Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Using herbivore induced plant volatiles and environmental factors for sensitive monitoring of pests infestation… Miresmailli, Saber 2009

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2009_fall_miresmailli_saber.pdf [ 11.58MB ]
Metadata
JSON: 24-1.0067692.json
JSON-LD: 24-1.0067692-ld.json
RDF/XML (Pretty): 24-1.0067692-rdf.xml
RDF/JSON: 24-1.0067692-rdf.json
Turtle: 24-1.0067692-turtle.txt
N-Triples: 24-1.0067692-rdf-ntriples.txt
Original Record: 24-1.0067692-source.json
Full Text
24-1.0067692-fulltext.txt
Citation
24-1.0067692.ris

Full Text

USING HERBIVORE INDUCED PLANT VOLATILES AND ENVIRONMENTAL FACTORS FOR SENSITIVE MONITORING OF PESTS INFESTATION IN GREENHOUSE TOMATO CROPS by SABER MIRESMAILLI B.Sc.,The University of Tehran, 2001 M.Sc., The university of British Columbia, 2006  A THESIS SUBMITED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Plant Science)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  September 2009  © Saber Miresmailli, 2009  ABSTRACT Suitability of herbivore-induced plant volatiles as indicators of cabbage looper (Trichoplusia ni, Noctuidae) infestation on tomato plants was assessed for developing a pest monitoring system inside tomato greenhouses. From volatile blends of infested tomato plants, four compounds have been selected as T. ni infestation indicators: (Z)-3-hexenyl acetate, (E)- !-ocimene, limonene and !caryophyllene. Laboratory results indicated significant quantitative differences in the emission level of these indicator chemicals from infested plants compared to clean plants. Research greenhouse trials confirmed these results and also indicated that these differences are detectable six hours after initiation of infestation. The research greenhouse trials also indicated that is it possible to obtain information about T.ni population density, their location within the plant canopy and their feeding duration. A survey was conducted inside a commercial greenhouse. Indicator chemical emission levels, pest infestation status, environmental factors and operational practices were recorded. Pest infestation was found to have significant effect on the emission of indicator chemicals. A portable gas chromatograph was used for most phases of this research. This instrument was found suitable for fieldwork and monitoring rapid changes in the emission levels of plant volatiles. In general, these four chemical volatiles were found to be suitable indicators of T. ni infestation in greenhouse tomato plants.  ii  TABLE OF CONTENTS ABSTRACT ..........................................................................................................................................................ii TABLE OF CONTENTS .................................................................................................................................. iii LIST OF TABLES ..............................................................................................................................................vi LIST OF FIGURES ...........................................................................................................................................vii ACKNOWLEDGEMENTS ...............................................................................................................................ix DEDICATION ......................................................................................................................................................x CO-AUTHORSHIP STATEMENT .................................................................................................................xi CHAPTER ONE: INTRODUCTION ...............................................................................................................1 1.1 PEST MONITORING AND IPM ...............................................................................................................3 1.2 PLANTS AS AN ALTERNATIVE SOURCE OF INFORMATION ....................................................6 1.3 ANALYZING PLANT VOLATILES.........................................................................................................9 1.4 SENSORY SYSTEMS.................................................................................................................................11 1.5 STATISTICAL AND COMPUTATIONAL METHODS .....................................................................12 1.6 PROJECT GOALS AND OBJECTIVES ................................................................................................13 1.7 REFERENCES.............................................................................................................................................16 CHAPTER TWO: QUALITATIVE ASSESSMENT OF AN ULTRA-FAST PORTABLE GAS CHROMATOGRAPH (ZNOSE!) FOR ANALYZING VOLATILE ORGANIC CHEMICALS IN THE LABORATORY AND GREENHOUSES.............................................................................................27 2.1 INTRODUCTION .......................................................................................................................................27  2.1.1 HOW DOES THE ZNOSE! WORK? .........................................................................29 2.1.2 CALIBRATION......................................................................................................34 2.1.3 REGULAR CARE AND MAINTENANCE ......................................................................35 2.2 MATERIALS AND METHODS ...............................................................................................................36  2.2.1 QUALITATIVE COMPARISON OF SAMPLES BY THE ZNOSE! AND A CONVENTIONAL GCMS.............................................................................................................................36 2.2.2 REAGENTS ..........................................................................................................36 2.2.3 PREPARATION OF INTERNAL AND EXTERNAL STANDARD SOLUTION............................37 2.2.4 SAMPLE PREPARATION .........................................................................................38 2.2.5 GC/MS PARAMETERS ..........................................................................................38 2.2.6 CALCULATION OF RESULTS ..................................................................................39 2.2.7 PREPARATION OF STANDARD SOLUTION FOR THE ZNOSE! AND SYSTEM CALIBRATION ...............................................................................................................39 2.2.8 SAMPLE PREPARATION FOR THE ZNOSE!..............................................................40 2.2.9 ZNOSE! GC PARAMETERS ..................................................................................40 2.2.10 COMPARING RETENTION INDICES BETWEEN MACHINES .........................................41 2.2.11 RETENTION INDEX VARIATION OF LIMONENE IN LABORATORY AND COMMERCIAL GREENHOUSES .............................................................................................................41 2.2.12 DATA ANALYSES ................................................................................................43  iii  2.3 RESULTS ......................................................................................................................................................44  2.3.1 COMPARING RETENTION INDICES OF SELECTED VOCS ANALYZED BY THE ZNOSE! AND CONVENTIONAL GC-MS...............................................................................................44 2.3.2 RETENTION INDEX VARIATION OF LIMONENE IN LABORATORY AND COMMERCIAL GREENHOUSES .............................................................................................................45 2.4 DISCUSSION ...............................................................................................................................................49 2.5 REFERENCES.............................................................................................................................................51 CHAPTER THREE: USING HERBIVORE-INDUCED PLANT VOLATILES FOR DETECTING CABBAGE LOOPER INFESTATION ON GREENHOUSE TOMATO PLANTS ...............................54 3.1 INTRODUCTION .......................................................................................................................................54  3.1.1 GREENHOUSE TOMATO ........................................................................................54 3.1.2 PEST MONITORING ..............................................................................................56 3.1.3 HERBIVORE-INDUCED PLANT VOLATILES ...............................................................57 3.1.4 PLANT VOLATILE COLLECTION AND ANALYSIS .........................................................59 3.1.5 THE ULTRA-FAST GAS CHROMATOGRAPH (ZNOSE!) ..............................................60 3.2 MATERIALS AND METHODS ...............................................................................................................61  3.2.1 CABBAGE LOOPER ...............................................................................................61 3.2.2 PLANT MATERIAL ................................................................................................61 3.2.3 PLANT VOLATILE COLLECTION SYSTEM ..................................................................62 3.2.4 LABORATORY EXPERIMENT FOR IDENTIFYING INDICATOR CHEMICALS .......................64 3.2.5 PLANT VOLATILE ANALYSIS WITH GC-MS ..............................................................64 3.2.6 GREENHOUSE EXPERIMENT ..................................................................................65 3.2.7 ZNOSE! CALIBRATION AND PROGRAM PROPERTIES ................................................67 3.2.8 DATA ANALYSES ..................................................................................................68 3.2.8.1 CHEMICAL BASELINES BEFORE INFESTATION .......................................................68 3.2.8.2 COMPARING LEVEL OF CHEMICALS IN CONTROL AND TREATMENT GROUPS .............69 3.2.8.3 CLASSIFICATION OF PLANTS INTO CLEAN AND INFESTED GROUPS ..........................70 3.3 RESULTS ......................................................................................................................................................71  3.3.1 LABORATORY EXPERIMENTS .................................................................................71 3.3.2 ZNOSE! CALIBRATION ........................................................................................71 3.3.3 GREENHOUSE EXPERIMENTS ................................................................................73 3.3.4 CLASSIFICATION OF PLANTS INTO CLEAN AND INFESTED GROUPS .............................77 3.4 DISCUSSION ...............................................................................................................................................78 3.5 REFERENCES.............................................................................................................................................81 CHAPTER FOUR: EFFECT OF PEST DENSITY, DISTRIBUTION OF PESTS IN THE PLANT CANOPY AND DAMAGE DURATION ON HERBIVORE-INDUCED PLANT VOLATILE EMISSION RATE IN THE CABBAGE LOOPER-TOMATO SYSTEM ...............................................90 4.1 INTRODUCTION .......................................................................................................................................90 4.2 MATERIALS AND METHODS ...............................................................................................................92  4.2.1 CABBAGE LOOPER ...............................................................................................92 4.2.2 PLANT MATERIAL ................................................................................................92 4.2.3 IDENTIFYING INDICATOR CHEMICALS ....................................................................92 iv  4.2.4 GREENHOUSE EXPERIMENT ..................................................................................93 4.2.5 EXPERIMENT ONE: EFFECT OF INFESTATION LEVELS ..............................................94 4.2.6 EXPERIMENT TWO: EFFECT OF LARVAL DISTRIBUTION WITHIN THE PLANT CANOPY ....95 4.2.7 EXPERIMENT THREE: EFFECT OF FEEDING DURATION BY LARVAE ............................95 4.2.8 ZNOSE! CALIBRATION AND PROGRAM PROPERTIES ................................................96 4.2.9 DATA ANALYSES ..................................................................................................97 4.2.9.1 CHEMICAL BASELINES BEFORE INFESTATION .......................................................97 4.2.9.2 COMPARING LEVEL OF CHEMICALS IN CONTROL AND TREATMENT GROUPS .............97 4.3 RESULTS ......................................................................................................................................................98  4.3.1 IDENTIFYING INDICATOR CHEMICALS ....................................................................98 4.3.2 EXPERIMENT ONE: EFFECT OF INFESTATION LEVEL ON VOLATILE EMISSION ..............99 4.3.3 EXPERIMENT TWO: EFFECT OF LARVAL DISTRIBUTION WITHIN PLANT ..................... 103 4.3.3 EXPERIMENT THREE: EFFECT OF FEEDING DURATION BY LARVAE .......................... 108 4.4 DISCUSSION .............................................................................................................................................112 4.5 REFERENCES...........................................................................................................................................118 CHAPTER FIVE: EFFECT OF PEST INFESTATION, CROP MAINTENANCE PRACTICES AND ENVIRONMENTAL FACTORS ON VOLATILE EMISSION RATE OF TOMATO PLANTS IN A COMMERCIAL GREENHOUSE.......................................................................................................123 5.1 INTRODUCTION .....................................................................................................................................123 5.2 MATERIALS AND METHODS .............................................................................................................125  5.2.1 COMMERCIAL GREENHOUSE .............................................................................. 125 5.2.2 MODEL PEST .................................................................................................... 125 5.2.3 INDICATOR CHEMICALS...................................................................................... 125 5.2.4 ZNOSE! PROGRAM PROPERTIES ........................................................................ 125 5.2.5 RECORDING ENVIRONMENTAL CONDITION ........................................................... 126 5.2.6 SELECTING SAMPLING SITES ............................................................................... 126 5.2.7 DATA ANALYSES ................................................................................................ 129 5.3 RESULTS ....................................................................................................................................................130 5.4 DISCUSSION .............................................................................................................................................138 5.5 REFERENCES...........................................................................................................................................141 CHAPTER SIX: SUMMARY AND DISCUSSION ...................................................................................144 6.1 REFERENCES...........................................................................................................................................153 APPENDIX ONE .............................................................................................................................................156 APPENDIX TWO ............................................................................................................................................160  !  v  LIST OF TABLES TABLE 2.1. STANDARD CONCENTRATE SOLUTION…………… .........................................................................37 TABLE 2.2 GC TEMPERATURE PROFILE………………………..........................................................................39 TABLE 2.3. COMPARING THE RETENTION INDICES OF FOUR CHEMICALS ANALYSED BY GC-MS AND ZNOSE! IN LABORATORY AND GREENHOUSE…………………… .........................................................................45 TABLE 2.4. EFFECT OF TEMPERATURE, RELATIVE HUMIDITY AND ANALYSES METHOD (GC-MS+ NEAT CHEMICAL, ZNOSE +NEAT CHEMICAL, ZNOSE+PLANT IN LABORATORY AND ZNOSE IN GREENHOSUE) ON RETENTION INDEX OF LIMONENE…………………….........................................................................45 TABLE 3.1. GEE REGRESSION COEFFICIENTS- (Z)-3-HEXENYL ACETATE EMISSION IN TOMATO PLANTS IN RESPONSE TO CABBAGE LOOPER INFESTATION OVER TIME .......................................................................74 TABLE 3.2. GEE REGRESSION COEFFICIENTS- (E)-"-OCIMENE EMISSION IN TOMATO PLANTS IN RESPONSE TO CABBAGE LOOPER INFESTATION OVER TIME……………. ........................................................................75 TABLE 3.3. GEE REGRESSION COEFFICIENTS- LIMONENE EMISSION IN TOMATO PLANTS IN RESPONSE TO CABBAGE LOOPER INFESTATION OVER TIME……………. ........................................................................76 TABLE 3.4. GEE REGRESSION COEFFICIENTS- "-CARYOPHYLLENE EMISSION IN TOMATO PLANTS IN RESPONSE TO CABBAGE LOOPER INFESTATION OVER TIME…………. .......................................................................76 TABLE 3.5. LINEAR DISCRIMINANT ANALYSIS OF TOMATO PLANT VOLATILES AFTER 6 HOURS OF INFESTATION WITH CABBAGE LOOPER. PRIOR PROBABILITIES OF GROUPS ARE : 0.99 FOR CLEAN AND 0.01 FOR INFESTED PLANTS……………………………………… .........................................................................77 TABLE 4.1. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER INFESTATION LEVEL ON THE (Z)3-HEXENYL A CETATE EMISSION FROM TOMATO PLANTS OVER TIME ....................................................100 TABLE 4.2. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER INFESTATION LEVEL ON THE (E)"-OCIMENE EMISSION FROM TOMATO PLANTS OVER TIME ......................................................................101 TABLE 4.3. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER INFESTATION LEVEL ON THE LIMONENE EMISSION FROM TOMATO PLANTS OVER TIME .......................................................................102 TABLE 4.4. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER INFESTATION LEVEL ON THE "CARYOPHYLLENE EMISSION FROM TOMATO PLANTS OVER TIME............................................................103 TABLE 4.5. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER DISTRIBUTION WITHIN THE TOMATO PLANT CANOPY ON THE EMISSION LEVEL OF (Z)-3-HEXENYL ACETATE OVER TIME ...............105 TABLE 4.6. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER DISTRIBUTION WITHIN THE TOMATO PLANT CANOPY ON THE EMISSION LEVEL OF (E)-"-OCIMENE OVER TIME ...............................105 TABLE 4.7. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER DISTRIBUTION WITHIN THE TOMATO PLANT CANOPY ON THE EMISSION LEVEL OF LIMONENE OVER TIME........................................106 TABLE 4.8. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER DISTRIBUTION WITHIN THE TOMATO PLANT CANOPY ON THE EMISSION LEVEL OF "-CARYOPHYLLENE OVER TIME ........................107 TABLE 4.9. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER FEEDING DURATION ON THE EMISSION LEVEL OF (Z)-3-HEXENYL ACETATE FROM TOMATO PLANTS OVER TIME ..............................109 TABLE 4.10. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER FEEDING DURATION ON THE EMISSION LEVEL OF (E)-"-OCIMENE FROM TOMATO PLANTS OVER TIME ...............................................110 TABLE 4.11. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER FEEDING DURATION ON THE EMISSION LEVEL OF LIMONENE FROM TOMATO PLANTS OVER TIME .......................................................110 TABLE 4.12. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER FEEDING DURATION ON THE EMISSION LEVEL OF "-CARYOPHYLLENE FROM TOMATO PLANTS OVER TIME ........................................111 TABLE 5.1. GEE REGRESSION COEFFICIENTS- (Z)-3-HEXENYL ACETATE .......................................................133 TABLE 5.2. GEE REGRESSION COEFFICIENTS- (E)-"-OCIMENE ........................................................................134 TABLE 5.3. GEE REGRESSION COEFFICIENTS- LIMONENE……........................................................................135 TABLE 5.4. GEE REGRESSION COEFFICIENTS- "-CARYOPHYLLENE .................................................................136 TABLE 6.1 FEEDBACK TO THE ZN OSE! MANUFACTURER……… ...................................................................151  vi  LIST OF FIGURES FIG 1.1 ACCUMULATION OF KNOWLEDGE AND COLLECTIVE WORK OF SCIENTISTS ............................................1 Fig 1.2 Some examples of field measurement technologies. (A) WATCHDOG WEATHER STATION MODEL 2900ET (B) WATERPROOF EC METER (C) CARDY TWIN PH METER (D) CARDY SODIUM METER (E) WATCHDOG WIRELESS CROP MONITOR 3540 (F) AELDSCOUT CM 1000 CHLOROPHYLL METER (G) WATCHDOG 2000 WEATHER TRACKER. SPECTRUM TECHNOLOGIES INC. PLAINFIELD, I LLINOIS. PICTURES OBTAINED FROM THE COMPANY WEBSITE: WWW. SPECMETERS.COM .......................................5 FIG 1.3 THE MOST COMMONLY USED HIPV COLLECTION METHODS DURING A TEN -YEARS PERIOD FROM 1995-2004. DATA FROM D'ALESSANDRO AND TURLINGS (2006) ...........................................................10 FIG 2.1 THE ZNOSE! VALVE, TRAP AND COLUMN………….. ..........................................................................30 FIG 2.2 THE ZNOSE! MICROSENSE! SOFTWARE- METHOD DEVELOPMENT...................................................31 FIG 2.3 COMPARING ZNOSE! RESULTS IN LABORATORY AND FIELD.................................................................42 FIG 2.4 ANALYZING TOMATO VOLATILES IN LABORATORY….. .........................................................................43 FIG 2.5 VARIATION OF RETENTION INDICES OF LIMONENE IN RESPONSE TO RELATIVE HUMIDITY AND TEMPERATURE IN LABORATORY AND GREENHOUSE…… .........................................................................48 FIG 2.6 ZNOSE! IN THE FIELD………………………………. ..........................................................................50 FIG 3.1. (A) ESTIMATED VALUE OF THE GREENHOUSE VEGETABLE INDUSTRY IN CANADA, (B) MAIN GREENHOUSE VEGETABLE CROPS IN CANADA IN 2003, (C) FARM GATE VALUE OF THE FOUR MAIN VEGETABLE CROPS PRODUCED IN 2003 IN FIELD VERSUS GREENHOUSE, (D) ESTIMATED GREENHOUSE TOMATO AREA IN MEXICO VERSUS CANADA. DATA FROM STATISTICS CANADA AND AGRICULTURE AND AGRI-FOOD CANADA (2006,2007)………………............................................................................55 FIG 3.2. VOLATILE COLLECTION SYSTEM. PQ = PORAPAK Q TUBES, V= VACUUM PUMP, F= AIR MICRO MIST FILTER AND ACTIVE CHARCOAL FILTER, P= AIR PUMP, HF= A IR PURIFIER FILTER ................................63 FIG 3.3 GREENHOUSE SETUP FOR COLLECTING VOLATILES FROM TOMATO PLANTS INFESTED WITH CABBAGE LOOPER………………………………………………… .........................................................................66 FIG 3.4 DIFFERENCES IN VOLATILE BLEND OF INFESTED (WITH CABBAGE LOOPER) TOMATO PLANTS (UPPER CHROMATOGRAM) AND CLEAN TOMATO PLANTS (LOWER CHROMATOGRAM). THERE ARE SEVERAL UNKNOWN COMPOUNDS IN THE BLEND………………… .........................................................................72 FIG 3.5 VOLATILE EMISSION (NANOGRAM) IN CONTROL AND TREATMENT (INFESTED WITH CABBAGE LOOPER) TOMATO PLANTS OVER TIME (HR) (MEAN±SD)………............................................................................74 FIG 4.1 EXPERIMENT ONE: EFFECT OF DIFFERENT PEST DENSITIES ON THE VOLATILE EMISSION OF TOMATO PLANTS. PESTS WERE PLACED ON UPPER PART OF CANOPY. 0= CONTROL, 1,3,9 = NUMBER OF PESTS PER PLANT…………………………………………………. ..........................................................................94 FIG 4.2 EFFECT OF LARVAL DISTRIBUTION WITHIN PLANT CANOPY. PESTS WERE PLACED ON UPPER OR LOWER PART OF TOMATO PLANT CANOPY . THE INFESTED AREA WAS COVERED WITH A SCREEN TO LIMIT PEST MOVEMENT. U=UPPER, L= LOWER……………………...........................................................................95 FIG 4.3 EFFECT OF FEEDING DURATION BY LARVAE ON VOLATILE EMISSION LEVEL OF TOMATO PLANTS. P=PEST ( CABBAGE LOPPER LARVAE) REMOVED AFTER 6, 12 OR 24 HOURS ............................................96 FIG 4.4 VOLATILE EMISSION LEVEL (NANOGRAM) IN CONTROL AND TREATMENT TOMATO PLANTS OVER TIME (HOURS AFTER PLACEMENT OF LARVAE) IN RESPONSE TO DIFFERENT CABBAGE LOOPER INFESTATION LEVEL (MEAN±SD)…………………………………… ..........................................................................99 FIG 4.5 VOLATILE EMISSION LEVEL (NANOGRAM) IN CONTROL AND TREATMENT TOMATO PLANTS OVER TIME (HOURS AFTER PLACEMENT OF LARVAE) IN RESPONSE TO DIFFERENT CABBAGE LOOPER DISTRIBUTION WITHIN THE PLANT CANOPY (MEAN±SD)……………...........................................................................104 FIG 4.6 VOLATILE EMISSION LEVEL (NANOGRAM) IN CONTROL AND TREATMENT TOMATO PLANTS OVER TIME (HOURS AFTER PLACEMENT OF LARVAE) IN RESPONSE TO DIFFERENT FEEDING DURATION OF CABBAGE LOOPER LARVAE (MEAN±SD)………………………… ........................................................................108 FIG 5.1 SAMPLING SITES IN THE HOUWELING’S HOT HOUSE ,DELTA, BC. .....................................................127 FIG 5.2 GPS COORDINATION OF SAMPLING SITES- SAMPLING TEAM ...............................................................128 FIG 5.3 VARIATION OF INDICATOR CHEMICALS AND ENVIRONMENTAL FACTORS AND RECORDS OF PEST INFESTATION AND MAINTENANCE PRACTICES AT SAMPLE SITE NUMBER TWO – DAMAGE SYMPTOMS, NO PEST AT WEEK 8…………………………………………. .....................................................................131  vii  FIG 5.4 VARIATION OF INDICATOR CHEMICALS AND ENVIRONMENTAL FACTORS AND RECORDS OF PEST INFESTATION AND MAINTENANCE PRACTICES AT SAMPLE SITE NUMBER FOUR- PEST PRESENCE AT WEEK EIGHT……………………………………………………......................................................................132 FIG 6.1. COMPARING CONVENTIONAL PEST MONITORING WITH A PROPOSED INTELLIGENT PEST MONITORING SYSTEM: PEST SCOUTS VARY IN EXPERIENCE AND PERFORMANCE, POSSIBLY RESULTING IN INCONSISTANCIES. INTELLIGENT PEST MONITORING ARE BASED ON MODERN SENSORY SYSTEMS AND CONSIDER A WIDE RANGE OF VARIABLES IN THE DECISION MAKING PROCESS. FEEDBACK FROM HUMAN EXPERTS WILL REMAIN PART OF THE SYSTEM AND WILL IMPROVE ITS PRECISION AND PERFORMANCE…………………………………………........................................................................150 MASS SPECTRA OF EXTERNAL STANDARDS AND NIST LIBRARY MATCH- MAJOR COMPOUNDS OF INSECT REPELLENT FORMULATION ……………………………… .....................................................................156 MASS SPECTRA OF LIMONENE EXTERNAL STANDARD AND VARIAN LIBRARY MATCH ....................................157 MASS SPECTRA OF INDICATOR CHEMICALS……………………......................................................................159  viii  ACKNOWLEDGEMENTS Many people contributed significantly to this project. My research supervisor and four committee members guided this study from its beginning to its final completion, and my gratitude for their valuable input and support is profound. I take this opportunity to thank Professor Murray Isman for his great assistance support and encouragement. Without his support, this study could not have been accomplished. I also would like to thank Dr. Gerhard Gries for allowing me to conduct parts of my research in his laboratory, supporting me throughout my research and providing critical feedbacks on my manuscripts; Dr. Ruben Zamar for his immense efforts in helping me with the experimental design and statistical analyses of my project; Dr. David Ehret for providing critical feedbacks on my project and acting as a liaison between me and the commercial greenhouse growers and Dr. Eduardo Jovel for teaching me the importance of looking at the big picture. I must express my warmest gratitude to Mrs. Regine Gries who generously helped me with the analytical chemistry aspect of my project and thaught me valuable lessons about chromatography. I also have to thank Professor Jorg Bohlmann for introducing me to the wonderful world of plant chemical ecology. It was in his class where I first came up with the idea of this project. Many thanks to Mr. Ruben Houweling and Houweling’s Hot House and Mr. Andre Lenders and Van Der Meulen greenhouse for providing plant materials and granting access to conduct a survey inside their greenhouses, and to Mr. Rod Bradbury from Ecosafe Natural Products Inc. for providing essential oils and analyzing samples (Chapter Two). Special thanks to Mr. Amandeep Bal, who maintained my communication link with the growers. My deepest appreciation also goes to my lovely wife Mrs. Maryam Antikchi who helped and supported me during my project. Finally, I would like to thank my research volunteer team for helping me with sampling in greenhouses and setting up experiments. This Project was funded in part by NSERC Industrial Postgraduate Scholarship in conjunction with BC Greenhouse Growers’ Association. ix  DEDICATION I dedicate this PhD thesis to my parents, Mehdi Miresmailli and Tahereh Gharib, the two most special people in my life. They not only gave me life, but also filled it with all the love and affection one can wish for, and to my sisters, Saba and Sanam Miresmailli who always loved me unconditionally. Thank you. I also would like to dedicate this thesis to the memory of my late grandparents. My grandpa used to walk me to school everyday when I was little. He always wanted me to become a doctor so I could fix his leg, which had been injured in an accident. Well papa! I finally become a Doctor…just not the kind who can cure people! but I am sure you would be fine with that anyhow. Miss you.  x  CO-AUTHORSHIP STATEMENT Chapter one: Modified version of chapter one is a book chapter accepted for publication. Dan Badulescu, Maryam Mahdaviani, Ruben Zamar and Murray Isman provided insights and contributed to the writing. I formulated the idea, designed the illustrations and wrote majority of the chapter. Chapter two: Modified version of chapter two is a manuscript submitted for publication. I designed the project in consultation with Murray Isman and Rod Bradbury. I conducted all laboratory work, data analyses and wrote the manuscript. Rod Bradbury helped with GC-MS analysis. Murray Isman and Rod Bradbury provided insights and contributed to the writing. Chapter three, four and five: Modified versions of Chapters 3,4 and 5 are manuscripts submitted for publication. I designed the project in consultation with Murray Isman, Gerhard Gries, Regine Gries, Dave Ehret and Ruben Zamar. I conducted all laboratory work, data analyses and wrote the manuscripts. Aline Tabet wrote the scripts for statistical analyses. Regine Gries analyzed plant volatile samples collected in laboratory with GC-MS. Murray Isman, Gerhard Gries, Regine Gries and Ruben Zamar provided insights and contributed to the writing.  xi  CHAPTER ONE: INTRODUCTION1 With the exclusion of serendipitous and accidental discoveries, science as I understand it, is the collective work of numerous researchers and scientists who generate small bits of information and put them together to improve our understanding of certain issues. Epiphanies are very rare in science. Everything must follow a very well established set of scientific rules and reasoning. We must ask appropriate questions, develop hypotheses based on those questions and prior knowledge, design experiments and choose proper statistical methods that enable us to test those hypotheses. Although this seems like a solid path that provides us with some degree of certainty, one should not forget that scientists are limited to their tools and instruments. Our knowledge goes as far as our instruments can take us.  FIG 1.1 ACCUMULATION OF KNOWLEDGE AND COLLECTIVE WORK OF SCIENTISTS  1  A version of this chapter has been accepted for publication. Miresmailli S., Badulescu D., Mahdaviani M., Zamar R.H. and Isman M.B. 2009. Integrating plant chemical ecology, sensors and artificial intelligence for accurate pest monitoring in: Columbus F. (Ed).Tomatoes: agricultural procedures, pathogen interaction and health effects. Nova Science Publishers, NY.  1  New tools and instruments can fundamentally change our understanding of certain issues. Microscopes, telescopes, fine analytical tools and tracing and tracking devices are a few examples of instruments that changed our way of seeing the world and understanding how it works. On the other hand, there were many occasions where a scientist developed a tool or a theory or introduced a concept for a specific situation or reason but other scientists over time realized that they could use this information or tool for purposes other than that originally intended. When President Eisenhower created the Advanced Research Projects Agency (ARPA) in 1958 in response to the launching of the Soviet Union’s Sputnik, the main goal was to promote research that would ensure that the USA would never again be beaten in any technological race. One of the ARPA’s offices was the Information Processing Technologies Office (IPTO) in which the first ideas of a Galactic Network were formed by J. C. R. Licklider of the Massachusetts Institute of Technologies in 1962. That was the origin of what we know today as the Internet and it was meant to create a mancomputer symbiosis (Kleinrock 2008). Another example is GPS (Global Positioning System) that was also developed as a military project for similar reasons (O'Brien and Griffin 2007). Now we carry cell phones in our pocket that provide us both Internet access and GPS functionality. In medical sciences, we have Viagra! that was initially developed as a treatment for high blood pressure and angina pectoris but now is used as a cure for erectile dysfunction (Boolell et al. 1996). Botox was first developed to treat crossed eyes and uncontrollable blinking but now is widely used for cosmetic surgeries (Dunlop et al. 1988).  2  Interdisciplinary projects are now quite common in many universities and research centers. We can see scientists, researchers and experts from completely different disciplines working together to solve a mystery or develop a new tool. A bizarre example of such collaboration happened in Britain's largest children's hospital, the Great Ormond Street Hospital. They changed their patient handoff techniques by copying the choreographed pit stops of Italy's Ferrari Formula One racing team (Naik 2006). My thesis results fall into the same category. Inspired by an introductory course in chemical ecology and utilizing prior knowledge of pest management and little information I had about sensory systems and electronics (thanks to my wife who is an engineer), I formulated this project which is an integration of chemical ecology and integrated pest management (IPM). This project is my humble effort to contribute to a relatively important issue in pest management using the great pool of knowledge accumulated in past decades in chemical ecology. I hope that other researchers and scientists who are interested in the same subject find this tiny piece of information useful.  !"!#$%&'#()*+'),+*-#.*/#0$(# For centuries, humans have been battling to protect crops from pests and diseases. Despite recent progress in crop protection tools and techniques, we still experience 26% to 40% crop losses in production and storage (Oerke and Dehne 2004). Before the advent of synthetic pesticides like DDT, most pest control depended on ecological knowledge of the pest and how its host environment could be altered to reduce damage (Van Den Bosch 1969). After the Second World War, pest management became synonymous with chemical pesticides worldwide. The publication of Silent Spring by Rachel Carson in  3  1962 raised awareness of the potentially deleterious effects of pesticides on the environment and health, launching a global environmental movement. Since then, integrated pest management or IPM, has become the alternative paradigm for crop protection. IPM stresses the interaction of multiple tools aimed at maintaining pest populations below the level where economic damage occurs. Pest management could benefit from a model based on systematic and automated tools and incorporating newer technologies for decision-making. Integrated pest management can be considered as an important part of a broader concept of integrated plant management that looks into the effect of the major elements of an agro-ecosystem on the crop (Cook 2000). These elements include climate, nutrition, soil, water and finally pests and diseases. Over the past few decades, several instruments have been developed for measuring different agro-ecosystem elements and provide information that enables us to develop decision-making models. As Alan Thomson described in his paper on indicator-based knowledge management, the three major components of participatory decision-making systems are knowledge, communication and reporting (Thomson 2005). We now have access to sophisticated tools that can measure the majority of environmental factors including temperature, relative humidity, airflow and light intensity. There are instruments that can measure the level of major nutrients in soil or assess the level of photosynthesis and gas exchange in plants. We have a variety of quality control systems for water and soil quality so we can easily obtain a lot of information about the cropping system (Ushada et al. 2007). The beauty of these instruments is that they can communicate with each other and provide real-time reports. We have wireless weather  4  stations and crop monitors that can provide detailed reports on the changes in our cropping system (Wang et al. 2006). Many of these instruments have been widely used in greenhouses for more than two decades (Pucheta et al. 2006).  Fig 1.2 Some examples of field measurement technologies. (A) WATCHDOG WEATHER STATION MODEL 2900ET (B) WATERPROOF EC METER (C) CARDY TWIN PH METER (D) CARDY SODIUM METER (E) WATCHDOG WIRELESS CROP MONITOR 3540 (F) AELDSCOUT CM 1000 CHLOROPHYLL METER (G) WATCHDOG 2000 WEATHER TRACKER. SPECTRUM TECHNOLOGIES INC. PLAINFIELD, I LLINOIS. PICTURES OBTAINED FROM THE COMPANY WEBSITE: WWW. SPECMETERS.COM  Monitoring pest populations is a cornerstone of the IPM philosophy but it has lagged in comparison to the development of other monitoring tools. Current pest monitoring methods and decision-making are built on a common model: detecting pests and/or signs of their presence or damage (Hughes 1999). Even modern technologies are following the  5  same path and base their monitoring on pest-related indicators ( Drake et al. 2002; Bange et al. 2004; Oerke et al. 2006; Skaloudova et al. 2006; Boissard et al. 2008). The majority of pest monitoring techniques enable the grower to estimate pest populations or predict pest outbreaks; however, we still use human scouts to establish the precise location of pests within a field or a greenhouse. Scouts are faced with the challenge of visually scanning a large representative number of plants, plant organs and a variety of pests. Fields and greenhouses pose problems due to limitations of human vision; some pests might be overlooked in their early stages of development. Besides, scouts need extensive training and their performance is never the same due to different levels of experience, individual values and considerations; thus making consistent decisions is a challenge (Lichtenberg and Berlind 2005). Therefore, my main goal in this project is to investigate the possibility of creating a new tool that can increase the efficiency and performance of human scouts based on an alternative source of information. For pest monitoring, one should not just look for the pests but rather detect the cues generated by plants, as they may be more informative.  !"1 $2.*'&#.&#.*#.2'%,*.'+3%#&)4,5%#)6#+*6),7.'