removed

100 80 60 40 20 0 20 40 60 80 100 % thrips responding

Plants damaged by 10 individuals Plants damaged by 50 individuals

T. tabaci T. tabaci ***

Two-spotted spider Mitesmites ***

100 80 60 40 20 0 20 40 60 80 100 % thrips responding

Two-spotted spider mite damaged T. tabaci damaged

AllAll arthropodsarthropods in in situ situ ***

All arthropods removed ***

100 80 60 40 20 0 20 40 60 80 100 % thrips responding

3.4. Discussion Thrips tabaci responded to cotton seedlings damaged by different arthropods in various ways. Specifically, they were more attracted to mechanically damaged plants, two spotted spider mite- damaged plants and plants damaged by conspecifics than to intact control plants (Figure 3.3). However, T. tabaci did not discriminate between control plants and those damaged by foliar feeding 54

H. armigera, F. occidentalis or F. schultzei or by root feeding by T. molitor (Figure 3.3). Attraction to plants infested by T. tabaci or two spotted spider mites increased with increased herbivore density (Figure 3.4) and plants infested by mites were significantly more attractive than plants infested by T. tabaci (Figure 3.5). Thus, as for F. schultzei and F. occidentalis (Figure 2.9, Chapter 2), T. tabaci was strongly attracted to seedlings damaged by its mite prey; this is consistent with field observations that show higher T. tabaci densities on two spotted spider mite infested cotton plants (Wilson et al., 1996). Previously, in Chapter 2, I demonstrated that the responses of F. occidentalis and F. schultzei to herbivore-damaged cotton seedlings were different from one another and that the two species of thrips often showed an opposite response to cotton seedlings that had been damaged in a particular manner. In this study, the responses of T. tabaci differed from those of both F. occidentalis and F. schultzei and they were often intermediate to the responses shown by the two Frankliniella species. For example, in dual choice tests between H. armigera damaged plants and undamaged plants F. schultzei was more attracted to H. armigera damaged plants and F. occidentalis was more attracted to undamaged plants (Chapter 2; Section 2.3.1, Figure 2.9). In this study, by contrast, T. tabaci showed no differential attraction to test plants (Figure 3.3). Based on the results of this and the previous study (Chapter 2), it can be concluded that damaged cotton seedlings induced different, species-specific behavioural responses in the three thrips species studied. The responses of T. tabaci to herbivore-induced plants were the same when plants were tested with herbivores in situ and when plants were tested following herbivore removal (Figure 3.3). Previously it was demonstrated that although F. schultzei and F. occidentalis were attracted to plants damaged by two spotted spider mites, they were not attracted to the odours of the arthropods themselves (Chapter 2). Based on this, it is likely that T. tabaci was also attracted to plant volatiles produced in response to herbivore feeding, rather than to volatiles produced by the herbivores themselves. The effect of root herbivory on T. tabaci behaviour has not been investigated previously, indeed, this and the previous study (Chapter 2) are the first to describe the behavioural responses of thrips to plants subject to below ground damage. Thrips tabaci did not discriminate between root damaged and undamaged plants (Figure 3.3), unlike F. occidentalis which was more attracted to undamaged plants than root-damaged plants and F. schultzei which showed increased attraction towards root damaged plants (Chapter 2; Section 2.3.1, Figure 2.9). Studies on other systems have shown wide-ranging responses of insect herbivores to root herbivory. For example, root damage to Brassica nigra L. (Brassicales: Brassicaceae) by Delia radicum L. (Diptera: Anthomyiidae) significantly reduced abundance of Phyllotrera spp. (Coleoptera: Alticinae) and Brevicoryne brassicae L. (Hemiptera: Aphidae) on root-infested plants but Pieris rapae L. 55

(Lepidoptera: Pieridae) and Myzus persicae Sulzer (Hemiptera: Aphidae) did not discriminate between root infested and uninfested plants (Soler et al., 2009). The mechanisms driving these interactions are still unclear; however, root herbivory can induce defence compounds in the foliage of cotton (Bezemer et al., 2004) and other plants (Bezemer & van Dam, 2005) as well as changes in volatile blends emitted by above ground plant parts (Soler et al., 2007). These outcomes highlight the difficulty in developing sound generalizations about the responses of insects to host plants that have been subjected to arthropod herbivory. Further, they show that such responses are clearly system specific and that they do not necessarily result in the avoidance of herbivore damaged plants. Indeed, the only consistent response across all thrips species was the attraction that they showed to cotton seedlings damaged by two spotted spider mite, a prey item for all three species investigated. Clearly the odours induced as a result of herbivory do not always result in reduced attraction to herbivore damaged plants and a recent review of published literature shows that herbivory resulted in reduced oviposition on damaged plants in only 54% of studies and reduced attractancy in only 41% of studies (Chapter 1; Table 1.1). Based on the results with thrips and the published studies reviewed, it can be argued that induced plant responses should not be viewed solely as plant defence mechanisms but rather they should be recognised as biochemical responses to herbivory and that the resultant volatiles are responded to in different ways by different species (Chapter 1). Clearly, the adaptive significance attributed to the biochemical changes induced by herbivory requires careful consideration, as do interpretations of the responses of the herbivores concerned. Further research is required to determine the volatile compounds, or combination of compounds, to which T. tabaci responds. Similar work is also required to improve understanding of the interactions between F. schultzei and F. occidentalis and their cotton host plants (Chapter 2). We do not yet know whether the cotton seedling responds differently to various forms of herbivore damage, whether T. tabaci, F. schultzei and F. occidentalis respond differently to the same cue, or whether the different species respond differently to different cues. What is surprising is the attraction of T. tabaci to cotton seedlings damaged by conspecifics, considering that they showed no differential attraction between undamaged cotton seedlings and those damaged by the other thrips tested, F. occidentalis or F. schultzei (Figure 3.3). Frankliniella schultzei produces species-specific attractive semiochemicals (Milne et al., 2002) and compounds with a similar function could mediate the attraction of T. tabaci to cotton seedlings infested with conspecifics. However, as T. tabaci- damaged cotton seedlings remained highly attractive to T. tabaci following the removal of herbivores (Figure 3.3) it is likely that the observed attraction is plant-mediated. This then raises the question of whether different thrips species induce different biochemical responses in cotton seedlings. Plant responses differ depending on the type of feeding damage 56 inflicted by different herbivores (Walling, 2000). Consistent quantitative differences have been demonstrated in cotton plant volatile profiles in response to feeding by Spodoptera frugiperda (Smith) (Lepidoptera: Noctuidae), Euschistus heros (Fabricius) (Hemiptera: Pentatomidae) and Anthonomus grandis (Boheman) (Coleoptera: Curculionidae), the differences were attributed to the different types of feeding damage (i.e. chewing vs. sucking) inflicted (Magalhães et al., 2012). Furthermore, quantitative differences in cotton volatile profiles have also been detected in plants infested with Heliothis virescens Fabricius (Lepidoptera: Noctuidae), S. exigua Hübner (Lepidoptera: Noctuidae) (Ngumbi et al., 2009) and H. zea Boddie (Lepidoptera: Noctuidae) (McCall et al., 1994), suggesting that even herbivores with similar modes of feeding can also induce different plant responses. However, as mouthparts of thrips are similar across all families (Lewis, 1997) it may be assumed that all three thrips species (T. tabaci, F. occidentalis or F. schultzei) have similar modes of feeding and so what exactly the thrips are doing and how similar feeding damage results in the cotton seedlings being induced in different ways remains unclear. Different insect species are known to produce different elicitor compounds in their saliva when feeding (Hilker & Meiners, 2010). Plant transcriptional responses can be specific to the elicitor compounds released by insect herbivores with similar modes of feeding (Frey et al., 2000), leading to different plant responses at feeding sites. For example, different transcriptional responses elicited in Nicotiana attenuata (Solanales: Solanaceae) by larvae of Manduca sexta L. (Lepidoptera: Sphingidae), H. virescens and S. exigua were correlated with the profile of larval elicitors in each species (Voeckel & Baldwin, 2004). Such different elicitors have yet to be demonstrated in different species of thrips, however, this presents one possible explanation for the different responses of cotton seedlings observed in response to feeding by different thrips species. Understanding the molecular and biochemical changes in plants and their perception by thrips is thus essential to fully understand these herbivore-host plant interactions; this is an exciting area of research that warrants additional study.

Chapter 4 Effects of single and dual species herbivory on the behavioural responses of three thrips species to cotton seedlings

A paper based on this chapter will be submitted to Entomologia Experimentalis et Applicata. The work has been reformatted and edited slightly, so that it is consistent with the rest of the thesis. For example, figures are now placed within the text, references are cited in a style consistent with the rest of the thesis and the bibliography has been incorporated into the thesis bibliography.

Abstract The responses of plants to arthropod herbivory are referred to as induced responses and such responses can ultimately impact the behaviours of herbivorous arthropods. However, the majority of studies on herbivore-induced plant responses have been mostly measured and interpreted from plant mediated effects caused by a single species of herbivore. Under natural conditions plants are often attacked by more than one herbivore species. The olfactory responses of three thrips species Frankliniella schultzei Trybom, F. occidentalis Pergrande and Thrips tabaci Lindeman (Thysanoptera: Thripidae) to cotton seedlings (Gossypium hirsutum L. (Malvales: Malvaceae)) simultaneously damaged by different combinations of herbivore species was investigated. Herbivores used to damage cotton seedlings were foliar feeding Tetranychus urticae Koch (Trombidiforms: Tetranychidae), Helicoverpa armigera Hübner (Lepidoptera: Noctuidae), Aphis gossypii Glover (Hemiptera: Aphididae), Bemisia tabaci B-Biotype Gennadius (Hemiptera: Aleyrodidae) and Trialeurodes vaporariorum Westwood (Hemiptera: Aleyrodidae) and root feeding Tenebrio molitor L. (Coleoptera: Tenebrionidae). Olfactometer experiments showed that F. occidentalis was more attracted to plants damaged by T. urticae or F. schultzei than to undamaged plants but that F. occidentalis was more attracted to undamaged plants than to plants damaged by either H. armigera, T. molitor or A. gossypii. Further, when the responses to plants damaged by two herbivores were investigated, herbivory by H. armigera, T. molitor or A. gossypii reduced attraction to plants that were simultaneously damaged by T. urticae. Thus, feeding by these herbivores negatively affected the location of T. urticae prey by F. occidentalis. Frankliniella schultzei was attracted to plants damaged by all herbivore species investigated except A. gossypii, B. tabaci and T. vaporariorum. These thrips did not discriminate between undamaged plants and plants damaged by these herbivores, and damage caused by these species had no negative effect on attraction to plants that were simultaneously damaged by other herbivores. Thrips tabaci did not discriminate between undamaged plants or plants damaged by H. armigera, T. molitor, B. tabaci or T. vaporariorum but T. tabaci was more attracted to plants damaged by T. urticae than to undamaged plants. However when the responses of T. tabaci to plants simultaneously damaged by two herbivores were investigated, these thrips were more attracted to, plants damaged by T. urticae and H. armigera, T. molitor, B. tabaci or T. vaporariorum than to undamaged plants. The work demonstrates that the responses of these three species of thrips to plants dually infested by different combinations of target herbivores were predictable based on their responses to plants damaged by single species alone and that infestation of cotton plants with more than one species of herbivore could disrupt the ability of F. occidentalis to locate its T. urticae prey.

Keywords: Thysanoptera, Frankliniella schultzei, Frankliniella occidentalis, Thrips tabaci, induced responses, simultaneous damage.

4.1 Introduction The responses of plants to arthropod herbivory are referred to as induced responses (Karban & Baldwin, 1997) and they are based upon the synthesis and emission of herbivore-induced plant volatiles (HIPVs) (Karban & Baldwin, 1997). These HIPVs can ultimately impact the behaviours of herbivorous arthropods (Agrawal, 1999; De Moraes et al., 2001; Karban, 2011). Many studies have examined the effects of induced plant responses on the behaviour of herbivores and their natural enemies and they have been shown to suppress herbivore feeding (Anderson & Agrell, 2005; Miyazaki et al., 2013), reduce herbivore attraction to and oviposition on induced plants (Bruinsma et al., 2007; Abe et al., 2009; Delphia et al., 2007; Bernasconi et al., 1998; De Moraes et al., 2001; Landolt, 1993) and attract herbivore natural enemies (Dicke et al., 1990; Turlings et al., 1990; Shiojiri et al., 2001; Vos et al., 2001; Dicke & Hilker, 2003). Consequently, these plant responses to herbivory have frequently been interpreted as plant defences against further herbivore attack (Karban & Myers, 1989; Karban & Baldwin, 1997; Agrawal & Sherriffs, 2001; Dicke & Baldwin, 2010). However studies have also documented increased herbivore attraction to and oviposition on induced plants (Silva et al., 2012; Rojas, 1999; Landolt, 1993; Lu et al., 2004), highlighting the fact that the ecological function and significance of these induced plant responses remains to be proven (Hare, 2011; Takabayashi & Dicke, 1996). To date, studies on herbivore-induced plant responses have been mostly measured and interpreted from plant mediated effects caused by a single species of herbivore whereas, under natural conditions, plants are often infested and simultaneously fed upon by more than one herbivore species (Thompson, 1998; Strauss, 1991; Vos et al., 2001). Plant responses to simultaneous feeding by multiple herbivore species and the impact that these may have on the behavioural responses of arthropod herbivores are poorly understood. A review of the literature examining the effects of herbivore-induced plant responses on the behavioural responses of herbivorous arthropods showed that out of a total of 147 studies, only 10 investigated the effects of multiple herbivory (Figure 4.1). Among the studies that have investigated the effects of multiple herbivory, the vast majority examined how the HIPV profiles of plants damaged by a single herbivore species differ from those of plants simultaneously damaged by more than one species of herbivore (Figure 4.1). Few studies have investigated the behavioural responses of herbivorous arthropods to plants simultaneously damaged by more than one species of herbivore (Figure 4.1).

1 CF + CF 10 3 behavioral responses 2 4 SF + CF 137 Volatile analysis SF + SF 7 CR + CF

3 10

Single species 3 behavioral herbivory responses Multiple species Volatile analysis

herbivoryherbivory species Multiple herbivory species Single 7

Figure 4.1 Studies that have investigated the effects of herbivory on the behavioural responses 137 (ovipositional and attractancy) of herbivorous arthropods. From a total of 147 published studies abstracted in Thomson Reuters Web of Science® between 1990 and 2013, only 10 investigated the effects of multiple herbivory on plants. Of these 10 studies, only 3 investigated the effects of plant responses on herbivore behaviour. The 10 studies that looked at the effects of multiple herbivory, herbivores with different modes of feeding were used to damage plants: herbivore chewing on foliage (CF), herbivore sucking on foliage (SF) and herbivore chewing on roots (CR).

Previous research has demonstrated that plant responses to feeding by a single herbivore species differ from responses to attack by multiple species of herbivores (Shiojiri et al., 2001; Rodriguez-Saona et al., 2003). Studies that have analyzed the HIPV profiles of plants simultaneously fed upon by more than one species of herbivore have shown that HIPV blends differ both quantitatively and qualitatively when compared to the HIPV blends induced by single herbivore species (Delphia et al., 2007; Rodriguez-Saona et al., 2003; Moayeri et al., 2007). For example, quantitative differences were detected in the HIPV profiles of Nicotiana tabacum plants simultaneously feed upon by F. occidentalis and Heliothis virescens Fabricius (Lepidoptera: Noctuidae) compared to plants fed upon by each species alone (Delphia et al., 2007). Qualitative differences in HIPV profiles induced by multiple herbivory have also been reported; feeding by two species of Lepidoptera, Plutella xylostella L. (Lepidoptera: Plutellidae) and Pieris rapae L. (Lepidoptera: Pieridae), caused cabbage plants to emit different HIPV blends from those emitted in response to feeding by either species alone (Shiojiri et al., 2001). This clearly demonstrates that even herbivores with similar modes of feeding can induce different responses in the same plant. Studies on herbivore-induced plant responses caused by single herbivore species have shown that different modes of feeding induce different plant metabolic systems from one another. For example, chewing arthropods activate the jasmonic acid-mediated signalling pathway whereas feeding by

62 phloem-sucking arthropods activates the salicylic acid pathway (Walling, 2000). If plants are challenged with one or more additional species of herbivores, these metabolic pathways may be altered (Fidantsef et al., 1999; Thaler et al., 2002). The HIPV analyses that have been conducted have confirmed the complexity of plant responses to multiple herbivory (Delphia et al., 2007; Rodriguez-Saona et al., 2003), but few studies have explored how these complex plant responses can affect the behavioural responses of herbivorous arthropods. Those studies that have investigated these behavioural responses have, not surprisingly, shown both positive and negative effects on the subsequent preference and performance of herbivores. For example, Rodriguez-Saona et al. (2003) reported that herbivory on Lycopersicon esculentum by Spodoptera exigua Hübner (Lepidoptera: Noctuidae) and Macrosiphum euphorbiae Thomas (Hemiptera: Aphididae) affected the ovipositional responses of S. exigua. Compared to undamaged plants, S. exigua laid fewer eggs on plants damaged by conspecifics but laid more eggs on plants damaged by M. euphorbiae. However, when plants were simultaneously damaged by both S. exigua larvae and M. euphorbiae, the number of eggs laid on undamaged and damaged plants did not differ (Rodriguez-Saona et al., 2003). This suggests that the discrimination of female moths between undamaged and conspecific-damaged plants was reduced in the presence of aphids. Thus the responses of herbivorous arthropods to plants damaged by multiple herbivore species can be quite different from the responses to plants damaged by single species of herbivores. In chapters 2 and 3, I demonstrated that the behavioural responses of three thrips species, F. occidentalis, F. schultzei and T. tabaci, to cotton seedlings induced by different herbivores varied enormously (Chapters 2 & 3). Thrips responses were both context dependent and unpredictable: F. schultzei was more attracted to cotton seedlings damaged by various arthropod pests (cotton bollworm, two-spotted spider mites and root feeding mealworms) than to intact seedlings, whereas F. occidentalis was more attracted to intact seedlings with the exception seedlings damaged by two- spotted spider mites. Similarly T. tabaci was attracted to seedlings damaged by two-spotted spider mites and showed no preference for intact plants or those damaged by the other herbivores investigated (Chapters 2 & 3). Given the unpredictability and specificity in the responses of thrips to cotton seedlings induced by single species of herbivore, I tested the olfactory responses of F. occidentalis, F. schultzei, T. tabaci to cotton seedlings damaged by different combinations of arthropod herbivores to determine how multiple herbivory might affect the responses of thrips to herbivore damaged plants.

4.2. Materials and methods

4.2.1. Literature search The Science Citation Index Expanded in the Thomson Reuters Web of Science® was used. The search terms “induced plant responses”, “oviposition” and “attraction” were used to search journal articles in the database that were published in English between 1990 and 2013. Output was scrutinized for studies which investigated the effects of host plant induction (by multiple herbivore feeding, single herbivore feeding or by the application of elicitors or mechanical damage).

4.2.2. Plants Cotton plants (G. hirsutum var. Sicott 71RRF) were grown from seed to seedling in individual pots (11 cm diameter) containing organic potting mix [Osmocote (N:P:K, 16:35:10); Scots Australia, Baulkham Hills, New South Wales]. Seedlings were watered daily and raised in a ventilated glasshouse under ambient conditions at The University of Queensland. Seedlings were raised to the two-leaf stage (BBCH code 12 (Munger et al., 1998)) (20 days after germination) for use in experiments.

4.2.3 Arthropods Thrips Laboratory populations of F. occidentalis, F. schultzei and Thrips tabaci (Figure 4.2) used in the study were established with individuals collected from the field. Frankliniella occidentalis were either collected from clover Trifolium repens L. (Fabales: Fabaceae), at the Gatton Research Station, Gatton, Queensland (27°32'31.6"S 152°19'45.9"E), or from cotton (G. hirsutum) flowers at the Australian Cotton Research Institute at Myall Vale near Narrabri, NSW (30°12'20.1"S 149°35'44.2"E), where T. tabaci were also collected from Ammi majus L. (Apiales: Apiaceae) flowers. Frankliniella schultzei were collected from Malvaviscus arboreus Cav. (Malvales: Malvacea) flowers at the St Lucia campus of The University of Queensland, in Brisbane, Queensland, Australia (27°29'43.9"S 153°00'38.2"E). Insects were reared in glass jars (17 cm in height, 8 cm diameter) (Figure 2.3, Chapter 2) held in an incubator (25 ± 2ºC, L12:D12) (light intensity 180µE). Green beans (Phaseolus vulgaris, cv. Stringless) were soaked in water for 16 h and then washed in warm soapy water to remove any insecticide residue before being provided as an adult food source and oviposition substrate. Cleaned beans were placed in glass jars lined with paper towel and female thrips were released to feed and oviposit. Glass jars were sealed with two layers of nylon mesh (22 × 22 × 22 cm, ≈ 1 mm2) and a black cardboard layer held between them to prevent thrips from escaping (Figure 2.3, Chapter 2). The green beans were left in the glass jars for 64 three days, after which time they were removed and replaced with new ones; old beans were placed in new glass jars lined with paper towel for larval rearing. Pollen collected from Hibiscus spp. (Malvales: Malvaceae) flowers, was also provided as a food supplement for both adult and larval stages of F. occidentalis and F. schultzei. All experiments were conducted on laboratory reared thrips 3-5 generations after field collection.

(A) (B) (C)

Figure 4.2 Thrips species investigated. (A) Frankliniella occidentalis. (B) Thrips tabaci. (C) F. schultzei.

Tetranychus urticae (Two spotted spider mites) To provide mites (Figure 4.3A) for use in experiments a culture was maintained on cotton seedlings (two-leaf stage) that originated from adults collected from cotton seedlings at the Australian Cotton Research Institute, Myall Vale, Narrabri, New South Wales, Australia. Mites were reared at 26 ± 2ºC (L12:D12) on 20 cotton seedlings in pots (20 cm diameter) in nylon mesh cages (45 × 45 × 45 cm). Cotton seedlings bearing mites were maintained in the incubator for five days after which leaves bearing mites and eggs were cut and placed on fresh cotton seedlings.

Helicoverpa armigera (cotton bollworm) Second instar H. armigera (Figure 4.3D) used in experiments were obtained from a laboratory culture that originated from eggs collected from various crops at Toowomba, Queensland, Australia. Helicoverpa armigera were reared on a soybean-based artificial diet (Teackle, 1991).

Tenebrio molitor (mealworms)

Mealworm larvae (Figure 4.3C) were purchased from a commercial supplier (Pisces Enterprises Pty Ltd., Brisbane, Australia). Larvae were held in plastic containers (6 cm height, 15 cm diameter) that contained vermiculite as a burrowing substrate. Larvae were supplied with small pieces of carrot as a food source but were starved for 24 h prior to use in experiments.

Aphis gossypii (cotton aphids) Cotton aphids (Figure 4.3B) used in experiments were obtained from a laboratory culture that originated from individuals collected from cotton plants at the Australian Cotton Research Institute, Myall Vale, Narrabri, New South Wales, Australia. Cotton aphids were reared at 26 ± 2ºC (L12:D12) on six leaf stage cotton plants in pots (20 cm diameter) in nylon mesh cages cages (45 × 45 × 45 cm).

(A) (B)

Bemisia tabaci (silverleaf whitefly) and Trialeurodes vaporariorum (greenhouse whitefly) Two species of whiteflies were used in experiments, silverleaf whitefly (Bemisia tabaci B-Biotype) (Figure 4.4A) and greenhouse whitefly (Trialeurodes vaporariorum) (Figure 4.4B). The two species of whiteflies were obtained from independent laboratory cultures that originated from individuals collected from cotton plants at the Australian Cotton Research Institute Myall Vale, Narrabri, New South Wales. Whiteflies were reared in the same manner as previously described for cotton aphids.

(A) (B)

Figure 4.4 Whitefly species used in experiments. (A) Adult silverleaf whitefly (Bemisia tabaci B- Biotype), note the absence of hairs on the silverleaf whitefly nymphs and adults have a split between the wings. (B) Adult greenhouse whitefly (Trialeurodes vaporariorum). Hairs are present on nymphs and the wings of adults overlap.

4.2.4 Standard olfactometer tests A Y-tube olfactometer was used to compare the responses of F. occidentalis, F. schultzei or T. tabaci to undamaged cotton seedlings and cotton seedlings either damaged by single herbivores species or seedlings damaged by dual herbivores infestations. The olfactometer set up consisted of a glass Y-tube (9 cm in length stem) and two side arms (9 cm in length, 0.9 cm diameter, 45º apart) that were connected to two sealed glass chambers (Figure 2.4, Chapter 2). A vacuum line was used 67 to draw air through the system at a rate of 1 L min-1. Before entering the olfactometer, air passed through activated charcoal filters attached to the glass chambers. Experiments were conducted in a temperature controlled room at 25ºC (± 2ºC), and a white light source (Osram L 15W) was placed directly above the Y junction of the olfactometer to avoid bias. Single adult female thrips were released at the stem entrance of the Y-tube and observed for 10 min. When a test thrips moved more than half way (4.5 cm) into one of the arms and remained there for more than 20 s, it was recorded as having been attracted to the odour source associated with that arm (Figure 2.4, Chapter 2). After testing five individuals, the chambers with plants were rotated and the Y- tubes cleaned with 100% ethanol and dried. After 10 thrips individuals had been screened, the test plants were changed. A minimum of 60 individual thrips was used in each test. The same method was used in all olfactometer experiments.

4.2.5 Standard method for damaging cotton seedlings Cotton seedlings were damaged by H. armigera larvae, female adult T. urticae, adult female thrips (F. occidentalis, F. schultzei or T. tabaci), adult cotton aphids (A. gossypii), adult whiteflies (B. tabaci or T. vaporariorum) and root feeding T. molitor larvae. Plants were either damaged by one species of herbivore alone, or they were damaged simultaneously by various combinations of the herbivores. Inverted plastic cups (9.7 cm tall, 8 cm diameter) that contained two windows (3.5×4.5 cm) covered with nylon mesh (≈1 mm² mesh) (Figure 2.5, Chapter 2) were used to enclose arthropod herbivores once they had been introduced onto the cotton seedlings. Cups were also placed over undamaged control plants but no arthropods were introduced. Arthropods were allowed to feed for 24 h and then cups were removed from damaged and control plants before they were used in olfactometer experiments.

4.2.6 Responses of thrips to T. urticae and H. armigera damage

Tetranychus urticae damage. Ten, 25 or 50 adult mites were independently introduced to cotton seedlings and enclosed as previously described. Mites were allowed to feed for 24 h, after which time standard olfactometer tests were performed to test the responses of the three test species of thrips.

Helicoverpa armigera damage. Two or 10 second instar larvae were allowed to feed on cotton foliage. Larvae were introduced onto the adaxial leaf surface and enclosed as previously described. Larvae were allowed to feed on foliage for 24 h after which time standard olfactometer experiments were performed. 68

Dual infestations. Ten, 25 or 50 adult T. urticae were independently introduced onto leaves of cotton seedlings and 10 H. armigera larvae were simultaneously introduced onto the mite infested plants (Figure 4.5). The herbivores were enclosed on the dually infested plants as previously described and allowed to feed for 24 h, after which time standard olfactometer tests were performed to test the responses of each of the three test species of thrips.

10 T. urticae + 10 25 T. urticae + 10 50 T. urticae + 10 H. armigera H. armigera H. armigera

4.2.7 Responses of thrips to Aphis gossypii and T. urticae damage

Aphis gossypii damage. Ten, 25 or 50 adult aphids were independently introduced onto leaves of cotton seedlings and enclosed on plants as previously described. Aphids were allowed to feed for 24 h after which time standard olfactometer tests were performed to test the responses of each of the three test species of thrips.

Dual infestations. Plants were simultaneously infested with 10 adult A. gossypii and 10 T. urticae, 25 adult A. gossypii and 25 T. urticae or 50 adult A. gossypii and 50 T. urticae; arthropods were simultaneously introduced onto leaves of cotton seedlings (Figure 4.6) and enclosed on plants as previously described. The herbivores were allowed to feed on the dually infested seedlings for 24 h, after which standard olfactometer tests were performed to test the responses of each of the three test species of thrips.

10 T. urticae + 10 A. gossypii 25 T. urticae + 25 A. gossypii 50 T. urticae + 50 A. gossypii

4.2.8 Responses of thrips to thrips and H. armigera damage

Thrips damage. F. occidentalis, F. schultzei and T. tabaci were used independently to damage cotton foliage. Ten, 25 or 50 adult thrips of each species were independently introduced onto the leaves of separate cotton seedlings and enclosed on plants as previously described. Thrips were

70 allowed to feed on seedlings for 24 h, after which time standard olfactometer tests were performed to test the responses of the three test species of thrips to damage caused by conspecifics and the two other test species of thrips. Thrips-thrips combinations were used that showed positive attraction based on results from Chapter 2 and 3.

Dual infestations. Ten, 25 or 50 of a given thrips species and 10 H. armigera were simultaneously introduced onto the leaves of cotton seedlings (Figure 4.7) and enclosed on plants as previously described. Herbivores were allowed to feed on seedlings for 24h, after which time standard olfactometer tests were performed to test the responses of the three test species of thrips to the various combinations.

10 Thrips spp + 10 25 Thrips spp + 10 50 Thrips spp + 10 H. armigera H. armigera H. armigera

4.2.9 Responses of thrips to whitefly and H. armigera damage

Whitefly damage. Ten, 25 or 50 adult B. tabaci or T. vaporariorum were independently introduced onto leaves of cotton seedlings and enclosed on plants as previously described. Whiteflies were allowed to feed on seedlings for 24 h, after which time standard olfactometer tests were performed to test the responses of the three test species of thrips.

Dual infestations. Ten, 25 or 50 whiteflies (B. tabaci or T. vaporariorum, used independently) and 10 H. armigera larvae were simultaneously introduced onto the leaves of cotton seedlings (Figure 4.8) and enclosed on plants as previously described. Herbivores were allowed to feed on seedlings for 24 h, after which time standard olfactometer tests were performed to test the responses of the three test species of thrips.

10 whitefly spp + 25 whitefly spp + 50 whitefly spp + 10 H. armigera 10 H. armigera 10 H. armigera

Figure 4.8 Schematic representation of the herbivore complexes used to damage cotton seedlings. Cotton seedlings were damaged by different densities of each whitefly species (B. tabaci or T. vaporariorum; n = 10, 25 or 50 per plant) and a single density of Helicoverpa armigera (n=10 per plant).