+)*# Plants cannot talk and they cannot walk but they can communicate through several types of cues and responses2. They can provide us with useful information about their health. Some even believe that we should put plants in charge of their own wellbeing and let them control the optimum conditions for their development and growth (Janssen et al.  2  I borrowed this phrase from Professor Jorg Bohlmann. It was the opening statement of his plant chemical ecology course that I took few years ago. 6  2004). Plants can defend themselves against threats both directly and indirectly and can actively manipulate their environment (Schoonhoven et al. 2006). Despite several controversial interpretations of plant-generated responses – whether they are intelligent or reflexive- or their evolutionary raison d'être (Heil et al. 2008; Trewavas 2003), it is an accepted fact that most plants are capable of responding to changes in their surroundings and can convey precise information about their overall health status through those responses (Volkov and Ranatunga 2006). There is a large body of evidence that supports this claim. For example, some plants are capable of showing the footsteps of insects crawling on their foliage (Bown et al. 2002), while some other plants react to pest oviposition (Hilker et al. 2005; Schroder et al. 2005) or feeding ( Kessler and Baldwin 2001; Arimura et al. 2004; Schoonhoven et al. 2006). Plants show various types of induced responses to organisms that range from viruses, bacteria, fungi, nematodes, mites, insects and mammals as part of their defense mechanisms (Dicke and Hilker 2003). Some plants also show the ability to alarm neighboring plants via their volatile emission (Arimura et al. 2002; Baldwin et al. 2002; Pickett et al. 2003). The ability of plants to sense external stimuli and translate them into signals that are transmitted to distant, non-stimulated organs has been known for a long time (Kaplan et al. 2008). One of the well-documented responses of plants to biotic stressors is the emission of herbivore induced plant volatiles (HIPVs) –also known as info-chemicals due to the fact that they carry some information about the status of the emitter (Arimura et al. 2005). HIPVs can strongly affect the behavior of both predatory and herbivorous arthropods in nature and some plants are under strong selection pressure to release these volatiles (Kessler and Baldwin 2001). HIPVs are known to be emitted by various parts of  7  plants including leaves (Dicke 1999; Turlings et al. 1995; Van Poecke et al. 2001) from both the abaxial and the adaxial side (Bergougnoux et al. 2007), flower buds (Rose and Tumlinson 2004) and roots (Rasmann et al. 2005). HIPV emission is not limited to the site of damage but also occurs systematically throughout the plant even in undamaged parts ( Turlings and Tumlinson 1992; Mattiacci et al. 2001; Neveu et al. 2002; Rose and Tumlinson 2005). Recent findings show that plants can recognize the herbivores and assess their threat via a series of chemical and electrical reactions that occur before activation of defensive mechanisms (Maffei et al. 2007a). Considering the vast knowledge of plant behavior and their responses to the environment, it is conceivable to use these plant-generated responses as indicators of herbivore presence in pest monitoring programs in addition to previously used indicators. The questions are: how reliable are these cues and how quickly can they convey information? When plants emit info-chemicals, they have no control over the receiver of these cues (Halitschke et al. 2008; Heil 2008) and although they can be found in higher concentrations closer to the emitter (Lennert et al. 1997), it is difficult to relate these volatiles that float in the air to their actual source. Environmental factors such as light intensity, temperature and moisture can profoundly affect the emission of plant volatiles (Gouinguene and Turlings 2002). However, in spite of the complexity of this system, predators can “learn” to associate these chemical cues in addition to other cues to locate their prey (Dicke 1999). Some of these volatile chemicals are emitted within minutes after tissue damage and can be considered a quick indicator of problems (Heil and Silva Bueno 2007), while other chemicals are released later as a complement to other types of defense (Kant et al. 2004). A pest monitoring system that is capable of harvesting  8  information from the environment through a series of sensory systems can also learn to associate different signal patterns with pest presence and perform the same task.  !"8 9*.2:;+*-#<2.*'#3)2.'+2%&# The literature on HIPVs is vast and continuously growing. In most cases, the researcher does not know the biological activity of the compounds assessed and therefore samples and analyzes a full range of HIPVs. Usually the volatile organic compounds in the headspace of plants that are enclosed in collection chambers are collected using an adsorbing material. Subsequently the collected volatiles are analyzed by gas chromatography (GC) and mass-spectrometry (MS) or a combination of both (GC/MS) (Turlings et al. 1990; Pickett et al. 1999). D’Alessandro and Turlings (2006) looked into the most commonly used HIPV collection methods from 1995 to 2004. They found that adsorbent/solvent desorption was the most popular method among the scientists who studied insect-plant interactions (Fig 1.3). There are different approaches to volatile analyses depending on the physiological state of the plant, whether the plant tissue is intact or if it has been detached from the plant. Choosing the method of analysis that will give the most accurate results takes some careful consideration. Inconsistencies and inherent methodological weaknesses will be unavoidable to some degree, but the investigator must examine certain parameters and try to minimize potential errors. First, the choice between using detached plant material or intact plants must be made.  9  FIG 1.3 THE MOST COMMONLY USED HIPV COLLECTION METHODS DURING A TEN -YEARS PERIOD FROM 1995-2004. DATA FROM D'ALESSANDRO AND TURLINGS (2006) .  For plant-herbivore interaction studies, using intact plants is generally considered to be most rigorous and accurate. Experiments that use detached plant parts often show more fluctuations in volatile profile than intact plants and are often verified against intact plant results to compensate for oscillations caused by the mechanical damage from detachment (Jakobsen 1997). For studies in chemical ecology, using intact plants is considered the most reliable method but additional considerations must be made to minimize other factors that can affect the accuracy of the results. Such parameters to consider are light intensity, relative humidity (RH), air temperature, and photoperiod ( Jakobsen 1997; Gouinguene and Turlings 2002). The investigator must try to standardize environmental conditions over the duration of the analysis so as to eliminate as much environmental influence as possible. Because some environmental elements can induce changes in volatile emission it can be difficult to ascertain if the volatiles are a result of insect damage or of abiotic factors. Recent improvements in analytical tools enable researchers to collect more accurate data about HIPVs and plant systems within their growing environment in short  10  periods of time (Kunert et al. 2002; Pasini et al. 2004; Lu et al. 2006; Oh et al. 2008; Zhong et al. 2009), which is a key element for developing an intelligent monitoring system.  !"=#>%*&),:#&:&'%7&# A good indicator of pest or disease presence in general should be (a) plant-generated, (b) caused directly or indirectly by the stressor and (c) measurable. In order to create a monitoring system, one needs to identify plausible indicators, select suitable corresponding sensors and finally develop computational techniques that can interpret the output of sensory systems. Environmental factors like temperature, moisture, light intensity and airflow can also be correlated to the fluctuations of the main indicators to create a robust decision making system. A whole array of sensory systems has been developed and improved in the past decade ranging from chemical sensors ( Pearce et al. 2003; Je et al. 2007; Si et al. 2007) to biological hybrids (Chen et al. 1999; Schütz et al. 2000; Mabeck and Malliaras 2006; Waggoner and Craighead 2007), physiological sensors (Eigenberg et al. 2008) and visual sensors ( Rind 2002; Balaji et al. 2005; Tellaeche et al. 2008). Despite great technological advances in recent years, there are still a number of challenges in developing these sensory systems for outdoor use (Mielle and Marquis 1999; Naik et al. 2008). There are a few successful cases (Baratto et al. 2005; Je et al. 2007; Laothawornkitkul et al. 2008) that have shown some promising results, but there is a great need for further improvement of these sensory technologies. Considering the fast pace of technological advances in this new century, it is not too ambitious to suggest that  11  within a few machine generations, we might have sensors that are accurate enough to be incorporated into an intelligent pest monitoring system.  !"?#>'.'+&'+5.2#.*/#5)7<4'.'+)*.2#7%'@)/&# To recognize patterns in the sensory data and predict whether a plant or a particular area is infested, a computational model needs to be developed. Pest-plant-environment systems are very complex; therefore we need a mechanism that learns cause and effect dependencies as well as the underlying noise in the environment (i.e. variability in temperature, RH%, light intensity, other sources of volatile chemicals). A model could be created by incorporating prior knowledge as well as on-site collected training data sets. After training the computational model with large amounts of sensory data, this system should be capable of decision-making based on the model and new observations. In recent years, certain statistical and computational methods have been applied to overcome challenges in modeling biological systems. In particular, machine learning and pattern recognition methods have been extensively used in various application domains such as medical diagnosis, bio-informatics and chemo-informatics (Baladi and Brunak 2001; Bishop 2007). Several statistical and machine learning methods can be adapted for intelligent pest monitoring. These methods range from simple Linear Discriminant Analysis (Peladan et al. 1984; Yu and Yang 2001; Siripatrawan et al. 2004) to Principal Component Analysis (Bravo-Linares and Mudge 2007; de Oliveira et al. 2008; Laothawornkitkul et al. 2008; Moularat et al. 2008), Neural networks (Schultz et al. 2000; Huang 2007; Movagharnejad and Nikzad 2007; Ehret et al. 2008), Support Vector  12  Machines (Karimin et al. 2006), logistic regression (Bielza et al. 2003) and more complicated techniques such as Bayesian Networks (Garrett et al. 2004; Huttenhower and Troyanskaya 2006; Hosack et al. 2008). From the statistical point of view, pest-related sensory data present problems similar to other sensor-based applications, although their domain-specific challenges are different.  !"A#$,)B%5'#-).2&#.*/#)CB%5'+3%&# Pest management programs will benefit from an interdisciplinary boost provided by developing a precise method for early detection of pests and locating them in the field, an important prerequisite for full implementation of IPM recommendations. In this project, I have suggested a different approach towards crop protection and pest monitoring by adding the idea of “monitoring plant-driven responses “ to the “pest and damage detection” mindset. In the past few decades, our knowledge of plant chemical ecology, artificial intelligence and sensory systems has increased considerably. I believe now is the time for us to use this wealth of knowledge and put it into practice. As I mentioned before, my ultimate goal in this project is to investigate the possibility of creating a new tool that can increase the efficiency and performance of human scouts based on an alternative source of information. To begin, I raise key questions in this project to prove the concept. These represent the preliminary steps I took to build a foundation for this idea. I choose to work with tomato greenhouses because greenhouses are more controlled systems compared to open fields and tomato plants are very well established model plants  13  for studying HIPVs. I used the cabbage looper, Trichoplusia ni Hübner, as the model insect pest. In this project, I raised three major questions: 1) Can we discriminate clean plants from infested plants based on their volatile emission pattern? This question is not novel and many studies have shown differences in volatile emissions of infested plants, although most of these studies were conducted in laboratories within closed chambers. I took this to a different level and tested plants inside a research greenhouse to see if I can observe the same responses. 2) What additional information can we get from the volatile emission pattern that can help us make better decisions for pest control? This question is an expansion of the first question. The more information we have about our pest(s) the better we can control them. At this stage, I looked into the volatile emissions of plants that were infested with (a) different densities of pests at (b) different parts of the canopy and experienced (c) different feeding duration. 3) What are the effects of environmental factors and greenhouse related production practices on the emission of the indicator volatiles in commercial greenhouses? This question addressed the important issue of complexity in commercial greenhouses and how it can affect the outcomes of this project. It is very difficult to accurately simulate the conditions of a commercial greenhouse in a research facility. Therefore, I went to a commercial greenhouse for this phase of the research. In addition to indicator chemical levels, I recorded main environmental factors (temperature, relative humidity, light intensity and air flow) 14  as well as greenhouse related practices (fruit picking, removing shoots, opening windows, etc.). Sample plants were scouted and their general health recorded. Research Objective: 1) To identify a few T.ni infestation indicator chemicals within tomato volatile blends and use them for classifying plants into infested and clean groups. 2) To investigate the possibility of obtaining additional information about the pest population and the severity of the damage on tomato plants based on the indicator chemicals’ emission pattern. 3) To investigate the effects of some environmental and maintenance related factors on the emission of indicator chemicals and test the suitability of the indicator chemicals for detecting infestations inside a commercial greenhouse. We might not immediately be able to precisely pinpoint the location of pests within a greenhouse by using indicator volatiles but we will be able to locate areas with higher probability of pest presence. Therefore, we can use our current resources more effectively. Like any novel concept, there are numerous complexities and challenges for this proposed approach, which require the close attention of the scientific community and the active collaboration of different disciplines.  15  !"D#E%6%,%*5%&# Arimura G, Kost C, Boland W. 2005. Herbivore-induced, indirect plant defences. Biochim Biophys Acta 1734:91-111. Arimura G, Ozawa R, Kugimiya S, Takabayashi J, Bohlmann J. 2004. Herbivore-induced defense response in a model legume. Two-spotted spider mites induce emission of (E)-beta-ocimene and transcript accumulation of (E)-beta-ocimene synthase in Lotus japonicus. Plant Physiol 135:1976-1983. Arimura G, Ozawa R, Nishioka T, Boland W, Koch T, Kuhnemann F, Takabayashi J. 2002. Herbivore-induced volatiles induce the emission of ethylene in neighboring lima bean plants. Plant J 29:87-98. Baladi P, Brunak S. 2001. Bioinformatics: The Machine Learning Approach. Cambridge: MIT Press. 476 p. Balaji T, Sasidharan M, Matsunaga H. 2005. Optical sensor for the visual detection of mercury using mesoporous silica anchoring porphyrin moiety. Analyst 130:11621167. Baldwin IT, Kessler A, Halitschke R. 2002. Volatile signaling in plant-plant-herbivore interactions: what is real? Curr Opin Plant Biol 5:351-354. Bange MP, Deutscher SA, Larsen D, Linsley D, Whiteside S. 2004. A handheld decision support system to facilitate improved insect pest management in Australian cotton systems. Comp Electron Agric 43:131-147. Baratto C, Faglia G, Pardo M, Vezzoli M, Boarino L, Maffei M, Bossi S, Sberveglieri G. 2005. Monitoring plants health in greenhouse for space missions. Sensors and Actuators 108:278-284.  16  Bergougnoux V, Caissard JC, Jullien F, Magnard JL, Scalliet G, Cock JM, Hugueney P, Baudino S. 2007. Both the adaxial and abaxial epidermal layers of the rose petal emit volatile scent compounds. Planta 226:853-866. Bielza C, Barreiro P, Rodriguez-Galiano MI, Martin J. 2003. Logistic regression for simulating damage occurrence on a fruit grading line. Comp Electron Agric 39:95-113. Bishop CM. 2007. Pattern Recognition and Machine Learning. Jordan M, Kleinberg J, Scholkopf B, editors. New York: Springer. 738 p. Boissard P, Martin V, Moisan S. 2008. A cognitive vision approach to early pest detection in greenhouse crops. Comp Electron Agric 62:81-93. Boolell M, Allen MJ, Ballard SA, Gepi-Attee S, Muirhead GJ, Naylor AM, Osterloh IH, Gingell C. 1996. Sildenafil: an orally active type 5 cyclic GMP-specific phosphodiesterase inhibitor for the treatment of penile erectile dysfunction. Int J Impot Res 8:47-52. Bown AW, Hall DE, MacGregor KB. 2002. Insect footsteps on leaves stimulate the accumulation of 4-aminobutyrate and can be visualized through increased chlorophyll fluorescence and superoxide production. Plant Physiol 129:14301434. Bravo-Linares CM, Mudge SM. 2007. Analysis of volatile organic compounds (VOCs) in sediments using in situ SPME sampling. J Environ Monit 9:411-418. Chen L, McBranch DW, Wang HL, Helgeson R, Wudl F, Whitten DG. 1999. Highly sensitive biological and chemical sensors based on reversible fluorescence quenching in a conjugated polymer. Proc Natl Acad Sci U S A 96:12287-12292.  17  Cook RJ. 2000. Advances in Plant Health Management in the Twentieth Century. Annu Rev Phytopathol 38:95-116. D'Alessandro M, Turlings TC. 2006. Advances and challenges in the identification of volatiles that mediate interactions among plants and arthropods. Analyst 131:2432. de Oliveira LS, Rodrigues FdeM, de Oliveira FS, Mesquita PR, Leal DC, Alcantara AC, Souza BM, Franke CR, Pereira PA, de Andrade JB. 2008. Headspace solid phase microextraction/gas chromatography-mass spectrometry combined to chemometric analysis for volatile organic compounds determination in canine hair: a new tool to detect dog contamination by visceral leishmaniasis. J Chromatogr B Analyt Technol Biomed Life Sci 875:392-398. Dicke M. 1999. Specificity of herbivore-induced plant defences. Novartis Found Symp 223:43-54; discussion 54-9, 160-165. Dicke M, Hilker M. 2003. Induced plant defences: from molecular biology to evolutionary ecology. Basic Appl Ecol 4:3-14. Drake VA, Wang HK, Harman IT. 2002. Insect monitoring radar:remote and network operation. Comp Electron Agric 35:77-94. Dunlop D, Pittar G, Dunlop C. 1988. Botulinum toxin in ophthalmology. Aust N Z J Ophthalmol 16:15-20. Ehret DL, Hill BD, Raworth DA, Estergaard B. 2008. Artificial neural network modeling to predict cuticle cracking in greenhouse peppers and tomatoes. Comp Electron Agric 61:108-116.  18  Eigenberg RA, Brown-Brandl TM, Nienaber JA. 2008. Sensors for dynamic physiological measurements. Comp Electron Agric 62:41-47. Garrett KA, Madden LV, Hughes G, Pfender WF. 2004. New applications of statistical tools in plant pathology. Phytopathol 94:999-1003. Gouinguene SP, Turlings TC. 2002. The effects of abiotic factors on induced volatile emissions in corn plants. Plant Physiol 129:1296-1307. Halitschke R, Stenberg JA, Kessler D, Kessler A, Baldwin IT. 2008. Shared signals 'alarm calls' from plants increase apparency to herbivores and their enemies in nature. Ecol Lett 11:24-34. Heil M. 2008. Indirect defence via tritrophic interactions. New Phytol 178:41-61. Heil M, Lion U, Boland W. 2008. Defense-inducing volatiles: in search of the active motif. J Chem Ecol 34:601-604. Heil M, Silva Bueno JC. 2007. Within-plant signaling by volatiles leads to induction and priming of an indirect plant defense in nature. Proc Natl Acad Sci U S A 104:5467-5472. Hilker M, Stein C, Schroder R, Varama M, Mumm R. 2005. Insect egg deposition induces defence responses in Pinus sylvestris: characterisation of the elicitor. J Exp Biol 208:1849-1854. Hosack GR, Hayes KR, Dambacher JM. 2008. Assessing model structure uncertainty through an analysis of system feedback and Bayesian networks. Ecol Appl 18:1070-1082. Huang K-Y. 2007. Application of artificial neural network for detecting Phalaenopsis seedling diseases using color and texture features. Comp Electron Agric 57:3-11.  