4.2.10 Responses of thrips to Tenebrio molitor and T. urticae damage

Mealworm (Tenebrio molitor) damage. Root damage was inflicted using late instar mealworm (T. molitor) larvae. Ten larvae were placed on the soil surface and allowed to burrow into the soil and feed on the seedling roots. The responses of thrips to root damaged and undamaged above ground foliage was measured in standard olfactometer experiments 48 h after release of larvae onto the soil surface. The degree of root damage in cotton seedlings was measured upon completion of the olfactometer experiments. Undamaged and damaged cotton seedlings were removed from the pots and the mealworms extracted from the soil surrounding damaged roots. Individual plants were placed into plastic boxes (45×25×10 cm) and the soil washed from the roots using slow running water (Figure 2.7, Chapter 2). The root mass was removed from the stem and all broken and dislodged pieces of root in the soil were collected separately and stored separately. Root material was then dried in an oven at 60ºC for 24h and root dry weights determined.

Dual infestations. Ten, 25 or 50 adult T. urticae were independently introduced onto leaves and 10 mealworm larvae were simultaneously placed on the soil surface and allowed to feed on the seedling roots. The herbivores were then enclosed on plants as previously described and allowed to feed on seedlings for 24 h, after which time standard olfactometer tests were performed to test the responses of the three test species of thrips.

4.2.11 Responses of thrips to single vs. dual herbivore damage The relative attraction of thrips to plants damaged by a single species of herbivore and plants damaged by two species was investigated using cotton seedlings that had either been damaged by 50 T. urticae or by 50 T. urticae and 10 H. armigera (damaged as previously described) in standard olfactometer tests to test the responses of the three test species of thrips.

4.2.12 Statistical analysis The number of thrips attracted to the different treatments in paired olfactometer tests was compared by a χ² test using StatView (SAS Institute, 1999). Responses were converted to percentages for presentation.

4.3 Results

4.3.1 Responses of thrips to T. urticae and H. armigera damage Frankliniella occidentalis was more attracted to plants damaged by 10, 25 or 50 T. urticae than to undamaged plants (χ2 = 4.27, P = 0.039, χ2 = 11.27, P<0.001, χ2 = 35.27, P<0.001; all d.f. = 1; Figure 4.9). However, they were more attracted to undamaged plants than to plants damaged by ten H. armigera (χ2 = 32.27, d.f. = 1, P<0.001; Figure 4.9) and did not discriminate between undamaged plants and plants that had been damaged by two H. armigera (χ2 = 0.27, d.f. = 1, P = 0.606; Figure 4.9). Herbivory by 10 T. urticae and 10 H. armigera resulted in F. occidentalis being more attracted to undamaged plants than to plants that were simultaneously damaged by both herbivores (χ2 = 29.40, d.f. = 1, P<0.001; Figure 4.9) but F. occidentalis did not discriminate between undamaged plants and plants simultaneously damaged by 25 T. urticae and 10 H. armigera (χ2 = 2.40, d.f. = 1, P = 0.121; Figure 4.9). However F. occidentalis was more attracted to plants simultaneously damaged by 50 T. urticae and 10 H. armigera than to undamaged plants (χ2 = 26.67, d.f. = 1, P<0.001; Figure 4.9). Thrips tabaci was more attracted to plants damaged by 25 or 50 T. urticae (χ2 = 6.67, P = 0.001; χ2 = 35.27, P<0.001; all d.f. = 1; Figure 4.9) but they did not discriminate between undamaged plants and plants that had been damaged by 10 T. urticae (χ2 = 0.60, d.f. = 1, P = 0.44; Figure 4.9). They did not discriminate between undamaged plants and plants that had been damaged by two (χ2 = 0.60, d.f. = 1, P = 0.44,) or 10 H. armigera larvae (χ2 = 0.267, d.f. = 1, P = 0.61) (Figure 4.9). Simultaneous herbivory by 25 T. urticae and 10 H. armigera or 50 T. urticae and 10 H. armigera resulted in T. tabaci being more attracted to damaged plants than to undamaged plants (χ2 = 21.60, χ2 = 29.40, all d.f. = 1, P<0.001; Figure 4.9). However, they did not discriminate between undamaged plants and plants simultaneously damaged by 10 T. urticae and 10 H. armigera (χ2 = 1.67, d.f. = 1, P = 0.197; Figure 4.9). Frankliniella schultzei was more attracted to plants damaged by ten, 25 or 50 T. urticae than to undamaged plants (χ2 = 8.07, P = 0.005; χ2 = 21.60, P<0.001, χ2 = 38.40, all d.f. = 1, P<0.001; Figure 4.9). Similarly, F. schultzei was also more attracted to plants damaged by 10 H. armigera larvae than to undamaged plants (χ2 = 26.67, d.f. = 1, P<0.001; Figure 4.9) but F. schultzei did not discriminate between undamaged plants and plants damaged by two H. armigera (χ2 = 0.60, d.f. = 1, P = 0.439; Figure 4.9). Similarly F. schultzei was more attracted to plants simultaneously damaged by either 10, 25 or 50 T. urticae and 10 H. armigera than to undamaged plants (χ2 = 13.07, P<0.001; χ2 = 24.07, P<0.001, χ2 = 45.07, P<0.001; all d.f. = 1; Figure 4.9).

Damaged Undamaged Frankliniella occidentalis 10 T. urticae *** 25 T. urticae *** 50 T. urticae *** 2 H. armigera NS 10 H. armigera *** 10 T. urticae + 10 H. armigera *** 25 T. urticae + 10 H. armigera NS 50 T. urticae + 10 H. armigera *** Thrips tabaci 10 T. urticae NS 25 T. urticae *** 50 T. urticae *** 2 H. armigera NS 10 H. armigera NS 10 T. urticae + 10 H. armigera NS 25 T. urticae + 10 H. armigera *** 50 T. urticae + 10 H. armigera *** Frankliniella schultzei 10 T. urticae *** 25 T. urticae *** 50 T. urticae *** 2 H. armigera NS 10 H. armigera *** 10 T. urticae + 10 H. armigera *** 25 T. urticae + 10 H. armigera *** 50 T. urticae + 10 H. armigera *** 100 80 60 40 20 0 20 40 60 80 100 % thrips responding

4.3.2 Responses of thrips to A. gossypii and T. urticae damage Frankliniella occidentalis was more attracted to plants damaged by 10 A. gossypii than to undamaged plants (χ2 = 26.7, d.f. = 1, P<0.001; Figure 4.10) but they were more attracted to undamaged plants than to plants damaged by 25 or 50 A. gossypii (χ2 = 15.00, P<0.001, χ2 = 35.27, P<0.001; all d.f. = 1; Figure 4.10). Similarly, following simultaneous feeding by 10 T. urticae and 10 A. gossypii, F. occidentalis was more attracted to damaged plants than to undamaged plants (χ2 = 29.40, d.f. = 1, P<0.001; Figure 4.10). However they did not discriminate between undamaged plants and plants simultaneously damaged by 25 T. urticae and 25 A. gossypii (χ2 = 1.67, d.f. = 1, P = 0.197; Figure 4.10). Thrips tabaci did not discriminate between undamaged plants and plants damaged by 10 A. gossypii (χ2 = 0.07, d.f. = 1, P = 0.796; Figure 4.10). However, they were more attracted to undamaged plants than to plants damaged by 25 or 50 A. gossypii (χ2 = 15.00, P = 0.001; χ2 = 41.67, P<0.001; all d.f. = 1; Figure 4.10). Following simultaneous feeding by 10 T. urticae and 10 A. gossypii, 25 T. urticae and 25 A. gossypii or 50 T. urticae and 50 A. gossypii, T. tabaci did not discriminate between damaged and undamaged plants (χ2 = 0.60, P = 0.437, χ2 = 0.067, P = 0.796, χ2 = 0.27, P = 0.606 respectively; all d.f. = 1; Figure 4.10). Frankliniella schultzei did not discriminate between undamaged plants and plants damaged by 10, 25 or 50 A. gossypii (χ2 = 3.27, P = 0.071; χ2 = 2.40, P = 0.121; χ2 = 4.27, P = 0.039; all d.f. = 1; Figure 4.10). Following simultaneous feeding by 10 T. urticae and 10 A. gossypii, 25 T. urticae and 25 A. gossypii or 50 T. urticae and 50 A. gossypii, F. schultzei being more attracted to damaged plants than undamaged plants (χ2 = 24.07, χ2 = 38.40, χ2 = 32.27 respectively; all d.f. = 1, P<0.001; Figure 4.10).

Damaged Undamaged Frankliniella occidentalis 10 A. gossypii *** 25 A. gossypii *** 50 A. gossypii *** 10 T. urticae + 10 A. gossypii *** 25 T. urticae + 25 A. gossypii NS 50 T. urticae + 50 A. gossypii *** Thrips tabaci 10 A. gossypii NS 25 A. gossypii *** 50 A. gossypii *** 10 T. urticae + 10 A. gossypii NS 25 T. urticae + 25 A. gossypii NS 50 T. urticae + 50 A. gossypii NS Frankliniella schultzei 10 A. gossypii NS 25 A. gossypii NS 50 A. gossypii NS 10 T. urticae + 10 A. gossypii *** 25 T. urticae + 25 A. gossypii *** 50 T. urticae + 50 A. gossypii *** 100 80 60 40 20 0 20 40 60 80 100 % thrips responding

4.3.3 Responses of thrips to thrips and H. armigera damage Frankliniella occidentalis did not discriminate between undamaged plants and plants damaged by 10 F. schultzei (χ2 = 0.27, d.f. = 1, P = 0.606; Figure 4.11) but they were more attracted to plants damaged by 25 or 50 F. schultzei than to undamaged plants (χ2 = 5.40, P = 0.020, χ2 = 29.40, P <0.001; all d.f. = 1; Figure 4.11). However, simultaneous feeding by 10 F. schultzei and 10 H. armigera resulted in greater attraction of F. occidentalis to undamaged plants than to damaged plants (χ2 = 26.67, d.f. = 1, P<0.001; Figure 4.11). Frankliniella occidentalis did not discriminate between undamaged plants and plants simultaneously damaged by 25 F. schultzei and 10 H. armigera (χ2 = 2.40, d.f. = 1, P = 0.121; Figure 4.11) but F. occidentalis was more attracted to plants simultaneously damaged by 50 F. schultzei and 10 H. armigera than to undamaged plants (χ2 = 24.07, d.f. = 1, P<0.001; Figure 4.11). Thrips tabaci did not discriminate between undamaged plants and plants damaged by 10 conspecifics (χ2 = 0.60, d.f. = 1, P = 0.439; Figure 4.11) but they were more attracted to plants damaged by 25 and 50 conspecifics than to undamaged plants (χ2 = 4.27, P = 0.039; χ2 =38.40, P<0.001; all d.f. = 1; Figure 4.11). Similarly T. tabaci did not discriminate between undamaged plants and plants simultaneously damaged by 10 conspecifics and 10 H. armigera (χ2 = 0.07, d.f. = 1, P = 0.796; Figure 4.11). However, T. tabaci was more attracted to plants simultaneously damaged by 25 conspecifics and 10 H. armigera or 50 conspecifics and 10 H. armigera than to undamaged plants (χ2 = 8.07, P = 0.005, χ2 = 26.67, P<0.001; all d.f. = 1; Figure 4.11). Frankliniella schultzei was more attracted to plants damaged by 10, 25 or 50 F. occidentalis than to undamaged plants (χ2 = 15.00, P = 0.001; χ2 = 24.07, P<0.001; χ2 = 26.67, P<0.001; all d.f. = 1; Figure 4.11). Similarly F. schultzei was more attracted to plants simultaneously damaged by 10, 25 or 50 F. occidentalis and 10 H. armigera than to undamaged plants (χ2 = 21.60, P<0.001; χ2 = 26.67, P<0.001; χ2 = 35.27, P<0.001; all d.f. = 1; Figure 4.11).

4.3.4 Responses of thrips to whitefly and H. armigera damage In all comparisons of plants damaged only by B. tabaci or T. vaporariorum and undamaged plants, no species of thrips discriminated between damaged and undamaged plants regardless of the density or species of whitefly tested (all test χ2<3.841, d.f. = 1, P>0.05; Figure 4.12). However, in experiments investigating the effects of dual infestations, F. occidentalis was more attracted to undamaged plants than to plants simultaneously damaged by 10, 25 or 50 silverleaf whitefly and 10 H. armigera (χ2 = 29.40, P<0.001; χ2 = 24.07, P<0.001; χ2 = 19.27, P<0.001; all d.f. = 1; Figure 4.12A). Similarly, F. occidentalis was also more attracted undamaged plants than to plants simultaneously damaged by 10, 25 or 50 greenhouse whitefly and 10 H. armigera (χ2 = 32.27, P<0.001; χ2 = 26.67, P<0.001; χ2 = 21.60, P<0.001; all d.f. = 1; Figure 4.12B).

Damaged Undamaged Frankliniella occidentalis 10 F. schultzei NS 25 F. schultzei *** 50 F. schultzei *** 10 F. schultzei + 10 H. armigera *** 25 F. schultzei + 10 H. armigera NS 50 F. schultzei + 10 H. armigera *** Thrips tabaci 10 T. tabaci NS 25 T. tabaci *** 50 T. tabaci *** 10 T. tabaci + 10 H. armigera NS 25 T. tabaci + 10 H. armigera *** 50 T. tabaci + 10 H. armigera *** Frankliniella schultzei 10 F. occidentalis *** 25 F. occidentalis *** 50 F. occidentalis *** 10 F. occidentalis + 10 H. armigera *** 25 F. occidentalis + 10 H. armigera *** 50 F. occidentalis + 10 H. armigera *** 100 80 60 40 20 0 20 40 60 80 100 % thrips responding

Thrips tabaci did not discriminate between undamaged plants and plants simultaneously damaged by 10, 25 or 50 silverleaf whitefly and 10 H. armigera (χ2 = 1.07, P = 0.302; χ2 = 1.07, P = 0.302; χ2 = 2.40, P = 0.121; all d.f. = 1; Figure 4.12A), nor did it discriminate between undamaged plants and plants simultaneously damaged by 10, 25 or 50 greenhouse whitefly and 10 H. armigera (χ2 = 3.27, P = 0.071; χ2 = 1.67, P = 0.197; χ2 = 0.07, P = 0.796; all d.f. = 1; Figure 4.12B). Frankliniella schultzei was more attracted to plants simultaneously damaged by 10, 25 or 50 silverleaf whitefly and 10 H. armigera than to undamaged plants (χ2 = 21.60, P<0.001; χ2 = 26.67, P<0.001; χ2 = 29.40, P<0.001; all d.f. = 1; Figure 4.12A) and it was also more attracted to plants simultaneously damaged by 10, 25 or 50 greenhouse whitefly and 10 H. armigera than to undamaged plants (χ2 = 19.27, P<0.001; χ2 = 24.07, P<0.001; χ2 = 41.67, P<0.001; all d.f. = 1; Figure 4.12B).

4.3.5 Responses of thrips to T. molitor and T. urticae damage Frankliniella occidentalis was more attracted to undamaged plants than to plants damaged by 10 T. molitor larvae (χ2 = 15.00, d.f. = 1, P<0.001; Figure 4.13). Similarly, herbivory by 10 T. urticae and 10 T. molitor resulted in F. occidentalis being more attracted to undamaged plants than to plants simultaneously damaged by both herbivores (χ2 = 11.27, d.f. = 1, P = 0.001; Figure 4.13) but they did not discriminate between undamaged plants and plants simultaneously damaged by 25 T. urticae and 10 T. molitor (χ2 = 1.07, d.f. = 1, P = 0.302; Figure 4.13). However, F. occidentalis was more attracted to plants simultaneously damaged by 50 T. urticae and 10 T. molitor than to undamaged plants (χ2 = 9.60, d.f. = 1, P = 0.001; Figure 4.13). Thrips tabaci did not discriminate between undamaged plants and plants that had been damaged by 10 T. molitor larvae (χ2 = 0.60, d.f. = 1, P = 0.439; Figure 4.13). Similarly, they did not discriminate between undamaged plants and plants simultaneously damaged by 10 T. urticae and 10 T. molitor larvae (χ2 = 0.07, d.f. = 1, P = 0.796; Figure 4.13). However, feeding by 25 T. urticae and 10 T. molitor or 50 T. urticae and 10 T. molitor resulted in T. tabaci being more attracted to damaged plants than to undamaged plants (χ2 = 9.60, P = 0.001, χ2 = 26.67, P<0.001; all d.f. = 1; Figure 4.13). Frankliniella schultzei was more attracted to plants damaged by 10 T. molitor larvae than to undamaged plants (χ2 = 29.40, d.f. = 1, P<0.001; Figure 4.13). Similarly, simultaneous herbivory by 10 T. urticae and 10 T. molitor, 25 T. urticae and 10 T. molitor or 50 T. urticae and 10 T. molitor also resulted in F. schultzei being more attracted to damaged plants than to undamaged plants (χ2 = 24.07, χ2 = 32.27, χ2 = 41.67, all P<0.001, d.f. = 1; Figure 4.13).

A Damaged Undamaged Frankliniella occidentalis 10 B. tabaci NS 25 B. tabaci NS 50 B. tabaci NS 10 B. tabaci + 10 H. armigera *** 25 B. tabaci + 10 H. armigera *** 50 B. tabaci + 10 H. armigera ***

Thrips tabaci 10 B. tabaci NS 25 B. tabaci NS 50 B. tabaci NS 10 B. tabaci + 10 H. armigera NS 25 B. tabaci + 10 H. armigera NS 50 B. tabaci + 10 H. armigera NS

Frankliniella schultzei 10 B. tabaci NS 25 B. tabaci NS 50 B. tabaci NS 10 B. tabaci + 10 H. armigera *** 25 B. tabaci + 10 H. armigera *** 50 B. tabaci + 10 H. armigera *** 100 80 60 40 20 0 20 40 60 80 100 % thrips responding

B Damaged Undamaged Frankliniella occidentalis 10 T. vaporariorum NS 25 T. vaporariorum NS 50 T. vaporariorum NS 10 T. vaporariorum + 10 H. armigera *** 25 T. vaporariorum + 10 H. armigera *** 50 T. vaporariorum + 10 H. armigera ***

Thrips tabaci 10 T. vaporariorum NS 25 T. vaporariorum NS 50 T. vaporariorum NS 10 T. vaporariorum + 10 H. armigera NS 25 T. vaporariorum + 10 H. armigera NS 50 T. vaporariorum + 10 H. armigera NS

Frankliniella schultzei 10 T. vaporariorum NS 25 T. vaporariorum NS 50 T. vaporariorum NS 10 T. vaporariorum + 10 H. armigera *** 25 T. vaporariorum + 10 H. armigera *** 50 T. vaporariorum + 10 H. armigera *** 100 80 60 40 20 0 20 40 60 80 100 % thrips responding

Figure 4.12 Olfactometer responses of Frankliniella occidentalis, Thrips tabaci and F. schultzei to undamaged cotton seedlings and seedlings damaged by either B. tabaci or T. vaporarium and cotton seedlings simultaneously damaged by B. tabaci and H. armigera or T. vaporariorum and H. armigera (n = 60 thrips per treatment in all cases). χ² tests: ns, not significant, ***P<0.001. 81

Frankliniella occidentalis Damaged Undamaged 10 T. molitor *** 10 T. urticae + 10 T. molitor *** 25 T. urticae + 10 T. molitor NS

50 T. urticae + 10 T. molitor *** Thrips tabaci 10 T. molitor NS 10 T. urticae + 10 T. molitor NS

25 T. urticae + 10 T. molitor *** 50 T. urticae + 10 T. molitor *** Frankliniella schultzei 10 T. molitor *** 10 T. urticae + 10 T. molitor *** 25 T. urticae + 10 T. molitor *** 50 T. urticae + 10 T. molitor *** 100 80 60 40 20 0 20 40 60 80 100 % thrips responding

4.3.6 Responses of thrips to single vs. dual herbivore damage When given a choice between plants damaged by single herbivores (50 T. urticae) and plants damaged by multiple herbivores (50 T. urticae and 10 H. armigera), F. occidentalis was more attracted to plants damaged by T. urticae alone (χ2 = 26.67, d.f. = 1, P<0.001; Figure 4.14), T. tabaci did not discriminate between single and dually damaged plants (χ2 = 0.60, d.f. = 1, P = 0.439; Figure 4.14) and F. schultzei was more attracted to plants damaged by T. urticae and H. armigera than to plants damaged by T. urticae alone (χ2 = 15.00, d.f. = 1, P = 0.001; Figure 4.14).

50 T. urticae 50 T. urticae + 10 H. armigera

F. occidentalis ***

T. tabaci NS

F. schultzei ***

100 80 60 40 20 0 20 40 60 80 100 % thrips responding

4.4 Discussion

4.4.1 Predictability of responses – a combination of additivity and amount of damage The main objective of this study was to investigate the olfactory responses of F. occidentalis, T. tabaci and F. schultzei to cotton seedlings simultaneously damaged by more than one species of herbivore. Olfactometer bioassays showed that the responses of the three species of thrips to plants dually infested by different combinations of target herbivores was additive based on their responses to plants damaged by single species infestations and the degree of damage inflicted on cotton seedlings. Frankliniella occidentalis was more attracted to plants damaged by T. urticae than to undamaged plants regardless of T. urticae density (Figure 4.9). Similarly, F. occidentalis was also attracted to plants damaged by 25 or 50 F. schultzei than to undamaged plants (Figure 4.11) or to plants damaged by 10 A. gossypii (Figure 4.10). But F. occidentalis was more attracted to

83 undamaged plants than to plants damaged by H. armigera, T. molitor larvae (Figures 4.9 & 4.13) or to plants damaged either 25 or 50 A. gossypii (Figure 4.10). Frankliniella occidentalis did not discriminate between undamaged plants and plants damaged by whitefly species (Figure 4.12). However, when the responses of F. occidentalis to plants infested by two herbivore species was investigated, F. occidentalis was not attracted to plants simultaneously damaged by low and intermediate infestations of T. urticae and H. armigera, T. molitor or A. gossypii (Figures 4.9, 4.13 & 4.10). Also, simultaneous damaged by H. armigera resulted in F. occidentalis being not attracted to plants damaged by low to intermediate infestations of F. schultzei (Figure 4.11). Frankliniella schultzei was attracted to plants damaged by most herbivore species (H. armigera, T. molitor, T. urticae and F. occidentalis) regardless of their type or density (Figures 4.9, 4.11 & 4.13) with the exception of the phloem feeding A. gossypii and whiteflies, where they showed no preference between damaged and undamaged plants (Figures 4.10 & 4.12). Predictably, when plants were simultaneously damaged by two herbivore species F. schultzei were more attracted to damaged plants than to undamaged plants (Figures 4.9, 4.10, 4.11, 4.12 & 4.13). Thrips tabaci was also attracted to plants damaged by 25 and 50 T. urticae (Figure 4.9) but they were not attracted to plants damaged by 25 and 50 A. gossypii (Figure 4.10). However T. tabaci did not discriminate between undamaged plants and plants damaged by H. armigera, T. molitor or whiteflies of either species (Figures 4.9, 4.12 & 4.13). The lack of discrimination between H. armigera damaged plants and undamaged plants, coupled with its strong attraction to T. urticae damaged plants, suggests that T. tabaci responds to plant damage in herbivore-specific ways. Thus the results presented in this study show that the responses of thrips to dual herbivory were generally predictable based on their responses to single species herbivory, including amount of damage and this is discussed in more detail below.

4.4.2 Additivity of damage Although only a few studies of multiple herbivory have been published, it has been suggested that knowledge of single attacker systems may not predict the response in a multiple attack (Dicke et al., 2009). However, what is apparent from the current study is that in some cases it is possible to predict herbivore responses to multiple herbivory based on their responses to plants damaged by a single herbivore species. Other studies have also shown similar effects of multiple herbivory (Rodriguez-Saona et al., 2003; Zhang et al., 2009; Vos et al., 2001; Moayeri et al., 2007). For example, Rodriguez-Saona et al. (2003) showed that compared to undamaged plants, S. exigua laid fewer eggs on L. esculentum, damaged by conspecifics, but laid more eggs on plants damaged by M. euphorbiae, however, when plants were simultaneously damaged by both S. exigua larvae and M. euphorbiae, the number of eggs laid on undamaged and damaged plants did not differ. This 84 suggests that the response was additive. The predatory mirid, Macrolophus caliginosus (Heteroptera: Miridae) was more attracted to sweet pepper plants (C. annuum L., cv. California Wonder) damaged by two-spotted spider mites (T. urticae) or by green peach aphids (Myzus persicae Sulzer (Hemiptera: Aphididae)) than to undamaged plants (Moayeri et al., 2006). However, the mirids were more attracted to plants that were simultaneously damaged by both herbivores than to plants damaged by either species alone (Moayeri et al., 2007). The current study also suggests the additive nature of the responses. For instance, F. schultzei was attracted to plants damaged by either T. urticae or H. armigera larvae (Figure 4.14), but was more strongly attracted to plants that were simultaneously damaged by both herbivore species than to plants damaged by only T. urticae (Figure 4.14). Thus the responses of some arthropods (carnivorous or herbivorous) to plants dually infested by different combinations of target herbivores can be predicted based on their responses to plants damaged by single species infestations.

4.4.3 Effect of degree of damage The results of the current study also show that in addition to multiple herbivory, the degree of damage inflicted on cotton seedlings was also an important factor in determining the responses of thrips. For example, in comparisons of dually damaged plants versus undamaged plants, F. occidentalis responses were dependent on the density of spider mites in those dual infestations during which density of H. armigera remained unchanged; at low densities of T. urticae (10) attraction was to undamaged plants, at moderate densities of T. urticae (25) attraction was unbiased, and at high densities of T. urticae (50) attraction was to dually damaged plants (Figure 4.9). Similarly, Zhang et al. (2009) also showed a similar effect of density on plants damaged by multiple herbivore species. The authors showed that as the number of whiteflies (B. tabaci) inflicting damage increased on plants also simultaneously damaged by T. urticae, the attraction of Phytoseiulus persimilis (Evans) (Acari: Phytoseiidae) decreased. The emission of HIPVs increases as the number of herbivores damaging the plant increases (Guerrieri et al., 1999; Girling et al., 2011; Sabelis & van de Baan, 1983; Maeda & Takabayashi, 2001). For example, headspace HIPV analysis revealed that the emission of HIPVs in Brassica oleracea increased as the density of P. xylostella larvae increased (Girling et al., 2011). Although Girling et al. (2011) did not investigate the effect of herbivore-density on the behavioural responses of herbivorous arthropods, the increase in HIPVs resulting from a higher density of P. xylostella larvae was correlated with a greater attractiveness of the parasitoid Cotesia vestalis (Hymenoptera: Braconidae) to infested plants.

4.4.4 Responses to arthropod feeding habits The current work also showed that the three species of thrips responded consistently to cotton seedlings damaged by the two species of whitefly (B. tabaci and T. vaporariorum) tested. Thrips species did not discriminate between undamaged plants and plants damaged by either whitefly species (Figure 4.13). Feeding by the silverleaf whitefly did not induce volatile production in cotton plants (Rodriguez-Saona et al., 2003). This could explain why none of the thrips species discriminated between undamaged plants and plants damaged by either of the whitefly species (Figure 4.13). Further support for this is provided by the fact that, in comparisons between undamaged plants and plants simultaneously damaged by H. armigera and whitefly species, the presence of whitefly damage in conjunction with H. armigera damage (Figure 4.13) did not alter the observed responses of thrips to plants damaged by H. armigera alone (Figure 4.9). In other words there is no evidence that whiteflies caused the production of any volatiles that influenced thrips behaviour. This once again shows the predictable and additive nature of the responses. Herbivory by H. armigera in the presence of whitefly elicited identical responses of thrips to those initiated by H. armigera alone. The inclusion of two phloem-feeding arthropods (whiteflies and A. gossypii) provides the opportunity to compare thrips responses to damage by different arthropods with the same feeding mode, as both whiteflies and A. gossypii are phloem-feeding arthropods (Moran & Thompson, 2001; Zarate et al., 2007; Zhang et al., 2009). Whereas whitefly damage has been shown not to induce volatile production in cotton, A. gossypii feeding on cotton does induce volatile production in cotton plants (Mahabaleshwar et al., 2011). This could explain why the three species of thrips showed a response to cotton seedlings damaged by A. gossypii but not to whitefly species (Figure 4.10). However what these results indicate is that even herbivores with similar modes of feeding can also induce different plant responses just as herbivores with different modes of feeding do. Others have also found that plant responses differ based on the type of feeding damage inflicted by a particular species of herbivore (Walling, 2000; Rodriguez-Saona et al., 2005; Rodriguez-Saona et al., 2003). For example, chewing caterpillars are known to elicit a different set of plant responses than phloem-sucking arthropods (Stout et al., 1998; Inbar et al., 1999; Walling, 2000; Bostock et al., 2001). Whitefly damage alone did not elicit a response from F. occidentalis and F. schultzei but simultaneous damage by H. armigera did elicit a response (Figure 4.13). These herbivore specific plant responses to feeding could be caused by a combination of factors resulting from herbivore specific physical damage and the type and amount of elicitors released during feeding (Hilker & Meiners, 2010). Thus understanding the molecular and biochemical changes in plants warrants further investigation.

4.4.5 Thrips as atypical herbivores All three thrips species have an additional aspect to their interaction with plants. Frankliniella schultzei, F. occidentalis and T. tabaci do not neatly fit into the ‘herbivore’ category as they are also facultative predators of two-spotted spider mites (Trichilo & Leigh 1986; Wilson et al., 1996; Milne & Walter, 1998a). Indeed, in cotton crops, F. occidentalis, F. schultzei and T. tabaci commonly occur on plant parts that are infested with their T. urticae prey (Wilson et al., 1996; Milne & Walter, 1998a; Agrawal & Kelin, 2000). Not only are thrips commonly associated with two-spotted spider mites but, as shown in Chapters 2, 3 and the current study, they were also attracted to cotton seedlings damaged by two-spotted spider mites. However, this interaction was affected by multiple species herbivory. For example, F. schultzei was more attracted to plants simultaneously damaged by T. urticae and H. armigera than to plants damaged only by T. urticae (Figure 4.14). In contrast, the attraction of F. occidentalis to plants damaged by T. urticae was disrupted by the presence of simultaneous damage by H. armigera, T. molitor or A. gossypii (Figures 4.9, 4.10 & 4.13). The presence of prey and non-prey herbivore species may decrease the detectability and reliability of the searching cues used by natural enemies and predators of herbivores (Vet & Dicke, 1992; De Boer et al., 2008; Moayeri et al., 2007). Thus the simultaneous presence of the two herbivores on the same plant weakens the prey–predator interaction between F. occidentalis or F. schultzei and T. urticae and such plants are considered to be enemy-free space (Jeffries & Lawton 1984; Holt & Lawton 1993; Feder, 1995; Shiojiri et al., 2001; Shiojiri et al., 2002; Shiojiri et al., 2010; Denno et al., 1990) for T. urticae. This study provides clear evidence that dual species herbivory affects the behavioural responses of herbivorous arthropods. Frequently the responses of plants to herbivory and its functional and evolutionary ecology have been interpreted and measured from damage caused by a single herbivore species. Attacks on plants by dual herbivore species are likely to be the norm in natural systems (Thompson, 1998; Strauss, 1991; Vos et al., 2001) thus the understanding and determining how arthropods respond to multiple herbivory is crucial and necessary and presents a more holistic picture of the way in which arthropod herbivores and plants interact. What the current study hasn’t shown is why dual species herbivory affected the responses of thrips in the way that it did; why one type of damage dimishes the response to another type of damage. It is possible that the induction of a specific plant metabolic system by a particular herbivore species be altered if the plant is simultaneously damaged by another herbivore species (Rodriguez-Saona et al., 2003). For instance it has been shown that the salicylic acid pathway (induced by phloem-feeding arthropods) may inhibit the jasmonic acid pathway (induced by leaf-chewing arthopods), and vice versa (Fidantsef et al., 1999; Thaler et al., 2002). However, understanding the molecular and biochemical changes in plants and their perception by thrips warrants further investigation. 87

Chapter 5 Herbivore induced plant responses: effect of the post-feeding interval

A paper based on this chapter has been accepted for publication in Insect Science, DOI: 10.1111/1744-7917.12252. The work has been reformatted and edited slightly, so that it is consistent with the rest of the thesis. For example, figures are now placed within the text, references are cited in a style consistent with the rest of the thesis and the bibliography has been incorporated into the thesis bibliography.