19  Hughes G. 1999. Sampling for decision making in crop loss assessment and pest management: introduction. Phytopathol 89:1080-1083. Huttenhower C, Troyanskaya OG. 2006. Bayesian data integration: a functional perspective. Comput Syst Bioinformatics Conf: 341-351. Jakobsen HB. 1997. The preisolation phase of in situ headspace analysis: methods and prespectives. In: H.F. L, Jackson JF, editors. Plant Volatile Analysis. Berlin: Springer-Verlag. p 1-22. Janssen K, Vermeulen K, Boonen C, Bleyaert P, Lemeur R, Berkcmans D. 2004. Introduction to speaking plant: let the crop control the greenhouse climate. Commun Agric Appl Biol Sci 69:151-153. Je CH, Stone R, Oberg SG. 2007. Development and application of a multi-channel monitoring system for near real-time VOC measurement in a hazardous waste management facility. Sci Total Environ 382:364-374. Kant MR, Ament K, Sabelis MW, Haring MA, Schuurink RC. 2004. Differential timing of spider mite-induced direct and indirect defenses in tomato plants. Plant Physiol 135:483-95. Kaplan I, Halitschke R, Kessler A, Sardanelli S, Denno RF. 2008. Effects of plant vascular architecture on aboveground-belowground-induced responses to foliar and root herbivores on Nicotiana tabacum. J Chem Ecol 34:1349-1359. Karimin Y, Prasher SO, Patel RM, Kim SH. 2006. Application of support vector machine technology for weed and nitrogen stress detection in corn. Comp Electron Agric 51:99-109.  20  Kessler A, Baldwin IT. 2001. Defensive function of herbivore-induced plant volatile emissions in nature. Science 291:2141-2144. Kleinrock L. 2008 February. History of the Internet and its flexible future. IEEE Wireless Communications: 8-18. Kunert M, Biedermann A, Koch T, Boland W. 2002. Ultra fast sampling and analysis of plant volatiles by a hand-held miniaturized GC with pre-concentration unit: Kinetic and quantitative aspects of plant volatile production. J Separation Sci 25:677-684. Laothawornkitkul J, Moore JP, Taylor JE, Possell M, Gibson TD, Hewitt CN, Paul ND. 2008. Discrimination of plant volatile signatures by an electronic nose: a potential technology for plant pest and disease monitoring. Environ Sci Tech 42:84338439. Lennert A, Nielsen F, Breum NO. 1997. Evaluation of evaporation and concentration distribution models--a test chamber study. Ann Occup Hyg 41:625-641. Lichtenberg E, Berlind AV. 2005. Does it matter who scouts? J Agric Resource Econ 30:250-267. Lu CJ, Jin C, Zellers ET. 2006. Chamber evaluation of a portable GC with tunable retention and microsensor-array detection for indoor air quality monitoring. J Environ Monit 8:270-278. Mabeck JT, Malliaras GG. 2006. Chemical and biological sensors based on organic thinfilm transistors. Anal Bioanal Chem 384:343-353. Maffei ME, Mithofer A, Boland W. 2007. Before gene expression: early events in plantinsect interaction. Trends Plant Sci 12:310-316.  21  Mattiacci L, Rocca BA, Scascighini N, D'Alessandro M, Hern A, Dorn S. 2001. Systemically induced plant volatiles emitted at the time of "danger". J Chem Ecol 27:2233-2252. Mielle P, Marquis F. 1999. An alternative way to improve the sensitivity of electronic olfactometers. Sensors and Actuators 58:526-535. Moularat S, Robine E, Ramalho O, Oturan MA. 2008. Detection of fungal development in a closed environment through the identification of specific VOC: demonstration of a specific VOC fingerprint for fungal development. Sci Total Environ 407:139-146. Movagharnejad K, Nikzad M. 2007. Modeling of tomato drying using artificial neural network. Comp Electron Agric 59:78-85. Naik G. 2006 November 14. Hospital races to learn lessons of Ferrari crew. The Wall Street Journal. Naik GR, Kumar DK, Palaniswami M. 2008. Identification of independent biological sensors-electromyogram example. Conf Proc IEEE Eng Med Biol Soc:1112-1115. Neveu N, Grandgirard J, Nenon JP, Cortesero AM. 2002. Systemic release of herbivoreinduced plant volatiles by turnips infested by concealed root-feeding larvae Delia radicum L. J Chem Ecol 28:1717-1732. O'Brien PJ, Griffin JM. 2007. Global Positioning system- system engineering case study. In: Engineering AFCfS, editor: AFIT/SY :http://www.afit.edu/cse/cases.cfm?case=17&a=detail. Oerke EC, Dehne HW. 2004. Safeguarding production-losses in major crops and the role of crop protection. Crop Protec 23:275-285.  22  Oerke EC, Steiner U, Dehne HW, Lindenthal M. 2006. Thermal imaging of cucumber leaves affected by downy mildew and environmental conditions. J Exp Bot 57:2121-2132. Oh SY, Ko JW, Jeong SY, Hong J. 2008. Application and exploration of fast gas chromatography-surface acoustic wave sensor to the analysis of thymus species. J Chromatogr A 1205:117-127. Pasini P, Powar N, Gutierrez-Osuna R, Daunert S, Roda A. 2004. Use of a gas-sensor array for detecting volatile organic compounds (VOC) in chemically induced cells. Anal Bioanal Chem 378:76-83. Pearce TC, Schiffman SS, H.T. N, Gardner JW. 2003. Handbook of Machine olfactionelectronic nose technology. Weinheim: WILEY-VCH. 624 p. Peladan F, Turlot JC, Monteil H. 1984. Discriminant analysis of volatile fatty acids produced in culture medium: a novel approach to the identification of Pseudomonas species. J Gen Microbiol 130:3175-3182. Pickett JA, Chamberlain K, Poppy GM, Woodcock CM. 1999. Exploiting insect responses in identifying plant signals. Novartis Found Symp 223:253-262; discussion 262-5, 266-9. Pickett JA, Rasmussen HB, Woodcock CM, Matthes M, Napier JA. 2003. Plant stress signalling: understanding and exploiting plant-plant interactions. Biochem Soc Trans 31:123-127. Pucheta JA, Schugurensky C, Fullana R, Patino H, Kuchen B. 2006. Optimal greenhouse control of tomato-seedling crops. Comp Electron Agric 50:70-82.  23  Rasmann S, Kollner TG, Degenhardt J, Hiltpold I, Toepfer S, Kuhlmann U, Gershenzon J, Turlings TC. 2005. Recruitment of entomopathogenic nematodes by insectdamaged maize roots. Nature 434:732-737. Rind FC. 2002. Motion detectors in the locust visual system: From biology to robot sensors. Microsc Res Tech 56:256-269. Rose US, Tumlinson JH. 2004. Volatiles released from cotton plants in response to Helicoverpa zea feeding damage on cotton flower buds. Planta 218:824-832. Rose US, Tumlinson JH. 2005. Systemic induction of volatile release in cotton: how specific is the signal to herbivory? Planta 222:327-235. Schoonhoven LM, van Loon JJA, Dicke M. 2006. Insect-Plant Biology. New York: Oxford University Press. 421 p. Schroder R, Forstreuter M, Hilker M. 2005. A plant notices insect egg deposition and changes its rate of photosynthesis. Plant Physiol 138:470-477. Schultz A, Wieland R, Lutze G. 2000. Neural networks in agroecological modellingstylish application of helpful tool? Comp Electron Agric 29:73-97. Schütz S, Schöning MJ, Schroth P, Malkoc Ü, Weißbecker B, Kordos P, Lüth H, Hummel HE. 2000. An insect-based BioFET as a bioelectronic nose. Sensors and Actuators 65:291-295. Si P, Mortensen J, Komolov A, Denborg J, Moller PJ. 2007. Polymer coated quartz crystal microbalance sensors for detection of volatile organic compounds in gas mixtures. Anal Chim Acta 597:223-230. Siripatrawan U, Linz JE, Harte BR. 2004. Solid-phase microextraction, gas chromatography, and mass spectrometry coupled with discriminant factor analysis  24  and multilayer perceptron neural network for detection of Escherichia coli. J Food Prot 67:1597-603. Skaloudova B, Krivan V, Zemek R. 2006. Computer-assisted estimation of leaf damage caused by spider mites. Comp Electron Agric 53:81-91. Tellaeche A, BurgosArtizzu XP, Pajares G, Ribeiro A, Fernandez-Quintanilla C. 2008. A new vision-based approach to differential spraying in precision agriculture. Comp Electron Agric 60:144-155. Thomson AJ. 2005. Indicator-based knowledge management for participatory decisionmaking. Comp Electron Agric 49:206-218. Trewavas A. 2003. Aspects of plant intelligence. Ann Bot (Lond) 92:1-20. Turlings TC, Loughrin JH, McCall PJ, Rose US, Lewis WJ, Tumlinson JH. 1995. How caterpillar-damaged plants protect themselves by attracting parasitic wasps. Proc Natl Acad Sci U S A 92:4169-4174. Turlings TC, Tumlinson JH. 1992. Systemic release of chemical signals by herbivoreinjured corn. Proc Natl Acad Sci U S A 89:8399-8402. Turlings TC, Tumlinson JH, Lewis WJ. 1990. Exploitation of herbivore-induced plant odors by host-seeking parasitic wasps. Science 250:1251-1253. Ushada M, Murase H, Fukuda H. 2007. Non-destructive sensing and its inverse model for canopy parameters using texture analysis and artificial neural network. Comp Electron Agric 57:149-165. Van Den Bosch R. 1969. Fewer pesticides--more control. Science 164:497.  25  Van Poecke RM, Posthumus MA, Dicke M. 2001. Herbivore-induced volatile production by Arabidopsis thaliana leads to attraction of the parasitoid Cotesia rubecula: chemical, behavioral, and gene-expression analysis. J Chem Ecol 27:1911-1928. Volkov AG, Ranatunga DRA. 2006. Plants as environmental biosensors. Plant Signal Behav 1:105-115. Waggoner PS, Craighead HG. 2007. Micro- and nanomechanical sensors for environmental, chemical, and biological detection. Lab Chip 7:1238-1255. Wang N, Zhang N, Wang M. 2006. Wireless sensors in agriculture and food industryrecent development and future perspective. Comp Electron Agric 50:1-14. Yu H, Yang J. 2001. A direct LDA algorithm for high-dimensional data- with application to face recognision. Pattern Recognition 34:2067-2070. Zhong Q, Steinecker WH, Zellers ET. 2009. Characterization of a high-performance portable GC with a chemiresistor array detector. Analyst 134:283-293.  26  CHAPTER TWO: QUALITATIVE ASSESSMENT OF AN ULTRAFAST PORTABLE GAS CHROMATOGRAPH (ZNOSE!) FOR ANALYZING VOLATILE ORGANIC CHEMICALS IN THE LABORATORY AND GREENHOUSES3 1"!#0*',)/45'+)*# Analysis of herbivore-induced plant volatiles (HIPVs) is a critical component of many research projects that study insect-plant interactions. The scientific literature on HIPVs is vast and continuously growing. In most cases, the researcher does not know the biological activity of the compounds assessed and therefore samples and analyzes a full range of HIPVs. Usually the volatile organic compounds in the headspace of plants that are enclosed in collection chambers are collected through use of an adsorbent material. Subsequently the collected volatiles are eluted and analyzed by gas chromatography (GC) and mass-spectrometry (MS) or a combination of both (GC/MS) (Turlings et al. 1990; Pickett et al. 1999). D'Alessandro and Turlings (2006) examined the most commonly used HIPV collection methods from 1995 to 2004. They found that adsorbent/solvent desorption was the most popular method among scientists who studied HIPVs in insectplant interactions. There are different approaches to volatile analyses depending on the physiological state of the plant, and whether the plant tissue is intact or has been detached from the plant. Choosing the method of analysis that will give the most accurate results takes some careful consideration. Inconsistencies and inherent methodological weaknesses will be unavoidable to some degree, but the investigator can examine certain  3  A version of this chapter has been submitted for publication. Miresmailli S., Bradbury R. and Isman M.B. 2009. Qualitative assessment of an ultra-fast portable gas chromatograph (zNose!) for analyzing volatile organic chemicals and essential oils in laboratory and greenhouses.  27  parameters so as to minimize potential errors. The choice between using detached plant material or intact plants must be made. For plant-herbivore interaction studies, using intact plants is generally considered to be most rigorous and accurate. Experiments that use detached plant parts often show more fluctuations in volatile profile than intact plants and are often verified against intact plant results to compensate for oscillations caused by the mechanical damage from detachment (Jakobsen 1997). For studies in chemical ecology, using intact plants is considered the most reliable method but additional considerations must be made to minimize other factors that can affect the accuracy of the results. Such parameters to consider are light intensity, relative humidity (RH), air temperature, and photoperiod (Jakobsen 1997; Gouinguene and Turlings 2002). The investigator must try to standardize environmental conditions over the duration of the collection period so as to eliminate as much environmental influence as possible. Because elements such as RH and temperature can induce changes in volatile emission it can be difficult to ascertain if the volatiles are a result of insect damage or of abiotic factors. Recent improvements in analytical tools enable researchers to collect more accurate data about HIPVs and plant systems within their growing environment in short periods of time (Kunert et al. 2002; Pasini et al. 2004; Lu et al. 2006; Oh et al. 2008; Zhong et al. 2009). In a very comprehensive review paper, Tholl et al. (2006) explored different practical methods of plant volatile collection and analyses. In addition to several conventional collection, separation and detection methods reviewed, they also discussed a relatively new instrument, the zNose !(Electronic Sensor Technology, Newbury Park, CA, USA),  28  which is a miniaturized ultra-fast portable gas chromatograph. They found this instrument a useful tool for fast quantitative estimation of known volatile profiles and monitoring rapid changes in volatile organic chemical (VOC) emissions. As a portable device, they suggested that it could also be used in field experiments. In this chapter, I provide a more detailed description of the zNose! and discuss its benefits and limitations for analyzing VOCs including HIPVs. I also used a conventional GC-MS for analyses of a commercial botanical insect repellent that consists mainly of plant essential oils and compared results with those obtain by the zNose!. Further I assessed the performance of the zNose! in a commercial greenhouse with variable temperature and humidity.  2.1.1 How does the zNose! work? This system is a miniature, high-speed gas chromatograph (GC). It is based on a 6-port valve and oven, a pre-concentrating trap, a short GC capillary column (DB-5, 1 m length, film thickness 0.25 µm, internal diameter 0.25 mm) (Fig 2.1) and a highly sensitive surface acoustic wave (SAW) quartz microbalance detector, in which VOC analytes are condensed on the surface of an oscillating crystal. A system controller, a laptop computer, operates the system, analyzes the data and provides the user interface (Fig 2.2).  29  Trap Column  Rotating Valve Inlet  FIG 2.1 THE ZNOSE! VALVE, TRAP AND COLUMN  Three sequences control the instrument operation: sampling, injecting and analysis. During the sampling sequence, the system draws a headspace sample into the inlet via a pump. The sample passes through the valve and onto the trap where the compounds are adsorbed. The valve is then rotated to put the trap in line with the column for the injection sequence. Once inline, the trap is heated quickly by a short burst of current that vaporizes the adsorbed material. The helium carrier gas then transports the material down to the capillary column. The column is heated under computer control facilitating separation of compounds. When materials sequentially exit the column, they land and stick on the SAW detector. The added mass of the material then lowers the oscillating frequency of the SAW crystal. This frequency is mixed with a reference frequency and the resulting IF (intermediate frequency) is counted by the system microprocessor board. The system controller interprets the detector responses and attempts to identify and quantify each material it has been programmed to recognize. The time of arrival at the detector identifies the compound, while the frequency shift caused by each analyte is characteristic of the amount of material deposited on the detector thus allowing quantitation (EST 2007).  30  Because of the short operation time and fully automated sampling and data acquisition, air samples can be collected in time intervals as low as 3 min over long periods without supervision.  FIG 2.2 THE ZNOSE! MICROSENSE! SOFTWARE- METHOD DEVELOPMENT.  The operating software enables the user to control different elements of the analytical method. Sensor temperature can range between 20°C to 100°C. Lowering the sensor temperature can increase its sensitivity. After each run, the sensor will bake itself up to 30 sec at 150°C. The user can also control the amount of sample by changing the pumping time. The operating software, MicroSense !(Electronic Sensor Technology, Newbury Park, CA, USA) provides users with the complete chromatogram of the compounds or display selected chemical levels in a bar graph. The system produces both the retention time of  31  each compound eluted from the column and also calculates its’ retention index based on a series of alkanes from C6 to C14 that is used for tuning the device. The manufacturer recommends that the device be tuned with the alkane mixture before each series of experiments. In 1958, Prof. Ervin Kovats introduced the Kovats retention index (Kovats 1958). The retention index is a method of quantifying the relative elution times of compounds in gas chromatography to help positively identify the compounds in a mixture. The purpose of the index system is to quantify the retention of a test compound by comparing it with a pair of n-alkane adjacent peaks. The method takes advantage of the linear relationship between the values of the logarithm of the adjusted retention time and the number of carbon atoms in a molecule. The value of the Kovats index is usually represented by I in mathematical expressions. Its applicability is restricted to organic compounds. The Kovats index is given by the equation: Equation 1. Kovats retention index  % $ $ ) )( log( t r(unknown ) ) # log( t r(n I = 100 " 'n + ( N # n ) * $ ) ) # log( t r(n $ )) ) log( t r(N & I = Kovats retention index,  ! !  n= the number of carbon atoms in the smaller alkane N= the number of carbon atoms in the larger alkane  t r" = the adjusted retention time  Because Kovats indices are related to a n-alkane series, they are relatively unique for !  most compounds, unlike retention times that change with linear flow velocity or temperature program. Kovats indices are mostly used when samples are analyzed isothermally. A few years later Van Den Dool and Kratz (1963) introduced a generalization of the Kovats indices for linear temperature-programmed analyses:  32  Equation 2. Van Den Dool and Kratz retention index  $t" # t" ' I = 100n + 100& r(unknown ) r(n ) ) " ) ) # t r(n " )) ( % t r(N I = Van Den Dool and Kratz retention index,  ! !  n= the number of carbon atoms in the smaller alkane N= the number of carbon atoms in the larger alkane  t r" = the adjusted retention time  To confirm identification of selected compounds, one should determine retention indices !  using at least two different conventional GC columns and match the mass spectrum of the target compound with that of an authentic standard. For the retention index system, homologous series of n-alkanes were chosen as standards because of their low polarity and freedom of H-bonding. Besides, in all stationary phases, their order of log vapor pressure ( t r" ) is directly proportional to the number of carbons. According to the manufacturer, the system can analyze compounds within the range of ! C4 to C24 however; the system has a short column which reduces the resolution of VOCs  with similar retention times. Despite this limitation, the zNose! has been successfully used in different studies. It has been employed for monitoring rhythmic VOC emissions from flowers and induced VOC emissions from herbivore-damaged plants (Kunert et al. 2002), honey classification (Lammertyn et al. 2004), grape aroma classification (Watkins and Wijesundera 2006), detection of lard adulteration in palm olein (Che Man et al. 2005), characterization of vegetable oils (Gan et al. 2005) and monitoring alarm pheromone emission by the pea aphid under predation (Schwartzberg et al. 2008). Part-per-billion sensitivity has been reported with this instrument for volatile compounds (Staples and Viswanathan 2008).  33  2.1.2 Calibration After tuning the system with the n-alkane series, the zNose ! requires careful calibration with relevant authentic standards. The 4300 analyzer uses from one to ten data points to calculate a calibration curve for up to 35 compounds in a given peak file. The operator should first prepare the standards with a known amount of authentic compounds (some standards are commercially available). Then with the zNose ! 3500 sample injector (or any automated injection tool that provides consistent delivery), the operator can inject different volumes of standards and correlate the sensor responses with the actual amount of compound in each injected sample to create the calibration curve. The operator needs to carefully measure and calculate the exact amount of the analyte in each volume that is injected into the machine. Alternatively, the operator can prepare a dilution series of the neat analyte (or standard stock), inject a fixed volume and measure the responses. The system can automatically calculate a scale factor based on sensor responses. The system can also be calibrated by measuring the response of the zNose! to headspace vapors of known concentration of chemicals in a closed container, but this method can be affected by environmental factors (i.e. ambient temperature) and the results might vary. Based on personal experience, I recommend direct injection of known chemicals in nvolume series with an automated delivery system for accurate calibration. With this method, we can decrease human errors that might occur during preparation of the dilution series. By using the automated sample injector, we can produce consistent results in a short period of time. Although the system has the capability of calibrating several compounds simultaneously, resolution of compounds with similar retention times can be problematic because of the  34  short column (1 m). Based on personal experience, loading the system with more than 10 compounds at a time is not advisable. The manufacturer claims that the machine can maintain the calibration for up to 30 days but for higher precision (especially in field experiments), I recommend calibration prior to each set of experiments.  2.1.3 Regular care and maintenance Based on the frequency of use, the device should be cleaned often of contaminants that might appear in the chromatogram as noise peaks. The system also contains a moisture filter which needs to be dried (baked at 200 °C) after operating in humid conditions such as greenhouses. If needed, the SAW detector can be cleaned by acetone as recommended by the manufacturer.  35  1"1#(.'%,+.2&#.*/#(%'@)/&# 2.2.1 Qualitative comparison of samples by the zNose! and a conventional GCMS4 I analyzed a commercial botanical insect repellent (EcoSMART! Insect Repellent, EcoSMART Technologies, Alpharetta, GA, USA) that consists mainly of plant essential oils, on a conventional GC-MS and compared it with the zNose! by calculating the Kovats indices on both machines using a C6-C14 alkane standard. The active ingredient oils in this insect repellent are rosemary oil, cinnamon leaf oil, lemongrass oil and geraniol. Rosemary oil is assayed by determining the level of the marker compound 1,8cineole; cinnamon leaf oil was assayed by determining the level of the marker compound, eugenol; lemongrass oil was assayed by determining the level of the marker compound, citral and geraniol, as a pure compound, was assayed directly. Samples were diluted in acetonitrile and analyzed by GC/MS and quantified against known standards. Cinnamic alcohol as an internal standard was used to improve method precision.  2.2.2 Reagents Cinnamic alcohol (internal standard) - Aldrich (98+% pure); 1,8-cineole - Aldrich (99+% pure); Eugenol – Pestanal grade analytical standard Supelco (99.6% pure); Citral – Aldrich (95.0% pure), Geraniol – Aldrich (98.0% pure); C6-C14 alkane standard – Electronic Sensor Technology (98% pure); Acetonitrile HPLC grade – VWR / Canlab.  4  The GC-MS analyses were conducted by Mr. Rod Bradbury at Ecosafe Natural Products Inc. Saanichton, BC  36  2.2.3 Preparation of internal and external standard solution An internal standard concentrate solution containing circa 100 µg/ml of cinnamic alcohol was prepared by weighing out 5 mg of analytical grade cinnamic alcohol and dissolving it in 500 ml of HPLC grade acetonitrile. The internal standard concentrate solution was used to prepare all external standards and in dilution of all unknowns. A working stock solution containing all reagents except the alkane standard was prepared from analytical grade material by dissolving the following amounts (table 2.1) in 10 ml of HPLC grade acetonitrile: TABLE 2.1. STANDARD CONCENTRATE SOLUTION  Compound  Amount (g)  1,8-cineole (99%)  0.1004  Eugenol (99%)  0.1011  Citral (95%)  0.1039  Geraniol (98%)  0.2071  The alkane standard was used without dilution. A working stock solution was prepared by diluting 0.2 ml of the standard concentrate solution in 10 ml of HPLC grade acetonitrile. A standard curve was then produced by transferring aliquots of the working stock solution together with the internal standard and acetonitrile to septum capped GC vials for analyses.  37  2.2.4 Sample preparation An aliquot of sample (0.2 ml), weighed to the nearest 0.1 mg of formulation was pipetted into a 10-ml volumetric flask and diluted to the mark with HPLC grade acetonitrile. The flask was then shaken and sonicated for 5 minutes to facilitate complete dissolution. A subsample (0.5 ml) was transferred to a septum-capped GC vial for analysis. To this sample was added 0.5ml of the internal standard solution.  2.2.5 GC/MS parameters A Varian 3900 GC system was used employing a Saturn 2100T ion trap, mass selective detector. Data were handled by a PC using Varian GC/MS Workstation software. The GC was used in temperature gradient mode. The GC temperature profile is listed in table 2.2. The column used was a FactorFour Capillary column VF-5ms, 30 m x 0.25 mm ID DF=0.25 with a low bleed/MS coating (equivalent to a standard J&W DB5 low bleed MS column). Injections were performed by a Varian CP-8410 autosampler with an injection volume of 1 µl. A split ratio of 100:1 was used in the injection port. The carrier gas was 99.999% UHP helium with a column flow rate of 1.0 ml / min. Total run time was 12 min per sample. The MS detector was used in EI (electron ionization) mode with a filament delay time of 2 min. Scan time was 1.0 sec per scan with a mass scan range of 40 – 300 m/z. Maximum ionization time in the ion trap was 25 milliseconds. Ion trap temperature was 220°C, manifold temperature was 80°C and transfer line temperature was 300°C.  38  TABLE 2.2 GC TEMPERATURE PROFILE  Temperature °C  Rate (°C/min)  Hold (min)  Total (min)  80  0.0  0.50  0.50  140  8.0  0.00  8.00  300  50.0  0.80  12.00  2.2.6 Calculation of results The chemicals were quantified against the standard curve on a PC using Varian GC/MS Workstation software. The retention indices were calculated for each compound using equation 2.  2.2.7 Preparation of standard solution for the zNose! and system calibration A working stock solution was prepared by diluting 1 gram (weighed to the nearest 0.1 mg) of each authentic standard into 50 ml of HPLC grade n- pentane solvent. An aliquot of each (100 µl) working stock was pipetted into a 10-ml volumetric flask and diluted to the mark with HPLC grade n-pentane. The flask was then shaken and sonicated for 5 min to facilitate complete dissolution. A subsample (0.5 ml) was transferred to a septumcapped GC vial for calibration. After tuning the machine with alkane standards four different volumes of reagent standards were injected into the zNose! using the zNose ! 3500 sample injector (0.1 µl, 0.2 µl, 0.4 µl and 0.8 µl- five times per volume). Calibration curves were developed for each chemical.  39  2.2.8 Sample preparation for the zNose! Four µl of commercial insect repellent were transferred into a 40-ml vial (98 mm L, 28 mm OD) sealed with a screw cap containing a silicon septum. The samples were allowed to equilibrate with the headspace in the vial for at least of 20 min at room temperature (24°C). Fifteen sample vials were prepared. The zNose ! was provided with a 5-cm needle at the inlet which was used for sampling through the septa of the vials. The sampling was set for 10 sec (Sample flow 20 ccm) after which the system switched to 20 sec of data acquisition mode.  2.2.9 zNose! GC parameters The inlet temperature was 200°C, the valve temperature was 165°C, and the initial column temperature was 40°C. During analysis the column temperature was increased at the rate of 10°C per sec to reach a final column temperature of 200°C. The SAW sensor was operated at 60°C and the trap was operated at 250°C. Helium flow was set at 3.00 ccm. After each data sampling period the system needed a 30-sec baking period, in which the sensor was heated briefly (~5-8 sec) to 150°C and after which the temperature conditions of the inlet, column, and sensor were reset to the initial conditions. In between each sample measurement at least one blank was run to ensure cleaning of the system and a stable baseline.  40  2.2.10 Comparing retention indices between machines Retention index values of 1,8-cineole, eugenol, citral and geraniol on both machines were analyzed by t-test (SPSS ver 16).  2.2.11 Retention index variation of limonene in laboratory and commercial greenhouses In this experiment, I examined the consistency of the zNose! in the field. It can be difficult to calibrate the machine at the site of analysis. In some situations, the investigator must calibrate the device the day prior to the analysis due to time and logistic limitations. Therefore, it is important to know whether the machine can provide consistent results in the field as it does in the laboratory. The simple act of transportation may cause the machine to go out of tune. Therefore, I designed an experiment to compare the retention index of limonene as a representative of the compounds that can be found in volatile blends of tomato plants (Baldwin et al. 2008; Mayer et al. 2008; Mathieu et al. 2009) both in the lab and in a commercial tomato greenhouse (Fig 2.3). As explained before, the retention indices are related to a n-alkane series and they are relatively unique for most compounds, unlike retention times that change with chromatographic methods. Therefore, they are a better option for comparing different GC machines. In addition to the retention index, I recorded changes in temperature and relative humidity as factors that might affect the performance of the device. The environmental factors were recorded using a Fisher Scientific Enviro-Meter!.  41  FIG 2.3 COMPARING ZNOSE! RESULTS IN LABORATORY AND FIELD.  The goal was to compare the retention indices both in the lab and greenhouse, therefore quantification was not part of this experiment. The experiment had four levels: analyzing (1) neat chemical by GC-MS, (2) neat chemical by the zNose !in laboratory, (3) tomato volatiles by the zNose! in laboratory and (4) tomato volatiles by the zNose! in a commercial greenhouse. One set of alkane standards (C6-C14 alkane standard – Electronic Sensor Technology) was used for all analyses. The zNose ! was calibrated with limonene as explained before. Fifteen samples were prepared and analyzed in the laboratory both by GC-MS and the zNose! and the retention indices were calculated. Ambient air temperature and relative humidity were recorded in laboratory. The same methods and setting were used for the GC as explained before. Fifteen tomato plants (6 weeks old) were obtained from Houwelings’ Hot House (Delta, BC) for this analysis. Plants were kept inside a growth chamber at 23°C with a16/8 light-dark period for 2 days prior to the experiment. Plants were then transferred to the laboratory (Fig 2.4). The same analysis method was used in the zNose!. The only difference was the sampling time, which was 30 sec (sample flow rate 20 ccm).  42  FIG 2.4 ANALYZING TOMATO VOLATILES IN LABORATORY.  After sampling plant volatiles in the laboratory, the zNose! was then turned off and stored in its carrying case. The greenhouse samples were collected the following day. Fifteen locations were randomly selected within the commercial tomato greenhouse (Houweling’s Hot Houses, Delta, BC) and plant volatiles were collected and analyzed by the zNose! using the same method and settings. Air temperature and relative humidity were recorded at each spot. The same analytical method was used for both the greenhouse and the laboratory samples. HPLC grade n-pentane was used as a blank between each sample both in the laboratory and in the greenhouse.  2.2.12 Data analyses The effect of temperature, relative humidity and method (GC-MS or zNose! in laboratory or greenhouse) on retention index of limonene were analyzed with linear multiple regression using SPSS software version 16.  43  1"8#E%&42'&# 2.3.1 Comparing retention indices of selected VOCs analyzed by the zNose! and conventional GC-MS. The calibration curves of all chemicals analyzed with GC-MS and the zNose! were linear. 1,8 Cineol in GC-MS (y=1.3859X+ 0.061803, R!= 0.9973) and in zNose! (y=190.33X+1504, R!= 0. 9721); eugenol in GC-MS (y=2.30343X+0.0469, R!= 0.9995 and in zNose! (y=295.07X+1421, R!= 0.9704); geraniol in GC-MS ( y=1.74479X + 0.019488, R!= 0.9993) and in zNose! (y=260.23X+88.435, R!= 0.9952); and citral in GC-MS (y=2.03368X+0.034854, R!= 0.9960) and in zNose !(y=237X+10503.3, R!= 0.9985) (y=kilo-count, X=Nanogram). No significant differences were found among the retention indices calculated from GC-MS data and the retention indices generated by the zNose! (Table 2.3).  44  TABLE 2.3. COMPARING THE RETENTION INDICES OF FOUR CHEMICALS ANALYSED BY GC-MS AND ZNOSE! IN LABORATORY AND GREENHOUSE  Chemical  Analysis  Mean  SD  SE m  1,8-cineole  GC-MS  1041.2  0.414  0.106  zNose!  1041.2  0.328  0.084  GC-MS  1367.7  0.457  0.118  zNose!  1367.8  0.414  0.106  GC-MS  1273.7  0.487  0.125  zNose!  1273.5  0.414  0.106  GC-MS  1253.3  0.457  0.118  zNose!  1253.4  0.736  0.196  Eugenol  Citral  Geraniol  F  F-sig  t  df  t-sig  95%CI (L-U)  0.758  0.391  0.332  28  0.742  -0.2342  0.3249  0.707  0.408  0.418  28  0.679  -0.3931  0.2597  2.635  0.116  0.807  28  0.426  -0.4718  0.2051  2.611  0.117  0.595  28  0.556  -0.1333  0.2239  SD= Standard deviation, SE m= standard error mean, F= Levene’s test for equality of variances, F. sig= significance of the Levene’s test, t = t-test for equality of means, df= degree of freedom, t-sig= significance of the t-test, 95% CI (L-U)= 95% confidence interval of the difference with lower and upper intervals  2.3.2 Retention index variation of limonene in laboratory and commercial greenhouses  TABLE 2.4. EFFECT OF TEMPERATURE, RELATIVE HUMIDITY AND ANALYSIS METHOD (GC-MS+ NEAT CHEMICAL, ZNOSE  +NEAT CHEMICAL, ZNOSE+PLANT IN LABORATORY AND ZNOSE IN  GREENHOSUE) ON RETENTION INDEX OF LIMONENE  Descriptive Statistics Mean  Std. Deviation  N  RI  1.0358E3  0.6195  60  Temperature  24.1897  0.77505  60  RH%  42.3542  3.74332  60  Method  2.5000  1.12747  60  45  Correlations  Pearson correlation  Sig. (1-tailed)  RI  Temperature  RH%  Method  RI  1.000  -0.054  -0.100  -0.163  Temperature  -0.054  1.000  0.869  0.638  RH%  -1.00  0.869  1.000  0.404  Method  -0.163  0.638  0.404  1.000  0.340  0.223  0.107  0.000  0.000  RI Temperature  0.340  RH%  0.223  0.000  Method  0.107  0.000  0.001 0.001  Model Summary Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  1  0.257*  0.066  0.016  0.41215  *Predictors: (Constant), method, RH%, Temperature  ANOVA: dependent variable = Retention index Model  Sum of Squares  df  Mean Square  F  Sig  Regression  0.671  3  0.224  1.316  0.278*  Residual  9.513  56  0.170  Total  10.183  59  *Predictors: (Constant), method, RH%, Temperature  Coefficients: dependent variable = Retention index Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig  B  Std. Error  Beta  (Constant)  1031.422  3.172  325.145  0.000  Temperature  0.273  0.181  0.509  1.508  0.137  RH%  -0.046  0.032  -0.413  -1.453  0.152  Method  -0.118  0.067  -0.320  -1.756  0.085  46  According to the results, temperature, relative humidity and analysis method do not have a significant correlation with the retention index. Only 6.6% of the variation of retention index can be explained by temperature, relative humidity or analytical method. The regression line predicted by the independent variables (temperature, RH% and method) does not explain a significant amount of variation in the retention index: F (3,56)= 1.316; p > 0.05. None of the independent variables are individually significant predictors of the retention index.  47  FIG 2.5 VARIATION OF RETENTION INDICES OF LIMONENE IN RESPONSE TO RELATIVE HUMIDITY AND TEMPERATURE IN LABORATORY AND GREENHOUSE  48  1"=#F+&54&&+)*# In this chapter I assessed a relatively new analytical technology. The zNose! is an instrument for rapid gas chromatography that is capable of repeated quantitative sampling of headspace volatiles (Kunert et al. 2002; Thollet al. 2006). My main goal was to investigate the detection and identification capability of the zNose! as well as its consistency within the field under variable environmental conditions. To do so, I compared retention indices (Van Den Dool and Kratz,1963) of known chemicals analyzed by the zNose! and compared them with the retention indices of the same chemicals analyzed by a conventional GC-MS. My results indicate that the zNose! is capable of correctly identifying the known compounds. The SAW sensor can produce quite consistent results in spite of the environmental conditions which makes this device a good option for field work. Because of the fast analysis time (about 3 minutes for a complete cycle), it can be used for monitoring rapid changes in volatile emissions of plants and other organisms (Schwartzberg et al. 2008). One of the major problems with this machine is its short column (1m) which can limit resolution of peaks when there are a large number of peaks in a sample. Based on my personal experience, I do not recommend overloading the system with more than 10 chemicals at the time during calibration. Calibration is another area which needs careful attention. Despite the manufacturer’s claim, I believe that the zNose! needs to be calibrated for each new analysis to improve the precision of the results. Overall, my results confirmed previous studies and show that the zNose! is a good option for volatile analysis both in the laboratory and the field. With proper care and maintenance it can provide consistent results and enable us to perform rapid volatile analysis in the field.  49  FIG 2.6 ZNOSE! IN THE FIELD  50  1"?#E%6%,%*5%&# Baldwin EA, Goodner K, Plotto A. 2008. Interaction of volatiles, sugars, and acids on perception of tomato aroma and flavor descriptors. J Food Sci 73:294-307. Che Man YB, Gan HL, Nor Aini I, Hamid NSA, Tan CP. 2005. Detection of lard adulteration in RBD palm olein using an electronic nose. Food Chem 90:829-835. D'Alessandro M, Turlings TC. 2006. Advances and challenges in the identification of volatiles that mediate interactions among plants and arthropods. Analyst 131:2432. EST. 2007. 4300 Ultra Fast GC Analyzer operation manual. Electronic Sensor Technology. Gan HL, Che Man YB, Nor Aini I, Tan CP, Hamid NSA. 2005. Characterization of vegetable oils by surface acoustic wave sensing electronic nose. Food Chem 89:507-518. Gouinguene SP, Turlings TC. 2002. The effects of abiotic factors on induced volatile emissions in corn plants. Plant Physiol 129:1296-1307. Jakobsen HB. 1997. The preisolation phase of in situ headspace analysis: methods and prespectives. In: H.F. L, Jackson JF, editors. Plant Volatile Analysis. Berlin: Springer-Verlag. p 1-22. Kovats E. 1958. Characterization of organic compounds by gas chromatography. part 1. retention indiced of aliphatic halides, alchohols, aldehdes and ketones. Helvetica Chimica Acta 41:1915-1932. Kunert M, Biedermann A, Koch T, Boland W. 2002. Ultra fast sampling and analysis of plant volatiles by a hand-held miniaturized GC with pre-concentration unit:  51  Kinetic and quantitative aspects of plant volatile production. J Separation Sci 25:677-684. Lammertyn J, Veraverbeke EA, Lrudayaraj J. 2004. zNose technology for the classification of honey based on rapid aroma profiling. Sensors and Actuators B: Chemical 98:544-562. Lu CJ, Jin C, Zellers ET. 2006. Chamber evaluation of a portable GC with tunable retention and microsensor-array detection for indoor air quality monitoring. J Environ Monit 8:270-278. Mathieu S, Cin VD, Fei Z, Li H, Bliss P, Taylor MG, Klee HJ, Tieman DM. 2009. Flavour compounds in tomato fruits: identification of loci and potential pathways affecting volatile composition. J Exp Bot 60:325-337. Mayer F, Takeoka GR, Buttery RG, Whitehand LC, Naim M, Rabinowitch HD. 2008. Studies on the aroma of five fresh tomato cultivars and the precursors of cis- and trans-4,5-epoxy-(E)-2-decenals and methional. J Agric Food Chem 56:3749-357. Oh SY, Ko JW, Jeong SY, Hong J. 2008. Application and exploration of fast gas chromatography-surface acoustic wave sensor to the analysis of thymus species. J Chromatogr A 1205:117-127. Pasini P, Powar N, Gutierrez-Osuna R, Daunert S, Roda A. 2004. Use of a gas-sensor array for detecting volatile organic compounds (VOC) in chemically induced cells. Anal Bioanal Chem 378:76-83. Pickett JA, Chamberlain K, Poppy GM, Woodcock CM. 1999. Exploiting insect responses in identifying plant signals. Novartis Found Symp 223:253-262; discussion 262-5, 266-9.  