Abstract Temporal changes in the responses of three thrips species; Frankliniella schultzei Trybom, Frankliniella occidentalis Pergrande and Thrips tabaci Lindeman (Thysanoptera: Thripidae), following herbivore damage to cotton seedlings (Gossypium hirsutum L. (Malvales: Malvaceae)) were investigated. Dual choice olfactometer experiments examined the responses of thrips to damaged and undamaged plants at a range of time intervals following damage by Tetranychus urticae (Koch), T. ludeni (Zacher) (Acari: Tetranychidae) or Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae) larvae. The behavioural responses of the three species of thrips demonstrated that the duration of response to damaged plants was ultimately determined by the nature of specific damage inflicted. All three species of thrips were attracted to two-spotted spider mite-damaged plants for longer than they were to bean spider mite-damaged plants. Duration of attraction was also affected by the degree of damage inflicted on cotton seedlings; F. schultzei was attracted to plants damaged by a higher density of two-spotted spider mites for much longer than to plants damaged by a lower density of two-spotted spider mites. In determining the duration of attraction to damaged plants, results clearly showed that the response of thrips to herbivore-induced plants decreased with time and ultimately disappeared, irrespective of the identity of the inducing herbivore and whether the effect of herbivory was to make the plant more or less attractive. The results re-enforce previous studies that demonstrate the complexity of responses of thrips to their cotton host plants and shows that these responses are variable in time following herbivore damage to cotton plants.

Keywords: Thysanoptera, cotton, olfactometer, duration attraction, arthropods, dual herbivory, HIPV, induction, density

5.1 Introduction The changes that occur in plants following feeding by herbivorous arthropods are referred to as induced plant responses (Karban & Baldwin, 1997). Such responses include the synthesis and emission of various blends of organic volatile compounds (Kugimiya et al., 2010; Arimura et al., 2009) and among these plant compounds, the most widely studied group are the herbivore-induced plant volatiles (HIPVs) (Agrawal, 1998; Zhuang et al., 2012). The composition of these HIPVs can vary considerably in both quality and quantity, depending on various biotic and abiotic factors (Kugimiya et al., 2010). For instance, factors that can affect the composition of HIPVs include the identity and densities of the damaging herbivores (Turlings et al., 1993; Powell et al., 1998), the developmental stage of the herbivore (Takabayashi et al., 1998), type of plant tissue being consumed (Textor & Gershenzon, 2009), and the composition of the herbivore infestation, where multiple herbivore species simultaneously feed on the same plant (De Moraes et al., 1998). The ecological function and significance of these HIPVs remains a topic of discussion and disagreement (Takabayashi & Dicke 1996; Hare, 2011). It is often assumed that induced emission of HIPVs is a defensive response against further arthropod herbivory, as HIPVs can attract carnivorous natural enemies of the herbivores (Bruessow et al., 2011). However, HIPVs can also mediate interactions between plants and micro-organisms (Bruessow et al., 2011), neighboring plants (Preston et al., 2004) and their herbivorous arthropods (Chapters 2 & 3). There is now considerable evidence that herbivorous arthropods also exploit HIPVs as cues to forage and/or assess food quality (Arimura et al., 2009; Miresmailli et al., 2011) and to locate suitable plants for oviposition (Lu et al., 2009; Silva & Furlong, 2012). The timing of these herbivore-induced plant responses varies considerably between plant species. In some plants, the responses become measurable within hours of damage (McCloud et al., 1995) while in other plants the effects are only apparent after years (Haukioja, 1980). However, the precise time course of induction is difficult to characterize and often there is no clear pattern (Karban, 2011). For example, when offered a choice between soybeans (Glycine max) damaged by Mexican bean beetles (Epilachna varivestis (Coleoptera: Coccinellidae)) and undamaged plants, E. varivestis conspecifics preferred undamaged plants over damaged plants after three days of testing but after twenty days of testing, they preferred damaged plants over undamaged plants (Underwood, 1998). Chemical analysis has revealed that plants can have temporally-dependent unique HIPVs, whereby the profiles of HIPVs change during the course of herbivory (Guerrieri et al., 1999; Hoballah & Turlings, 2005). These temporal changes in HIPV profiles increase variability in the interactions between arthropods and plants. Recently, Kugimiya et al., (2010) screened HIPV profiles from Brassica rapa following herbivory by larvae of the diamondback moth (Plutella xylostella L. (Lepidoptera: Plutellidae)), and found that there were significant differences in the 90 relative concentrations of various HIPVs over a period of 24h, indicating that B. rapa plants emit temporally-dependent unique HIPV profiles. However, the effects of these temporally unique HIPV profiles on the behavioural responses of herbivorous arthropods were not investigated in the study, and how long after induction plants continue to attract/ repel arthropods remains largely unknown. To date the majority of studies that have investigated the duration of attraction to induced plant responses have been measured from the behaviorual responses of carnivorous arthropods and parasitoids. Previously the responses of F. occidentalis, F. schultzei and T. tabaci to cotton seedlings that were damaged for 24 h were investigated (Chapters 2 & 3). Studies showed that the behavioural responses of thrips were context dependent and unpredictable but consistent. Frankliniella schultzei was more attracted to cotton seedlings damaged by various arthropod pests (cotton bollworm, two-spotted spider mites, and root-damaging mealworms) than to intact seedlings, whereas F. occidentalis was more attracted to intact seedlings with the exception of seedlings that were damaged by two-spotted spider mites (Chapter 2). However, T. tabaci showed no preference between damaged or intact seedlings but was attracted to two-spotted spider mite damaged seedlings (Chapter 3). These responses, however, may change over time if temporal changes in HIPV profiles of herbivore-induced plants influence thrips attraction and behaviour. The day-to-day changes in the attractancy responses of F. occidentalis, F. schultzei and T. tabaci when presented with intact and herbivore-induced cotton seedlings was examined. Specifically, the study aimed to determine (1) how long thrips remained attracted to either spider mite or H. armigera damaged seedlings following the cessation of herbivore feeding (2) whether the density of spider mites affected the duration of attraction and (3) the effect of simultaneous feeding by two herbivores on the duration of attraction.

5.2 Materials and methods

5.2.1 Plants Cotton seedlings (G. hirsutum var. Sicott 71RRF) were grown to the two-leaf stage (BBCH code 12 (Munger et al., 1998)) in individual pots (11 cm diameter) containing organic potting mix [Osmocote (N:P:K, 16:35:10); Scots Australia, Baulkham Hills, New South Wales]. Seedlings were raised in a ventilated glasshouse under ambient conditions at The University of Queensland, St Lucia, Brisbane, Qld, Australia.

5.2.2 Arthropods Thrips 91

The study utilized laboratory populations of F. occidentalis, F. schultzei and T. tabaci (Figure 5.1) established from field-collected individuals. Frankliniella occidentalis were collected from cotton (G. hirsutum) flowers and T. tabaci were obtained from Ammi majus (Apiales: Apiaceae) flowers at the Australian Cotton Research Institute at Myall Vale near Narrabri, NSW (30°12'20.1"S 149°35'44.2"E). Frankliniella schultzei were collected from Malvaviscus arboreus Cav. (Malvales: Malvacea) flowers, The University of Queensland, St Lucia, Brisbane, Queensland, Australia (27°29'43.9"S 153°00'38.2"E). All three species of thrips were reared separately in sealed (black cardboard sandwiched between two layers of nylon mesh (22×22×22 cm, ≈ 1 mm²)) glass jars (17 cm, 8 cm diameter), lined with paper towel (Figure 2.3, Chapter 2) at 25 ± 2ºC, L12:D12. Green beans (Phaseolus vulgaris) that had been soaked in water for 16h, and then washed in warm soapy water to remove any residual insecticides, were provided as an adult food source and oviposition substrate. Beans were enclosed in glass jars for 3 days, after which time they were removed and replaced with fresh material; beans exposed to thrips were then transferred to clean glass jars for larval rearing. All experiments were conducted on laboratory-reared thrips 3-5 generations after field collection.

(A)

(B)

(C)

Figure 5.1 Thrips species investigated. (A) Frankliniella occidentalis. (B) Thrips tabaci. (C) Frankliniella schultzei. Spider mites

Two species of spider mites were used in experiments, two-spotted spider mites (Tetranychus urticae) and bean spider mites (T. ludeni). A separate culture of each species was maintained on cotton seedlings (two-leaf stage). The two-spotted spider mite colony was established from individuals that were collected from cotton seedlings at the Australian Cotton Research Institute, Myall Vale, Narrabri, New South Wales, Australia and bean spider mite colony was established from individuals collected from glass house grown cotton at The University of Queensland.

Cotton bollworm Second instar H. armigera used in experiments were obtained from a laboratory culture that originated from eggs collected from various crops at Toowomba, Queensland, Australia. Helicoverpa armigera were reared on a soybean-based artificial diet (Teackle, 1991).

(A)

(B)

(C)

Figure 5.2 Herbivore species used to damage cotton seedlings. (A) Two-spotted spider mites (T. urticae). (B) Bean spider mites (T. ludeni). (C) Second instar H. armigera larvae.

5.2.3 Standard olfactometer bioassays The responses of F. occidentalis, F. schultzei or T. tabaci were examined in a Y-tube olfactometer. The olfactometer consisted of a glass Y-tube (9 cm in length stem) and two side arms (9 cm in

93 length, 0.9 cm diameter, 45º apart) both of which were connected to two glass chambers (Chapter 2; Figure 2.4). Single plant treatments were placed in each of the chambers. Air was drawn through the device using a vacuum line pump at a rate of 1 L min-1 and monitored using a flow meter. Activated charcoal filters attached to glass chambers filtered air before entry into the olfactometer. The olfactometer bioassays were conducted in a temperature controlled room (25ºC ± 2ºC) and a white light source (Osram L 15W) was placed directly above the Y junction of the olfactometer to avoid bias. Using a fine brush single adult female thrips were individually introduced at the stem entrance of the Y-tube and observed for 10 min. When a single thrips moved more than half way (4.5 cm) into one of the arms (Chapter 2; Figure 2.4) and remained there for more than 20s, it was recorded as being attracted to the odour source associated with that arm. After testing five individuals, the chambers with plants were rotated and the Y- tubes cleaned with 100% ethanol and dried. After 10 individual thrips had been screened, the test plants were changed. A minimum of 60 individual thrips was used in each test. The same method was used in all olfactometer experiments.

5.2.4 Standard method for damaging cotton seedlings Prior to the olfactometer bioassays, cotton seedlings were damaged by H. armigera larvae, adult female two-spotted spider mites or adult female bean spider mites. Seedlings were damaged by enclosing the arthropod herbivores on seedlings in cages constructed from inverted plastic cups (9.7 cm in height, 8 cm diameter) that contained two windows (3.5×4.5 cm), covered with nylon mesh (≈1 mm²) (Chapter 2; Figure 2.5). The cups were placed over the plant and its rim pushed into the soil to enclose the seedling and relevant arthropods. Cups were also placed over undamaged control plants but no arthropods were introduced. Ultimately, cups were removed from damaged and undamaged control plants before they were used in olfactometer experiments.

5.2.4 Effect of post-feeding interval on responses of thrips to damaged cotton seedlings A series of dual choice olfactometer experiments was conducted to measure the effect of the post- feeding interval on the responses of F. occidentalis, F. schultzei or T. tabaci to undamaged cotton seedlings and herbivore-damaged cotton seedlings that had been damaged previously in one of the following ways:

Helicoverpa armigera damage. Plants were damaged by introducing ten second instar larvae onto the adaxial leaf surface; they were then enclosed on plants as described above. Groups of infested plants were held at 25°C ± 1°C, 70% RH, 12D:12L, but in separate rooms from the undamaged plants, which were maintained under identical conditions of temperature, relative humidity and photoperiod. Larvae were allowed to feed for either 5 h or 24 h before they were removed (Figure 94

5.3). Immediately upon removal of larvae after 5 h of feeding, damaged plants were tested against undamaged control plants in standard olfactometer tests against F. occidentalis and F. schultzei. Similarly, immediately upon removal of larvae after 24 h of feeding, damaged plants were tested against undamaged control plants in standard olfactometer tests against F. occidentalis and F. schultzei (Figure 5.3). The remaining plants that had been subjected to 24 h feeding by 10 H. armigera larvae were maintained at 25°C ± 1°C, 70% RH, 12D:12L, while control undamaged plants were maintained under identical conditions in a different room. At daily intervals a sub- sample of the two groups of plants was removed and subject to standard olfactometer tests using F. occidentalis and F. schultzei. Plants were discarded after testing. Daily testing continued until test species of thrips stopped responding differently to the damaged and undamaged control seedlings.

Two-spotted spider mite damage or bean spider mites. Plants were independently damaged by either 50 adult female two-spotted spider mites or 50 adult female bean spider mites. Mites species were introduced onto leaves and enclosed as described above. Mites were allowed to feed on plants for either 5 h or 24 h, after which all mites were removed using a fine paint brush. All three thrips species were subjected to standard olfactometer tests over time as described for the H. armigera tests.

5.2.6 Effect of degree of damage and post-feeding interval on responses of thrips to damaged cotton seedlings To investigate the effects of the degree of herbivore feeding on the temporal effects of the post- feeding interval on the responses of thrips to the degree of damage inflicted on cotton seedlings, F. schultzei was exposed to volatiles from cotton seedlings that had been subject to high (damaged by 100 two-spotted spider mites) levels of herbivory. Two-spotted spider mites were allowed to feed for 5 h or 24 h (Figure 5.3) before they were removed and subjected to standard olfactometer tests over time as described for the H. armigera tests.

5.2.7 Effect of dual herbivory and post-feeding interval on responses of thrips to damaged cotton seedlings Plants were simultaneously damaged by either two or 10 second instar H. armigera larvae and fifty adult female two-spotted spider mites. Arthropods were introduced onto the leaves and enclosed as described above and allowed to feed on plants for either 5 h or 24 h, after which time all arthropods were removed. Frankliniella occidentalis and F. schultzei were subjected to standard olfactometer tests over time as described above.

Attraction tested after 5 hrs of herbivore feeding Remove herbivores

5 hrs

Attraction tested Attraction tested 48 hrs Attraction tested 72 hrs B) immediately after 24 hrs after 24 hrs of herbivore after 24 hrs of herbivore of herbivore feeding feeding feeding

Remove herbivores Continuing…. 24 hrs

24 48 72 Time (h)

5.2.8 Statistical Analyses Preferences of thrips measured in various pairwise comparisons of herbivore damaged and undamaged plants in olfactometer tests were analysed using a G-test for each comparison, each day. Differences in these preferences over time were analysed using replicated goodness of fit (G’) tests (Sokal & Rohlf, 1995).

5.3 Results 5.3.1 Effect of post-feeding interval on responses of thrips to damaged cotton seedlings Helicoverpa armigera damage Frankliniella occidentalis and F. schultzei did not discriminated between undamaged and H. armigera damaged plants five hours after feeding began (G = 0.27, d.f. = 1, P = 0.61; G = 0.27, d.f. = 1, P = 0.61; Figure 5.4). Frankliniella occidentalis was more attracted to undamaged plants compared to damaged plants on the first three days following herbivory (Days 1 to 3: G>3.8, d.f. = 1, P<0.001; Figure 5.4) but did not discriminate between undamaged plants and H. armigera damaged plants on day four (G = 1.07, d.f. = 1, P = 0.30; Figure 5.4). In contrast, F. schultzei was more attracted to damaged plants than undamaged plants on the first four days (Days 1 to 4: G>3.8, d.f. = 1, P<0.001; Figure 5.4) but did not discriminate between undamaged and H. armigera damaged plants on day five (G = 0.60, d.f. = 1, P = 0.438; Figure 5.4). Overall the responses of thrips towards H. armigera damaged plants declined over time, F. occidentalis (G’ = 20.5, d.f. = 4, P = 0.003; Figure 5.4) and F. schultzei (G’ = 58.8, d.f. = 5, P<0.001; Figure 5.4).

Two-spotted spider mite damage or bean spider mites When the responses of F. occidentalis, T. tabaci or F. schultzei were tested five hours after damage by 50 T. urticae, thrips did not show differential attraction between undamaged and damaged plants (F. occidentalis, G = 0.27, d.f. = 1, P = 0.61; T. tabaci, G = 0.60, d.f. = 1, P = 0.44; F. schultzei, G = 0.60, d.f. = 1; P = 0.44; Figure 5.5). Significant preferences for damaged plants over undamaged plants were observed 24 h after the onset of herbivory and for the next three days for both F. occidentalis and T. tabaci (Days 1 to 4: G>3.8, d.f. = 1, P<0.001; Figure 5.5) and for the next four days for F. schultzei (Days 1 to 5: G>3.8, d.f. = 1, P<0.001; Figure 5.5). However on day five neither F. occidentalis nor T. tabaci discriminated between undamaged and damaged plants (G = 0.60, d.f. = 1, P = 0.44; G = 0.27, d.f. = 1, P = 0.61; Figure 5.5) and on day six F. schultzei did not discriminate between undamaged and damaged plants (G = 1.67, d.f. = 1, P = 0.19; Figure 5.5). Overall the responses of thrips towards T. urticae damaged plants declined over time; F. occidentalis (G’ = 45.4, d.f. = 5, P<0.001; Figure 5.5), T. tabaci (G’ = 35.5, d.f. = 5, P<0.001; Figure 5.4) and F. schultzei (G’ = 27.4, d.f. = 6, P = 0.001; Figure 5.5). 97

10 H. armigera Damaged Undamaged Frankliniella occidentalis 5 hours G = 0.27, d.f. = 1, P = 0.61

Day 1 G = 36.1, d.f. = 1, P<0.001 G-test between 5 hours to Day 2 G = 5.48, d.f. = 1, P = 0.02 day 4 G=20.5, d.f.=4, P=0.003 Day 3 G = 11.7, d.f. = 1, P = 0.006

Day 4 G = 1.07, d.f. = 1, P = 0.30

Frankliniella schultzei 5 hours G = 0.27, d.f. = 1, P = 0.61

Day 1 G = 44.2, d.f. = 1, P<0.001 G-test between 5 hours to Day 2 G = 32.5, d.f. = 1, P<0.001 day 5 G=58.8, d.f.=5, P<0.001 Day 3 G = 39.9, d.f. = 1, P<0.001

Day 4 G = 29.1, d.f. = 1, P<0.001

Day 5 G = 0.60, d.f. = 1, P = 0.438

100 80 60 40 20 0 20 40 60 80 100 % thrips responding

Figure 5.4 Responses of female thrips (Frankliniella occidentalis and F. schultzei) in olfactometer assays when presented with undamaged and damaged cotton seedlings 5 h after the onset of feeding by 10 Helicoverpa armigera larvae, and also (in a separate test) daily following 24 h of feeding damage to the cotton plants by 10 larvae. Assays were conducted every day until a given species of thrips stopped discriminating between previously damaged and undamaged plants. Data are presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness-of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test.

50 T. urticae Damaged Undamaged Frankliniella occidentalis 5 hours G = 0.27, d.f. = 1, P = 0.61 Day 1 G = 48.8, d.f. = 1, P<0.001 Day 2 G = 39.9, d.f. = 1, P<0.001 G-test between 5 hours to day 5 Day 3 G = 29.1, d.f. = 1, P<0.001 G=45.4, d.f.=5, P<0.001 Day 4 G = 23.1, d.f. = 1, P<0.001 Day 5 G = 0.60, d.f. = 1, P = 0.44 Thrips tabaci 5 hours G = 0.60, d.f. = 1, P = 0.44 Day 1 G = 26.1, d.f. = 1, P<0.001 G-test between 5 hours to Day 2 G = 32.5, d.f. = 1, P<0.001 day 5 Day 3 G = 39.9, d.f. = 1, P<0.001 G=35.5, d.f.=5, P<0.001 Day 4 G = 15.7, d.f. = 1, P<0.001 Day 5 G = 0.27, d.f. = 1, P = 0.61 Frankliniella schultzei 5 hours G = 0.60, d.f. = 1, P = 0.44 Day 1 G = 23.1, d.f. = 1, P<0.001 Day 2 G = 29.1, d.f. = 1, P<0.001 G-test between 5 hours to Day 3 G = 26.0, d.f. = 1, P<0.001 day 6 Day 4 G = 32.5, d.f. = 1, P<0.001 G=27.4, d.f.=5, P=0.001 Day 5 G = 15.7, d.f. = 1, P<0.001 Day 6 G = 1.67, d.f. = 1, P = 0.19 100 80 60 40 20 0 20 40 60 80 100 % thrips responding

When responses of F. occidentalis, T. tabaci or F. schultzei were tested against damage caused by bean spider mites, thrips did not show differential attraction between undamaged and damaged plants after five hours of feeding (F. occidentalis: G = 1.07, d.f. = 1, P = 0.30; T. tabaci: G = 1.07, d.f. = 1, P = 0.30; F. schultzei: G = 0.27, d.f. = 1; P = 0.61; Figure 5.6). Significant attraction toward damaged plants over undamaged plants was found on the first two days of testing for both F. occidentalis and T. tabaci (Days 1 to 4: G>3.8, d.f. = 1, P<0.001; Figure 5.6) and on the first three days for F. schultzei (Days 1 to 5: G>3.8, d.f. = 1, P<0.001; Figure 5.6). However, neither F. occidentalis nor T. tabaci discriminated between undamaged and damaged plants on day three after feeding (G = 0.27, d.f. = 1, P = 0.61; G = 2.42, d.f. = 1; P = 0.12; Figure 5.6) and F. schultzei stopped discriminating between undamaged and damaged plants on day four after feeding (G = 0.61, d.f. = 1, P = 0.44; Figure 5.6). Overall the responses of thrips towards bean spider mites damaged plants declined over time; F. occidentalis (G’ = 11.7, d.f. = 3, P = 0.008; Figure 5.6), T. tabaci (G’ = 20.2, d.f. = 3, P = 0.001; Figure 5.6) and F. schultzei (G’ = 25.1, d.f. = 4, P<0.001; Figure 5.6).

5.3.2 Effect of degree of damage and post-feeding interval on responses of thrips to damaged cotton seedlings The degree of T. urticae feeding damage inflicted on cotton seedlings also significantly affected the time period over which herbivore damaged plants affected the orientation behaviour of F. schultzei (Figure 5.7). When the responses of F. schultzei were tested when the density of T. urticae was increased to 100 per plant, F. schultzei remained attracted to damaged plants for much longer than when damaged by 50 T. urticae. Frankliniella schultzei was more attracted to damaged plants than to undamaged plants for the first 11 days following damage (Days 1 to 11: G>3.8, d.f. = 1, P<0.001; Figure 5.7). However, F. schultzei did not discriminate between undamaged plants and plants damaged by 100 T. urticae after five hours (G = 0.60, d.f. = 1, P = 0.44; Figure 5.7) or on day 12 (G = 0.27, d.f. = 1, P = 0.61; Figure 5.7). Overall the responses of F. schultzei towards damaged plants declined over time (G’ = 54.8, d.f. = 12, P<0.001; Figure 5.7)

100

50 T. ludeni Damaged Undamaged Frankliniella occidentalis 5 hours G = 1.07, d.f. = 1, P = 0.30 Day 1 G = 23.1, d.f. = 1, P<0.001 G-test between 5 hours to day 3 Day 2 G = 6.79, d.f. = 1, P = 0.009 G=11.7, d.f.=3, P=0.01 Day 3 G = 0.27, d.f. = 1, P = 0.61

Thrips tabaci 5 hours G = 1.07, d.f. = 1, P = 0.30 Day 1 G = 39.9, d.f. = 1, P<0.001 G-test between 5 hours to day 3 Day 2 G = 17.9, d.f. = 1, P<0.001 G=20.2, d.f.=3, P=0.001 Day 3 G = 1.07, d.f. = 1, P = 0.30 Frankliniella schultzei 5 hours G = 0.27, d.f. = 1, P = 0.61

Day 1 G = 32.5, d.f. = 1, P<0.001 G-test between 5 hours to Day 2 G = 23.1, d.f. = 1, P<0.001 day 4 G=25.1, d.f.=4, P<0.001 Day 3 G = 17.9, d.f. = 1, P<0.001 Day 4 G = 0.60, d.f. = 1, P = 0.44

100 80 60 40 20 0 20 40 60 80 100 % thrips responding

101

100 T. urticae Damaged Undamaged

5 hours G = 0.60, d.f. = 1, P = 0.44 Day 1 G = 23.1, d.f. = 1, P<0.001

Day 2 G = 29.1, d.f. = 1, P<0.001 Day 3 G = 32.5, d.f. = 1, P<0.001 Day 4 G = 39.9, d.f. = 1, P<0.001

Day 5 G = 44.2, d.f. = 1, P<0.001 G-test between 5 hours to Day 6 G = 29.1, d.f. = 1, P<0.001 day 12 G=54.8, d.f.=12, P<0.001 Day 7 G = 32.5, d.f. = 1, P<0.001

Day 8 G = 20.5, d.f. = 1, P<0.001

Day 9 G = 29.1, d.f. = 1, P<0.001

Day 10 G = 17.9, d.f. = 1, P<0.001

Day 11 G = 6.79, d.f. = 1, P = 0.009

Day 12 G = 0.27, d.f. = 1, P = 0.61

100 80 60 40 20 0 20 40 60 80 100 % thrips responding

Figure 5.7 Responses of female Frankliniella schultzei thrips in olfactometer assays when presented with undamaged and damaged cotton seedlings 5 h after the onset of feeding by 100 two-spotted spider mites (Tetranychus urticae) and (in a separate test) daily following 24 h feeding by two-spotted spider mites. Assays were conducted every day until F. schultzei stopped discriminating between previously damaged and undamaged plants. Data are presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness-of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test.

102

5.3.3 Effect of dual herbivory and post-feeding interval on responses of thrips to damaged cotton seedlings When the responses of F. occidentalis or F. schultzei were tested five hours after simultaneous damage by ten H. armigera and fifty T. urticae, neither species of thrips discriminated between undamaged plants and dually damaged plants (G = 0.27, d.f. = 1, P = 0.61; G = 0.27, d.f. = 1, P = 0.61; Figure 5.8). Frankliniella occidentalis was more attracted to dually damaged plants than to undamaged plants for the first two days after herbivore feeding (Days 1 to 2: G>3.8, d.f. = 1, P<0.001; Figure 5.8) and F. schultzei was more attracted to dually damaged plants than to undamaged plants for the first eight days after herbivore feeding (Days 1 to 8: G>3.8, d.f. = 1, P<0.001; Figure 5.8). Frankliniella occidentalis did not show differential attraction between undamaged and dually damaged plants on day three after herbivore feeding (G = 1.67, d.f. = 1, P = 0.19; Figure 5.8) and F. schultzei did not discriminate between damaged and undamaged plants nine days after herbivore feeding (G = 1.07, d.f. = 1, P = 0.30; Figure 5.8). Overall the responses of thrips declined over time; F. occidentalis (G’ = 18.6, d.f. = 6, P = 0.003; Figure 5.8) and F. schultzei (G’ = 45.6, d.f. = 6, P<0.001; Figure 5.8). When the responses of F. schultzei were subsequently tested when the number of H. armigera larvae were reduced, it did not discriminate between undamaged plants and plants simultaneously damaged by two H. armigera and fifty T. urticae after five hours of feeding (G = 0.60, d.f. = 1, P = 0.44; Figure 5.9) but it was more attracted to damaged plants 24 h after the initiation of feeding (G = 29.1, d.f. = 1, P<0.001; Figure 5.9) and remained attracted to these plants until day six (G = 0.27, d.f. = 1, P = 0.61; Figure 5.9). Frankliniella schultzei was however, more attracted to simultaneously damaged plants than to undamaged plants for the first eight days (Days 1 to 8: G>3.8, d.f.=1, P<0.001; Figure 5.9). Overall there were significant differences between experimental days (G’ = 37.3, d.f. = 6, P = 0.001; Figure 5.9).