52  Schwartzberg EG, Kunert G, Stephan C, David A, Rose US, Gershenzon J, Boland W, Weisser WW. 2008. Real-time analysis of alarm pheromone emission by the pea aphid (Acyrthosiphon pisum) under predation. J Chem Ecol 34:76-81. Staples EJ, Viswanathan S. 2008. Development of a novel odor measurement system using gas chromatography with surface acoustic wave sensor. J Air Waste Manag Assoc 58:1522-1528. Tholl D, Boland W, Hansel A, Loreto F, Rose US, Schnitzler JP. 2006. Practical approaches to plant volatile analysis. Plant J 45:540-560. Turlings TC, Tumlinson JH, Lewis WJ. 1990. Exploitation of herbivore-induced plant odors by host-seeking parasitic wasps. Science 250:1251-1253. Van Den Dool H, Kratz PD. 1963. A generalization of the retention index system including linear temperature programmed gas-liquid partition chromatography. J Chromatography 11:463-471. Watkins P, Wijesundera c. 2006. Application of zNose for the analysis of selected grape aroma compounds. Talanta 70:595-601. Zhong Q, Steinecker WH, Zellers ET. 2009. Characterization of a high-performance portable GC with a chemiresistor array detector. Analyst 134:283-293.  53  CHAPTER THREE: USING HERBIVORE-INDUCED PLANT VOLATILES FOR DETECTING CABBAGE LOOPER INFESTATION ON GREENHOUSE TOMATO PLANTS5  8"!#0*',)/45'+)*# 3.1.1 Greenhouse tomato The greenhouse vegetable industry is an important and growing segment of Canadian agriculture. Canada has a short summer, which limits the quantity of field vegetables that can be grown. Through greenhouse production, Canadian grown tomatoes are available from March to December with peak production in May. There is a move toward trying to provide a year round supply; however, the economics of producing a crop when light levels and temperatures are at their lowest will increase costs and limit supplies from December to February. Almost all greenhouse vegetable production uses some form of hydroponics. The most common systems use rock-wool and coconut fiber slabs as the growing medium. Computerized production facilities and new varieties have increased the diversity of products and improved their quality. Most greenhouses in Ontario and British Columbia are heated with natural gas, usually purchased through producer-owned cooperatives. In warmer climates, cooling becomes a significant cost. The estimated value of the greenhouse vegetable industry was $80 M in 1988, reaching $600 M in 2000. The main greenhouse vegetable crops in Canada are tomatoes (468 ha), cucumbers (190 ha), sweet peppers (144 ha) and lettuce (21 ha) (Statistics Canada 2006).  5  A version of this chapter has been submitted for publication. Miresmailli S., Gries G.J., Gries R. M., Zamar, R.H. and Isman M.B. 2009. Using herbivore-induced plant volatiles for detecting cabbage looper infestation on greenhouse tomato plants.  54  FIG 3.1. (A) ESTIMATED VALUE OF THE GREENHOUSE VEGETABLE INDUSTRY IN CANADA, (B) MAIN GREENHOUSE VEGETABLE CROPS IN CANADA IN 2003, (C) FARM GATE VALUE OF THE FOUR MAIN VEGETABLE CROPS PRODUCED IN 2003 IN FIELD VERSUS GREENHOUSE, (D) ESTIMATED GREENHOUSE TOMATO AREA IN MEXICO VERSUS CANADA. DATA FROM STATISTICS CANADA AND AGRICULTURE AND AGRI-FOOD CANADA (2006, 2007).  During the 1990s, the total area under glass and plastic more than doubled to nearly 1500 ha and by 2003, it had reached nearly 1900 ha. In 2003, revenues from greenhouse sales reached a record high of almost $2.1 billion, nearly double that of six years earlier. Ornamentals accounted for about 70% of sales and vegetables was the remaining 30%. In the early 1990s, revenues from comparable greenhouse and field-grown vegetables were roughly the same. However, since 1996, revenues from greenhouse vegetables have increased at a much more rapid rate than field vegetables. For example, in 2003, the farm gate value of the four main vegetable crops produced under glass or plastic (tomatoes,  55  cucumbers, lettuce and peppers) amounted to $605.8 million. This was more than three times higher than the value of $171.7 million for the same four vegetable crops produced in the field. Farmers grow more tomatoes than any other vegetable crop, both in the greenhouse and in the field. Tomatoes alone account for over one-half of revenues from the sale of greenhouse vegetables. According to Statistics Canada, the main greenhouse vegetable producing provinces in 2007 were Ontario with farm cash receipts (FCR) of $418.2 million and British Columbia with FCR of $201.6 million. The 2007 export value of the three major greenhouse crops (tomatoes, cucumbers and peppers) to the U.S. was down by 11% from the previous year to $510.7 million. This is mostly due to the higher value of the Canadian dollar versus the U.S currency. Tomato exports showed a 21% decrease in volume in 2007, due to a significant increase in Mexico’s exports to the U.S. at reportedly lower prices. The estimated greenhouse tomato area in Mexico in 2008 was 850 hectares which was substantially more than what it was in 2003 (182 ha) while in Canada it went down from 444 hectares in 2003 to 432 hectares in 2008 (Canada 2008). Due to the current global economic crisis and its significant effects on the greenhouse tomato market, it is very important to put more effort on protecting the crops and reducing the crop loss caused by pests and diseases.  3.1.2 Pest monitoring Monitoring pest populations is a cornerstone of the integrated pest management (IPM) philosophy but it has lagged in comparison to the development of other monitoring tools. Many of the current pest monitoring systems are based on presence-absence methods and  56  rely on pest counts (Alatawiet al. 2005) or signs of their presence or damage (Hughes 1999). Even modern technologies follow the same path and base their monitoring on pest-related indicators (Drake et al. 2002; Bange et al. 2004; Oerke et al. 2006; Skaloudova et al. 2006; Boissard et al. 2008). The majority of pest monitoring techniques enable the grower to estimate pest populations or predict pest outbreaks; however, we still use human scouts to establish the precise location of pests within a field or a greenhouse (McCornacket al. 2008). Scouts are faced with the challenge of visually scanning a large representative number of plants, plant organs and a variety of pests. Fields and greenhouses pose problems due to limitations of human vision; some pests might be overlooked in their early stages of development. Besides, scouts need extensive training and their performance is never the same due to different levels of experience, individual values and considerations; thus making consistent decisions is a challenge (Lichtenberg and Berlind 2005). By using new statistical and computational tools, we can now incorporate different parameters in our decision making process (Bhattacharyya and Bhattacharya 2007; Soubeyrandet al. 2009). My main goal in this project is to investigate the possibility of creating a new tool that can increase the efficiency and performance of human scouts based on an alternative source of information. For pest monitoring, one should not just look for the pests but rather detect the signals generated by plants, as they may be more informative.  3.1.3 Herbivore-induced plant volatiles One of the well-documented responses of plants to biotic stressors is the emission of herbivore-induced plant volatiles (HIPVs) also known as info-chemicals due to the fact  57  that they carry some information about the status of the plant (Arimura et al. 2005). HIPVs can strongly affect the behavior of both predatory and herbivorous arthropods and some plants are under strong selection pressure to release these volatiles (Kessler and Baldwin 2001). HIPVs are known to be emitted by various parts of plants including leaves (Turlings et al. 1995; Dicke 1999; Van Poecke et al. 2001) from both the abaxial and the adaxial side (Bergougnoux et al. 2007), flower buds (Rose and Tumlinson 2004) and roots (Rasmann et al. 2005). HIPV emission is not limited to the site of damage but also occurs systematically throughout the plant even in undamaged parts (Turlings and Tumlinson 1992; Mattiacci et al. 2001; Neveu et al. 2002; Rose and Tumlinson 2005). Recent findings show that plants can recognize the herbivores and assess their threat via a series of chemical and electrical reactions that occur before the activation of defensive mechanisms (Maffei et al. 2007a). Considering the vast knowledge of plant responses to the environment, it is conceivable to use these plant-generated cues as indicators of herbivore presence in pest monitoring programs in addition to previously used indicators. The questions are: how reliable are these cues and how quickly can they convey information? When plants emit info-chemicals, they have no control over the receiver of these signals (Halitschke et al. 2008; Heil 2008) and although they can be found in higher concentrations closer to the emitter (Lennert et al. 1997), it is difficult to relate these volatiles that float in the air to their actual source. Environmental factors such as light intensity, temperature and moisture can profoundly affect the emission of plant volatiles (Gouinguene and Turlings 2002). However, in spite of the complexity of this system, predators can associate these chemical cues in addition to other cues to locate their prey  58  (Dicke 1999). Some of these volatile chemicals are emitted within minutes after tissue damage and can be considered a quick indicator of problems (Heil and Silva Bueno 2007), while other chemicals are released later as a complement to other types of defense (Kant et al. 2004). A pest monitoring system that is capable of harvesting information from the environment through a series of sensory systems can also be capable of associating different signal patterns with pest presence and perform the same task.  3.1.4 Plant volatile collection and analysis The scientific literature on HIPVs is vast and continuously growing. In most cases, the researcher does not know the biological activity of the compounds assessed and therefore samples and analyzes a full range of HIPVs. Usually the volatile organic compounds in the headspace of plants that are enclosed in collection chambers are collected using an adsorbing material. Subsequently the collected volatiles are analyzed by gas chromatography (GC) and mass-spectrometry (MS) or a combination of both (GC/MS) (Turlings et al. 1990; Pickett et al. 1999). (D'Alessandro and Turlings 2006) looked into the most commonly used HIPV collection methods from 1995 to 2004. They found that adsorbent/solvent desorption was the most popular method among the scientists who studied insect-plant interactions. For plant-herbivore interaction studies, using intact plants is generally considered to be most rigorous and accurate. Experiments that use detached plant parts often show more fluctuations in volatile profile than intact plants and are often verified against intact plant results to compensate for oscillations caused by the mechanical damage from detachment (Jakobsen 1997).  59  For studies in chemical ecology, using intact plants is considered the most reliable method but additional considerations must be made to minimize other factors that can affect the accuracy of the results. Such parameters to consider are light intensity, relative humidity (RH), air temperature, and photoperiod (Jakobsen 1997; Gouinguene and Turlings 2002). Recent improvements in analytical tools enable researchers to collect more accurate data about HIPVs and plant systems within their growing environment in short periods of time (Kunert et al. 2002; Pasini et al. 2004; Lu et al. 2006; Oh et al. 2008; Zhong et al. 2009).  3.1.5 The ultra-fast gas chromatograph (zNose!) In a very comprehensive review paper, Tholl et al. (2006) explored different practical methods of plant volatile collection and analysis. In addition to several conventional collection, separation and detection methods reviewed, they also discussed a relatively new instrument, the zNose !(Electronic Sensor Technology, Newbury Park, CA, USA), which is a miniaturized ultra-fast portable gas chromatograph. They found this instrument a useful tool for fast quantitative estimation of known volatile profiles and for monitoring rapid changes in volatile organic chemical (VOC) emissions. As a portable device, they suggested that it could be used in field experiments. Although it is a relatively new technology, the zNose! has been successfully used in several studies. It has been employed for monitoring rhythmic VOC emissions from flowers and induced VOC emissions from herbivore-damaged plants (Kunert et al. 2002), honey classification (Lammertyn et al. 2004), grape aroma classification (Watkins and Wijesundera 2006), detection of lard adulteration in palm olein (Che Man et al. 2005),  60  characterization of vegetable oils (Gan et al. 2005) and monitoring alarm pheromone emission by the pea aphid under predation (Schwartzberg et al. 2008). In this project, I asked one key question: Can we discriminate clean plants from infested plants based on their volatile emission pattern? This question is not novel and many studies have shown differences in volatile emissions of infested plants, although most of these studies were conducted in laboratories within closed chambers. I took this to a different level and tested the plants inside a research greenhouse to see if I can observe the same responses. The main objective was to identify some infestation indicator volatiles and use them for classifying plants into infested or clean groups.  8"1#(.'%,+.2&#.*/#(%'@)/&# 3.2.1 Cabbage looper Cabbage loopers (Trichoplusia ni, Noctuidae) were obtained from a research colony maintained without pesticide exposure at the University of British Columbia.  3.2.2 Plant material Tomato plants (Lycopersicon esculentum Mill cv. Clearence) were obtained from Houweling’s Hot House (Delta, BC). Intact tomato plants (30-40 cm high) were selected carefully and stored in a growth chamber at 25 °C ±1 and a 16:8h L:D photoperiod.  61  3.2.3 Plant volatile collection system Following the closed-loop stripping system developed by Boland et al. (1984), I designed and built a system with some modifications (Fig 3.2). The volatile collection system consisted of two vacuum pumps (GAST Miniature diaphragm 15D 1150 series – Max pressure 15psi (1.03421 bar), Max Vacuum 14psi (0.965266 bar) – IDEX corporation, Benton Harbor, MI, USA), vinyl tubing (2236 PL-4-NT, inside diameter 4 mm, operating pressure 0.95-9 bar, Food-industry approved 90/128/EWG FDA, FESTO Ltd. Mississauga, Ontario), an air purifier filter (PISCO FTA 300 MD-B, Bensenville, IL, USA), an air micro-mist filter (PISCO UADR 300 AD, Bensenville, IL, USA), active charcoal filter, Pyrex tubes for introducing air to the chambers and a digital timer which controlled the vacuum pumps and allowed continuous volatile collection for different periods of time. The timer was constructed using a NE555 chip, a highly stable controller capable of producing accurate timing pulses. The timer could turn the pump on and off within the period of 0.5 to 72 hours. Customized Pyrex tubes were designed and made to carry adsorbent materials. Fifty milligrams of Porapak Q adsorbent (Supelco, 100-120 mesh) were weighted and poured inside each tube. Lab grade glass wool pieces were put on both sides of each tube to fix the Porapak Q. Before use in experiments, each tube was washed by n-pentane several times and heated in an oven at 60°C for five min while a current of pure nitrogen passed through the filter to remove all impurities. The tubes were then wrapped individually in aluminum foil. This process was repeated before each experiment.  62  PQ  FIG 3.2. VOLATILE COLLECTION SYSTEM. PQ = PORAPAK Q TUBES, V= VACUUM PUMP, F= AIR MICRO MIST FILTER AND ACTIVE CHARCOAL FILTER, P= AIR PUMP, HF= A IR PURIFIER FILTER  The collection chambers consisted of 12 glass jars (20 " 15 cm wide). Each jar had a small hole (7 mm) to insert the Porapak tube. Six tables with two 12-cm diameter holes in each were made as the base for the system. The surface of the tables was covered with aluminum foil. Two glass plates covered each hole on the table. Once plants were placed inside the hole, two glass plates were placed around the plant stem and then the open area sealed with lab-grade parafilm. The plants were then covered with the glass jars and jars were sealed with lab film to make them airtight. By using industry grade push-in fitting connectors (FESTO Ltd. Mississauga, Ontario), the vacuum pump and air pump were attached to tubes in a way that provided equal pressure (1 psi ~ 0.0689 bar). The pressure at the end of each tube was measured and required adjustments were made to ensure equal vacuum pressure. With this system, it has been possible to collect volatiles from 6 clean and 6 infested plants (control and treatment, respectively) randomly placed inside each chamber. This system pumps purified air into the airtight chambers containing test  63  plants. The clean air blows over plant foliage and gets sucked out by the vacuum pump. Plant volatiles were then trapped on the Porapak Q filters and extracted by a solvent (2 ml of HPLC grade n-Pentane solvent) immediately after collection.  3.2.4 Laboratory experiment for identifying indicator chemicals This experiment was designed to test for differences between volatiles emitted from clean plants versus infested plants and to identify the indicator chemical cues. This experiment was conducted inside a growth chamber (16:8 light-dark regime, 25°C). I placed ten third-instar T. ni larvae (starved for 3 hours) on each treatment plant (total of six plants) and no insects on control plants (total of six plants) and collected volatiles for 24 hours. Both treatments and controls were replicated six times. Plants were randomly assigned to treatment or control groups. This experiment was repeated five times.  3.2.5 Plant volatile analysis with GC-MS After each experiment, volatile chemicals from Porapak Q adsorbent filters were extracted using 2 ml of HPLC grade n-Pentane solvent. The samples were then analyzed by GC-MS6. The GC mass spectrometer that was used for this experiment was a Varian Saturn 2000 Ion Trap. The GC column was a DB-5 (30 meters long - 0.25mm inner diameter, J&W Scientific, Folsom, CA). The GC program was 50 º C for 2 minutes, then increased 10 º per minute to 280 º C. Heptyl acetate was used as an internal standard. The following compounds were used as authentic external standards7: (Z)-3-hexenyl acetate (obtained by esterification of (Z)-3-hexen-1-ol (Aldrich 98%) with acetic anhydride and 6  All the GC-MS analysis were conducted in the Gries-laboratory at Simon Fraser University by Mrs. Regine Gries 7 Authentic standards were obtained from the Gries-laboratory at Simon Fraser University  64  pyridin), limonene (Aldrich 96%), ocimene (International Flavour and Fragrances, 69% trans, 27% cis), !-caryophyllene (Sigma 98%).  3.2.6 Greenhouse experiment After the major indicator chemicals in infested plant volatiles were identified, I designed an experiment inside the UBC horticulture greenhouse. The main objectives of this experiment were (a) to investigate the differences in volatile emission of clean versus infested plants inside a research greenhouse using the zNose! and (b) to monitor the changes in the volatile emission rate over time and explore the possibility of discriminating infested plants at an early stage of damage. In each set of experiments, 50 tomato plants (eight weeks old) were placed on a greenhouse bench inside the greenhouse 48 hours prior to the experiment. The waiting period was included to allow the plants to acclimatize to greenhouse conditions. From these 50 plants, 40 plants that had no sign of mechanical or pest damage were selected for the experiment. The plants then were spaced 1 m from each other on the greenhouse bench (45 by 1.5 meter). Twenty plants were randomly assigned as treatments and the other 20 as controls (completely randomized design- 20 replicates per group). On top of each plant, I hung a clear plastic bag (Fisher precision ultra-clean Bag, Fisher Scientific, Canada) that covered the upper half of the plant canopy. I made a cross-shaped frame with wooden sticks wrapped in aluminum foil and attached it to the exterior upper part of the bags. The frame kept the bag in a cube-shaped form, which minimized contact of the plant part with the bag and therefore, minimized condensation. The lower part of the canopy was covered by a fine mesh screen which allowed ventilation but prevented pests  65  from leaving the plants in treatment groups (Fig 3.3). A short tube was attached to the upper part of the bag to form an opening for volatile collection (no glue or tape was used).  