103

50 T. urticae + 10 H. armigera Damaged Undamaged Frankliniella occidentalis 5 hours G = 0.267, d.f. = 1, P = 0.605 Day 1 G = 15.70, d.f. = 1, P<0.001 G-test between 5 hours to day 3 Day 2 G = 32.45, d.f. = 1, P<0.001 G=18.6, d.f.=3, P=0.003 Day 3 G = 1.674, d.f. = 1, P = 0.196

Frankliniella schultzei 5 hours G = 0.267, d.f. = 1, P = 0.605 Day 1 G = 17.99, d.f. = 1, P<0.001 Day 2 G = 29.11, d.f. = 1, P<0.001 Day 3 G = 32.45, d.f. = 1, P<0.001 Day 4 G = 44.17, d.f. = 1, P<0.001 G-test between 5 hours to day 9 Day 5 G = 39.95, d.f. = 1, P<0.001 G=45.6, d.f.=9, P<0.001 Day 6 G = 29.11, d.f. = 1, P<0.001 Day 7 G = 20.46, d.f. = 1, P<0.001 Day 8 G = 15.70, d.f. = 1, P<0.001 Day 9 G = 1.07, d.f. = 1, P = 0.301

100 80 60 40 20 0 20 40 60 80 100 % Thrips responding

Figure 5.8 Responses of female thrips (Frankliniella occidentalis and F. schultzei) in olfactometer assays when presented with undamaged and damaged cotton seedlings 5h after the onset of feeding by 50 two-spotted spider mites (Tetranychus urticae) plus 10 Helicoverpa armigera larvae, and (in a separate test) daily following 24 h feeding. Assays were conducted every day until a given species of thrips stopped discriminating between previously damaged and undamaged plants. Data are presented as percentage of respondents. G, G’, and P-values calculated from a replicated goodness-of-fit test, where P<0.05 indicates statistical significance. N = 60 thrips/test. 104

50 T. urticae + 2 H. armigera Damaged Undamaged

5 hours G = 0.601, d.f. = 1, P = 0.438

Day 1 G = 29.11, d.f. = 1, P<0.001

Day 2 G = 39.95, d.f. = 1, P<0.001 G-test between 5 hours to Day 3 G = 29.11, d.f. = 1, P<0.001 day 6 G=37.3, d.f.=6, P<0.001

Day 4 G = 26.01, d.f. = 1, P<0.001

G = 15.70, d.f. = 1, P<0.001 Day 5

G = 0.267, d.f. = 1, P = 0.605 Day 6

100 80 60 40 20 0 20 40 60 80 100 % thrips responding

105

5.4 Discussion The main objective of this study was to investigate the duration for which herbivorous arthropods respond to herbivore-induced plants. The observed duration of attraction of thrips to herbivore- induced cotton seedlings was ultimately determined by the specific damage caused, including the type of damage (chewing vs sucking) inflicted on cotton seedlings and the density of the damaging herbivore. For example F. schultzei remained attracted to T. urticae damaged plants for much longer compared to H. armigera larvae damaged plants (Figures 5.5 & 5.4). Also, all three species of thrips remained attracted to plants damaged by T. urticae for much longer (Figure 5.4) than to plants damaged by T. ludeni (Figure 5.6), suggesting that even herbivores with similar modes of feeding, induced different plant responses. The degree of damage inflicted on cotton seedlings also affected the duration of attraction of thrips. Frankliniella schultzei remained attracted to seedlings for much longer when the density of T. urticae increased (Figure 5.7). Similarly when two herbivore species simultaneously damaged cotton seedlings the duration of attraction for both F. occidentalis and F. schultzei was affected (Figure 5.8). Thus the results from the current study reveal that the identity, the density of the damaging herbivore species (inducer) and the identity of the responding herbivore were all important factors in determining how long herbivorous arthropods remained attracted to the damaged plants post herbivore feeding.

5.4.1 Timing of response The timing of induced plant responses following herbivory varies between plant species (McCloud et al., 1995; Haukioja, 1980). For example, in corn plants the response is fast and the induced emission of HIPVs can be detected immediately after leaf damage (Hanyu et al., 2009). In cotton the response takes longer as gas chromatography and spectroscopy studies have found the emission of HIPVs from leaves fed upon by Lepidoptera larvae only becomes detectable 24-48 h after feeding (Loughrin et al., 1994; Röse & Tumlinson, 2005; Röse et al., 1996). Similarly, McCall et al. (1994) only detected an increase in HIPV emission in cotton seedlings after 16 h, following feeding by Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae). The herbivore-induced responses of the cotton seedlings were not instantaneous but rather, as the behavioral responses of thrips demonstrate, they were gradual but could be detected after 24 h, though it is possible that responses could have occurred at any stage between 5 h and 24 h. None of the three thrips species discriminated between undamaged plants and plants that had been fed upon for five hours, thrips species only showed a preference after 24 h (Figures 5.4-5.9). After five hours of herbivory, cotton seedlings may have only received a fraction of the total damage compared to those plants damaged for 24 h and HIPVs may either not have been released or, more likely, they may have been

106 produced in such low quantities that they were insufficient to elicit a behavioral response in the three thrips species.

5.4.2 Duration of response Results of the current study also showed that, thrips preferences between damaged and undamaged plants were not only significant within each experimental day, but also significantly different between experimental days, regardless of the damaging herbivore species. This suggests that thrips attraction or repulsion to induced plants diminished over time (Figures 5.4-5.9). Thus plants must have recovery mechanisms in response to herbivory, which would bring the plant back to its constitutive state (Richards, 1993) once herbivore attack ceases. This suggests that HIPV emissions change over the time course of the induction (Loughrin et al., 1994; Turlings et al., 1998; Scascighini et al., 2005). Recent studies have shown that the relative concentrations of HIPVs vary over time such that unique HIPV blends can be present at specified intervals post-herbivory (Loughrin et al., 1994; Röse & Tumlinson, 2005; Kugimiya et al., 2010; Maeda & Takabayashi, 2001). Specifically in herbivore-induced cotton seedlings, changes in the relative portions of compounds within HIPV blends over time have been detected (McCall et al., 1994). This could explain why thrips attraction/repulsion to herbivore-induced cotton seedlings diminished in the later stages of plant recovery, as after the effects of herbivory were alleviated, HIPV blends changed and/or decreased in quantity over time. This pattern of induction was evident across all cotton seedlings that were damaged by different herbivore species. However, it also has to be acknowledged that this recovery phase may be rapid or a much slower mechanism (Karban, 2011). For instance, Gomez et al. (2010) showed that, Trifolium repens damaged by Mamestra brassicae (Li.) (Lepidoptera: Noctuidae) took 28 days to recover.

5.4.3 Comparison in responses between different sources of herbivory Previously, in Chapters 2 and 3 it was suggested that the various behavioural responses observed among herbivorous arthropods to herbivore-induced plants were due to the different species and ways in which various herbivores feed on plants. Such forms of feeding can ultimately lead to the emission of quantitatively and qualitatively unique blends of HIPVs (Delphia et al., 2007; Rodriguez-Saona et al., 2003). Based on the results of the current study it can also be concluded that these different types of feeding damage may not only influence preferences of herbivores to herbivore-induced plants but also influence the duration which herbivores remain attracted to or not attracted todamaged plants. For instance, F. schultzei remained attracted to plants damaged by two- spotted mites (T. urticae) for longer (five days; Figure 5.4) than it did to plants damaged by H. armigera larvae (4 days; Figures 5.7). Previous studies have shown these differences in HIPV 107 profiles between herbivores so it is reasonable to assume that there were unique differences in the HIPV profiles of two-spotted spider mites damaged plants and H. armigera damaged plants. For example, quantitative differences have been reported in HIPV profiles in cotton in response to feeding by Spodoptera frugiperda (Smith) (Lepidoptera: Noctuidae), Euschistus heros (Fabricius) (Hemiptera: Pentatomidae) and Anthonomus grandis (Boheman) (Coleoptera: Curculionidae), these differences were attributed to the different types of feeding damage (i.e. chewing vs. sucking) inflicted (Magalhães et al., 2012).

5.4.4 Comparison of responses from the same mode of herbivory Results also provides evidence that even herbivores with similar modes of feeding, affected the duration of attraction of thrips species to damaged cotton seedlings differently. All three thrips species remained attracted to plants damaged by two-spotted spider mites for much longer compared to plants damaged by bean spider mites (Figure 5.4 & 5.6). Other studies have shown that herbivores with similar modes of feeding do produce different HIPV profiles. Quantitative differences have been detected in cotton HIPV profiles when cotton plants were damaged by Heliothis virescens Fabricius (Lepidoptera: Noctuidae), S. exigua Hübner (Lepidoptera: Noctuidae) (Ngumbi et al., 2009) and H. zea Boddie (Lepidoptera: Noctuidae) (McCall et al., 1994) even though these arthropods have similar modes of feeding. Different arthropod species are known to produce different elicitor compounds in their saliva when feeding (Hilker & Meiners, 2010). Plant transcriptional responses can be specific to the elicitor compounds released by insect herbivores with similar modes of feeding (Frey et al., 2000), leading to different plant responses at feeding sites. For example, different transcriptional responses elicited in Nicotiana attenuata (Solanales: Solanaceae) by larvae of Manduca sexta L. (Lepidoptera: Sphingidae), H. virescens and S. exigua were correlated with the profile of larval elicitors in each species (Voeckel & Baldwin, 2004). This may explain why different modes of herbivory may result in different plant responses, which influence duration of attraction correspondingly.

5.4.5 Effect of damage intensity of duration of response The current study also showed that the degree of damage inflicted on cotton seedlings determined how long thrips remained attracted to or not attracted to damaged plants. Frankliniella schultzei remained attracted to plants that were damaged by a higher density of two-spotted spider mites (100 mites per plant) for much longer (Figure 5.5), compared to plants that were damaged by a lower density (50 mites per plant) (Figure 5.4). This suggests that the higher densities of two-spotted spider mites resulted in increased damage to cotton seedlings and correspondingly higher amounts of HIPV’s over an extended period which prolonged F. schultzei attraction to damaged plants. 108

Other studies have shown that the emission of HIPVs do increase as the density of a given herbivore damaging the plant increases (Guerrieri et al., 1999; Girling et al., 2011; Sabelis & van de Baan, 1983; Sabelis et al., 1984; Maeda & Takabayashi, 2001). For example, headspace HIPV analysis revealed that the emission of HIPVs in Brassica oleracea increased as the density of P. xylostella larvae increased (Girling et al., 2011). Girling et al. (2011) did not investigate the effect of herbivore-density on the behavioural responses of herbivorous arthropods, but the increase in HIPVs resulting from a higher density of P. xylostella larvae were correlated with greater attractiveness of infested plants to the parasitoid Cotesia vestalis (Hymenoptera: Braconidae).

5.4.6 Responses to dual herbivory Dual herbivory also affected the duration of attraction of both F. occidentalis and F. schultzei (Figure 5.8). Frankliniella schultzei remained attracted to plants simultaneously damaged by 10 H. armigera larvae and 50 T. urticae for much longer (eight days, Figure 5.8) compared to plants damaged by only H. armigera larvae (4 days, Figure 5.4) or T. urticae (5 days, Figure 5.5). The intensity of damage was also important in these dual herbivore comparisons as when the number of damaging H. armigera larvae was reduced 10 to two along with 50 T. urticae, the duration over which F. schultzei remained attracted to these plants was shortened from eight to five days (Figures 5.8 & 5.9). Frankliniella occidentalis was also more attracted to plants simultaneously damaged by both H. armigera and T. urticae, but they remained attracted to these plants for a much shorter time period compared to those plants only damaged by T. urticae (Figure 5.5). In the case of F. schultzei dual herbivory, prolonged its attraction towards damaged plants but in the case of F. occidentalis it reduced to the duration of attraction towards damaged plants. It can thus be argued that responses of thrips to dual herbivory was equal to the sum of the responses to single species herbivory (Figure 5.4 & 5.5), as reported in Chapter 4. In the current study the evidence is clear that temporal changes occur in a plant’s profile. Not only did changes in plant’s HIPV profile occur, but these changes also influenced the behavioural responses of the three species thrips in highly specific ways. This resulted in unique patterns being observed which describe differences in the responses of thrips at specific times post- herbivore feeding. Understanding and determining these responses present a more holistic picture of the temporal changes in the interaction between plants and herbivorous arthropods, and more studies which measure temporal changes in profiles of plants should also be complimented with studies on the responses of herbivores to these changes. In general the majority of studies in this field revolve around determining an herbivorous arthropod’s response to damage after a fixed time period (Table 1.1, Chapter 1), rarely is the full time course of induction investigated. It is also the case that the majority of studies aim to determine the chemical temporal changes in plants, in 109 particular HIPVs (for example; Shiojiri et al., 2001; Shiojiri et al., 2002; Kugimiya et al., 2010; Shiojiri et al., 2010), and though this is important, more attention should be given to temporal changes in other aspects of a plant’s profile. An exciting area of research that warrants further investigation is the fact that the current study only investigated the temporal changes in the pre- alighting responses of thrips, what would also be interesting is to determine the temporal changes in the post-alighting oviposition responses of thrips. An interesting future study would be to determine if the observed pre-alighting (olfactometer) responses of thrips be reliable indicators of post-alighting (oviposition) responses.

110

Chapter 6 Temporal host plant use by three thrips species (Thysanoptera: Thripidae) in an Australian cotton growing region: identifying sources of thrips invasions

A paper based on this chapter will be submitted to Environmental Entomology. The work has been reformatted and edited slightly, so that it is consistent with the rest of the thesis. For example, figures are now placed within the text, references are cited in a style consistent with the rest of the thesis and the bibliography has been incorporated into the thesis bibliography.

111

Abstract The seasonal abundance of Thrips tabaci Lindeman, Frankliniella schultzei Trybom and F. occidentalis Pergrande (Thysanoptera: Thripidae), regarded as pests of early season cotton, were studied on both cultivated and non-cultivated host plants in the cotton producing region centered around the Namoi valley, NSW, Australia. The aim of this chapter was to determine the causes of the seasonal pheneology of the three species of thrips on cotton (Gossypium hirsutum L. (Malvales: Malvaceae)). Patterns of host plant utilization by thrips associated with cotton were quantified through a structured sampling program that determined the presence and estimated the abundance of both adult and larval thrips. The floral and vegetative parts of different plant species were sampled regularly over two years during which 69 different plant species from 20 families were sampled. Thrips tabaci was the most numerous of the thrips species and was recorded from 31 plant species, while F. occidentalis was recorded from 35 plant species. Frankliniella schultzei was recorded from 25 plant species, but at very low densities. The seasonal composition of thrips on cotton changed from predominately T. tabaci on early season seedling cotton to F. schultzei and F. occidentalis on mature flowering cotton, later in the season. The predominately high T. tabaci abundance on early season cotton was attributed to the high numbers of T. tabaci on the surrounding weed species. During this time weed hosts on which T. tabaci was recorded were in plentiful supply and thus contributed to the high T. tabaci abundance on cotton. Results indicated that the pattern of abundance of F. occidentalis on cotton was not correlated with their abundance on the weeds, as during this time period, the weed species on which they were found were in low abundance. It was likely that the low numbers of F. occidentalis present on the few surrounding weed species, moved onto cotton flowers and increased in abundance, indicating that cotton flowers were the determining factor for the high F. occidentalis abundance on late season cotton. This study couldn’t however determine reasons for the pattern of abundance of F. schultzei on cotton. The genetic relationships of thrips on cotton relative to those on the other host plant species sampled was investigated by analysis of mitochondrial CO1 gene sequences. The T. tabaci from cotton and from weeds were genetically identical, which is consistent with these insects moving onto cotton plants from nearby source populations on weeds. The same is true of F. occidentalis. An important conclusion from this research is that weeds play an important role in the population biology of thrips and the specific insights related to this point need to be incorporated into pest management strategies. Results of this study indicate that it could be possible to predict large out breaks of thrips (particularly T. tabaci) on cotton. It could be argued that whenever there is a high abundance of weeds, it could result in a high abundance of thrips that move onto cotton. Knowledge

112 of, or monitoring the arthropod pests hosted by weeds can permit a manager to better anticipate arthropod problems in adjacent crops.

Keywords Thrips tabaci, Frankliniella schultzei, Frankliniella occidentalis, Thysanoptera, Gossypium hirsutum, weed hosts, seasonal abundance, flowers,

113

6.1 Introduction Understanding the sources of infestation of crops by polyphagous pests is important in developing effective integrated pest management systems. This is because such knowledge may be used to predict likely future pest abundance, thereby assisting in developing pro-active management plans, or may allow the sources to be manipulated to reduce the magnitude of the infesting pest population, thereby reducing the risk that it will need to be controlled (Wilson, 1995). In many cotton-producing regions, thrips (Thysanoptera: Thripidae) are often abundant and damaging (Atakan et al., 1998). In Australia three species of thrips, Thrips tabaci Lindeman, Frankliniella schultzei Trybom and F. occidentalis Pergande are regarded as early season pests of cotton crops (Williams et al., 2011). These thrips primarily damage seedling cotton. Thrips generally feed on the undersides of the leaves, causing damage that can lead to dramatic deformation of seedling leaves (Chapter 1; Figure 1.4) (Williams et al., 2011). In extreme cases thrips damage can result in the death of the plant terminal which delays plant development and the production of fruiting bodies and can result in reduced yield and/or delayed maturity (Sadras & Wilson, 1998; Lei & Wilson, 2004). Most cotton growers reduce the risk of thrips damage by using seed treated with neonicotinoid seed treatments or granular insecticides, such as phorate, applied into the seed furrow at planting. Thrips can also sometimes damage plants in mid to late season crops, causing distortion of the upper leaves. However, as this damage does not usually result in economic loss and the thrips help control spider mites they are rarely sprayed with insecticide (Williams et al., 2011). In Australia cotton is planted in October (Figure 6.1) and begins to mature in March (Figure 6.2A). The plants are chemically defoliated (Figure 6.2B) when about 60% of bolls have opened (Wilson, 1995). In the Namoi valley, New South Wales, T. tabaci has been recorded as the most abundant species of thrips on cotton early in the season; this is typically followed by a rapid decline in abundance during mid-December, while the abundance of F. schultzei increases after this period (Wilson & Bauer 1993). The processes behind the decline in T. tabaci abundance and the later increase in F. schultzei abundance are poorly understood. Further, F. occidentalis has entered Australia and spread to cotton producing regions since the report of Wilson & Bauer, (1993) was published. However, little is known of its seasonal abundance, host use or ecology. Indeed, the ecology of thrips in relation to cotton crops more broadly is not well understood, despite their pest status. For example, host plant use and the impact of seasonal conditions on their abundance and the species composition of the thrips pest complex remains unclear. This study was designed to identify the host plants from which thrips invade cotton. Wilson and Bauer (1993) suggested that the observed changes in seasonal abundance and species composition of thrips in cotton were probably due to the senescence of nearby winter weeds or cereal crops, resulting in an influx of thrips into cotton plants. Support for this claim was 114 provided by Milne and Walter (1998b) who reported high densities of T. tabaci adults and larvae on inflorescences and vegetative parts of five common weed species near cotton fields, suggesting the possibility of T. tabaci moving into cotton crops from nearby alternative hosts. However, the abundance of adults and larvae on wheat was very low, calling into question the common folklore in the cotton industry that wheat is a major source of T. tabaci populations. The three species of thrips that invade seedling stage cotton are generally regarded as polyphagous (Milne & Walter, 2000; Milne & Walter, 1998b). That is, they utilize several host plant species belonging to different families to survive (Bernays & Chapman, 1994). Many pest thrips species have wide host ranges that include both non-cultivated and cultivated plants and the relative attraction of the different host plant species is important during periods of dispersal as they can be affected by plant senescence (Doederlein & Sites, 1993). Movement of individuals of any particular species from one locality to another can be brought about by unfavorable conditions in the habitat of origin or when a species responds to certain environmental stimuli (Walter, 2003; Loxdale & Lushai, 1999; Dingle & Drake, 2007). For example, North & Shelton (1986) demonstrated that populations of T. tabaci developed on cereal and forage crops and colonized nearby cabbage fields when cereal and forage crops were harvested or became senescent. Similarly, F. occidentalis feeding damage on chrysanthemum inflorescences caused senescence and this, in turn, caused F. occidentalis to colonize healthy plants (Rhainds & Shipp, 2003). These studies suggest that thrips populations shift from one host type to another and that movement is driven by deterioration in host plant quality. However, there is no direct evidence from Australia that the thrips on proximate weeds or crops subsequently colonize cotton crops. In this chapter I quantified the relative importance of the major host species in the Namoi valley for F. occidentalis, F. schultzei and T. tabaci by sampling insects across various crop and weed species in the area. Specifically I aimed to determine (1) on which weeds and crops the different thrips species were found, (2) which host plants were the most important, (3) how the abundance of thrips was affected by the abundance of a particular host plant species, (4) the species composition of the thrips complex in cotton and the relative abundance of each species in the crop and (5) how the abundance of thrips in weed or crop host plants affected their abundance in cotton crops.

115

Figure 6.1 In Australia cotton is planted in October, seedling cotton in the Namoi valley, New South Wales, 27 October 2011.

116

(A)

(B)

Figure 6.2 Cotton begins to mature in March (A) Mature cotton, 23rd February 2012 (B) Chemically defoliated plants when 60% of bolls have opened, 8th April 2013.

117

6.2 Materials and Methods

6.2.1 Sampling sites and methods A structured sampling program was conducted in the area centred on the Namoi valley, NSW. Sampling was conducted for two days in each of the following months: October and November 2011, February, August, October, November and December 2012, February, April, May, June, July, September and November 2013. Twelve sample sites that spanned a distance of about 35 km from east to west (Figure 6.3) were selected across the Namoi valley. The same 12 sites were sampled on each sampling occasion (Figure 6.4). Each site contained a variety of different flowering plants (Figure 6.5). At each site the range of flowering weed, crop and grass species was sampled for the presence of thrips; the availability of plants changed between sampling events due to rainfall and seasonality. At each site 10 samples of inflorescences and 10 samples of leaves (~20 cm long and with associated leaves) of each plant species available were collected haphazardly and stored separately 100% ethanol in (Figure 6.6 & 6.7). These samples were stored before thrips were extracted in the laboratory and then identified and quantified. For some weed species it was not practical to separate inflorescences and leaves, so in these cases sections of stem (including leaves and flowers) were collected.

6.2.2 Thrips and plant identification Thrips were extracted from plant material by vigorously shaking the specimen containers. The ethanol and plant parts were then searched for thrips under a dissecting microscope. Larval thrips were separated from the adults for identification. Adults were initially sorted to suborder, Terebrantia or Tubulifera; any Tubulifera were discarded. In samples that contained more than 50 terebrantian larvae or adults, 10% of individuals were subsampled for identification. In all other samples all individuals were identified. Individual adults were slide mounted in a drop of Hoyer’s medium, viewed under a compound light microscope (Figure 6.8) and morphologically identified to species (Mound et al., 2014). Larvae were also mounted in Hoyer’s medium, examined under a phase contrast microscopy and identified using the keys and descriptions of Kirk (1987) and Milne et al. (1997). All plant specimens were identified to species by staff from CSIRO Agriculture Flagship or New South Wales Department of Primary Industries at the Australian Cotton Research Institute, Myall Vale, Narrabri, New South Wales, Australia.

118

11 9 5

6 10 7 4 3 8 Site number GPS coordinates Plant type present at site 1 12 1 30°17'09.2"S 149°47'49.5"E Non-crop 2 2 30°16'21.9"S 149°43'59.8"E Non-crop 3 30°15'04.9"S 149°42'25.5"E Non-crop 4 30°13'29"S 149°25'11"E Non-crop 5 30°11'48"S 149°38'46"E Non-crop 6 30°11'44"S 149°30'0"E Non-crop 7 30°12'20.1"S 149°35'44.2"E Non-crop & crop 8 30°15'57.1"S 149°44'08.8"E Non-crop 9 30°11'25.5"S 149°19'51.1"E Non-crop 10 30°11'37.3"S 149°32'38.4"E Non-crop 11 30°09'47.1"S 149°35'17.5"E Non-crop & crop 12 30°16'11.3"S 149°48'16.9"E Non-crop & crop

Figure 6.3 Location of the sampling areas in the Namoi valley. A total of 12 sites were chosen, each site contained either non-crop plants, crop plants or a combination of both. 119

Site 1 Site 2

Site 3 Site 4

Site 5 Site 6

Figure 6.4 Pictures of individual sampling sites. Sites were different from one another in terms of the availability of host plants and topography of the land.

120

Site 7 Site 8

Site 9 Site 10

121

(A) (B)

Rapistrum rugosum

Echium plantagineum Echium plantagineum

Phyla canescens Verbena tenuisecta (C) (D) Brassica x napus

Ammi majus

10 flowers

10 leaves

123

Figure 6.7 Sampling different plant species for thrips. From top left; sampling canola (Brassica napus), sampling wheat (Triticum aestivum), canola field and sampling cotton.

124

Plant material removed Specimen container containing plants and Different thrips species collected from a single thrips specimen container

Thrips mounted on slides using Slide mounted thrips hoyers mounting media

Figure 6.8 Individual adult and larval thrips were slide mounted in a drop of Hoyers medium and viewed under a compound light microscope . 125

6.2.3 Seasonal abundance of potential host plants On each sample date the abundance of each plant species at a given site was estimated visually and assigned a score based on % ground cover (Figure 6.9). At the same time, the total plant abundance across all hosts sampled was summed for each site. This provided an accurate estimate of overall plant abundance as it accounted for both the abundance of each species and the number of species present. Mean abundance was found to be misleading as two sites could have similar mean abundance but very different numbers of species present, and hence very different levels of total ground cover. All sites were on fertile black soils (Isbell et al., 1997) that support growth of many plant species. Seasonal conditions that support abundant plant growth (plentiful rainfall) should therefore also support growth of a larger number of species, while in less suitable conditions (low rainfall) there will be less abundant growth and only a small number of species will persist. To test this, the count of the number of species present was regressed against the total plant abundance.

6.2.4 Effect of rainfall on abundance of thrips’ host plants Monthly rainfall data was obtained for the Australian Cotton Research Institute which is located roughly central to the sample sites. Total plant abundance was regressed against monthly rainfall, but the rainfall was offset by 1 month to allow for the delay in plant growth following rain.

6.2.5 Seasonal abundance of thrips on non-cultivated hosts Seasonal patterns of abundance were assessed by calculating the total number of adult and larval thrips collected per site per date, for each of the three thrips species. This was done for each site from non-crop plant species. The total number of adult and larval thrips collected per site per date was calculated by first dividing the number of thrips per species by the number of units sampled (eg divide by 10 if 10 flowers sampled) then summing this value across all of the non-crop species on that date. Total number of adult and larval thrips collected per site per date is a better estimate of abundance than mean abundance because two sites or dates could have the same mean, but at one site or date this mean could be derived from two hosts and at another from 15 hosts, hence the second site or date actually has far higher overall thrips abundance.

6.2.6 Seasonal use of non-cultivated hosts by thrips The number of hosts on which either adult or larval thrips were found was also calculated for each site for each sample date. To further understand the significance of these non-crop hosts, the year was divided into seasons; Winter (June, July and August), Spring (September, October and November), Summer (December, January and February) and Autumn (March, April and May). For each season the number of times a plant species was recorded with thrips present was plotted 126 against the mean number of thrips recorded from that plant species. This captures both the abundance of the plant species and the abundance of adult or larval thrips on that species.

< 5 cover 5-10 cover

Host Hostplant plant

10-20 cover 20-40 cover

40-60 cover + 60 cover

127

6.2.7 Seasonal abundance of thrips on cotton and other crops Seasonal patterns of abundance were assessed by calculating the total number of adult and larval thrips collected, for each of the three thrips species, from each site from each crop plant species for each sample date. A variety of crops were sampled as they became available in the Namoi valley. These included summer crops; G. hirsutum, Cajanus cajan, Zea mays, Helianthus annuus and winter crops; Triticum aestivum, Brassica x napus, Carthamus tinctorius, Vicia faba and Cicer arietinum.

6.2.8 DNA extraction and COI sequencing: Molecular analysis of Thrips tabaci and Frankliniella occidentalis Genomic DNA was extracted with 10% Chelex (Walsh et al., 1991). For T. tabaci a 690-bp fragment of the COI gene was amplified with the primers LCO-1490 and HCO-2198 (Folmer et al.,

1994). PCR was performed using one unit of Ampli Taq™ (Life technologies), 0.2 μM of each primer and 2.5 mM of MgCl2. PCR cycling conditions consisted of 40 cycles and 52°C annealing temperature. Denaturation was 95°C for 30 s, and elongation was 72°C for 45 s. For F. occidentalis, a 380-bp fragment of COI gene was amplified with the primers C1-N-2329 andC1-J-1751 or HCO- 2198 and C1-J-1718 (Simon et al., 1994). PCR was performed using Ampli Taq™ under the same conditions described for T. tabaci. Amplicons for COI were sequenced bidirectionally on an ABI 3730 (Macrogen Inc., Seoul, South Korea). Sequences were edited using CodonCode Aligner and aligned using Geneious™ (Drummond et al., 2010). Haplotype networks were constructed using the TCS method (Clement et al., 2000) implemented in the Popart computer program. Additional COI sequences were then obtained for T. tabaci and F. occidentalis from Genbank using a blastn search, so that the global phylogeny could be constructed for each. Sequences were aligned using the Geneious aligner, and Bayesian trees constructed with MrBayes (Huelsenbeck & Ronquist, 2001; Ronquist & Huelsenbeck, 2003), using the GTR+I+G model, which was implicated as the most likely after analysis with jmodeltest (Darriba et al., 2012; Guindon & Gascuel, 2003).

6.2.9 Statistical analysis A two way analysis of variance was used to analyze variations in total plant abundance and total number of species sampled between sampling events and between sites. Interactions between site and sampling event could not be analysed replicate samples for each species at each site were not collected. The mean total abundance of T. tabaci per unit and total number of hosts from which it was recorded were analysed for differences between sampling events and sites using a two-way analysis of variance.

128

To understand the overall importance of each host species the mean number of adult thrips per sample was plotted against the number of times thrips were recorded on that species. When the year was divided into seasons this analysis shows which hosts each species of thrips were most commonly recorded from and which hosted the largest numbers of thrips. This is valuable as some hosts may be abundant but host many or few thrips while others may be rare but host many or few thrips.

6.3 Results

6.3.1 Seasonal abundance of potential host plants A total of 69 plant species from 20 families was sampled in the Namoi Valley (Tables 6.1-6.3).

Total plant abundance differed significantly between sampling events (F12,126 = 5.56, P < 0.001) and between sites (F11,126 = 5.38, P < 0.001) (Figure 6.10A) (See Figure 6.11 & 6.12 showing examples of how sites differed through seasons). Plant abundance declined progressively through the sampling period, this decline was indicated by a significant negative relationship between total plant 2 abundance and date of sampling event (F1,148 = 24.41, P < 0.001, r = 13.6, slope = -0.21). Variation in total plant abundance showed that sites 1, 4 and 7 tended to have higher plant abundance while sites 8, 11 and 12 had lower plant abundance (Figure 6.10B). The number of plant species showed similar patterns to those for total plant abundance. The number of species available for sampling differed between sample events (F12,126 = 8.79, P < 0.001) and declined over time (Figure 6.13A). Similarly, there was variability between sites in the number of plant species sampled (F12,126 = 5.56, P < 0.001) with sites 1 and 4 having larger numbers of species and sites 8, 11 and 12 having fewer (Figure 6.13B). There was a strong positive relationship between the mean number of plant species and the mean total abundance of plants (F1,148 = 385.2, P < 0.001, r2 = 72.1, slope = 0.36).

6.3.2 Effect of rainfall on abundance of thrips’ host plants

There was a strong positive relationship between the mean total plant abundance and rainfall (F1,11 = 7.075, P = 0.022, r2 = 0.33; Figure 6.14).

129

Table 6.4 Total numbers of adult and larval Thrips tabaci, Frankliniella occidentalis and F. schultzei, collected from host plant species, summed across all 12 sample sites (Figure 6.4) in the Namoi valley, 2011-2013.