Wooden frame  Clear Bag  Mesh screen  FIG 3.3 GREENHOUSE SETUP FOR COLLECTING VOLATILES FROM TOMATO PLANTS INFESTED WITH CABBAGE LOOPER  Despite the careful selection process which I used for plants, I was aware of the fact that the plants are not uniform and their sizes might vary. Therefore, to eliminate the effect of size variation in the amount of volatiles emitted from plants, I measured the level of all indicator chemicals in all treatment and control groups one hour before introducing pests to the treatment groups. I formed a baseline for each plant and compared all the changes in volatile emission with this baseline. In addition to this technique, I measured the fresh weight of all plants (including the feces of cabbage loopers in treatment groups) immediately after each experiment. 66  As a negative control, I hung 10 empty bags and screens randomly in various locations among the bags containing the plants. The comparison of the levels of indicator volatiles at the baseline was compared in control, treatment and negative control (empty bags) groups using analysis of variance (ANOVA). After establishing the baseline in all groups, I placed 20 third-instar T. ni larvae (starved for 3 hours) on each plant (randomly on all parts of the canopy) in the treatment group. I then measured the level of indicator chemicals in all groups 6, 12 and 24 hours after infestation. Time was considered as a factor (with 3 levels) as opposed to a continuous variable. This allowed more flexibility to my model (I didn't assume a monotone linear effect). This experiment was repeated two times.  3.2.7 zNose! calibration and program properties The system was tuned with alkane standards (C4-C24 as explained in Chapter Two) prior to each set of experiments. Authentic standards were obtained from the Gries-laboratory at SFU. A working stock solution was prepared by diluting 1 g (weighed to the nearest 0.1 mg) of each authentic standard into 50 ml of HPLC grade n-pentane (total of four working stocks). An aliquot (100 µl) of each working stock was pipetted into a 10 ml volumetric flask and diluted to the mark with HPLC grade n-pentane. The flask was then shaken and sonicated for 5 min to facilitate complete dissolution. A subsample (0.5 ml) was transferred to a septum-capped GC vial for calibration. Four different volumes of reagent standards were injected into the zNose! using the zNose ! 3500 sample injector (0.1 µl, 0.2 µl, 0.4 µl and 0.8 µl - five times per volume). Calibration curves were  67  developed for each chemical. I completely baked, cleaned and calibrated the system for each set of experiments (two times for greenhouse experiment). The zNose! was programmed as follows: The inlet temperature was 200°C, the valve temperature was 165°C, and the initial column temperature was 40°C. During analysis the column temperature was increased at the rate of 10°C per second to reach a final column temperature of 200°C. The SAW sensor was operated at 50°C and the trap was operated at 250°C. Helium flow was set at 3.00 ccm. The sampling was set for 10 seconds (Sample flow 20 ccm) after which the system switched to 20 seconds of data acquisition mode. After each data sampling period the system needed a 30-sec baking period, in which the sensor was heated briefly to 150°C and after which the temperature conditions of the inlet, column, and sensor were reset to the initial conditions. In between each sample measurement at least one blank (one ml HPLC grade n-pentane in a 40-ml GC vial with silicon septa) was run to ensure cleaning of the system and a stable baseline.  3.2.8 Data analyses 3.2.8.1 Chemical baselines before infestation Chemical levels at one hour prior to infestation in control and treatment groups were analyzed by ANOVA (SPSS ver.16) to compare the level of volatiles at the baseline before treatment.  68  3.2.8.2 Comparing level of chemicals in control and treatment groups In the model used to analyze the data, it was necessary to take into account the temporal nature of the results. Classical regression using least squares makes an important assumption that observations are independent. When a study involves repeated measurement on the same plant over time, this independence assumption is no longer met, because the data are characterized by repeated observations over time on the same set of units, and data obtained from the same unit can be closely correlated. If the correlation of data within a unit is ignored, it may produce incorrect standard errors, resulting in invalid hypothesis tests and confidence intervals. Furthermore, failure to incorporate correlation of responses can lead to incorrect estimation of regression model parameters, particularly when such correlations are large. The regression estimates are less efficient, that is, they are more widely scattered around the true population value than they would be if the within unit correlation were incorporated in the analysis. There are many methods available for the analysis of longitudinal data. Examples include repeated measure ANOVA, generalized linear mixed models (GLMM) and generalized estimating equations (GEE). Repeated measure ANOVA assumes compound symmetry which implies constant variances and covariances over time. Such an assumption has little if any validity for longitudinal data. The model allows each experimental unit to have its own trend line, however, the trend lines can only differ in terms of their intercepts which implies that experimental units deviate at baseline but are consistent thereafter. Generalized Linear mixed-effects regression models are another common approach  69  used for the analysis of longitudinal data. They are robust for missing data and irregularly spaced measurement occasions, however, they inherently make distributional assumptions which may not hold, and it is difficult to verify their validity. Generalized estimating equations use methods of inference which take into account the correlated nature of the within unit correlation, but are robust to misspecification of the covariance structure and make no distributional assumptions on the data. GEE estimates are the same as those produced by regression when the dependent variable is normally distributed and no correlation within response is assumed (Ballinger 2004). A generalized estimating equations (GEE) regression technique was used to analyze the data (Using R ver. 2.9.0) 8.  3.2.8.3 Classification of plants into clean and infested groups Linear discriminant analysis (LDA) was used for data classification (Using R ver.2.9.0)9. In order to detect infected plants, it is important to classify them depending on the level of the volatile chemicals they emit. I used linear discriminant analsyes, a method used in statistics and machine learning to find the linear combination of features which best separate two or more classes of objects or events. The resulting combination may be used as a linear classifier (Peladan et al. 1984; Yu and Yang 2001; Siripatrawan et al. 2004).  8  GEE analysis has been conducted by the statistical consulting and research laboratory (SCARL) at the Department of Statistics, The University of British Columbia. 9 LDA has been conducted by the statistical consulting and research laboratory (SCARL) at the Department of Statistics, The University of British Columbia.  70  8"8#E%&42'&# 3.3.1 Laboratory experiments My results showed quantitative differences between chemical volatiles emitted from infested versus clean plants. From these volatiles, I chose the following four chemicals as pest (cabbage looper) infestation indicators: (Z)-3-hexenyl acetate, (E)-!-ocimene, limonene and !-caryophyllene (Fig 3.4). All of these chemicals have been previously reported in tomato plants volatile blend (Buttery et al. 1987; Baldwin et al. 2008; Mayer et al. 2008; Mathieu et al. 2009).  3.3.2 zNose! calibration The calibration curves for all chemicals analyzed with the zNose! were linear: (Z)-3hexenyl acetate (y=214.78X-805.78, R!= 0.9984); (E)-!-ocimene (y= 189.39X+1098.6, R!= 0.9905); limonene (y=201X+402.91, R!= 0.9838) and !-caryophyllene (y=162.3X+3055.1, R!= 0.9703) (y=kilo-counts, X=Nanogram).  71  !-caryophyllene  (E)-!-ocimene  limonene  (Z)-3-hexenyl acetate  FIG 3.4 DIFFERENCES IN VOLATILE BLEND OF INFESTED (WITH CABBAGE LOOPER) TOMATO PLANTS (UPPER CHROMATOGRAM) AND CLEAN TOMATO PLANTS (LOWER CHROMATOGRAM). THERE ARE SEVERAL UNKNOWN COMPOUNDS IN THE BLEND.  72  3.3.3 Greenhouse experiments No significant quantitative differences were found among the amount of volatiles emitted from clean versus infested plants at the baseline. However, strong significant differences were found in the levels of all chemicals at 6,12 and 24 hours. Both treatment and time were found to be effective in creating these differences (Fig 3.5 and Table 3.1, 3.2, 3.3, 3.4). The following regression model was fitted to data using GEE. The objective is to try to explain chemicals changes over time with consideration of the treatment that plants received. "Chemical = # 0 + #1 (treatment) + # 2 (time1) + # 3 (time2) + error  For everything being held constant "1 for example represents the change in the chemical ! level with a unit change in treatment. We want to see which beta is significantly different ! from zero. None of the indicator volatiles was detected in empty bags. Time was not a  continuous variable. The lines in the graphs are simply aids for following the direction of changes in chemical volatiles.  73  (Z)-3-hexenyl acetate  (E)-!-ocimene  limonene  !-caryophyllene  FIG 3.5 VOLATILE EMISSION (NANOGRAM) IN CONTROL AND TREATMENT (INFESTED WITH CABBAGE LOOPER) TOMATO PLANTS OVER TIME (HR) (MEAN±SD)  TABLE 3.1. GEE REGRESSION COEFFICIENTS- (Z)-3-HEXENYL ACETATE EMISSION IN TOMATO PLANTS IN RESPONSE TO CABBAGE LOOPER INFESTATION OVER TIME  Estimate  Std Error  Wald test  Sig  7.28  5.50  1.75  0.19  Treatment  708.13  7.22  9627.68  p<0.001  Time1 (6-12)  98.76  12.99  57.79  p<0.001  -136.86  18.37  55.51  p<0.001  Intercept  Time2 (12-24)  74  In case of (Z)-3-hexenyl acetate, treatment had a strong significant effect (GEE coefficient 708.13- the significance level included all time intervals) on the emission of volatiles. The first time period had a strong positive effect on the amount of volatiles while the second time period had strong negative effect on the volatile emission level (GEE coefficient -136.86). It means that the level of (Z)-3- hexenyl acetate significantly increased for about 12 hours after infestation but then it decreased between 12 to 24 hours after infestation. However, it never reached the level of control.  TABLE 3.2. GEE REGRESSION COEFFICIENTS- (E)-!-OCIMENE EMISSION IN TOMATO PLANTS IN RESPONSE TO CABBAGE LOOPER INFESTATION OVER TIME  Estimate  Std Error  Wald test  Sig  -49.46  6.83  52.5  p<0.001  Treatment  2435.27  44.16  3040.8  p<0.001  Time1 (6-12)  126.95  15.41  67.9  p<0.001  Time2 (12-24)  106.19  12.79  18.9  p<0.001  Intercept  In case of (E)-!-ocimene, treatment had the strongest effect on the level of volatile. The first time interval (6-12hr) had greater effect on the emission level than the second time interval (12-24hr). This means that plants’ responses were slightly stronger at the early stages of damage.  75  TABLE 3.3. GEE REGRESSION COEFFICIENTS- LIMONENE EMISSION IN TOMATO PLANTS IN RESPONSE TO CABBAGE LOOPER INFESTATION OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  -166.2  20.0  68.8  p<0.001  Treatment  744.8  19.2  1497.4  p<0.001  Time1 (6-12)  160.2  18.0  79.5  p<0.001  Time2 (12-24)  335.9  43.0  68.4  p<0.001  Treatment had a significant effect on the emission of limonene. Unlike (E)-!-ocimene, the second time interval had greater effect on the emission of limonene.  TABLE 3.4. GEE REGRESSION COEFFICIENTS- "-CARYOPHELLENE EMISSION IN TOMATO PLANTS IN RESPONSE TO CABBAGE LOOPER INFESTATION OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  -1220.1  139.4  76.6  p<0.001  Treatment  18191.6  71.2  65353.5  p<0.001  Time1 (6-12)  1913.7  220.6  75.3  p<0.001  Time2 (12-24)  1665.5  189.2  77.5  p<0.001  Like other chemicals, treatment had a very strong effect on the emission of !caryophyllene followed with the first and second time interval.  76  3.3.4 Classification of plants into clean and infested groups The misclassification rate of all chemicals were zero at 6 hours, meaning that given the level of chemicals, we are able to classify plants into clean or infested groups only six hours after pest damage begins (Table 3.5).  TABLE 3.5. LINEAR DISCRIMINANT ANALYSIS OF TOMATO PLANT VOLATILES AFTER 6 HOURS OF INFESTATION WITH CABBAGE LOOPER. PRIOR PROBABILITIES OF GROUPS ARE :  0.99 FOR CLEAN AND  0.01 FOR INFESTED PLANTS  (Z)-3-hexenyl acetate (E)-!-ocimene limonene !-caryophyllene  Clean  Infested  Mean  Mean  LDA  Misclassification  Coefficient  rate  458.25  1176.70  0.01139335  0%  1173.775  3515.600  0.00427449  0%  120.70  525.85  0.02519481  0%  1310.175  17062.375  0.00370752  0%  #  77  8"=#F+&54&&+)*# In this chapter I explored the possibility of using HIPV-based indicators for detecting pest infestations. Plants cannot talk and they cannot walk but they can communicate through several types of cues and responses. They can provide us with useful information about their health. Some even believe that we should put plants in charge of their own well-being and let them control the optimum conditions for their development and growth (Janssen et al. 2004). Plants can defend themselves against threats both directly and indirectly and can actively manipulate their environment (Schoonhoven et al. 2006). Despite several controversial interpretations of plant-generated cues and responses – whether they are intelligent or reflexive- or their evolutionary raison d'être (Trewavas 2003; Heil et al. 2008), it is an accepted fact that most plants are capable of responding to changes in their surroundings and can convey information about their overall health status through those responses (Volkov and Ranatunga 2006). There is a large body of evidence that supports this claim. For example, some plants are capable of showing the footsteps of insects crawling on their foliage (Bown et al. 2002), while some other plants react to pest oviposition (Hilker et al. 2005; Schroder et al. 2005) or feeding (Kessler and Baldwin 2001; Arimura et al. 2004; Schoonhoven et al. 2006). Plants show various types of induced responses to organisms that range from viruses, bacteria, fungi, nematodes, mites, insects to mammals as part of their defense mechanisms (Dicke and Hilker 2003). Emitting HIPVs is one of these responses. Following the most commonly used technique for volatile collection and analysis (D'Alessandro and Turlings 2006), I managed to identify four chemical volatiles as pest indicator signals.  78  The first compound is (Z)-3-hexenyl acetate which is a green leaf volatile. Green leaf volatiles (GLVs) are C6 aldehydes, alcohols, and their esters formed through the hydroperoxide lyase pathway of oxylipin metabolism. Plants start to form GLVs after disruption of their tissues and after suffering biotic or abiotic stresses (Matsui 2006). GLVs have been shown to significantly activate jasmonic acid production in exposed corn (Zea mays) seedlings, thereby priming these plants specifically against subsequent herbivore attack (Engelberth et al. 2007). In addition to the major ecological functions of GLV biosynthesis which is related to resistance against both herbivores and pathogens, the genetic modification of GLV biosyntheses could be a unique approach for improving plant resistance against such biotic stresses (Shiojiri et al. 2006). The second indicator chemical is (E)-!-ocimene which is an acyclic monoterpene. Terpenes are the most abundant and varied class of HIPVs. For instance, it has been shown that in response to the beet armyworm, Spodoptera exigua Hubner, feeding, the volatile (E)-!-ocimene was released from leaves of both undamaged and insect-damaged Medicago truncatula plants, but at levels two-fold higher in insect-damaged M. truncatula. The strong release of volatile (E)-!-ocimene, suggest that it plays an active role in indirect insect defenses in M. truncatula (Navia-Gine et al. 2009). The defensive property of (E)-!-ocimene is not limited to plants. It has been shown that the presence of (E)-!-ocimene may serve to provide some protection of box elder bugs, Boisea trivittata, from predation (Palazzo and Setzer 2009). The third indicator compound is limonene which is a cyclic monoterpene. Limonene is also known to have defensive properties against both pests and diseases either directly  79  (Byun-McKay et al. 2006; Ben-Yehoshua et al. 2008) or indirectly (Verheggen et al. 2008). The last indicator chemical was !-caryophyellene which is a sesquiterpene. !-Caryophyllene has been reported as an herbivore-induced plant volatile both above ground (Abel et al. 2009) and below ground (Hiltpold and Turlings 2008). My results indicated the suitability of these volatile chemicals as indicators of pest presence. They not only can enable us to discriminate clean and infested plants, but as I showed in my greenhouse experiment, they can indicate the problem at early stages of infestation. Both precision and early detection are important factors for a successful IPM program. Compared to current methods of monitoring, HIPVs can provide us more reliable information about the health of the plant. By using modern sensory systems, we can detect these volatiles and use them as an additional source of information for rapid decision making in crop protection.  80  8"?#E%6%,%*5%&# Abel C, Clauss M, Schaub A, Gershenzon J, Tholl D. 2009. Floral and insect-induced volatile formation in Arabidopsis lyrata ssp. petraea, a perennial, outcrossing relative of A. thaliana. Planta 230:1-11. Alatawi FJ, Opit GP, Margolies DC, Nichols JR. 2005. Within-plant distribution of twospotted spider mites (Acari: Tetranychidae) on impatiens: development of a presence-absence sampling plan. J Econ Entomol 98:1040-1047. Arimura G, Kost C, Boland W. 2005. Herbivore-induced, indirect plant defences. Biochim Biophys Acta 1734:91-111. Arimura G, Ozawa R, Kugimiya S, Takabayashi J, Bohlmann J. 2004. Herbivore-induced defense response in a model legume. Two-spotted spider mites induce emission of (E)-!-ocimene and transcript accumulation of (E)-!-ocimene synthase in Lotus japonicus. Plant Physiol 135:1976-83. Baldwin EA, Goodner K, Plotto A. 2008. Interaction of volatiles, sugars, and acids on perception of tomato aroma and flavor descriptors. J Food Sci 73:294-307. Ballinger GA. 2004. Using generalized estimatin equations for longitudinal data analysis. Organizational Res. Metd. 7:127-150. Bange MP, Deutscher SA, Larsen D, Linsley D, Whiteside S. 2004. A handheld decision support system to facilitate improved insect pest management in Australian cotton systems. Comp Electron Agric 43:131-147. Ben-Yehoshua S, Rodov V, Nafussi B, Feng X, Yen J, Koltai T, Nelkenbaum U. 2008. Involvement of limonene hydroperoxides formed after oil gland injury in the  81  induction of defense response against Penicillium digitatum in lemon fruit. J Agric Food Chem 56:1889-1895. Bergougnoux V, Caissard JC, Jullien F, Magnard JL, Scalliet G, Cock JM, Hugueney P, Baudino S. 2007. Both the adaxial and abaxial epidermal layers of the rose petal emit volatile scent compounds. Planta 226:853-866. Bhattacharyya S, Bhattacharya DK. 2007. A more realistic approach to pest management problem. Bull Math Biol 69:1277-1310. Boissard P, Martin V, Moisan S. 2008. A cognitive vision approach to early pest detection in greenhouse crops. Comp Electron Agric 62: 81-93. Boland W, Ney P, Jaenicke L, Gassmann G. 1984. A closed-loop-striping technique as a versatile tool for metabolic studies of volatiles. In: Schreier P, editor. Analysis of Volatiles. Berlin: Walter de Gruyter. p 371-373. Bown AW, Hall DE, MacGregor KB. 2002. Insect footsteps on leaves stimulate the accumulation of 4-aminobutyrate and can be visualized through increased chlorophyll fluorescence and superoxide production. Plant Physiol 129:14301434. Buttery RG, Liang LC, Light DM, 1987, tomato leaf volaile aroma component. J Agric Food Chem 35: 1039-1042 Byun-McKay A, Godard KA, Toudefallah M, Martin DM, Alfaro R, King J, Bohlmann J, Plant AL. 2006. Wound-induced terpene synthase gene expression in Sitka spruce that exhibit resistance or susceptibility to attack by the white pine weevil. Plant Physiol 140:1009-1021. Statistics Canada. 