Thrips species1 Plant family Plant species Common name FO FS TT Adult Larvae Adult Larvae Adult Larvae Apiaceae Ammi majus L. Bishops weed 65 2 3365 2188 Asteraceae Sonchus oleraceus L. Common sowthistle 52 21 24 2 Xanthium occidentale Bertol. Noogoora burr 25 6 3 Conyza bonariensis (L.) Levyns Flaxleaf fleabane 2 Arctotheca calendula (L.) Levyns Cape weed 68 60 19 12 15 5 Bidens pilosa L. Cobbler’s peg 34 1 1 53 14 Helianthus annuus L. Sunflower 4 21 3 25 Hypochaeris radicata L. Catsear 5 2 4 Cotula australis Carrot weed 19 Brassicaceae Capsella bursa-pastoris (L.) Medik Shepherd’s purse 2 10 Rapistrum rugosum (L.) Turnip weed 770 941 71 39 293 92 Lepidium bonariense Peppercress 7 49 10 Echium plantagineum L. Patterson’s curse 210 113 28 5 190 135 Convolvulaceae Convolvulus erubescens Sims Australian bindweed 15 1 32 14 Operculina aequisepala (Domin) Onion vine 1 1 37 20 Cucurbitaceae Citrillus lanatus Camel melon 3 Fabaceae Medicago polymorpha L. var. Burr medic 28 2 2 34 1

130

vulgaris Vicia sativa L. Common vetch 53 22 51 25 28 5 Medicago sativa L. ssp. Sativa Lucerne 324 290 28 37 82 83 Sesbania canabina (Retz.) Pers. Sesbania 12 1 1 Fumariaceae Fumaria capreolata L. White flowered 2 1 1 4 1 fumitory Lamiaceae Lamium amplexicaule L. Deadnettle 1 3 1 2 Malvaceae Hibiscus tridactylites Lindley Bladder ketmia 8 51 Malva parviflora L. Marshmallow 4 1 1 Sida fibulifera Lindl. Pin sida 1 Nyctaginaceae Boerhavia dominii Meikle & Tarvine 2 4 1 1 1 Hewson Poaceae Hordeum leporinum Link Barley grass 1 Cynodon dactylon (L.) Pers. Couch 2 Lolium perenne Perennial Ryegrass 1 3 Bromus catharticus Vahl. Prairie grass 1 Avena fatua L. Wild oats 12 14 1 Polygonaceae Polygonum aviculare L. Wireweed 1 29 13 Polygonum orientale L. Prince’s plume 1 10 5 Rumex crispus L. Curled dock 1 Solanaceae Solanum nigrum L. Blackberry nightshade 48 2 16 126 2 Verbenaceae Phyla canescens (Kunth) Greene Lippia 30 1 8

131

Verbena bonariensis L. Purpletop 14 4 6 Verbena officinalis L. Vervain 39 35 17 24 Verbena tenuisecta Moss verbena 49 8 36 19 Zygophyllaceae Tribulus terrestris L. Cathead 35 4 59 31 3 Tribulus micrococcus Domin Yellow vine 35 20

1 FO = Frankliniella occidentalis, FS = Frankliniella schultzei, TT = Thrips tabaci.

132

Table 6.5 Total numbers of other thrips species collected from the plants species listed in table 6.1. The only plant species listed here that does not also appear in Table 6.1 is Centaurea melitensis.

Other species of thrips1 Plant family Plant species Common name TI PAA PPA TF MA Apiaceae Ammi majus Bishops weed 126 161 Asteraceae Sonchus oleraceus Common sowthistle 96 635 Xanthium occidentale Noogoora burr 4 3 Centaurea melitensis Maltese cockspur 10 Conyza bonariensis Flaxleaf fleabane 4 Arctotheca calendula Cape weed 25 Bidens pilosa Cobbler’s peg 8 15 6 Helianthus annuus Sunflower 21 10 Hypochaeris radicata Catsear 1 3 Brassicaceae Capsella bursa-pastoris Shepherd’s purse 1 Rapistrum rugosum Turnip weed 2632 379 Lepidium bonariense Peppercress 6 Echium plantagineum Patterson’s curse 165 75 Convolvulaceae Convolvulus erubescens Australian bindweed 5 62 Operculina aequisepala Onion vine 2 Cucurbitaceae Citrillus lanatus Camel melon 1 Fabaceae Medicago polymorpha Burr medic 4 10 Vicia sativa Common vetch 21 12

133

Medicago sativa Lucerne 23 65 6 Sesbania canabina Sesbania 2 Lamiaceae Lamium amplexicaule Deadnettle 1 Malvaceae Hibiscus tridactylites Bladder ketmia 2 1 Malva parviflora Marshmallow 2 Poaceae Cynodon dactylon Couch 7 Lolium perenne Perennial Ryegrass 6 Avena fatua Wild oats 20 Polygonaceae Polygonum aviculare Wireweed 2 Rumex crispus Curled dock 17 Polygonum orientale Prince’s plume 1 4 1 Solanaceae Solanum nigrum Blackberry nightshade 8 1 Verbenaceae Phyla canescens Lippia 21 Verbena bonariensis Purpletop 2 Verbena officinalis Vervain 16 11 verbena tenuisecta Moss verbena 25 83 Zygophyllaceae Tribulus terrestris Cathead 7 70

1 TI = Thrips imaginis, PAA = Pseudanaphothrips achaetus, PPA = Pseudanaphothrips parvus, TF = Tenothrips frici and MA = Microcephalothrips abdominalis

134

Table 6.6 Plant species from which no thrips species were collected during the surveys detailed in Tables 6.1 and 6.2.

Plant family Plant species Common names Amaranthaceae Gomphrena celosioides Mart. Gomphrena weed Aizoaceae Tetragonia tetragonioides (Pallos) Kuntze New Zealand spinach Trianthema portulactastrum L. Black pigweed Lactuca serriola L. Prickly lettuce Chenopodiaceae Salsola australis R.Br. Soft roly-poly Einadia hastata (R.Br.) A.J.Scott Berry saltbush Convolvulaceae Ipomoea lonchophylla J.M.Black Cowvine Cyperaceae Cyperus bifax C.B. Clarke Downs nutgrass Cyperus eragrostis Lam. Umbrella sedge Euphorbiaceae Ricinus communis L. Castor oil Fabaceae Acacia farnesiana (L.) Willd. Mimosa bush Rhynchosia minima (L.) DC. Ryncho Malvaceae Malvastrum coromandelianum (L.) Garcke Prickly malvastrum Sida rhombifolia Paddy’s lucerne Mimosaceae Neptunia gracilis Benth. Native sensitive plant Poaceae Chloris virgata Sw. Feathertop rhodes grass Chloris gayana Kunth Rhodes grass Dactyloctenium radulans (R.Br) P.Beauv Button grass Dichanthium sericeum S.T. Blake ssp. Sericeum Silky bluegrass Echinochloa crus galli Barnyard grass Sorghum halapense (L.) Pers. Johnson grass Cenchrus ciliaris Buffel grass Lolium perenne L. Perennial ryegrass Paspalum dilatatum Poir. Caterpillar grass Solanaceae Solanum esuriale Quena

135

(A) Sp Su W Sp Su A W Sp 40

10 SE) plant abundance (% groundcover) abundance plant SE)

0 Mean Mean (

Sample event

(B) 40

at each siteeach at 15 plant abundance (% groundcover) (% groundcover) abundance plant

SE) SE)

5 Mean Mean ( 0 1 2 3 4 5 6 7 8 9 10 11 12 Sites

Figure 6.10 Mean plant abundance (% groundcover). (A) Mean plant abundance across all sites. Sp=Spring, Su=Summer, A=Autumn, W=Winter (B) Mean plant abundance at each site.

136

Site 1: October 2011 Site 1: October 2012

Site 1: November 2011 Site 1: November 2012

Site 1: February 2012 Site 1: December 2012

137

Site 6: October 2011 Site 6: December 2012

Site 6: August 2012 Site 6: February 2013

Site 6: October 2012 Site 6: May 2013

138

(A) Sp Su W Sp Su A W Sp 20 18 16 14 12 10

number of of number species plant 8

SE) SE) 6

Mean Mean ( 2 0

Sample event

(B) 18 16

8 number of of number species plant

SE) SE) 6

Mean Mean ( 2

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

Figure 6.13 Mean number of plant species. (A) Number of plant species across all sites (B) Number of plant species at each site.

139

160

140

120

100

Rainfall (mm) (mm) Rainfall 60

0 0 10 20 30 40 Mean plant abundance (% groundcover)

Figure 6.14 Relationship between mean plant abundance and monthly rainfall, but the rainfall was offset by 1 month rainfall, y = 2.6506x, R² = 0.3274.

6.3.3 Seasonal abundance of thrips on non-cultivated hosts Thrips species were recorded on a total of 42 plant species (Tables 6.1 and 6.2). Twenty seven plant species yielded no thrips at all (Table 6.3). In addition to the three thrips species that were targeted in sampling (F. occidentalis, F. schultzei and T. tabaci) a further five thrips species were identified in the samples: Microcephalothrips abdominalis (Crawford DL), Tenothrips frici (Uzel), Thrips imaginis (Bagnall), Pseudanaphothrips achaetus (Bagnall) and P. parvus (Bagnall) (Table 6.2). Thrips tabaci

The total number of T. tabaci adults was significantly different between both sample events (F12,127

= 24.3; P < 0.001; Figure 6.15A) and sample sites (F11,127 = 2.63; P=0.004; Figure 6.16A).

Similarly, significant differences were observed for larval T. tabaci between sample events (F12,127

= 29.4; P < 0.001; Figure 6.17A) and sample sites (F11,127 = 1.97; P = 0.036; Figure 6.18A). The total abundance of both adult and larval T. tabaci broadly declined over time, but showed seasonal peaks of abundance in spring and summer each year (Figures 6.15A, 6.17A). Across the sites the abundance of adult T. tabaci was significantly higher at sites 1, 4 and 9 than at other sites (Figures 6.16A) and the abundance of larval T. tabaci was higher at sites 1, 3, 8 and 9 than at other sites

140

(Figure 6.18A). The total number of hosts from which adult T. tabaci was recorded showed significant differences between sample events (F12,127 = 12.1, P < 0.001; Figure 6.19A) and sample sites (F11,127 = 1.93, P=0.040; Figure 6.20A). There was a significant positive relationship between mean plant abundance and the number of hosts on which T. tabaci adults were found (F1,11= 6.43, P=0.028, r2= 0.31; Figure 6.21A). Similarly significant differences were observed for larval thrips between sample events (F12,127 = 21.6; P < 0.001; Figure 6.22A) and sample sites (F11,127 = 1.91; P= 0.044; Figure 6.23A). Across sample dates there was an overall trend towards T. tabaci adults and larvae being recorded on progressively fewer hosts (Figure 6.19 A & 6.22A). The number of host plant species from which adult T. tabaci was recorded was highest at sites 1 and 9 and lowest at sites 5 and 12 (Figure 6.20A). The number of host plant species from which T. tabaci larvae was recorded was highest at sites 1, 3 and 9 and lowest on site 12 (Figure 6.23A). Frankliniella occidentalis The total number of F. occidentalis adults was significantly different between both sample events

(F12,127 = 17.8, P < 0.001; Figure 6.15B) and sample sites (F11,127 = 2.1, P = 0.023; Figure 6.16B).

Similarly significant differences were observed for larval thrips between sample events (F12,127 =

14.0; P < 0.001; Figure 6.17B) and sample sites (F11,127 = 2.76; P = 0.003; Figure 6.18B).The total abundance of both adult and larval F. occidentalis broadly declined over time, but showed seasonal peaks of abundance in spring (Figures 6.15B, 6.17B). Across the sites the abundance of adult F. occidentalis was significantly higher at Sites 3, 5 and 7 than at other sites (Figure 6.16B) and the abundance of larval F. occidentalis was higher at Sites 1, 3, 5 and 7 than at other sites (Figure 6.18B). The total number of host plant species from which adult F. occidentalis was recorded showed significant differences between sample events (F12,127 = 21.3; P < 0.001; Figure 6.19B) but not for sample sites (F12,127 = 0.92; P=0.515; Figure 6.20B). Similar to adult T. tabaci, there was a strong positive relationship between the mean plant abundance and the number of hosts on which F.

2 occidentalis adults were found (F1,11= 10.6, P < 0.001, r = 0.44; Figure 6.21B). Significant differences were also observed for larval thrips between sample events (F12,127 = 19.8; P < 0.001; Figure 6.22B) and sample sites (F11,128 = 3.51; P<0.001; Figure 6.23B). Again, similar to T. tabaci, across sample dates there was an overall trend toward F. occidentalis adults and larvae being recorded on progressively fewer hosts (Figure 6.19B & 6.22B). The number of host plant species from which adult F. occidentalis was recorded was highest at Sites 1, 4 and 9 and lowest at Site 12 (Figure 6.20B). The number of host plant species from which F. occidentalis larvae was recorded was highest at Sites 1 and 6 and lowest at Sites 4 and 12 (Figure 6.20B). Frankliniella schultzei The total number of F. schultzei adults per site per sample event showed significant differences between sample events (F12,126 = 7.64, P < 0.001; Figure 6.15C) and sample sites (F11,126 = 2.06, P = 141

0.027; Figure 6.16C). Similarly significant differences were observed for larval thrips between sample events (F12,127 = 14.0; P < 0.001; Figure 6.17C) and sample sites (F11,127 = 2.76; P = 0.003; Figure 6.18C). The total abundance of adult and larval F. schultzei was much lower than for T. tabaci or F. occidentalis and showed small seasonal peaks of abundance in late spring or autumn each year (Figure 6.15C). The total number of hosts from which adult F. schultzei was recorded showed significant differences between sample events (F12,124 = 7.38; P < 0.001; Figure 6.19C) and sample sites (F12,124 = 1.96; P = 0.039; Figure 6.20C). However, there wasn’t a strong relationship between the total number total plant abundance and the number of hosts on which F. schultzei

2 adults were found (F1,11= 2.47, P=0.143, r = 0.10; Figure 6.21C). The total number of hosts from which larval F. schultzei was recorded showed significant differences between sample events (F12,126 =

2.31, P = 0.010, Figure 6.22C) however, no significant differences were found between sites (F11,126 = 1.16, P = 0.317, Figure 6.23C).

142

(A) Sp Su W Sp Su A W Sp (B) Sp Su W Sp Su A W Sp

250 250 adults adults 200 200

adults per sample 150 150

100 100

per sample per

Thripstabaci

Frankliniella occidentalis occidentalis Frankliniella SE)

50 SE) 50

Mean Mean (

0 Mean ( 0

Sample event Sample event

adults per per adults 40 35 30 Figure 6.15. Mean number of adult thrips per host plant 25 species across all sites, Namoi valley, 2011-2013. (A) sample 20 Thrips tabaci (B) Frankliniella occidentalis (C) F. 15 Frankliniella schultzei schultzei Frankliniella schultzei. Sp=Spring, Su=Summer, A=Autumn,

SE) SE) 10 W=Winter.

Mean Mean ( 0

Sample event

143

(A) 80 (B) 80

70 70 adults adults

60 60

50 50 adults per per adults sample

40 40

30 per sample 30

Thrips tabaci tabaci Thrips

Frankliniella occidentalis occidentalis Frankliniella SE) SE)

20 SE) 20

10 10

Mean Mean ( Mean Mean ( 0 0 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Sites Sites

40 adults per per adults

sample Figure 6.16. Mean number of adult thrips per host plant 20

species at each site, Namoi valley, 2011-2013. (A) Frankliniella schultzei schultzei Frankliniella

Thrips tabaci (B) Frankliniella occidentalis (C) F. SE) SE)

10 schultzei. Mean Mean ( 0 1 2 3 4 5 6 7 8 9 10 11 12 Sites

144

(A) Sp Su W Sp Su A W Sp (B) Sp Su W Sp Su A W Sp 160 160

140 larvae 140

120 120

larvae per sample larvae 100 100

80 80

60 per sample 60 Thrips tabaci tabaci Thrips

40 Frankliniellaoccidentalis 40

SE) SE)

SE)

20 20

Mean Mean ( 0 0 Mean Mean (

Sample event Sample event

40 larvae per larvae

30 Figure 6.17. Mean number of larval thrips per host

sample 20 plant species across all sites, Namoi valley, 2011-2013. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. Frankliniella schultzei schultzei Frankliniella schultzei. Sp=Spring, Su=Summer, A=Autumn,

SE) SE) 10

W=Winter.

Mean Mean ( 0

Sample event

145

(A) 50 (B) 50 larvae larvae 40 40

larvae larvae per sample 30 30

20 per sample 20

Thripstabaci

Frankliniella occidentalis occidentalis Frankliniella

SE)

SE) SE)

10 10

Mean Mean ( Mean Mean ( 0 0 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Sites Sites

40 larvae per larvae

Figure 6.18. Mean number of larval thrips per host sample 20 plant species at each site, Namoi valley, 2011-2013.

Frankliniella schultzei schultzei Frankliniella (A) Thrips tabaci (B) Frankliniella occidentalis (C) F.

SE) SE) schultzei.

10 Mean Mean ( 0 1 2 3 4 5 6 7 8 9 10 11 12 Sites

146

(A) Sp Su W Sp Su A W Sp (B) Sp Su W Sp Su A W Sp 8 8

7 7

6 6

5 5

4 4

3 were recorded adults 3 adults were recorded were adults

2 2

number of plant species from from of which number species plant from of which number species plant

T. tabaci tabaci T. SE) SE)

SE) SE) 1 1

F. occidentalis occidentalis F.

0 0

Mean Mean ( Mean (

Sample event Sample event

8 F. F. 7

4 Figure 6.19. Mean number of plants species from which adult thrips were recorded across all sites, adults were recorded were adults 3 Namoi valley, 2011-2013. (A) Thrips tabaci (B)

number of plant species which which of number species plant 2 Frankliniella occidentalis (C) F. schultzei. Sp=Spring,

SE) SE) schultzei schultzei 1 Su=Summer, A=Autumn, W=Winter.

0 Mean Mean (

Sample event

147

(A) 3 (B) 3

2 2

adults were were adults recorded adults were recorded were adults

1 1

number of plant species from from of number which species plant

number of plant species from from of which number species plant

T. tabaci tabaci T.

SE) SE)

F. occidentalis occidentalis F. Mean Mean ( 0 Mean ( 0 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Sites Sites

adults were recored were adults Figure 6.20. Mean number of plants species from

1 which adult thrips were recorded at each site, Namoi

number of plant species from from of which number species plant valley, 2011-2013. (A) Thrips tabaci (B) Frankliniella

SE) SE) F. schultzei schultzei F.

occidentalis (C) F. schultzei.

Mean Mean ( 0 1 2 3 4 5 6 7 8 9 10 11 12 Sites

148

(A) 7 (B) 7

T. T. F. 6 6

5 5 y = 0.1358x - 1.5152 4 R² = 0.3692 4 y = 0.0766x - 1.0299

3 recorded were adults 3 R² = 0.4917 adults were recorded were adults

2 2

tabaci tabaci occidentalis occidentalis

1 1

Mean number of plant species species from plant of which Mean number species from plant of which Mean number 0 0 0 10 20 30 40 0 10 20 30 40 Mean plant abundance (% groundcover) Mean plant abundance (% groundcover)

5 Figure 6.21. Relationship between mean number of hosts on 4 which thrips were found and number of hosts available (A) Thrips tabaci, y = 0.1358x – 1.5152, R² = 0.3692 (B) 3 adults were recorded were adults Frankliniella occidentalis, y = 0.0766x – 1.0299, R² = y = 0.0086x - 0.0638 0.4917 (C) F. schultzei, y = 0.0086x – 0.0638, R² = 0.1839.

2 R² = 0.1839 schultzei schultzei

1 Mean number of plant species species from plant of which Mean number 0 0 10 20 30 40 Mean plant abundance (% groundcover)

149

(A) Sp Su W Sp Su A W Sp (B) Sp Su W Sp Su A W Sp 7 7

6 6

5 5

4 4

3 3

larvae recorded were larvae larvae were were larvae recorded

2 2

number of plant species from from of which number species plant number of plant species from from of which number species plant

T. tabaci tabaci T. 1 1

SE) SE)

F. occidentalis occidentalis F.

0 0

Mean Mean ( Mean Mean (

Sample event Sample event

4 Figure 6.22. Mean number of plants species from 3 which larval thrips were recorded across all sites, larvae were were larvae recorded Namoi valley, 2011-2013. (A) Thrips tabaci (B) 2

number of plant species from from of which number species plant Frankliniella occidentalis (C) F. schultzei. Sp=Spring, 1

SE) SE) Su=Summer, A=Autumn, W=Winter.

F. schultzei schultzei F.

0 Mean Mean (

Sample event

150

(A) 3 (B) 3

2 2

larvae larvae recorded were larvae were were larvae recorded

1 1

number of plant species from from of which number species plant

T. tabaci tabaci T.

SE) SE)

F. occidentalis occidentalis F. Mean Mean ( 0 Mean ( 0 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Sites Sites

Figure 6.23. Mean number of plants species from larvae were were larvae recorded 1 which larval thrips were recorded at each site, Namoi

number of plant species from from of which number species plant valley, 2011-2013. (A) Thrips tabaci (B) Frankliniella

SE) SE) F. schultzei schultzei F.

occidentalis (C) F. schultzei.

Mean Mean ( 0 1 2 3 4 5 6 7 8 9 10 11 12 Sites

151

6.3.4 Seasonal use of non-cultivated hosts by thrips Thrips tabaci A total of 4545 T. tabaci adults were collected from 31 different plant species and a total of 2638 larvae were collected from 20 different plant species (Table 6.1). The vast majority of adults (89%) were collected from just four plant species, Ammi majus (n=3365 individuals), Rapistrum rugosum (n=293), Echium plantagineum (n=190), Solanum nigrum (n=126) with the remaining 11% collected from the 27 other plant species (n=541) (Table 6.1). Most larvae (92%) were collected from four plant species, A. majus (n=2188), R. rugosum (n=92), E. plantagineum (n=135), Medicago sativa (n=83) while the remaining 8% collected from the 17 other plant species (n=201) (Table 6.1). Significantly more adults were collected from flowers than leaves for the species on which thrips were most abundant, A. majus (t= 6.335, d.f. = 54, P<0.001; Figure 6.24A), R. rugosum (t = 4.706, d.f. = 88, P<0.001; Figure 6.24A), E. plantagineum (t = 3.609, d.f. = 49, P<0.001; Figure 6.24A), S. nigrum (t = 2.664, d.f. = 22, P=0.014; Figure 6.24A) but no significant difference was found for other plant species (t = 3.609, d.f. = 49, P<0.001; Figure 6.24A). Similarly more larvae were also collected from flowers than from leaves; A. majus (t= 6.335, d.f. = 54, P<0.001; Figure 6.25A), R. rugosum (t = 4.706, d.f. = 88, P<0.001; Figure 6.25A), E. plantagineum (t = 3.609, d.f. = 49, P<0.001; Figure 6.25A), S. nigrum (t = 2.664, d.f. = 22, P=0.014; Figure 6.25A). Further analysis of seasonal patterns considered differences between sample events and sites for the four plant species (A. majus, E. plantagineum, S. nigrum, R. rugosum) and the remaining plant species combined (other species). Thrips tabaci adults showed significant differences in abundance between plant species (F12, 326 = 57.6, P < 0.001; Figure 6.26A). Similarly within each species group there were significant differences between sample events (A. majus ; F7, 42 = 15.9, P <

0.001, R. rugosum ; F10, 76 = 4.0, P < 0.001, S. nigrum ; F9, 12 = 6.5, P = 0.002, E. plantagineum ; F11,

35 = 6.7, P < 0.001, other species.; F12, 120 = 9.82, P < 0.001) (Figure 6.26A). However, there were no significant differences between sites. Overall, T. tabaci adults were more abundant on A. majus than on S. nigrum, E. plantagineum, R. rugosum or other species. Analysis of the seasonal abundance of thrips showed that in spring A. majus was the most common host plant and hosted large numbers of T. tabaci adults (Figure 6.27) and larvae (Figure 6.28). Further, in spring, R. rugosum was also a key host along with E. plantagineum and S. nigrum. Amongst the ‘other spp.’ B. pilosa was an important spring host for adult T. tabaci (Figure 6.27). Rapistrum rugosum and E. plantagineum were key spring hosts for larval T. tabaci and amongst the ‘other spp.’ B. pilosa and M. sativa were also important. In summer, A. majus was the most common host plant and once again hosted large numbers of T. tabaci adults (Figure 6.27) and larvae (Figure 6.28). In autumn, species such as B. pilosa, E. plantagineum, S. oleraceus and B. dominii were the most important 152 hosts for adult T. tabaci, while in winter it is predominantly R. rugosum (Figure 6.26). For larval T. tabaci, in autumn, Xanthium occidentale was the only host that recorded thrips and in winter, R. rugosum and Capsella bursa-pastoris were the most important hosts (Figure 6.28). Overall both adult and larval T. tabaci were more abundant in spring and summer.

Frankliniella occidentalis A total of 1937 adult F. occidentalis were collected from 36 different plant species and 1496 larvae from 20 different plant species (Table 6.2). The majority of adult thrips (68%) were collected from just three plant species; R. rugosum (n=770), E. plantagineum (n=210), M. sativa (n=324) although smaller numbers were also found from a range of other plant species (n=623, = 32% of total) (Table 6.2). The majority of larval thrips (90%) were collected from R. rugosum (n=941), E. plantagineum (n=113), M. sativa (n=290) and smaller numbers were collected from the 13 other plant species (n=142, 10% of total) (Table 6.2). Significantly more adults were collected from the flowers than the leaves of plants: R. rugosum (t = 4.706, d.f. = 88, P<0.001; Figure 6.24B) E. plantagineum (t = 3.609, d.f. = 49, P<0.001; Figure 6.23B), other plant species (t = 7.585, d.f. = 819, P<0.001; Figure 6.24B); however no significant differences were found for M. sativa (t = 1.654, d.f. = 10, P=0.129; Figure 6.24B). Similarly more larvae were also collected from flowers than the leaves: R. rugosum (t= 5.292, d.f. = 88, P<0.001; Figure 6.25B), E. plantagineum (t= 2.615, d.f. = 49, P=0.011; Figure 6.25B), other plant species (t = 4.173, d.f. = 817, P<0.001; Figure 6.25B) however no significant differences were found for M. sativa (t = 1.509, d.f. = 10, P=0.162; Figure 6.25B). Further analysis of seasonal patterns considered differences between sample events and sites for the three plant species that had the highest numbers of thrips (R. rugosum, E. plantagineum, M. sativa) and the remaining plant species combined (other species). There were significant differences in the abundance of F. occidentalis adults between plant species (F12, 281 = 17.46, P < 0.001; Figure 6.26B). Similarly, within each species group there were significant differences between sample events (R. rugosum; F11, 77 = 7.24, P < 0.001, E. plantagineum; F11, 41 = 12.18, P < 0.001, other species; F12, 129 = 11.45, P < 0.001; Figure 6.26B). However no significant differences were found between sample events for M. sativa (F4, 6 = 0.63, P = 0.654; Figure 6.26B). Overall, F. occidentalis adults were more abundant on R. rugosum than on E. plantagineum, M. sativa or other species. The seasonal analysis showed that in spring R. rugosum, M. sativa, E. plantagineum and Vicia sativa were the most significant hosts of adult F. occidentalis (Figure 6.29). For larval thrips, in spring, R. rugosum, E. plantagineum, V, sativa, Verbena officinalis and M. sativa were the most significant hosts (Figure 6.30). In summer, R. rugosum and M. sativa were the most important hosts for both adult (Figure 6.29) and larval thrips (Figure 6.30). In autumn, R. rugosum, M. sativa, B. pilosa and X. occidentale were the most important for adults (Figure 6.29), while R. rugosum, M. sativa, B. 153 pilosa and S. nigrum were the most important for larvae (Figure 6.30). In winter, R. rugosum, Medicago polymorpha and Hypochaeris radicata were the most important for adults (Figure 6.29), while larval thrips were only recorded on R. rugosum (Figure 6.30). Overall both adult and larval F. occidentalis were more abundant in spring and summer.

Frankliniella schultzei A total of 494 adult F. schultzei were collected from 25 different plant species and a total of 143 F. schultzei larvae were collected from 8 different plant species (Table 6.1). Most adult thrips (53%) were collected from Hibiscus tridactylites (n=51), R. rugosum (n=71), Tribulus terrestris (n=59), V. sativa (n=51) and smaller numbers (n=262, 47% of total) were found from a range of 22 other plant species. Most larval thrips were collected from R. rugosum (n=39), V. sativa (n=25), M. sativa (n=37), Tribulus micrococcus (n=20) and 5 other plant species (n= 22) (Table 6.1). Significantly more adults were collected from the flowers than the leaves of some plant species, R. rugosum (t = 2.217, d.f. = 86, P=0.029; Figure 6.24C), T. terrestris (t = 2.555, d.f. = 21, P=0.018; Figure 6.24C) and other plant species (t = 4.710, d.f. = 135, P<0.001; Figure 6.24C) but no significant differences were found between the numbers on leaves and flowers of H. tridactylites (t = 1.511, d.f. = 9, P=0.165; Figure 6.24C), and V. sativa (t = 1.149, d.f. = 8, P=0.283; Figure 6.24C). Similarly more larvae were also collected from flowers than leaves; R. rugosum (t = 1.759, d.f. = 86, P=0.082; Figure 6.25C). But no significant differences were found between the numbers on leaves and flowers of V. sativa (t = 1.044, d.f. = 12, P=0.317; Figure 6.25C) and M. sativa (t = 1.247, d.f. = 9, P=0.244; Figure 6.25C). Further analysis of seasonal patterns considered differences between sample events and sites for the four plant species (H. tridactylites, R. rugosum, T. terrestris, V. sativa) and the remaining plant species combined (other species). F. schultzei adults showed significant differences in abundance between plant species (F12, 257 = 6.361, P < 0.001; Figure 6.26C). Within each species group there were significant differences between sample events (H. tridactylites: F3, 7 = 0.46; P =

0.719; R. rugosum: F10, 77 = 6.121, P < 0.001; T. terrestris: F6, 16 = 3.344, P= 0.025; other species:

F12, 129 = 11.45, P < 0.001; Figure 6.26C). However no significant differences were found between sample events for H. tridactylites (F3, 7 = 0.46, P = 0.719; Figure 6.26C). Analysis of seasonal host use by F. schultzei adults shows that in spring and summer, this species is spread across a range of host plants including T. terrestris, R. rugosum, Convolvulus erebescens and E. plantagineum at low densities (Figure 6.31). F. schultzei adults were most abundant in autumn and were mainly on R. rugosum, while in winter they were very rare and only recorded from Verbena tenuisecta. F. schultzei larvae were similarly only recorded in low numbers in spring, summer and autumn and not at all in winter. In spring the most significant hosts were M. 154

80 sativa, Arctotheca calendula70 and V. sativa, in summer on R. rugosum and M. sativa and in Autumn on T. micrococcus and60 R. rugosum.

recorded adults 50 40 80

(A) tabaci T. 30 Flowers Leaves 70 20 60 10

adultsrecorded 50

Mean Mean number of 0 40 Ammi majus Echium Other spp. Rapistrum Solanum

T. tabaci tabaci T. 30 plantagineum rugosum nigrum Plant species 20

0 Mean number number of Mean Ammi majus Echium Other spp. Rapistrum Solanum nigrum plantagineum rugosum Plant species

(B) 80

70 adults 60

40 F. occidentalis occidentalis F.

recorded 30

Mean number number of Mean 0 Echium Medicago sativa Other spp. Rapistrum rugosum plantagineum Plant species

adultsrecorded 50

30 F. schultzei schultzei F. 20

Mean number number of Mean Hibiscus Other spp. Rapistrum Tribulus Vicia sativa tridactylites rugosum terrestris Plant species

adults recorded adults 50

40 80 (A) tabaci T. 30 Flowers Leaves 70 20 60 10

larvae recorded larvae 50

Mean Mean number of 0 40Ammi majus Echium Other spp. Rapistrum Solanum plantagineum rugosum nigrum T. tabaci tabaci T. 30 Plant species 20

0 Mean number number of Mean Ammi majus Echium Other spp. Rapistrum Solanum nigrum plantagineum rugosum Plant species

(B) 80

70 larvae larvae 60

40 F. occidentalis occidentalis F.

recorded 30

Mean number number of Mean 0 Echium Medicago sativa Other spp. Rapistrum rugosum plantagineum Plant species

larvae recorded larvae 50

30 F. schultzei schultzei F. 20

Mean number number of Mean Hibiscus Other spp. Rapistrum Tribulus Vicia sativa tridactylites rugosum terrestris Plant species

156

(A) Sp Su W Sp Su A W Sp (B) Sp Su W Sp Su A W Sp

200 200 on host on

160 160 on host plants plants host on

120 120

F. occidentalis occidentalis F.

T. tabaci tabaci T. per per sample

80 80

per per sample plants

40 40 SE) number of of number SE)

SE) number of of number SE) 50

0 0 Mean Mean ( Mean Mean ( 40 160 35 sample event Sample event 50 30 Hibiscus tridactylites 120 (C) Sp Su W Sp Su 200 A W Sp 2545 Other spp. 50 per per 180 2040 Rapistrum rugosum 45 Ammi majus Tribulus terrestris 16080 35

on host on 15 40 Echium plantagineum Vicia sativa

event 140 1030 35 Other species Hibiscus tridactylites 120 F. occidentalis occidentalis F. 25 30 40 5 Rapistrum rugosum Other spp. F. schultzei schultzei F. 100 SolanumEchium plantagineum nigrum Rapistrum rugosum 25 200

SE) number of T. tabaci per sampling per tabaci T. of number SE) 80 Medicago sativa Tribulus terrestris

20 event sampling 15

Other species Jul-13 Vicia sativa Oct-11

60 Oct-12

Feb-12 Feb-13 Sep-13

0 Apr-13

Dec-12

Nov-11 Aug-12 Nov-12 Nov-13

May-13 plants per per plants sample

15 of number SE) 10 Rapistrum rugosum Mean ( Mean 40

SE) number of of number SE) Figure 6.26. Mean number of adult thrips recorded Jul-13

Oct-12 Oct-11

Feb-12 Feb-13 Sep-13 Apr-13

5Dec-12

Nov-11 Aug-12 Nov-12 Nov-13 20 May-13 on host plant species across all sites. (A) Thrips 5 Mean ( sample event 0 0 Mean Mean ( 0 tabaci (B) Frankliniella occidentalis (C) F.

schultzei. Sp=Spring,Jul-13 Su=Summer, A=Autumn,

Oct-11 Oct-12

Jul-13

Feb-12 Feb-13 Sep-13

Apr-13

Dec-12

Nov-11 Aug-12 Nov-12 Nov-13

Oct-11 Oct-12

Feb-12 Feb-13 Sep-13

May-13

Apr-13

Dec-12

Nov-11 Aug-12 Nov-12 Nov-13 May-13 Sampling event W=Winter. sample event

157

Spring Summer 45 12 Ammi majus 40 Ammi majus 10 35

30 8 adultssample per

Bidens pilosa adultssample per 25 Boerhavia dominii Polygonum aviculare 6

20 T. tabaci tabaci T. Solanumnigrum tabaci T. Solanum nigrum 15 4 Medicago polymorpha Rapistrum rugosum 10 Rapistrum rugosum Polygonum aviculare 2

5 Echium plantagineum Sonchus oleraceus Mean number number of Mean 0 number of Mean 0 0 5 10 15 20 25 30 35 40 45 0 2 4 6 8 10 12 14 16 Number of samples with T. tabaci adults present Number of samples with T. tabaci adults present

Autumn Winter 5 5

4 4 adultssample per adultssample per 3 3

Sonchus oleraceus T. tabaci tabaci T. T. tabaci tabaci T. 2 2 Boerhavia dominii Solanum nigrum Rapistrum rugosum 1 Echium plantagineum Bidens pilosa 1

Polygonum aviculare Mean number number of Mean Mean number number of Mean 0 0 0 1 2 3 4 5 6 0 2 4 6 8 10 12 Number of samples with T. tabaci adults present Number of samples with T. tabaci adults present Figure 6.27 Relative host use by Thrips tabaci adults based on the number of times they were recorded on that host and the mean numbers recorded on that host. Hosts in the upper right of each season plot are likely to be highly significant. 158

Spring Summer 30 6 Ammi majus Ammi majus 25 Operculina aequisepala 5

20 4 larvae sample per larvae

15 Medicago sativa 3 T. tabaci tabaci T. 10 2 Sonchus oleraceus

Rapistrum rugosum 5 Bidens pilosa Echium plantagineum 1 Polygonum aviculare

Rapistrum rugosum per sample tabaci larvae T. of Mean number Mean of Mean number 0 0 0 10 20 30 40 50 60 0 2 4 6 8 10 12 14 Number of samples with T. tabaci larvae present Number of samples with T. tabaci larvae present

Autumn Winter 6 6

5 5

4 4

larvae sample per larvae larvae sample per larvae

3 Capsella bursa-pastoris

T. tabaci tabaci T. T. tabaci tabaci T. 2 2

1 1

Xanthium occidentale Rapistrum rugosum Mean of Mean number 0 of Mean number 0 0 1 2 3 0 1 2 3 Number of samples with T. tabaci larvae present Number of samples with T. tabaci larvae present

Spring Summer 35 7 Medicago sativa 30 6 Medicago sativa

25 5

adults per sample per adults adults per sample per adults

20 4 Rapistrum rugosum

15 3 Rapistrum rugosum F. occidentalis occidentalis F. F. occidentalis occidentalis F. Vicia sativa 10 2

5 Echium plantagineum 1

0 0 Mean number number of Mean Mean number number of Mean 0 5 10 15 20 25 30 35 0 1 2 3 4 5 Number of samples with F. occidentalis adults present Number of samples with F. occidentalis adults present

Autumn Winter 7 7

6 6 Hypochaeris radicata

5 Rapistrum rugosum 5

adults per sample per adults adults per sample per adults

4 Medicago sativa 4

Xanthium occidentale 3

F. occidentalis occidentalis F. F. occidentalis occidentalis F. 2 2 Bidens pilosa Medicago polymorpha 1 1 Rapistrum rugosum

0 Mean number number of Mean

Mean number number of Mean 0 0 2 4 6 8 10 12 14 16 0 1 2 3 4 5 6 7 8 Number of samples with F. occidentalis adults present Number of samples with F. occidentalis adults present

Spring Summer 35 6 Medicago sativa 30 5

4 larvae per sample larvae Arctotheca calendula per sample larvae 20 Medicago sativa Rapistrum rugosum Rapistrum rugosum 3 15

Vicia sativa 2 F. occidentalis F. 10 occidentalis F. Verbena officinalis 1 5 Echium plantagineum

0 0 Mean of Mean number 0 5 10 15 20 25 30 35 of Mean number 0 1 2 3 4 5 6 Number of samples with F. occidentalis larvae present Number of samples with F. occidentalis larvae present

Autumn Winter 6 6

Rapistrum rugosum 5 5

4 4

larvae per sample larvae larvae per sample larvae

3 3

2 Medicago sativa 2

F. occidentalis F. F. occidentalis F.

1 Solanum nigrum 1 Rapistrum rugosum Bidens pilosa

0 0 Mean of Mean number Mean of Mean number 0 2 4 6 8 10 12 0 1 2 3 4 5 6 Number of samples with F. occidentalis larvae present Number of samples with F. occidentalis larvae present

Spring Summer 6 6 Vicia sativa 5 5 Arctotheca calendula Verbena tenuisecta

4 Tribulus terrestris 4

adultssample per adultssample per Vicia sativa Medicago sativa 3 3 Medicago sativa

Convolvulus erubescens F. schultzei schultzei F. F. schultzei schultzei F. Tribulus terrestris 2 Sonchus oleraceus Convolvulus erubescens 2 Rapistrum rugosum Verbena tenuisecta 1 1 Echium plantagineum Echium plantagineum Sonchus oleraceus Rapistrum rugosum

0 0

Mean number number of Mean number of Mean 0 1 2 3 4 5 6 7 8 9 0 2 4 6 8 10 12 14 Number of samples with F. schultzei present Number of samples with F. schultzei present

Autumn Winter 20 6 18 5 16 Verbena tenuisecta 14

4 adultssample per adultssample per 12 Helianthus annuus

10 Tribulus terrestris 3

8 Medicago sativa F. schultzei schultzei F. F. schultzei schultzei F. 2 6 Verbena tenuisecta Sonchus oleraceus 4 Echium plantagineum Rapistrum rugosum 1 2 Boerhavia dominii

0 0

Mean number number of Mean number of Mean 0 2 4 6 8 10 12 0 1 2 3 4 5 6 7 8 9 10 Number of samples with F. schultzei present Number of samples with F. schultzei present

Spring Summer 6 3

5 Echium plantagineum

4 2

larvae sample per larvae larvae sample per larvae Vicia sativa Rapistrum rugosum 3

Rapistrum rugosum Arctotheca calendula F. schultzei F. 2 schultzei F. 1 Medicago sativa 1 Medicago sativa

Lamium amplexicaule Mean of Mean number 0 of Mean number 0 0 1 2 3 4 5 6 0 1 2 3 Number of samples with F. schultzei larvae present Number of samples with F. schultzei larvae present

Autumn 6

4 Rapistrum rugosum larvae sample per larvae

3 Helianthus annuus F. schultzei F. 2 Medicago sativa Tribulus micrococcus 1 Boerhavia dominii

Mean of Mean number 0 0 2 4 6 8 10 12 Number of samples with F. schultzei larvae present

Figure 6.32 Relative host use by Frankliniella schultzei larvae based on the number of times they were recorded on that host and the mean numbers recorded on that host. Hosts in the upper right of each season plot are likely to be highly significant. 163

6.3.5 Seasonal abundance of thrips on cotton and other crops Cotton The total number of T. tabaci adults and larvae on cotton was significantly different between sampling events (adults: F7,17 = 21.32, P < 0.001; Figure 6.33A, larvae: F7,12 = 10.98, P < 0.001; Figure 6.33B). Similarly, significant differences were also observed between sample events for F. schultzei (F6,17 = 8.19; P < 0.001; Figure 6.33A) and F. occidentalis larvae ((F7,12 = 2.507, P=0.078;

Figure 6.33B). However no significant differences were found for F. occidentalis adults (F7,12 = 1.814, P=0.174; Figure 6.33A). Cotton phenology had a significant effect on thirps numbers in that, significantly more adult T. tabaci were collected from leaves than flowers (t = 6.201, d.f. = 128, P<0.001; Figure 6.34A) however, more F. occidentalis (t = 3.733, d.f. = 36, P<0.001; Figure 6.34B) and F. schultzei (t = 5.764, d.f. = 35, P<0.001; 6.34C) were collected from flowers than leaves.

Other crop species No thrips were found on Carthamus tinctorius, Helianthus annuus and Zea mays (Table 6.4). Thrips were also recorded on five other crop plants (Table 6.4). The total number of T. tabaci adults was significantly different between sample events (T. aestivum: F5,10 = 7.199, P < 0.001, B. napus: F2,5 = 469.5, P<0.001; Figure 6.35). Similarly the total number of F. occidentalis adults was significantly different between sample events (T. aestivum: F5,5 = 4.017, P=0.077, B. napus: F5,5 = 4.017, P=0.077; Figure 6.35). Also significantly more adult T. tabaci (t = 3.507, d.f. = 40, P = 0.001; Figure 6.36) and F. occidentalis (t = 2.563, d.f. = 40, P = 0.014; Figure 6.36) were collected from flowers than leaves of T. aestivum. No F. schultzei was recorded on either T. aestivum or B. napus. Small numbers of T. tabaci, F. occidentalis and F. schultzei were also recorded on Vicia faba and Cicer arietinum (Figure 6.35 & Table 6.4).

164

(A) Sp Su W Sp Su A W Sp 35

15 sample

10 SE) number of thrips recorded recorded per of number thrips SE)

Mean Mean ( 0

sample event

18 Sp Su W Sp (B) Su A W Sp 35 16 Sp Su W Sp Su A W Sp 14 35 30 Sp Su W Sp Su A W Sp 35 12 30 25 10 30 8 25 sample 20 6 25

20 4 SE) number of thrips recorded recorded per of number thrips SE)

sample 15

20 2 sample 15 10

Mean Mean ( 0 sample SE) number of thrips recorded recorded per of number thrips SE) 15 10

5 recorded per of number thrips SE)

10 SE) number of thrips recorded recorded per of number thrips SE) Mean Mean ( 5 Sample event 0 Frankliniella schultzei

Mean Mean ( 5 Oct-11 Nov-11 Feb-12 Aug-12 Oct-12 Nov-12 Dec-12 Feb-13Figure Apr-136.33 MeanMay-13 numbersJul-13 ofSep-13 adult thripsNov-13 per sample across all cotton crop sites, Namoi valley, 0 Frankliniella occidentalisFrankliniella schultzei Mean Mean ( sample event Oct-11 Nov-11 Feb-12 Aug-12 Oct-12 Nov-12 Dec-12 Feb-13 Apr-13 May-13 Jul-13 Sep-13ThripsNov-13 tabaci 0 2011-2013. (A) adult thrips (B) larval thrips. Thrips tabaci Frankliniella occidentalis occidentalis F.Frankliniella schultzei Oct-11 Nov-11sampleFeb-12 eventAug-12 Oct-12 Nov-12 Dec-12 Feb-13 Apr-13 May-13 Jul-13 Sep-13 Nov-13 Thrips tabaci Frankliniella occidentalis schultzei. Sp=Spring, Su=Summer, A=Autumn,sample event W=Winter. Thrips tabaci

165

(A) T. 9 8 7 6

5 per per sample number of of number adult 4

SE) SE) 3

tabaci tabaci 2 1

Mean Mean ( 0 leaves flowers Plant structure

(B) F. 12 10

8 per per sample

number of of number adult 6

SE) SE) 4

occidentalis occidentalis 2

Mean Mean ( 0 leaves flowers Plant structure

15 per per sample

number of of number adult 10 SE) SE)

5 schultzei schultzei

Mean Mean ( 0 leaves flowers Plant structure

Figure 6.34 Comparisons of thrips density on cotton flowers and cotton leaves. (A) Thrips tabaci (B) Frankliniella occidentalis (C) F. schultzei.

166

Thrips species1 Plant species Plant family Common name FO FS TT Adult Larvae Adult Larvae Adult Larvae Carthamus tinctorius L. Asteraceae Safflower 0 0 0 0 0 0 Helianthus annuus L. Sunflower 0 0 0 0 0 0 Brassica x napus L. var. napus Brassicaceae Canola 60 69 0 0 211 209

Gossypium hirsutum L. Malvaceae Cotton 483 181 623 332 902 259 Vicia faba L. Fabaceae Faba bean 39 10 6 1 84 19 Cicer arietinum L. Chickpea 3 0 12 0 26 0 Cajanus cajan (L.) Millsp. Pigeon pea 69 9 87 15 1 0 Triticum aestivum L. Poaceae Wheat 49 0 0 0 280 9 Zea mays Maize 0 0 0 0 0 0

1 FO = Frankliniella occidentalis, FS = Frankliniella schultzei, TT = Thrips tabaci

167

(A) Sp Su W Sp Su A W Sp (B) Sp Su W Sp Su A W Sp 160 80 140 120 60 100 80 40 60

SE) number of of number SE) thrips 20

40 SE)number of thrips

recorded sample recorded per recorded recorded sample per

0 0

Mean Mean ( Mean Mean (

Sample event Sample event

60 60

40 40

20 of number thrips SE)

SE) number of of number thrips SE)

recorded sample recorded per recorded sample recorded per

0 0

Mean Mean ( Mean Mean (

Sample event Sample event

(A) 10 T. T.

per per sample 4

SE) number of of number adult SE)

tabaci 2

Mean Mean ( 0 flowers leaves Plant structure

(B) F.

6 s per per s sample

4 SE) number of of number SE) adult

2 occidentali

Mean Mean ( 0 flowers leaves Plant structure

Figure 6.36 Comparisons of thrips density on Triticum aestivum flowers and leaves. (A) Thrips tabaci (B) Frankliniella occidentalis.

169

6.3.6 DNA extraction and COI sequencing: Molecular analysis of Thirps tabaci and Frankliniella occidentalis For T. tabaci four haplotypes were returned with most individuals represented by two haplotypes separated by a single mutation (Figure 6.29). Insects from all four host plants were represented by these two haplotypes. Similarly for F. occidentalis most individuals were represented by two haplotypes, with individuals from all ten host plants represented in these two haplotypes (Figure 6.30). Thus, no genetic differentiation was found across host plant species. This indicates that both T. tabaci and F. occidentalis are all from one breeding population.

170

Figure 6.38 Most Frankliniella occidentalis individuals were represented by two haplotypes, with individuals from all ten host plants represented in these two haplotypes.

CO1 sequence analysis has also revealed that both T. tabaci and F. occidentalis have multiple divergent lineages globally (Figure 6.39 & 6.40). Thrips tabaci was represented by at least three lineages globally; samples collected from the Namoi valley (current study) belonged to the polyphagous thelytokous lineage (Figure 6.39). Frankliniella occidentalis was represented by two lineages globally; individuals collected from the Namoi valley, belong to one of these lineages (Figure 6.40).

171

Figure 6.39 Global phylogeny of Thrips tabaci showing the three main lineages samples from current study are shown in blue (continued on following page).

172

173

Figure 6.40 Global phylogeny of Frankliniella occidentalis showing the two main lineages samples from current study are shown in blue (continued on following page).

174

175

6.4 Discussion

6.4.1 Seasonal abundance of weed species and thrips The aim of the study was to investigate the ecology of T. tabaci, F. occidentalis and F. schultzei in the cotton producing region of the Namoi valley. More specifically, it was aimed at determining how different host plants and seasonal conditions affected thrips abundance and composition. Results from the sampling surveys indicated that T. tabaci was the most numerous of the thrips species, followed by F. occidentalis, however F. schultzei was found in low densities. Most plant species from which thrips species were recorded were weeds. Thrips tabaci was recorded on 31 plant species from 15 families, F. occidentalis was recorded from 35 plants from 14 families and F. schultzei was recorded from 25 plant species from 13 families but at low densities (Table 6.1). Both T. tabaci and F. occidentalis were more abundant during the spring (September, October and November) and summer months (December, January and February) (Figures 6.15A & B) and decreased during the autumn (March, April, and May) and winter months (June, July and August). However the pattern of abundance of F. schultzei was less obvious (Figure 6.15C) on weed hosts, though it tended to be more abundant in autumn but virtually absent in winter. Collectively, the results demonstrate that the spatio-temporal abundance of both T. tabaci and F. occidentalis was ultimately related to the time of year and the abundance and diversity of the sampled weed species (Figures 6.15A & B, 6. 19A & B). The reason why both T. tabaci and F. occidentalis were more abundant during the spring months was because weed hosts on which they were recorded flowered during spring (Figures 6.19A & B) and senesced with the onset of hot dry weather (Cunningham et al., 1981). Similar trends in seasonal abundance of a variety of thrips species have been reported in different parts of the world (Chellemi et al., 1994; Atakan & Uygur, 2005; Katayama, 2006; Doederlein & Sites, 1993; Toapanta et al., 1996; Yudin et al., 1988; Chyzik & Ucko, 2002). For example, Atakan & Uygur (2005) reported that in the eastern Mediterranean region of Turkey, T. tabaci and Frankliniella spp were more abundant when flowering weed species on which they were recorded were in plentiful supply but numbers of thrips declined when the abundance of flowering weeds declined. Similarly Chellemi et al. (1994) also found that the abundance and diversity of weed host plants affected the abundance of thrips in north Florida. Thus, the results of the current study and other studies show that the abundance of thrips is heavily dependent of the abundance of flowering weeds. In the Australian cotton system the high abundance of weed hosts that flower in the spring probably explains the high abundance of thrips at this time. The abundance of weed species was also correlated with rainfall patterns (Figure 6.14). Thus it can be argued that, higher rainfall results in higher weed abundance that ultimately results in higher thrips abundance. 176

6.4.2 Host plant use by thrips The study shows that Ammi majus L. (Apiales: Apiaceae) played an important role in the seasonal abundance of T. tabaci, as both adult and larvae were found in high densities on this species (Figures 6.26, 6.27 & 6.28). When the abundance of A. majus decreased (autumn and winter months) the abundance of T. tabaci also decreased (Figures 6.26 & 6.27). This could suggest that A. majus is a primary host of T. tabaci and the other host plant species being only relatively minor hosts. Reasons for this pattern of association could be attributed to the colour of the flowers on the different weeds. Although the colour preference of T. tabaci on weeds is not clearly understood, it has been argued that T. tabaci responds most positively to white flowers (Atakan & Uygur, 2005), A. majus has large white flowers. Surveys conducted by Atakan and Uygur (2005) showed that T. tabaci was more abundant on Ammi visnaga L. (Apiales: Apiaceae) and Daucus carota (Apiales: Apiaceae), both of which are from the same family as A. majus and they have large white flowers of a similar structure. Thrips tabaci thus could have a strong association with certain species of plants in the family Apiaceae. Unlike T. tabaci which was more associated with one host plant species, F. occidentalis did not have a strong association with one particular host plant species (Figures 6.26B, 6.29 & Table 6.2). Doederlein & Sites (1993) also showed a similar pattern of host use by F. occidentalis, in that it was not associated with one host plant more strongly. Franklinella occidentalis has a wide host plant range (Atakan & Uygur, 2005; Katayama, 2006; Chyzik & Ucko, 2002), however, the current study did record relatively high densities of F. occidentalis on Rapistrum rugosum L. (Brassicales: Brassicaceae) compared to other host plant species (Figures 6.26B, 6.29 & Table 6.2). The flowers of R. rugosum are yellow and yellow-flowered weed species are known to be particularly attractive to F. occidentalis (Yudin et al., 1986; Blumthal et al., 2005). However patterns of host plant use by T. tabaci are different in other parts of the world and could even be different within Australia. It is evident that T. tabaci is also strongly associated with host plant species different to the ones reported in this study (Atakan & Uygur, 2005; Milne & Walter, 1998b; Yudin et al., 1988; Doederlein & Sites, 1993). Global analysis of CO1 sequences revealed three lineages of T. tabaci (Figure 6.39). One of these lineages is restricted to tobacco and has only been sampled in Greece and Iran (Fekrat et al., 2014; Brunner et al., 2004). The other two lineages are polyphagous and associated with reproductive mode, one thelytokous and one arrhenotokous. These two lineages are recorded from multiple countries, and share some hosts (i.e. onion and leek, Figure 6.39). It is likely, however, that the thelytokous and the arrhenotokous lineages have different host plant associations. Brunner et al. (2004) showed that the two lineages of T. tabaci have different host associations. Thrips tabaci from the Namoi valley examined in the current study belong to the polyphagous thelytokous parthenogenetic reproduction lineage (Figure 177

6.39). All samples in the current study were females, and no male T. tabaci have been collected in Australia (pers. comm. Desley Tree) supporting the proposition that at this stage Australia only has the thelytokous lineage present. What the current study and all other studies mentioned demonstrate is that host plant availability determines the abundance of T. tabaci, different host plants are available at different sites and different parts of the world and that different T. tabaci lineages could be associated with different host plant species. For pest management strategies for T. tabaci (or any other insect pest for that matter), it is important to make sure that one knows which linage of T. tabaci, one is dealing with.

6.4.3 The importance of flowers in determining abundance of thrips Patterns of thrips distributions within individual weed species were also evident in the current study. Both larval and adult thrips of all species were predominantly found on inflorescences of their host plants rather than leaves (Figures 6.23 & 6.24). Previous studies have shown that most thrips species, including the ones investigated in the current study feed primarily on pollen and are associated with flowers (Yudin et al., 1986; Gerin et al., 1999; Milne et al., 1996). This could be the reason why large numbers of thrips were found on the flowers of weed species compared to the leaves. Gerin et al. (1999) showed that flowers impacted the population growth of F. occidentalis; in the presence of flowers, the populations grew exponentially while in the absence there was no growth. Flowers are essential for the success of thrips development (Gerin et al., 1999; Lewis, 1997), for example, F. schultzei has been shown to perform better on floral tissue than on leaf tissue (Milne et al., 1996). It has even been shown that in the absence of flowers, thrips colonization is reduced, implying that thrips respond directly to flowers (Yudin et al., 1988). Thus both T. tabaci and F. occidentalis were found in high numbers during the spring months as there were more flowering weeds present during this time (Cunningham et al., 1981), enabling thrips populations to expand. Another factor could be the fact that many species of thrips are thigmotactic and seek protection in narrow crevices provided by the host plants (Lewis & Navas, 1962). Flowers with dense inflorescences provide crevices which protects thrips (Doederlein & Sites, 1993). Thus the structures and colour of inflorescences may influence the attractiveness to thrips. However these observations and extrapolations warrant further investigation. Understanding the role of flowers in maintaining and sustaining thrips populations is essential to understanding the ecology of thrips.

6.4.4 Thrips abundance and diversity on cotton and other crops The pattern of abundance of T. tabaci and F. schultzei on cotton in the Namoi valley was similar to that reported by Wilson and Bauer (1993). Thrips tabaci was the most abundant species of thrips on cotton early in the season followed by F. schultzei and F. occidentalis later in the season (Figure 178

6.33). The abundance of both F. schultzei and F. occidentalis increased significantly when cotton flowers emerged (Figures 6.33 & 6.34). It can be argued that the seasonal changes in both the abundance and composition of thrips on cotton are related to the abundance of thrips on the surrounding weed species. Thrips tabaci was most abundant during the spring months (October and November) (Figures 6.15A). The predominately high abundance of T. tabaci on early season cotton (Figure 6.33) could be attributed to the high abundance to T. tabaci on the surrounding weed species (Figure 6.15A). During the summer months (December and February), when cotton mature and flower, T. tabaci numbers collapses (Figure 6.33), this can be correlated with the decline of T. tabaci numbers on the surrounding weeds, also and more importantly, the preferred host of T. tabaci, A. majus also declined at this time (Figures 6.15A & 6.26A). Thus the decline in abundance of weed hosts associated with T. tabaci, results in an overall decline in T. tabaci that result in less T. tabaci on mature cotton. It can thus be argued that the abundance to weeds played an important part in determining the abundance of T. tabaci on cotton crops. Similar patterns have been observed in other studies, for example, few flowering weed species in summer resulted in low numbers of thrips in cotton fields in Turkey (Atakan et al., 1998). What was interesting about the pattern of abundance reported in the current study was the fact that T. tabaci was found in low numbers during the cotton flowering stage, which was surprising as they were found in high numbers on the flowering weed species, most importantly cotton flowers are white and presumably highly attractive. A similar pattern was observed with Frankliniella fusca (Hinds); they were also more abundant on seedling cotton than on the flowers in northern Florida (Osekre et al., 2009). This underscores the fact that the ecology of thrips is highly complex. Reasons for the pattern of abundance of F. occidentalis and F. schultzei on cotton crops are less obvious than with T. tabaci. Frankliniella occidentalis was abundant on the surrounding weeds during the spring months (Figure 6.12) but this wasn’t represented by a high abundance on early season seedling cotton (Figure 6.27). Abundance of F. occidentalis on cotton only became high during the summer months when cotton flowers emerged (Figure 6.33). This indicates that the pattern of abundance of F. occidentalis on cotton was not correlated with their abundance on the weeds but rather the presence of cotton flowers. It could be that F. occidentalis moved from the few surrounding flowering weeds available during the summer months, onto cotton flowers and increased in abundance. Apart from high densities of adults, high densities of F. occidentalis larvae were also recorded on cotton flowers (Figure 6.33). This indicates that unlike T. tabaci, the seasonal abundance of F. occidentalis on cotton is primarily driven by the cotton crop itself, more so the availability of cotton flowers. A similarly explanation could be given to the observed pattern of abundance of F. schultzei, which was also more abundant when there were cotton flowers (Figure 6.33). Although F. schultzei was recorded at low densities on the surrounding weed species (Figure 179

6.15), the emergence of cotton flowers probably enabled F. schultzei to increase in abundance. A similar pattern was also observed for Frankliniella tritici (Fitch), it was more abundant on cotton flowers than on seedling cotton in northern Florida (Osekre et al., 2009). Teulon and Penman, (1991) found that pollen and nectar are important food sources for thrips development and reproduction and hence this could explain the high abundance of both F. occidentalis and F. schultzei on cotton flowers rather than on seedling cotton. It has been reported that cotton flower pollen significantly improved the fitness and fecundity of F. occidentalis (Trichilo & Leigh, 1988). Collectively these results demonstrate that there are variations in the aggregation behaviors of T. tabaci, F. occidentalis and F. schultzei on cotton plant parts. All three species of thrips were also found on other crop plants (Table 6.4). A common folklore in the cotton industry is that wheat (Triticum aestivum) is a major source of T. tabaci populations. However the results of the current study indicate that wheat is unlikely to be a major source of T. tabaci, F. occidentalis or F. schultzei as very low densities of T. tabaci larvae and no F. occidentalis larvae or F. schultzei adults and larvae were recorded on wheat. It is most likely that thirps adults move onto wheat when its flowers, as predominantly both T. tabaci and F. occidentalis were found on flowers (Figure 6.36). These adults could ultimately move onto cotton. Thus it is more likely that wheat is only an incidental host of T. tabaci and F. occidentalis. It was also evident that both T. tabaci and F. occidentalis were predominately recorded on wheat when the flowers appeared (Figure 6.36), no thrips were recorded during the leaf stages of wheat (Figure 6.35A) indicating the attractiveness of flowers to thrips. Significant numbers of both adult and larval T. tabaci and F. occidentalis were however found on canola (Brassica x napus) (Figure 6.35B) thus raising the possibility of B. napus being an important source for both thrips.

6.4.4 Sources from which thrips invade cotton and management tactics The findings of the current study suggest that the surrounding weed species may be the sources from which thrips populations migrate onto nearby cotton fields in the Namoi area. COI sequence analysis has revealed that T. tabaci and F. occidentalis collected from a selected group of weed, crop species and cotton suggest they are all from one breeding population (Figures 6.29 & 6.30). It can be inferred that it is these thrips that move into the nearby cotton fields. This is consistent with the suggestion of Wilson and Bauer (1993) that T. tabaci migrates into cotton from nearby weed species, the same can be said about F. occidentalis as well. An important conclusion of this research was the fact that weeds play an important role in the population biology of thrips and in the potential pest management strategies. Results of this study indicate that it could be possible to predict large out breaks of thrips (particularly T. tabaci) on cotton. It was evident that high weed abundance and diversity produced more thrips that ultimately resulted in more thrips 180

(predominately T. tabaci) being recorded in cotton. From this, a strategy can be devised, whereby managers can monitor weed abundance to determine whether the influx of thrips to cotton seedlings could be high or low. Knowledge of, or even monitoring the arthropod pests hosted by weeds can permit a manager to better anticipate arthropod problems in adjacient crops (Norris & Kogan, 2000). It could be argued that whenever there is a high abundance of weeds, cotton seedlings should be more intensively protected. Also the current study was also able to determine that rainfall could play an important part in the spatio-temporal distribution of thrips. Results showed that weed abundance was high when rainfall was plentiful. Thus whenever rainfall is high, it results in a higher abundance of weeds that ultimately results in higher thrips abundance, once again it can be possible to predict higher outbreaks of thrips. However this warrants further investigation, as it predominantly depends on the amount of weeds relative to cotton and the timing of the weeds senescing.

181

Chapter 7 General Discussion

Three species of thrips, Thrips tabaci Lindeman, Frankliniella schultzei Trybom and F. occidentalis Pergrande (Thysanoptera: Thripidae), are pests of cotton in Australia (Williams et al., 2011). Their damage to stems and leaves can lead to deformation in new growth that can delay development, result in the early production of fruiting bodies and increase susceptibility to disease (Williams et al., 2011). Wilson and Bauer (1993) showed that in the Namoi valley, T. tabaci was the most abundant species on cotton early in the season, but that numbers declined rapidly by mid-December. Whereas the abundance of F. schultzei was low on seedling cotton, its abundance increased once cotton flowers appeared later in the season (Wilson & Bauer 1993). Frankliniella occidentalis entered Australia and spread to cotton producing regions sometime after Wilson & Bauer’s (1993) paper was published. The reasons for the seasonal changes in the species composition and the relative abundances of these thrips species in cotton remain poorly understood. An understanding of the invasion ecology of these thrips in relation to seedling cotton, especially with regard to the origin of the thrips populations that invade cotton has practical implications. Answers would help determine whether a broader integrated pest management scheme is appropriate based on the relative abundance of thrips on weed species and the possible management of weeds to manage invasions into cotton. In this Chapter I discuss how damage inflicted by various herbivore species on cotton affects the attractiveness of these herbivore damaged plants to thrips and propose how this might affect the local movement of the thrips concerned and their distribution within the crop. I also outline how the presence of host weed species ultimately affects the population dynamics and relative abundance of these three species of thrips in the cotton growing region of the Namoi valley. Finally I summarize the implications for the management of thrips in cotton that arise from these results and highlight aspects of their ecology that warrant further investigation, with particular reference to further understanding the molecular and biochemical changes in plants, the perception of the associated volatile compounds by thrips, and the associations between thrips and flowers in their weed and cotton host plants.

7.1 Responses of thrips to herbivore-induced responses in cotton This research presented in Chapters 2-5 demonstrates that the behavioural responses of the thrips to herbivore-induced cotton seedlings are species-specific. Frankliniella schultzei was more attracted to herbivore-damaged plants than to undamaged plants, whereas F. occidentalis was more attracted undamaged plants than to herbivore-damaged plants (Figure 2.8, Chapter 2). However, the 182 responses of T. tabaci differed from those of both F. occidentalis and F. schultzei and they were often intermediate to the responses shown by the two Frankliniella species (Figure 3.3, Chapter 3). Collectively, the results from Chapters 2-5 demonstrate that herbivore-induced responses should not be viewed solely as plant defensive strategies. The responses of cotton to herbivore damage did not always result in reduced attraction of thrips to these plants; rather, the responses of the three species of thrips to herbivore-induced cotton plants varied dramatically (Figures 2.8 & 3.3). Frequently, plant responses to herbivory have been interpreted as plant defences against further herbivore attack (Karban & Myers, 1989; Karban & Baldwin, 1997; Agrawal & Sherriffs, 2001; Dicke & Baldwin, 2010). If the herbivore-induced responses in cotton correspond to a direct defence strategy against further herbivory, why then would thrips be attracted to herbivore damaged cotton seedlings? Clearly induced responses do not always result in reduced attraction to herbivore-damaged plants. Thus herbivore-induced plant responses should not be solely viewed as direct plant defence responses but rather they should be recognized as biochemical responses to herbivory and the resultant volatiles are responded to in different ways by different herbivore species. The role of such induced responses in indirect defences that attract the natural enemies of herbivores is well established (e.g. Dicke & Baldwin, 2010). Studies investigating the behaviour of natural enemies of thrips in response to thrips-induced plants would provide another dimension to this work and contribute to a better overall understanding of chemical ecology that underpins the interactions between thrips and their cotton host plants When the thrips species were tested against plants damaged by various arthropod herbivores (Helicoverpa armigera, Tetranychus urticae, T. ludeni, Tenebrio molitor, Bemisia tabaci, Trialeurodes vaporariorum, Aphis gossypii), it was possible to predict how each species would respond once the response of a given thrips species to a given type of damage had been established. This predictability in the responses of thrips to herbivore-induced plants was also apparent when tested against cotton seedlings that were simultaneously damaged by more than one herbivore species. The responses of thrips to dually infested plants were predictable based on their responses to plants damaged by single species infestations (Chapter 4). Plant responses to arthropod herbivory differ based on the type of feeding damage inflicted by the species of herbivore (Chapters 2-5). For example, F. occidentalis was more attracted to cotton seedlings damaged by T. urticae or F. schultzei than to undamaged plants (Figures 2.8) but more attracted to undamaged plants than plants damaged by H. armigera larvae or T. molitor (Figure 2.8). Furthermore, different modes of herbivory not only resulted in different responses from each thrips species but also influenced the duration of attraction of thrips (Chapter 5). All three species of thrips remained attracted to plants damaged by T. urticae for much longer than to plants damaged by T. ludeni (Figures 5.5 & 5.6). Similarly F. schultzei remained attracted to T. 183 urticae damaged plants for much longer than to plants damaged by H. armigera (Figures 5.4 & 5.5). Thus not only were the responses thrips different and species specific, they also changed at different rates determined by the type of damage and the responding thrips species, adding extra complexity to these herbivore-host plant interactions. These herbivore specific plant responses to feeding could be caused by a combination of factors resulting from herbivore specific physical damage and the type and amount of elicitors released during feeding (Hilker & Meiners, 2010). However the study did not determine the molecular and biochemical changes in plants and their perception by thrips. Determining what exactly the thrips are responding to is essential to fully understand these herbivore-host plant interactions and warrants future investigation.

7.2 The sources of the three thrips species that invaded cotton In the Namoi valley the seasonal pattern of abundance of T. tabaci, F. occidentalis and F. schultzei on cotton differed (Chapter 6). The predominately high T. tabaci abundance on early season cotton is attributable to the high numbers of T. tabaci on the surrounding weed species during the spring months (Figure 6.15). During the spring months weed hosts on which T. tabaci was recorded were in plentiful supply and thus contributed to the high T. tabaci abundance on cotton. As the weeds (and wheat) senesce in spring, thrips are forced to seek new hosts. In these semi-arid environments availability of other hosts may be limited, and irrigated cotton planting may essentially be ‘islands’ of host plant in what is otherwise a host poor environment at that time. This is consistent with the arguments made by Wilson and Bauer, (1993) and Milne & Walter, (1998b), who suggested the possibility of T. tabaci moving into cotton from nearby alternative hosts. Analysis of mitochondrial CO1 sequences in the current study also revealed that individual T. tabaci collected from selected weed hosts, cotton and other crop hosts showed no genetic differentiation, indicating that the T. tabaci collected on its various hosts did not form discrete and independent breeding populations (Figure 6.37). This molecular analysis lends weight to the argument that T. tabaci moves from weeds and non-cotton crop hosts onto cotton. However, why these T. tabaci populations do not persist and build on cotton crops through the growing season is still not understood. Possible explanations are that cotton leaf tissue is a poor host with few T. tabaci larvae surviving. These results demonstrate that knowledge of the weed host ranges of thrips and their abundance is essential to understanding the population biology of thrips and the timing of infestation of cultivated crops. Such results also highlight the need for weed scientists and entomologists to work together to more thoroughly understand the effect of weed management on arthropod population dynamics, as suggested by Weber et al. (1990). It is practical for growers to control weeds on their farms but they can’t do much about what happens in the areas outside of the areas for which they are responsible. However, a genuine weed control strategy that removed relevant weed species (i.e. 184

Ammi majus L. (Apiales: Apiaceae) or Rapistrum rugosum L. (Brassicales: Brassicaceae)) could potentially reduce thrips numbers. The seasonal patterns of abundance and the interplant movements of F. occidentalis and F. schultzei were less clear. Both Frankliniella species seem to be only occasional pests on seedling cotton but are more common once flowering starts (Figure 6.33). Results indicate that the pattern of abundance of F. occidentalis on cotton was not correlated with their abundance on the weeds; this was because when F. occidentalis numbers were high on cotton, the weed species on which they were found were scarce. What mostly likely occurred is that the few F. occidentalis present, on the few surrounding weed species, moved onto cotton flowers, and increased in abundance. This implies that unlike T. tabaci the seasonal abundance of F. occidentalis on cotton was primarily driven by cotton flowers. Teulon & Penman, (1991) found that pollen to be an important source of food for thrips development and reproduction and this could explain the high abundance of F. occidentalis on cotton flowers rather than on seedling cotton. Cotton flowers could play an important role in the population biology of F. occidentalis and thus warrants further investigation, with comparative tests of their survival and reproduction rates on cotton seedlings and cotton flowers. The same could be said for F. schultzei, although F. schultzei was recorded at low densities on the surrounding weed species (Figure 6.15), the emergence of cotton flowers probably enabled F. schultzei populations to increase. The field sampling of weeds and cotton also showed the ecological significance of the association between all three species of thrips and flowers as opposed to the leaves of the plants (Chapter 6). Not only was the abundance of thrips related to the presence of flowering weeds but also both larval and adult thrips were predominantly found on flowers of their host plants rather than leaves (Figures 6.24 & 6.25). This aspect of their host association raises questions about the invasion of thrips, especially T. tabaci, onto seedling cotton. Thus, understanding the role of flowers in maintaining and sustaining thrips populations is essential to understanding the ecology of thrips. This is especially important when it comes to determining the reasons as to why T. tabaci was found in low numbers on cotton flowers compared to high numbers of both F. occidentalis and F. schultzei and thus warrants further investigation.

7.3 Implications for management of thrips in relation to cotton An important conclusion of this research is that weeds play a key role in the population biology of thrips and that a consideration of the ecology of the relevant weeds must be integral to the future development of pest management strategies. Results of this study indicate that it could be possible to predict large out breaks of thrips (particularly T. tabaci) on cotton. It was evident that high weed abundance and diversity produced more thrips, and this ultimately resulted in more thrips 185

(predominantly T. tabaci) being recorded in cotton. From this, a strategy could be devised whereby managers can monitor weed abundance and phenology to determine whether the influx of thrips to cotton seedlings could be high or low. Knowledge of (or even monitoring of) the arthropod pests hosted by weeds should/would permit a manager to better anticipate arthropod problems in adjacent crops (Norris & Kogan, 2000). It could be argued that whenever there is a high abundance of weeds, cotton seedlings should be more intensively protected. Also the current study was also able to determine that rainfall likely plays an important part in the population ecology of thrips. Results showed that weed abundance was high when rainfall was plentiful. Thus, whenever rainfall is high it results in a higher abundance of weeds that ultimately results in higher thrips abundance, once again it may be possible to predict the risk of thrips outbreaks based on rainfall events and the presence of favourable weed species. Results however, showed that, both F. occidentalis and F. schultzei only appeared to be occasional pests on seedling cotton. The study species of thrips also provide valuable control of mites (Milne & Walter, 1997; Milne & Walter, 1998a; Williams et al., 2011) and occur on cotton plants infested with T. urticae. This study has shown that all three thrips species were attracted to T. urticae damaged plants but the presence of other pests on T. urticae infested plants disrupted the attraction to T. urticae damaged cotton seedlings. This could potentially impact the success of thrips being used as biocontrol agents against T. urticae.

7.4 Future research It is important to understand the phenology and requirements of each relevant weed species, to determine when and why the thrips move away from them. For example, field sampling of weeds has shown that A. majus played an important role in the seasonal abundance of T. tabaci (Figure Figures 6.26, 6.27 & 6.28). When the abundance of A. majus decreased (autumn and winter months) the abundance of T. tabaci also decreased (Figures 6.26 & 6.27). So it is important to understand what drives the seasonal patterns of A. majus. Additionally, in the current study T. tabaci was found in low numbers during the cotton flowering stage, which was surprising as they were found in high numbers on the flowering weed species. It is important to find out why T. tabaci was found in greater abundance on seedling cotton but not on cotton flowers. Conversely, it is also important to find out why F. occidentalis and F. schultzei was more abundant on cotton flowers and not so on cotton seedlings. Finally, olfactometer experiments have revealed that the behavioural responses of the thrips to herbivore-induced cotton seedlings are species-specific. However, these experiments were conducted under laboratory conditions, future research should be also conducted in the field.

186

References

Abe H, Shimoda T, Ohnishi J, Kugimiya S, Narusaka M et al. (2009) Jasmonate-dependent plant defense restricts thrips performance and preference. BMC Plant Biology 9: 1-12.

Addesso KM, McAuslane HJ & Alborn HT (2011) Attraction of pepper weevil to volatiles from damaged pepper plants. Entomologia Experimentalis et Applicata 138: 1-11.

Agrawal AA (1998) Induced responses to herbivory and increased plant performance. Science 279: 1201-1202.

Agrawal AA (1999) Induced responses to herbivory in wild radish: effects on serval herbivores and plant fitness. Ecology 80: 1713-1723.

Agrawal AA (2000) Specificity of induced resistance in wild radish: causes and consequences for two specialist and two generalist caterpillars. Oikos 89: 493-500.

Agrawal AA & Colfer RG (2000) Consequences of thrips-infested plants for attraction of conspecifics and parasitoids. Ecological Entomology 25: 493-496.

Agrawal AA & Klein CN (2000) What omnivores eat: direct effects of induced plant resistance on herbivores and indirect consequences of diet selection by omnivores. Journal of Animal Ecology 69: 525-535.

Agrawal AA, Kobayashi C & Thaler JS (1999) Influence of prey availability and induced host-plant resistance on omnivory by western flower thrips. Ecology 80: 518-523.

Agrawal AA & Sherriffs MF (2001). Induced plant resistance and susceptibility to late-season herbivores of wild radish. Annals of the Entomological Society of America 94: 71-75.

Alborn HT, Röse USR & McAuslane HJ (1996) Systemic induction of feeding deterrents in cotton plants by feeding of Spodoptera spp. larvae. Journal of Chemical Ecology 22: 919-932.

Aliakbarpour H & Salmah MRC (2011) Seasonal abundance and spatial distribution of larval and adult thrips (Thysanoptera) on weed host plants in mango orchards in Penang Malaysia. Applied Entomology and Zoology 46: 185-194.

Andersen DC (1987) Below-ground herbivory in natural communities: a review emphasizing fossorial animals. Quarterly Review of Biology 62: 261-286.

Anderson P & Alborn H (1999) Effects on oviposition behaviour and larval development of Spodoptera littoralis by herbivore-induced changes in cotton plants. Entomologia Experimentalis et Applicata 92: 45-51.

Anderson P & Agrell J (2005) Within-plant variation in induced defense in developing leaves of cotton. Oecologia 144: 427-434.

Ananthakrishnan TN, Daniel AM & Kumar NS (1982) Spatial ans seasonal distribution patterns of some phytophagous thrips (Thysanoptera: Insecta) infesting Ricinus communis Linn. (Euphorbiaceae) and Achyranthes aspera Linn. (Amaranthaceae). Proceedings of the Indian National Science Academy B48: 183-189. 187

Anderson P, Sadek MM & Wäckers FL (2011) Root herbivory affects oviposition and feeding behavior of a foliar herbivore. Behavioral Ecology 22: 1272-1277.

Attique MR & Ahmad Z (1990) Investigation of Thrips tabaci Lind. as a cotton pest and the development of strategies for its control in Punjab. Crop Protection 9: 469-473.

Atakan E, Özgür AF & Kersting U (1998) Frankliniella occidentalis (Thysanoptera: Thripidae) on cotton in Çukurova region, pp. 7-12. Proceedings of the Sixth International Symposium on Thysanoptera, 27 April-1 May 1998, Antalya, Turkey.

Atakan E & Uygur S (2005) Winter and spring abundance of Frankliniella spp. and Thrips tabaci Lindeman (Thysan., Thripidae) on weed host plants in Turkey. Journal of Entomology and Nematology 129: 17-26.

Arimura GI, Matsui K & Takabayashi J (2009) Chemical and molecular ecology of herbivore- induced plant volatiles: proximate factors and their ultimate function. Plant Cell Physiology 50: 911-923.

Bernays EA (2001) Neural limitations in phytophagous insects: implications for diet breadth and evolution of host affiliation. Annual Review of Entomology 46: 703-727.

Bernays EA & Chapman RF (1994). Host-plant selection by phytophagous insects. Chapman & Hall, New York.

Bernasconi ML, Turlings TCJ, Ambrosetti L, Bassetti P & Silvia D (1998) Herbivore-induced emissions of maize volatiles repel the corn leaf aphid, Rhopalosiphum maidis. Entomologia Experimentalis et Applicata 87: 133-142.

Bezemer TM & van Dam NM (2005) Linking aboveground and belowground interactions via induced plant defences. Trends in Ecology and Evolution 20: 617-624.

Bezemer TM, Wagenaar R, van Dam NM & Wäckers FL (2003) Interactions between above- and belowground insect herbivores as mediated by the plant defense system. Oikos 101: 555-562.

Bezemer TM, Wagenaar R, van Dam NM, van Der Putten WH & Wäckers FL (2004) Above- and below-ground terpenoid aldehyde induction in cotton, Gossypium herbaceum, following root and leaf injury. Journal of Chemical Ecology 30: 53-67.

Blumthal MR, Cloyd RA, Spomer LA & Warnock DF (2005) Flower color preferences of western flower thrips. HortTechnology 15: 846-853.

Bostock RM, Karban R, Thaler JS, Weyman PD & Gilchrist D (2001) Signal interactions in induced resistance to pathogens and insect herbivores. European Journal of Plant Pathology 107: 103-111.

Brunner PC, Chatzivassiliou EK, Katis NI & Frey JE (2004) Host-associated genetic differentiation in Thrips tabaci (Insecta; Thysanoptera), as determined from mtDNA sequence data. Heredity 93: 364-370.

Bruinsma M, van Dam NM, van Loon JJA & Dicke M (2007) Jasmonic acid-induced changes in Brassica oleracea affect oviposition preference of two specialist herbivores. Journal of Chemical Ecology 33: 655-668. 188

Brecke BJ, Funderbuk JE, Teare ID & Gobert DW (1996) Interaction of early-season herbicide injury, tobacco thrips injury and cultivar on peanut. Agronomy Journal 88: 14-18.

Bruessow F, Gouhier-Darimont C, Buchala A, Jean-Pierre M & Reymond P (2010) Insect egg suppress plant defence against chewing herbivores. The Plant Journal 62: 876-885.

Bryan DE & Smith RF (1956) The Frankliniella occidentalis (Pergande) complex in California (Thysanoptera: Thripidae). University of California Publications in Entomology 10: 359-410.

Buteler M, Weaver DK & Peterson RKD (2009) Oviposition behavior of the wheat stem sawfly when encountering plants infested with cryptic conspecifics. Environmental Entomology 38: 1707- 1715.

Childers CC (1997) Feeding and oviposition injuries to plants. Thrips as Crop Pests (ed. by T. Lewis), pp. 505-537, CAB International, Wallingford, United Kingdom.

Chellemi DO, Funderburk JE & Hall DW (1994) Seasonal abundance of flower-inhabiting Frankliniella species (Thysanoptera: Thripidae) on wild plant species. Environmental Entomology 23: 337-342.

Chyzik R & Ucko O (2002) Seasonal abundance of the western flower thrips Frankliniella occidentalis in the Arava valley of Israel. Phytoparasitica 30: 335-346.

Clement M, Posada D & Crandall K (2000) TCS: a computer program to estimate gene genealogies. Molecular Ecology 9: 1657-1660.

Cook DR, Allen CT, Burris E, Freeman BL, Herzog GA et al. (2003) A survey of thrips (Thysanoptera) species infesting cotton seedlings in Alabama, Arkansas, Georgia, Louisiana, Mississippi and Tennessee. Journal of Entomological Science 38: 669-681.

Cook D, Herbert A, Akin DS & Reed J (2011) Biology, Crop Injury, and Management of Thrips (Thysanoptera: Thripidae) Infesting Cotton Seedlings in the United States. Journal of Integrated Pest Management 2: B1-B9.

Crawley MJ (1983) Herbivory: the dynamics of animal-plant interactions. Blackwell Scientific Publications, Oxford, UK.

Cunningham GM, Mulham WE, Milthorpe PL & Leigh JH (1981) Plants of western New South Wales. NSW Government Printing Office, Sydney.

Davidson J & Bald JG (1930) Description and bionomics of Frankliniella insularis Franklin (Thysanoptera). Bulletin of Entomological Research 21: 365-385. van Dam NM, Tytgat T & Kirkegaard J (2009) Root and Shoot glucosinolates: a comparison of their diversity, function and interactions in natural and managed ecosystems. Phytochemistry Reviews 8: 171-186.

Darriba D, Taboada GL, Doallo R & Posada D (2012) jModelTest 2: more models, new heuristics and parallel computing. Nature Methods 9: 772.

189

Deligeorgidis PN, Athanassiou CG & Kavallieratos NG (2002) Seasonal abundance, spatial distribution and sampling indices of thrip populations on cotton; a 4-year survey from central Greece. Journal of Applied Entomology 126: 343-348.

De Boer JG, Hordijk CA, Posthumus MA & Dicke M (2008) Prey and non-prey arthropods sharing a host plant: effects on induced volatile emission and predator attraction. Journal of Chemical Ecology 34: 281-290.

Delphia CM, Mescher MC & De Moraes CM (2007) Induction of plant volatiles by herbivores with different feeding habits and the effects of induced defenses on host-plant selection by thrips. Journal of Chemical Ecology 33: 997-1012.

De Moraes CM, Lewis WJ, Paré PW, Alborn HT & Tumlinson JH (1998) Herbivore-infested plants selectively attract parasitoids. Nature 393: 570-573.

De Moraes CM, Mescher MC & Tumlinson JH (2001) Caterpillar-induced nocturnal plant volatiles repel conspecific females. Nature 410: 577-580.

Denno RF, Larsson S & Olmstead KL (1990) Role of enemy free space and plant quality in host- plant selection by willow beetles. Ecology 71: 124–137.

De Vries EJ, Vos RA, Jacobs G & Breeuwer HAJ (2006) Western flower thrips (Thysanoptera: Thripidae) preference for thrips-damaged leaves over fresh leaves enable uptake of symbiotic gut bacteria. European Journal of Entomology 103: 779-786.

Dicke M (2009) Behavioural and community ecology of plants that cry for help. Plant, Cell and Environment 32: 654-665.

Dicke M & Baldwin IT (2010) The evolutionary context for herbivore-induced plant volatiles: beyond the ‘cry for help’. Trends in Plant Science 15: 167-175.

Dingle H & Drake VA (2007) What is migration? BioScience 57: 113-121.

Dicke M & Hilker M (2003) Induced plant defences: from molecular biology to evolutionary ecology. Basic and Applied Ecology 4: 3-14.

Dicke M & van Loon JJA (2000). Multitrophic effects of herbivore-induced plant volatiles in an evolutionary context. Entomologia Experimentalis et Applicata 97: 237-249.

Dicke M, van Loon JJA & Soler R (2009) Chemical complexity of volatiles from plants induced by multiple attack. Nature Chemical Biology 5: 317-324.

Dicke M, Sabelis MW, Takabayashi J, Bruin J & Posthumus MA (1990) Plant strategies of manipulating predator–prey interactions through allelochemicals: prospects for application in pest control. Journal of Chemical Ecology 16: 3091-3118.

Doederlein TA & Sites RW (1993) Host plant preferences of Frankliniella occidentalis and Thrips tabaci (Thysanoptera: Thripidae) for onions and associated weeds on the southern High Plains. Journal of Economic Entomology 86: 1706-1713.

Drummond AK, Held J, Moir R, Theierer T, Ashton B & Wilson A (2010) Geneious v.5.0, p.

190

Du YJ, Poppy GM, Powell W, Pickett JA, Wadhams LJ & Woodcock CM (1998) Identification of semiochemicals released during aphid feeding that attract parasitoid Aphidius ervi. Journal of Chemical Ecology 24: 1355–1368.

Edwards PJ & Wratten SD (1983) Wound induced defences in plants and their consequences for patterns of insect grazing. Oecologia 59: 88-93.

Feder JL (1995) The effects of parasitoids on sympatric host races of Rhagoletis pomonella (Diptera: Tephritidae). Ecology 76: 801–813.

Fekrat L, Manzari S & Shishehbor P (2014) Morphometric and molecular variation in Thrips tabaci Lindeman (Thysanoptera: Thripidae) populations on onion and tobacco in Iran. Journal of Agricultural Science and Technology 16: 1505-1516.

Fidantsef AL, Stout MJ, Thaler JS, Duffey SS & Bostock RM (1999) Signal interactions in pathogen and insect attack: Expression of lipoxygenase, proteinase inhibitor II, and pathogenesis- related protein P4 in tomato, Lycopersicon esculentum. Physiological and Molecular Plant Pathology 54: 97-114.

Folmer O, Black M, Hoeh W, Lutz R & Vrihenhoek R (1994) DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Molecular Marine Biology and Biotechnology 3: 294–303.

Forrester NW & Wilson AGL (1988) Insect pests of cotton. New South Wales Department of Agriculture Agfact P5.AE.1, 18 pp.

Frey M, Stettner C, Paré PW, Schmelz EA, Tumlinson JH & Gierl A (2000) An herbivore elicitor activates the gene for indole emission in maize. Proceedings of the National Academy of Sciences, U.S.A 97: 14801-14806.

Fritz RS (1992) Community structure and species interactions of phytophagous insects on resistant and susceptible host plants. Plant Resistance to Herbivores and Pathogens (ed. By RS Fritz & EL Simms), pp. 240-277. The University of Chicago Press, Chicago, Illinois, USA.

Gerin C, Hance Th & van Impe G (1999) Impact of flowers on the demography of western flower thrips Frankliniella occidentalis (Thysan., Thripidae). Journal of Applied Entomology 123: 569- 574.

Girling RD, Stewart-Jones A, Dherbecourt J, Staley JT, Wright DJ & Poppy GM (2011) Parasitoids select plants more heavily infested with their caterpillar hosts: a new approach to aid interpretation of plant headspace volatiles. Proceedings of the Royal Society B 278: 2646-2653.

Gomez S, van Dijk W & Stuefer JF (2010) Timing of induced resistance in a clonal plant network. Plant Biology 12: 512-517.

Guindon S & Gascuel O (2003) A simple, fast and accurate method to estimate large phylogenies by maximum-likelihood. Systematic Biology 52: 696-704.

Guerrieri E, Poppy GM, Powell W, Tremblay E & Pennacchio F (1999) Induction and systemic release of herbivore-induced plant volatiles mediating in-flight orientation of Aphidius ervi. Journal of Chemical Ecology 25: 1247-1261.

191

Green TR & Ryan CA (1972) Wound-induced proteinase inhibitor in plant leaves-possbile defense mechanism against insects. Science 175: 776-777.

Haukioja E (1980) On the role of plant defences in the fluctuations of herbivore populations. Oikos 35: 202-213.

Hare DJ (2011) Ecological role of volatiles produced by plants in response to damage by herbivorous insects. Annual Review Entomology 56: 161-180.

Harmel N, Almohamad R, Fauconnier ML, Du Jardin P, Verheggen F et al. (2007) Role of terpenes from aphid-infested potato on searching and oviposition behaviour of Episyrphus balteatus. Insect Science 14: 57-63.

Haukioja E & Hanhimäki S (1985) Rapid wound-induced resistance in white birch (Betula pubescens) foliage to the geometrid Epirrita autumnata: a comparison of tress and moths within and outside the outbreak range of the moth. Oecologia 65: 223-228.

Hanyu K, Ichiki RT, Nakamura S & Kainoh Y (2009) Duration and location of attraction to herbivore-damaged plants in the tachinid parasitoid Exorista japonica. Applied Entomology Zoology 44: 371-378.

Haukioja E & Niemelä P (1977) Retarded growth of a geometrid larva after mechanical damage to leaves of its host tree. Annales Zoologici Fennici 14: 48-52.

Hawkins BS, Peacock HA & Steele TE (1966) Thrips injury to upland cotton (Gossypium hirsutum L.) varieties. Crop Science 6: 256-258.

Heard TA (1995) Oviposition and feeding preferences of a flower-feeding weevil, Coelocephalapion aculeatum, in relation to conspecific damage to its host plant. Entomologia Experimentalis et Applicata 76: 203-209.

Hern A & Dorn S (2002) Induction of volatile emissions from ripening apple fruits infested with Cydia pomonella and the attraction of adult females. Entomologia Experimentalis et Applicata 102: 145-151.

Hilker M & Meiners T (2010) How do plants “notice” attack by herbivorous arthropods? Biological Reviews 85: 267-280.

Holt RD & Lawton JH (1993) Apparent competition and enemy-free space in insect host–parasitoid communities. American Naturalist 142: 623–645.

Hoballah ME & Turlings TCJ (2005) The role of fresh versus old leaf damage in the attraction of parasitic wasps to herbivore-induced maize volatiles. Journal of Chemical Ecology 31: 2003-2018.

Hunt-Joshi TR & Blossey B (2005) Interactions of root and leaf herbivores on purple loosestrife (Lythrum salicaria). Oecologia 142: 554-563.

Huelsenbeck, JP & Ronquist F (2001) MRBAYES: Bayesian inference of phylogeny. Bioinformatics 17: 754-755.

192

Huang CH, Yan FM, Byers JA, Wang RJ & Xu CR (2009) Volatiles induced by the larvae of the Asian corn borer (Ostrinia furnacalis) in maize plants affect behavior of conspecific larvae and female adults. Insect Science 16: 311-320.

Inbar M, Doostdar H & Mayer RT (1999) Effects of sessile whitefly nymphs (Homoptera: Aleyrodidae) on leaf-chewing larvae (Lepidoptera: Noctuidae). Environmental Entomology 28: 353-357.

Isbell RF (1996) The Australian soil classification. CSIRO Publishing, Collingwood, Victoria.

ICAC (2014) Supply and use of cotton by country.

Jacobsen LA (1945) The effect of stinkbug feeding on wheat. Canadian Entomologist 77: 200.

Jeffries MJ & Lawton JH (1984) Enemy-free space and the structure of ecological communities. Biological Journal of the Linnean Society 23: 269–286.

Johnson MTJ (2011) Evolutionary ecology of plant defences against herbivores. Functional Ecology 25: 305-311.

Karban R (1986) Induced resistance against spider mites in cotton: field verification. Entomologia Experimentalis et Applicata 42: 239-242.

Karban R (1993) Costs and Benefits of Induced Resistance and Plant Density for a Native Shrub, Gossypium Thurberi. Ecology 74: 9-19.

Karban R (2011) The ecology and evolution of induced resistance against herbivores. Functional Ecology 25: 339-347.

Katayama H (2006) Seasonal prevalence of the occurrence of western flower thrips Frankliniella occidentalis (Pergande) (Thysanoptera: Thripidae) on weed hosts growing around ornamental fields. Applied Entomology and Zoology 41: 93-98.

Karban R & Baldwin IT (1997) Induced Responses to Herbivory. Chicago: University of Chicago Press.

Karban R & Carey JR (1984) Induced resistance of cotton seedlings to mites. Science 225: 53-54.

Karban R & Myers JH (1989) Induced plant responses to herbivory. Annual Review of Ecology and Systematics 20: 331-348.

Kessler A & Baldwin IT (2001) Defensive function of herbivore-induced plant volatile emissions in nature. Science 291: 2141-2144.

Kessler A, Halitschke R & Baldwin IT (2004) Silencing the Jasmonate Cascade: Induced Plant Defenses and Insect Populations. Science 305: 665-668.

Kessler A & Halitschke R (2007) Specificity and complexity: the impact of herbivore-induced plant responses on arthropod community structure. Current Opinion in Plant Biology 10: 409-414.

Kessler A & Heil M (2011) The multiple faces of indirect defences and their agents of natural selection. Functional Ecology 25: 348-357. 193

Kirk WDJ (1984) Pollen-feeding in thrips (Insecta: Thysanoptera). Journal of Zoology (London) 204: 107-118.

Kirk WDJ (1987) A key to the larvae of some common Australian flower thrips (Insecta: Thysanoptera), with a host-plant survey. Australian Journal of Zoology 35: 173-185.

Kirk WDJ (1997) Feeding. Thrips as Crop Pests (ed. by T Lewis), pp 119-174. CAB International, New York, USA.

Kirk WDJ (2002) The pest and vector from the west: Frankliniella occidentalis: Thrips and Tospoviruses. In proceedings of the seventh international symposium on Thysanoptera (eds. By R Marullo & LA Mound), pp. 33-44. Australian National Insect Collection, Canberra, Australia.

Kiritani K, Hokyo N, Kimura K & Nakasuji F (1965) Imaginal dispersal of the southern green stink bug, Nezara viridula L., in relation to feeding and oviposition. Japanese Journal of Applied Entomology and Zoology 9: 291-296.

Klose MJ, Sdoodee R, Teakle DS, Milne JR, Greber RS & Walter GH (1996) Transmission of three strains of tobacco streak ilarvirus by different thrips species using virus-infected pollent. Journal of Phytopathology 144: 281-284.

Kranthi S, Kranthi KR & Wanjari RR (2003) Influence of semilooper damage on cotton host-plant resistance to Helicoverpa armigera (Hub). Plant Science 164: 157-163.

Kugimiya S, Shimoda T, Tabata J & Takabayashi J (2010) Present or past herbivory: a screening of volatiles released from Brassica rapa under caterpillar attacks as attractants for the solitary parasitoid, Cotesia vestalis. Journal of Chemical Ecology 36: 620-628.

Landolt PJ (1993) Effects of host plant leaf damage on cabbage looper moth attraction and oviposition. Entomologia Experimentalis et Applicata 67: 79-85.

Landolt PJ, Tumlinson JH & Alborn DH (1999) Attraction of Colorado potato beetle (Coleoptera: Chrysomelidae) to damaged and chemically induced potato plants. Environmental Entomology 28: 973-978.

Lewis T (1973) Thrips: their Biology, Ecology and Economical importance. Academic Press, London and New York. P. 349.

Lewis T (1997) Thrips as Crop Pests. CAB International, Wallingford, UK.

Lewis T & Navas DE (1962) Thysanopteran populations overwintering in hedge bottoms, grass litter and bark. Annals of Applied Biology 50: 299-311.

Lei TT & Wilson LJ (2004) Recovery of leaf area through accelerated shoot ontogeny in thrips- damaged cotton seedlings. Annals of Botany 94: 179-186.

Lin L, Shen TC, Chen YH & Hwang SY (2008) Responses of Helicoverpa armigera to tomato plants previously infested by ToMV or damaged by H. armigera. Journal of Chemical Ecology 34: 353-361.

194

Loxdale HD & Lushai G (1999) Slaves of the environment: the movement of herbivorous insects in relation to their ecology and genotype. Philosophical Transactions of the Royal Society B: Biological Sciences 354: 1479-1495.

Loughrin JH, Manukian A, Heath RR, Turlings TCJ & Tumlinson JH (1994) Diurnal cycle of emission of induced volatile terpenoids by herbivore-injured cotton. Proceedings of the National Academy of Sciences, U.S.A 91:11836-11840.

Lu YB, Liu SS, Liu YQ, Furlong MJ & Zalucki MP (2004). Contrary effects of jasmonate treatment of two closely related plant species on attraction of and oviposition by a specialist herbivore. Ecology Letters 7: 337-345.

Masters GJ & Brown VK (1992) Plant-mediated interactions between 2 spatially separated insects. Functional Ecology 6: 175-179.

Masters GJ, Brown VK & Gange AC (1993) Plant mediated interactions between above- and below-ground insect herbivores. Oikos 66: 148-151.

Magalhães DM, Borges M, Laumann RA, Sujii ER, Mayon P et al. (2012) Semiochemicals from herbivory induced cotton plants enhance the foraging behavior of the cotton boll weevil, Anthonomus grandis. Journal of Chemical Ecology 38: 1528-1538.

Mayer RT, Inbar M, McKenzie CL, Shatters R, Borowicz V et al. (2002) Multitrophic interactions of the Silverleaf Whitefly host plants, competing herbivores and phytopathogens. Archives of Insect Biochemistry and Physiology 51: 151-169.

Mahabaleshwar H, Oliveira JN, da Costa JG, Bleicher E, Santana AEG et al. (2011) Identification of semiochemicals released by cotton, Gossypium hirsutum, upon infestation by the cotton aphid, Aphis gossypii. Journal of Chemical Ecology 37: 741-750.

Maddox GD & Root RB (1990) Structure of the encounter between goldenrod (Solidago altissima) and its diverse insect fauna. Ecology 71: 2115-2124.

Maeda T & Takabayashi J (2001) Production of herbivore-induced plant volatiles and their attractiveness to Phytoseius persimilis (Acari: Phytoseiidae) with changes of Tetranychus urticae (Acari: Tetranychidae) density on a plant. Applied Entomology and Zoology 36: 47-52.

Matthews GA & Tunstall JP (1994) Insect pests of cotton. CAB International, Wallingford, UK.

McCall PJ, Turlings TCJ, Loughrin J, Proveaux AT & Tumlinson JH (1994) Herbivore-induced volatile emissions from cotton (Gossypium hirsutum L.) seedlings. Journal of Chemical Ecology 20: 3039-3050.

McCloud ES, Tallamy DW & Halaweish FT (1995) Squash beetle trenching behaviour: avoidance of cucurbitacin induction or mucilaginous plant sap? Ecological Entomology 20: 51-59.

Miresmailli S, Gries R, Gries G, Zamar RH & Isman MB (2011) Population density and feeding duration of cabbage looper larvae on tomato plants alter the levels of plant volatile emissions. Pest Management Science 68: 101-107.

195

Milne JR, Walter GH, Kaonga D & Sabio GC (1996) The importance of non-pollen plant parts as food sources for the common blossom thrips, Frankliniella schultzei. Entomologia Experimentalis et Applicata 78: 271-281.

Milne JR, Milne M & Walter GH (1997) A key to larval thrips (Thysanoptera) from granite belt stonefruit trees and a first description of Pseudanaphothrips achaetus (Bagnall) larvae. The Australian Journal of Entomology 36: 319-326.

Milne M & Walter GH (1997) The significance of prey in the diet of the phytophagous thrips, Frankliniella schultzei. Ecological Entomology 22: 74-81.

Milne M & Walter GH (1998a) Significance of mite prey in the diet of the onion thrips Thrips tabaci Lindeman (Thysanoptera: Thripidae). Australian Journal of Entomology 37: 120-124.

Milne M & Walter GH (1998b) Host species and plant part specificity of the polyphagous onion thrips, Thrips tabaci Lindeman (Thysanoptera: Thripidae), in an Australian cotton-growing area. Australian Journal of Entomology 37: 115-119.

Milne M & Walter GH (2000) Feeding and breeding across host plants within a locality by the widespread thrips Frankliniella schultzei, and the invasive potential of polyphagous herbivores. Diversity and Distributions 6: 243-257.

Milne M, Walter GH & Milne JR (2002) Mating aggregations and mating success in the flower thrips, Frankliniella schultzei (Thysanoptera: Thripidae), and a possible role for pheromones. Journal of Insect Behavior 15: 351-368.

Milne M, Walter GH & Milne JR (2007) Mating behavior and species status of host-associated populations of the polyphagous thrips, Frankliniella schultzei. Journal of Insect Behavior 20: 331- 346.

Miyazaki J, Wilson LJ & Stiller WN (2013) Fitness of twospotted spider mites is more affected by constitutive than induced resistance traits in cotton (Gossypium spp.). Pest Management Science 69: 1187-1197.

Mound LA (2005) Thysanoptera: Diversity and Interactions. Annual Review of Entomology 50: 247-269.

Mound LA (2011) Order Thysanoptera Haliday, 1836. In: Zhang, Z.-Q. (Ed.) Animal biodiversity: An outline of higher-level classification and survey of taxonomic richness. Zootaxa 3148: 1–237.

Moayeri HRS, Ashouri A, Brødsgaard HF & Enkegaard A (2006) Odour-mediated responses of a predatory mirid bug and its prey. Experimental and Applied Acarology 40: 27-36.

Moayeri HRS, Ashouri A, Poll L & Enkegaard A (2007) Olfactory response of a predatory mirid to herbivore induced plant volatiles: multiple herbivory vs. single herbivory. Journal of Applied Entomology 131: 326-332.

Mound LA & Houston KJ (1987) An annotated check-list of Thysanoptera from Australia. British Museum (Natural History), London.

Morse JG & Hoddle MS (2006) Invasion biology of thrips. Annual Review of Entomology 51: 67- 89. 196

Moritz G, Kumm S & Mound LA (2004) Tospovirus transmission depends on thrips ontogeny. Virus Research 100: 143-149.

Mound LA & Morris DC (2007) The insect order Thysanoptera: classification versus systematics. Zootaxa 1668: 395-411.

Mound LA & Teulon DAJ (1995) Thysanoptera as phytophagous opportunists. Thrips Biology and Management (ed. by BL Parker, M Skinner & T Lewis), pp 3-19. Plenum Press, London, UK.

Moran PJ & Thompson GA (2001) Molecular responses to aphid feeding in Arabidopsis in relation to plant defense pathways. Plant Physiology 125: 1074–1085.

Mound LA, Tree DJ. & Paris D (2014) OzThrips –Thysanoptera in Australia. http://www.ozthrips. org/ [Accessed 2014-2015].

Moran NA & Whitham TG (1990) Interspecific competition between root-feeding and leaf-galling aphids mediated by host-plant resistance. Ecology 71: 1050-1058.

Munger P, Bleiholder H, Hack H, Hess M, Stauss R et al. (1998) Phenological growth stages of the cotton plant (Gossypium hirsutum L.): codification and description according to the BBCH scale. Journal of Agronomy & Crop Science 180: 143-149.

Ngumbi E, Chen L & Fadamiro HY (2009) Comparative GC-EAD responses of a specialist (Microplitis croceipes) and a generalist (Cotestia marginiventris) parasitoid to cotton volatiles induced by two caterpillar species. Journal of Chemical Ecology 35: 1009-1020.

Niemelä P, Aro E-M & Haukioja E (1979) Birch leaves as a resource for herbivores. Damage- induced increase in leaf phenolics with trypsin-inhibiting effects. Reports from the Kevo Subarctic Research Station 15: 37-40.

Norris RF & Kogan M (2000) Interactions between weeds, arthropods pests, and their natural enemies in managed ecosystems. Weed Science 48: 94-158.

North RC & Shelton AM (1986) Colonization and intraplant distribution of Thrips tabaci (Thysanoptera: Thripidae) on cabbage. Journal of Economic Entomology 79: 219-223.

Ohgushi T (2008) Herbivore-induced indirect interaction webs on terrestrial plants: the importance of non-trophic, indirect, and facilitative interactions. Entomologia Experimentalis et Applicata 128: 217-229.

Oiver AD, Granett J, Karban R & Villa EM (2001) Chemically-induced resistance against multiple pests in cotton. International Journal of Pest Management 47: 49-54.

Oluwafemi S, Bruce TJA, Pickett JA, Ton J & Birkett MA (2011) Behavioral responses of the leafhopper Cicadulina storey China, a major vector of maize streak virus, to volatile cues from intact and leafhopper-damaged maize. Journal of Chemical Ecology 37: 40-48.

Osekre EA, Wright DL, Marois JJ & Funderburk J (2009) Population Dynamics and Within-Plant Distribution of Frankliniella spp. Thrips (Thysanoptera: Thripidae) in Cotton. Environmental Entomology 38: 1205-1210.

197

Pearsall IA & Myers JH (2000) Population dynamics of western flower thrips (Thysanoptera : Thripidae) in nectarine orchards in British Columbia. Journal of Economic Entomology 93: 264- 275.

Poelman EH, Broekgaarden C, van Loon JJA & Marcel D (2008) Early season herbivore differentially affects plant defence responses to subsequently colonizing herbivores and their abundance in the field. Molecular Ecology 17: 3352-3365.

Poelman EH, van Loon JJA, van Dam NM, Vet LM & Dicke M (2010) Herbivore-induced plant responses in Brassica oleracea prevail over effects of constitutive resistance and result in enhanced herbivore attack. Ecological Entomology 35: 240-247.

Powell W, Pennacchio F, Poppy GM & Tremblay E (1998) Strategies involved in the location of hosts by the parasitoid Aphidius ervi Haliday (Hymenoptera: Brachonidae: Aphidiinae). Biological Control 11: 104-112.

Preston CA, Laue G & Baldwin IT (2004) Plant–plant signalling: application of trans- or cis-methyl jasmonate equivalent to sagebrush releases does not elicit direct defenses in native tobacco. Journal of Chemical Ecology 30: 2193–2214. van der Putten WH, Vet LEM, Harvey JA & Wackers FL (2001) Linking above- and below-ground multitrophic interactions of plants, herbivores, pathogens, and their antagonists. Trends in Ecology and Evolution 16: 547-554.

Reddy GVP, Tabone E & Smith MT (2004). Mediation of host plant selection and oviposition behaviour in the diamondback moth Plutella xylostella and its predator Chrysoperla carnea by chemical cues from cole crops. Biological Control 29: 270-277.

Reitz SR, Yu-Lin G & Zhong-ren L (2011) Thrips: pests of concern to China and the United States. Agricultural Sciences in China 10: 867-892.

Rhainds M & Shipp L (2003) Dispersal of adult western flower thrips (Thysanoptera: Thripidae) on Chrysanthemum plants: impact of feeding-induced senescence of inflorescences. Environmental Entomology 32: 1056-1065.

Richards JH (1993) Physiology of plants recovering from defoliation. Proceedings of the XVII International Grassland Congress, 85-94.

Rojas JC (1999) Influence of host plant damage on the host-finding behavior of Mamestra brassicae (Lepidoptera: Noctuidae). Environmental Entomology 28: 588-593.

Rodriguez-Saona C, Crafts-Brandner SJ & Canas LA (2003) Volatile emissions triggered by multiple herbivore damage: beet armyworm and whitefly feeding on cotton plants. Journal of Chemical Ecology 29: 2539-2550.

Rodriguez-Saona C, Chalmers JA, Raj S & Thaler JS (2005) Induced plant responses to multiple damagers: differential effects on an herbivore and its parasitoid. Oecologia 143: 566-577.

Rodriguez-Saona C, Crafts-Brandner SJ, Williams III L & Paré PW (2002) Lygus hesperus feeding and salivary gland extracts induce volatile emissions in plants. Journal of Chemical Ecology 28: 1733–1747.

198

Ronquist F & Huelsenbeck JP (2003) MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572-1574.

Rostás M & Hilker M (2002) Feeding damage by larvae of the mustard leaf beetle deters conspecific females from oviposition and feeding. Entomologia Experimentalis et Applicata 103: 267-277.

Röse USR, Manukian A, Heath RR & Tumlinson JH (1996) Volatile semiochemicals released from undamaged cotton leaves. Plant Physiology 111: 487-496.

Röse USR & Tumlinson (2005) Systemic induction of volatile release in cotton: How specific is the signal to herbivory? Planta 222: 327-335.

Rugman-Jones PF, Hoddle MS & Stouthamer R (2010) Nuclear-mitochondrial barcoding exposes the global pest western flower thrips (Thysanoptera: Thripidae) as twp sympatric cryptic species in its native California. Molecular Entomology 103: 877-886.

Sabelis MW, Afman BP & Slim PJ (1985) Location of distant spider mite colonies by Phytoseiulus persimilis: Localization and extraction of a kairomone. Acarology 6: 431-440.

Sabelis MW & van de Baan HE (1984) Location of distant spider mite colonies by phytoseiid predators: Demonstration of specific kairomones emitted by Tetranychus urticae and Panonychus ulmi. Entomologia Experimentalis et Applicata 33: 303-314.

Sadras VO & Wilson LJ (1998) Recovery of cotton crops after early season damage by thrips (Thysanoptera). Crop Science 38: 399-409.

Scascighini N, Mattiacci L, D’Alessandro M, Hern A, Rott AS & Dorn S (2005) New insights in analysing parasitoid attracting synomoes: early volatile emission and use of stir bar sorptive extraction. Chemoecology 15: 97-104.

Shelton AM, Nyrop JP, North RC, Petzold C & Foster R (1987) Development and use of dynamic sequential sampling program for onion thrips, Thrips tabaci (Thysanoptera: Thripidae), on onions. Journal of Economic Entomology 80: 1051-1056.

Shiojiri K, Ozawa R, Kugimiya S, Uefune M, van Wijk M, et al. (2010) Herbivore-specific, density-dependent induction of plant volatiles: honest or “cry wolf” signals? PLoS ONE 5: e.12161.

Shiojiri K, Takabayashi J, Yano S & Takafuji A (2001). Infochemically mediated tritrophic interaction webs on cabbage plants. Population Ecology 43: 23-29.

Shiojiri K, Takabayashi J, Yano S & Takafuji A (2002) Oviposition preferences of herbivores are affected by tritrophic interaction webs. Ecology Lettters 5: 186-192.

Silva R & Furlong MJ (2012) Diamondback moth oviposition: effects of host plant and herbivory. Entomologia Experimentalis et Applicata 143: 218-230.

Simon C, Frati F, Beckenbach A, Crespi B, Liu H & Flook P (1994) Evolution, weighting, and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved polymerase chain reaction primers. Annals of the Entomological Society of America 87: 651–701.

199

Sites RW, Chambers WS & Nichols BJ (1992) Diel periodicity of thrips (Thysanoptera: Thripidae) dispersion and the occurrence of Frankliniella williamsi on onions. Journal of Economic Entomology 85: 100-105.

Soler R, Harvey JF, Bezemer TM & Stuefer JF (2008) Plants as green phones. Novel insights into plant-mediated communication between below- and aboveground insects. Plant Signaling and Behavior 3: 519-520.

Soler R, Harvey JA, Kamp AFD, Vet LEM, van der Putten WH et al. (2007) Root herbivores influence the behaviour of an aboveground parasitoid through changes in plant-volatile signals. Oikos 116: 367-376.

Soler R, Harvey JA, Rouchet R, Schaper SV & Bezemer TM (2010) Impacts of belowground herbivory on oviposition decisions in two congeneric butterfly species. Entomologia Experimentalis et Applicata 136: 191-198.

Sokal RR & Rohlf FJ (1995) Biometry. Third Edition. New York: W.H. Freeman and Company.

Soler R, Schaper SV, Bezember MT, Cortesero AM, Hoffmeister TS et al. (2009). Influence of presence and spatial arrangement of belowground insects on host-plant selection of aboveground insects: a field study. Ecological Entomology 34: 339-345.

Spence KO, Bicocca VT, Rosenheim JA (2007) Friend or foe? A plant’s induced response to an omnivore. Environmental Entomology 36: 623-630.

Strauss SY (1991) Direct, indirect, and cumulative effects of three native herbivores on a shared host plant. Ecology 72: 543–558.

StatView (1999) User’s Guide, Version 5.0. SAS Institute, Cary, NC, USA.

Stout MJ, Workman KV, Bostock RM & Duffey SS (1998) Specificity of induced resistance in the tomato, Lycopersicon esculentum. Oecologia 113: 74-81.

Takabayashi J & Dicke M (1996) Plant-carnivore mutualism through herbivore-induced carnivore attractants. Trends in Plant Science 1: 109-113.

Takabayashi J, Sato Y, Horikoshi M, Yamaoka R, Yano S, et al. (1998) Plant effects on parasitoid foraging: differences between two tritrophic systems. Biological Control 11: 97-103.

Tabuchi K, Ueda A & Ozaki K (2010) Contrasting effects of deer browsing on oviposition preference, neonate survival and potential fecundity of galling insect. Ecoscience 17: 379-386.

Teakle RE (1991) Laboratory culture of Heliothis spp. and identification of diseases. Heliothis: Research Methods and Prospects (ed. By MP Zalucki), pp. 11-21. Springer, New York, NY, USA.

Textor S & Gershenzon J (2009) Herbivore induction of the glucosinolate-myrosinase defense system: major trends, biochemical bases and ecological significance. Phytochemistry Reviews 8: 149-170.

200

Tebayashi SI, Horibata Y, Mikagi E, Kashiwagi T, Mekuria DB et al. (2007) Induction of resistance against the leafminer, Liriomyza trifolii, by jasmonic acid in sweet pepper. Bioscience, Biotechnology and Biochemistry 71: 1521-1526.

Teulon DAJ and Penman DR (1991) Effects of temperature and diet on oviposition rate and development time of New Zealand flower thrips, Thrips obscuratus. Entomologia Experimentalis et Applicata 60: 143-155.

Thaler JS, Fidantsef AL & Bostock RM (2002) Antagonism between jasmonate- and salicylate- mediated induced plant resistance: Effects of concentration and timing of elicitation on defense- related proteins, herbivore, and pathogen performance in tomato. Journal of Chemical Ecology 28: 1143-1171.

Thompson JN (1998) Coping with multiple enemies: 10 years of attack on Lomatium dissectum plants. Ecology 79: 2550-2554.

Titmarsh IJ, Storey RI & Strickland GR (1990) The species composition of Helicoverpa Hardwick (Heliothis Ochsenheimer) (Lepidoptera: Noctuidae) infestations on tobacco in far North Queensland. Journal of Australian Entomological Society 29: 81-86.

Toapanta M, Funderburk J, Webb S, Chellemi D & Tsai J (1996) Abundance of Frankliniella spp. (Thysanoptera: Thripidae) on winter and spring host plants. Environmental Entomology 25: 793- 800.

Trichilo PJ & Leigh TF (1986) Predation on spider mite eggs by the Western flower thrips, Frankliniella occidentalis (Thysanoptera:Thripidae), an opportunist in a cotton agroecosystem. Environmental Entomology 15: 821-825.

Trichilo PJ & Leigh TF (1988) Influence of resource quality on the reproductive fitness of flower thrips (Thysanoptera: Thripidae). Annals of the Entomological Society of America 81: 64-70.

Turlings TCJ, Lengwiler UB, Bernasconi ML & Wechsler D (1998) Timing of induced volatile emissions in maize seedlings. Planta 207: 146-152.

Turlings TCJ, Tumlinson JH & Lewis WJ (1990) Exploitation of herbivore-induced plant odors by host seeking parasitic wasps. Science 250: 1251-1253.

Turlings TCJ, Wackers FL, Vet LEM, Lewis WJ & Tumlinson JH (1993) Learning of host-finding cues by hymenopterous parasitoids. Insect Learning: ecological and evolutionary perspectives (ed. by DR Papaj & AC Lewis), pp. 51-78. Chapman and Hall Inc, London, UK.

Underwood N (1998) The timing of induced resistance and induced susceptibility in the soybean Mexican bean beetle system. Oecologia 114: 376-381.

Underwood N & Rausher MD (2002) The effects of host-plant genotype on herbivore population dynamics in a model system. Ecology 81: 1565-1576.

Velasco LRI & Walter GH (1993) Potential of host-switching in Nezara viridula (Hemiptera: Pentatomidae) to enhance survival and reproduction. Environmental Entomology 22: 326-333.

Vet LEM & Dicke M (1992) Ecology of infochemical use by natural enemies in a tritrophic context. Annual Review of Entomology 37: 141-172. 201

Vierbergen G & Mantel WP (1991) Contribution to the knowledge of Frankliniella schultzei (Thysanoptera: Thripidae). Entomologische Berichten (Amsterdam) 51: 7-12.

Viswanathan DV, McNickle G & Thaler JS (2008) Heterogeneity of plants phenotypes caused by herbivore-specific induced responses influences the spatial distribution of herbivores. Ecological Entomology 33: 86-94.

Viswanathan DV, Narwani AJT & Thaler JS (2005) Specificity in induced plant responses shapes patterns of herbivore occurrence on Solanum dulcamara. Ecology 86: 886-896.

Voelckel C & Baldwin IT (2004) Generalist and specialist lepidopteran larvae elicit different transcriptional responses in Nicotiana attenuata, which correlate with larval FAC profiles. Ecology Letters 7: 770-775.

Vos M, Berrocal SM, Karamaouna F, Hemerik L & Vet LEM (2001) Plant-mediated indirect effects and the persistence of parasitoid herbivore communities. Ecology Letters 4: 38-45.

Walling LL (2000) The myriad plant responses to herbivores. Journal of Plant Growth Regulation 19: 195-216.

Walsh PS, Metzger DA & Higuchi R (1991) Chelex 100 as a medium for simple extraction of DNA for PCR-based typing from forensic material. BioTechniques 10: 506–513.

Walter GH (2003) Insect pest management and ecological research. Cambridge University Press, Cambridge, UK.

Weber DC, Mangan FX, Ferro DN & Marsh jr HV (1990) Effect of weed abundance on European borer (Lepidoptera: Pyralidae) infestation of sweet corn. Environmental Entomology 19: 1858- 1865.

White JA & Andow DA (2006) Habitat modification contributes to associational resistance between herbivores. Oecologia 148: 482-490.

Wilson EO (2001) The diversity of life. Penguin, Toronto, Canada.

Wilson LJ (1995) Habitats of twospotted spider mites (Acari: Tetranychidae) during winter and spring in a cotton-producing region of Australia. Environmental Entomology 24: 332-340.

Wijkamp I, Almarza R, Goldbach R & Peters D (1995) Distinct levels of specificity in thrips transmission of tospoviruses. Phytopathology 85: 1069-1074.

Wilson LJ & Bauer LR (1993) Species composition and seasonal abundance of thrips (Thysanoptera) on cotton in the Namoi Valley. Journal of the Australian Entomological Society. 32: 187-192.

Wilson LJ, Bauer LR & Walter GH (1996) ‘Phytophagous’ thrips are facultative predators of twospotted spider mites (Acari: Tetranychidae) on cotton in Australia. Bulletin of Entomological Research 86: 297-305.

202

Wimmer D, Hoffmann & Schausberger P (2008) Prey suitability of western flower thrips, Frankliniella occidentalis, and onion thrips, Thrips tabaci, for the predatory mite Amblyseius swirskii. Biocontrol Science and Technology 18: 533-542.

Williams III L, Rodriguez-Saona C, Paré PW & Crafts-Brandner SJ (2005) The piercing-sucking herbivores Lygus hesperus and Nezara viridulainduce volatile emissions in plants. Archives of Insect Biochemistry and Physiology 58: 84-96.

Wilson LJ, Sadras VO & Bauer L (1994) Effects of thrips on growth, maturity and yield of cotton – Preliminary results. pp. 113-120. Proceedings of the seventh Australian Cotton Conference, 10-12 August 1994, Broadbeach, Australia. Australian Cotton Growers Research Association, Wee Waa, NSW. Australia.

Wilson LJ, Sadras VO, Heimoana SC & Gibb D (2003) How to succeed by doing nothing: cotton compensation after simulated early season pest damage. Crop Science 43: 2125-2134.

Williams S, Wilson LJ & Vogel S (2011) Pests & beneficial in Australian cotton landscapes. A publication of the Cotton Catchment Communities Cooperative Research Centre, Narrabri, NSW. Greenmount Press, Toowoomba, QLD.

Wise MJ & Weinberg AM (2002) Prior flea beetle herbivory affects oviposition preference and larval performance of a potato beetle on their shared host plant. Ecological Entomology 27: 115- 122.

Wold EN & Marquis RJ (1997) Induced defense in white oak: effects on herbivores and consequences for the plant. Ecology 78: 1356-1369.

Yudin LS, Tabashnik BE, Cho JJ & Mitchell WC (1986) Host range of western flower thrips Frankliniella occidentalis (Thysanoptera: Thripidae), with special reference to Leucaena glauca. Environmental Entomology 15: 1292-1295.

Yudin LS, Tabashnik BE, Cho JJ & Mitchell WC (1988) Colonization of weeds and lettuce by thrips (Thysanoptera: Thripidae). Environmental Entomology 17: 522-526.

Zarate SI, Kempema LA & Walling LL (2007) Silverleaf whitefly induces salicylic acid defenses and suppresses effectual jasmonic acid defenses. Plant Physiology 143: 866–875.

Zhuang X, Köllner TG, Zhao N, Li G, Jiang Y et al. (2012) Dynamic evolution of herbivore- induced sesquiterpene biosynthesis in sorghum and related grass crops. The Plant Journal 69: 70-80.

Zhang PJ, Zheng SJ, van Loon JJA, Boland W, David A et al. (2009) Whiteflies interfere with indirect plant defense against spider mites in Lima bean. Proceedings of the National Academy of Sciences 106: 21202-21207.

203