2008. Canada's Tomato industry Report.  82  Statistics Canada. 2006. Horticulture and Greenhouse Products by Province. Che Man YB, Gan HL, Nor Aini I, Hamid NSA, Tan CP. 2005. Detection of lard adulteration in RBD palm olein using an electronic nose. Food Chem 90:829-835. D'Alessandro M, Turlings TC. 2006. Advances and challenges in the identification of volatiles that mediate interactions among plants and arthropods. Analyst 131:2432. Dicke M. 1999. Specificity of herbivore-induced plant defences. Novartis Found Symp 223:43-54; discussion 54-9, 160-5. Dicke M, Hilker M. 2003. Induced plant defences: from molecular biology to evolutionary ecology. Basic Appl Ecol 4:3-14. Drake VA, Wang HK, Harman IT. 2002. Insect monitoring radar:remote and network operation. Comp Electron Agric 35:77-94. Engelberth J, Seidl-Adams I, Schultz JC, Tumlinson JH. 2007. Insect elicitors and exposure to green leafy volatiles differentially upregulate major octadecanoids and transcripts of 12-oxo phytodienoic acid reductases in Zea mays. Mol Plant Microbe Interact 20:707-716. Gan HL, Che Man YB, Nor Aini I, Tan CP, Hamid NSA. 2005. Characterization of vegetable oils by surface acoustic wave sensing electronic nose. Food Chem 89:507-518. Gouinguene SP, Turlings TC. 2002. The effects of abiotic factors on induced volatile emissions in corn plants. Plant Physiol 129:1296-307.  83  Halitschke R, Stenberg JA, Kessler D, Kessler A, Baldwin IT. 2008. Shared signals 'alarm calls' from plants increase apparency to herbivores and their enemies in nature. Ecol Lett 11:24-34. Heil M. 2008. Indirect defence via tritrophic interactions. New Phytol 178:41-61. Heil M, Lion U, Boland W. 2008. Defense-inducing volatiles: in search of the active motif. J Chem Ecol 34:601-604. Heil M, Silva Bueno JC. 2007. Within-plant signaling by volatiles leads to induction and priming of an indirect plant defense in nature. Proc Natl Acad Sci U S A 104:5467-5472. Hilker M, Stein C, Schroder R, Varama M, Mumm R. 2005. Insect egg deposition induces defence responses in Pinus sylvestris: characterisation of the elicitor. J Exp Biol 208:1849-1854. Hiltpold I, Turlings TC. 2008. Belowground chemical signaling in maize: when simplicity rhymes with efficiency. J Chem Ecol 34:628-635. Hughes G. 1999. Sampling for decision making in crop loss assessment and pest management: introduction. Phytopathology 89:1080-1083. Jakobsen HB. 1997. The preisolation phase of in situ headspace analysis: methods and prespectives. In: Linskens HF and Jackson JF, editors. Plant Volatile Analysis. Berlin: Springer-Verlag. p 1-22. Janssen K, Vermeulen K, Boonen C, Bleyaert P, Lemeur R, Berkcmans D. 2004. Introduction to speaking plant: let the crop control the greenhouse climate. Commun Agric Appl Biol Sci 69:151-153.  84  Kant MR, Ament K, Sabelis MW, Haring MA, Schuurink RC. 2004. Differential timing of spider mite-induced direct and indirect defenses in tomato plants. Plant Physiol 135:483-495. Kessler A, Baldwin IT. 2001. Defensive function of herbivore-induced plant volatile emissions in nature. Science 291:2141-2144. Kunert M, Biedermann A, Koch T, Boland W. 2002. Ultra fast sampling and analysis of plant volatiles by a hand-held miniaturized GC with pre-concentration unit: Kinetic and quantitative aspects of plant volatile production. J Separation Sci 25:677-684. Lammertyn J, Veraverbeke EA, Lrudayaraj J. 2004. zNose technology for the classification of honey based on rapid aroma profiling. Sensors and Actuators B: Chemical 98:544-62. Lennert A, Nielsen F, Breum NO. 1997. Evaluation of evaporation and concentration distribution models--a test chamber study. Ann Occup Hyg 41:625-41. Lichtenberg E, Berlind AV. 2005. Does it matter who scouts? J Agric Resource Econ 30:250-267. Lu CJ, Jin C, Zellers ET. 2006. Chamber evaluation of a portable GC with tunable retention and microsensor-array detection for indoor air quality monitoring. J Environ Monit 8:270-278. Maffei ME, Mithofer A, Boland W. 2007. Before gene expression: early events in plantinsect interaction. Trends Plant Sci 12:310-316.  85  Mathieu S, Cin VD, Fei Z, Li H, Bliss P, Taylor MG, Klee HJ, Tieman DM. 2009. Flavour compounds in tomato fruits: identification of loci and potential pathways affecting volatile composition. J Exp Bot 60:325-337. Matsui K. 2006. Green leaf volatile: hydroperoxide lyase pathway of oxylipin metabolism. Current Opin Plant Biol 9:274-280. Mattiacci L, Rocca BA, Scascighini N, D'Alessandro M, Hern A, Dorn S. 2001. Systemically induced plant volatiles emitted at the time of "danger". J Chem Ecol 27:2233-2252. Mayer F, Takeoka GR, Buttery RG, Whitehand LC, Naim M, Rabinowitch HD. 2008. Studies on the aroma of five fresh tomato cultivars and the precursors of cis- and trans-4,5-epoxy-(E)-2-decenals and methional. J Agric Food Chem 56:37493757. McCornack BP, Costamagna AC, Ragsdale DW. 2008. Within-plant distribution of soybean aphid (Hemiptera: Aphididae) and development of node-based sample units for estimating whole-plant densities in soybean. J Econ Entomol 101:14881500. Navia-Gine WG, Yuan JS, Mauromoustakos A, Murphy JB, Chen F, Korth KL. 2009. Medicago truncatula (E)-!-ocimene synthase is induced by insect herbivory with corresponding increases in emission of volatile ocimene. Plant Physiol Biochem 47:416-425. Neveu N, Grandgirard J, Nenon JP, Cortesero AM. 2002. Systemic release of herbivoreinduced plant volatiles by turnips infested by concealed root-feeding larvae Delia radicum L. J Chem Ecol 28:1717-1732.  86  Oerke EC, Steiner U, Dehne HW, Lindenthal M. 2006. Thermal imaging of cucumber leaves affected by downy mildew and environmental conditions. J Exp Bot 57:2121-2132. Oh SY, Ko JW, Jeong SY, Hong J. 2008. Application and exploration of fast gas chromatography-surface acoustic wave sensor to the analysis of thymus species. J Chromatogr A 1205:117-127. Palazzo MC, Setzer WN. 2009. Monoterpene hydrocarbons may serve as antipredation defensive compounds in Boisea trivittata, the box elder bug. Nat Prod Commun 4:457-459. Pasini P, Powar N, Gutierrez-Osuna R, Daunert S, Roda A. 2004. Use of a gas-sensor array for detecting volatile organic compounds (VOC) in chemically induced cells. Anal Bioanal Chem 378:76-83. Peladan F, Turlot JC, Monteil H. 1984. Discriminant analysis of volatile fatty acids produced in culture medium: a novel approach to the identification of Pseudomonas species. J Gen Microbiol 130:3175-3182. Pickett JA, Chamberlain K, Poppy GM, Woodcock CM. 1999. Exploiting insect responses in identifying plant signals. Novartis Found Symp 223:253-262; discussion 262-5, 266-9. Rasmann S, Kollner TG, Degenhardt J, Hiltpold I, Toepfer S, Kuhlmann U, Gershenzon J, Turlings TC. 2005. Recruitment of entomopathogenic nematodes by insectdamaged maize roots. Nature 434:732-737. Rose US, Tumlinson JH. 2004. Volatiles released from cotton plants in response to Helicoverpa zea feeding damage on cotton flower buds. Planta 218:824-832.  87  Rose US, Tumlinson JH. 2005. Systemic induction of volatile release in cotton: how specific is the signal to herbivory? Planta 222:327-335. Schoonhoven LM, van Loon JJA, Dicke M. 2006. Insect-Plant Biology. New York: Oxford University Press. 421 p. Schroder R, Forstreuter M, Hilker M. 2005. A plant notices insect egg deposition and changes its rate of photosynthesis. Plant Physiol 138:470-477. Schwartzberg EG, Kunert G, Stephan C, David A, Rose US, Gershenzon J, Boland W, Weisser WW. 2008. Real-time analysis of alarm pheromone emission by the pea aphid (Acyrthosiphon pisum) under predation. J Chem Ecol 34:76-81. Shiojiri K, Kishimoto K, Ozawa R, Kugimiya S, Urashimo S, Arimura G, Horiuchi J, Nishioka T, Matsui K, Takabayashi J. 2006. Changing green leaf volatile biosynthesis in plants: an approach for improving plant resistance against both herbivores and pathogens. Proc Natl Acad Sci U S A 103:16672-16676. Siripatrawan U, Linz JE, Harte BR. 2004. Solid-phase microextraction, gas chromatography, and mass spectrometry coupled with discriminant factor analysis and multilayer perceptron neural network for detection of Escherichia coli. J Food Prot 67:1597-1603. Skaloudova B, Krivan V, Zemek R. 2006. Computer-assisted estimation of leaf damage caused by spider mites. Comp Electron Agric 53:81-91. Soubeyrand S, Neuvonen S, Penttinen A. 2009. Mechanical-statistical modeling in ecology: from outbreak detections to pest dynamics. Bull Math Biol 71:318-38. Trewavas A. 2003. Aspects of plant intelligence. Ann Bot (Lond) 92:1-20.  88  Turlings TC, Loughrin JH, McCall PJ, Rose US, Lewis WJ, Tumlinson JH. 1995. How caterpillar-damaged plants protect themselves by attracting parasitic wasps. Proc Natl Acad Sci U S A 92:4169-4174. Turlings TC, Tumlinson JH. 1992. Systemic release of chemical signals by herbivoreinjured corn. Proc Natl Acad Sci U S A 89:8399-8402. Turlings TC, Tumlinson JH, Lewis WJ. 1990. Exploitation of herbivore-induced plant odors by host-seeking parasitic wasps. Science 250:1251-1253. Van Poecke RM, Posthumus MA, Dicke M. 2001. Herbivore-induced volatile production by Arabidopsis thaliana leads to attraction of the parasitoid Cotesia rubecula: chemical, behavioral, and gene-expression analysis. J Chem Ecol 27:1911-1928. Verheggen FJ, Arnaud L, Bartram S, Gohy M, Haubruge E. 2008. Aphid and plant volatiles induce oviposition in an aphidophagous hoverfly. J Chem Ecol 34:301307. Volkov AG, Ranatunga DRA. 2006. Plants as environmental biosensors. Plant Sig Behav 1:105-115. Watkins P, Wijesundera C. 2006. Application of zNose for the analysis of selected grape aroma compounds. Talanta 70:595-601. Yu H, Yang J. 2001. A direct LDA algorithm for high-dimensional data with application to face recognision. Pattern Recognition 34:2067-2070. Zhong Q, Steinecker WH, Zellers ET. 2009. Characterization of a high-performance portable GC with a chemiresistor array detector. Analyst 134:283-293.  89  CHAPTER FOUR: EFFECT OF PEST DENSITY, DISTRIBUTION OF PESTS IN THE PLANT CANOPY AND DAMAGE DURATION ON HERBIVORE-INDUCED PLANT VOLATILE EMISSION RATE IN THE CABBAGE LOOPER-TOMATO SYSTEM10  ="!#0*',)/45'+)*# In previous chapters, I discussed the importance of the greenhouse vegetable industry as a growing segment of Canadian agriculture. I mentioned that due to the current global economic crisis and its significant effects on the greenhouse tomato market, it is very important to put more effort on protecting our crops and reducing the crop loss caused by pests and diseases. I also talked about pest monitoring as an important pillar of a successful IPM program and discussed the possibility of using herbivore-induced plant volatiles as an indicator of pest presence for pest monitoring. I introduced an ultra fast gas chromatograph (zNose!) that can be used for monitoring changes in the HIPVs in the field and finally, I showed that we could discriminate clean and infested plants based on their volatile emission pattern both in the laboratory and in a research greenhouse at early stages of damage. In this chapter I took this concept to a different level and investigated the possibility of obtaining additional information about the pest population by monitoring emission of HIPVs. My focus was on the pest population density, distribution of the pest within plant canopy and damage duration.  10  A version of this chapter has been submitted for publication. Miresmailli S., Gries G.J., Gries R. M., Zamar, R.H. and Isman M.B. 2009. Effect of pest density, distribution of pests in the plant canopy and damage duration on herbivore-induced plant volatile emission rate in the cabbage looper-tomato system.  90  Martel and Malcolm (2004) investigated the density effects of the oleander aphid, Aphis nerii, on cardenolide expression in two milkweed species, Asclepias curassavica and A. incarnata. They found that cardenolide concentration (µg/g) of A. curassavica in both aphid-treated leaves and opposite, herbivore-free leaves decreased initially in comparison with aphid-free controls, and then increased significantly with A. nerii density. They found significantly higher cardenolide concentration in aphid-treated leaves compared to herbivore-free leaves. On the other hand, by studying the same aphid species on 18 species of milkweed, Asclepias spp., Agrawal (2004) showed that plant traits that probably evolved for primary and defensive functions contribute to the ecological dynamics of herbivore populations. I also studied the effect of pest density on volatile emission of tomato plants. My objective was to obtain information about the infestation level through monitoring volatile emission levels over time. The other objective was to obtain information about the location of pest within the plant canopy which can be important for designing an effective pest management plant. According to optimal defense theory (ODT) plants should invest in stronger defense in the most valuable parts, such as reproductive or young tissues (Pavia et al. 2002). Anderson and Agrell (2005) found an increase in the concentration of terpenoids in developing leaves of cotton plants (both at the top of the plant and on the side shoots) in response to feeding by larvae of the generalist moth Spodoptera littoralis. I designed an experiment to investigate the difference in volatile emission in response to infestation on upper or lower part of tomato plant canopy. Plants emit HIPVs at different rates. Some of these volatile chemicals are emitted within minutes after tissue damage and can be considered a quick indicator of problems (Heil  91  and Silva Bueno 2007), while other chemicals are released later as a complement to other types of defense (Kant et al. 2004). My last question was about persistence of plant responses after the herbivore leaves the plant. It is of great importance to know if plants change their responses when there is no more damage from the herbivore. I used cabbage looper as the model pest and greenhouse tomato as the model plant.  ="1#(.'%,+.2&#.*/#(%'@)/&# 4.2.1 Cabbage looper Cabbage loopers (Trichoplusia ni, Noctuidae) were obtained from a research colony maintained for 50 generations without any pesticide exposure at the University of British Columbia.  4.2.2 Plant material Tomato plants (Lycopersicon esculentum Mill cv. Clearence) were obtained from Houweling’s Hot House (Delta, BC). Intact tomato plants (30-40 cm high) were selected carefully and stored in a growth chamber at 25 °C ±1 and a 16:8h L:D photoperiod.  4.2.3 Identifying indicator chemicals Indicator HIPVs were collected and identified by GC-MS in the laboratory as explained in Chapter Three.  92  4.2.4 Greenhouse experiment After I identified four major indicator chemicals in infested plant volatiles, I designed three experiments inside the UBC Horticulture Greenhouse. The main objectives of these experiments were to investigate the differences in volatile emission of plants using the zNose!in response to (a) different densities of pests T. ni, (b) different distribution of larvae within the plant canopy and (c) feeding duration. In each set of experiments, tomato plants (eight weeks old) were placed on a greenhouse bench inside the greenhouse 48 hours prior to the experiment. The waiting period was included to allow the plants to acclimatize to the greenhouse conditions. From these plants, those that had no sign of mechanical or pest damage were selected for the experiment. The plants were then spaced from each other (~50cm) on the greenhouse bench (45 by 1.5 meter). Twenty plants were randomly assigned to each treatment group and 20 plants were assigned as controls (completely randomized design- total number of plants varied from 60 to 80 plants based on the experiment). Above each plant I hung a clear plastic bag (Fisher precision ultraclean Bag, Fisher Scientific, Canada) that covered the upper half of the plant canopy. I made a cross-shaped frame with wooden sticks wrapped in aluminum foil and attached it to the exterior upper part of the bags. The frame kept the bag in a cube-shaped form, which minimized contact of the plant part with the bag and therefore, minimized condensation. I covered the lower part of the canopy by a fine mesh screen which allow ventilation but prevented pests from leaving the plants in treatment groups (Fig 3.3). I attached a short tube to the upper part of the bag through a small hole (no glue or tape was used) to form an opening for volatile collection.  93  To eliminate the effect of size variation in the amount of volatiles emitted from plants, I measured the level of all indicator chemicals in all treatment and control groups one hour before introducing T. ni to the treatment groups. I formed a baseline for each plant and compared changes in volatile emission with this baseline. In addition to this technique, I measured the fresh weight of all plants (including the feces of cabbage loopers in treatment groups) immediately after each experiment. As a negative control, I hung 10 empty bags and screens randomly in various locations among the bags containing the plants. I compared the level of indicator volatiles at the baseline in control, treatment and negative control (empty bags) groups and analyzed them by ANOVA.  4.2.5 Experiment one: Effect of infestation levels In experiment one, I had three treatment groups and one control group. After establishing the baseline in all groups, I placed one, three or nine third-instar T.ni larvae (starved for 3 hours) on each plant in the treatment groups (Fig 4.1). I then measured the level of indicator chemicals in all groups 6, 12 and 24 hours after placement of larvae. This experiment was repeated two times.  0  1  3  9  FIG 4.1 EXPERIMENT ONE: EFFECT OF DIFFERENT PEST DENSITIES ON THE VOLATILE EMISSION OF TOMATO PLANTS. PESTS WERE PLACED ON UPPER PART OF CANOPY. 0= CONTROL, 1,3,9 = NUMBER OF PESTS PER PLANT  94  4.2.6 Experiment two: effect of larval distribution within the plant canopy In experiment two, I had two treatment groups (pests in the upper canopy and pests in the lower canopy) and one control group without pest. After establishing the baseline in all groups, I placed 10 third-instar T. ni larvae (starved for 3 hours) on each plant in the treatment groups. In treatment groups, lower or upper parts of the canopy were covered by additional screens to limit the movement of pests on the plants (Fig 4.2). I then measured the level of indicator chemicals in all groups 6, 12 and 24 hours after placement of larvae. This experiment was repeated two times.  U L  FIG 4.2 EFFECT OF LARVAL DISTRIBUTION WITHIN PLANT CANOPY. PESTS WERE PLACED ON UPPER OR LOWER PART OF TOMATO PLANT CANOPY . THE INFESTED AREA WAS COVERED WITH A SCREEN TO LIMIT PEST MOVEMENT. U=UPPER, L= LOWER  4.2.7 Experiment three: effect of feeding duration by larvae In experiment three, I had three treatment groups and one control group. After establishing the baseline in all groups, I placed 20 third-instar T. ni larvae (starved for 3 hours) on each plant in the treatment groups. In treatments 1, 2 and 3, larvae were removed 6,12 and 24 hours, respectively, after their placement on plant. No larvae were placed on control groups (Fig 4.3). I measured the level of indicator chemicals in all groups 6, 12 and 24 hours after placement of larvae. This experiment was repeated two times.  95  P  P 6hr  P 12hr  24hr  FIG 4.3 EFFECT OF FEEDING DURATION BY LARVAE ON VOLATILE EMISSION LEVEL OF TOMATO PLANTS. P=PEST ( CABBAGE LOPPER LARVAE) REMOVED AFTER 6, 12 OR 24 HOURS  4.2.8 zNose! calibration and program properties The system was tuned with alkane standards (C4-C24 as explained in Chapter Two) prior to each experiment. Authentic standards ((Z)-3-hexenyl acetate (obtained by esterification of (Z)-3-hexen-1-ol (Aldrich 98%) with acetic anhydride and pyridine), limonene (Aldrich 96%), ocimene (International Flavour and Fragrances, 69% trans, 27% cis), !caryophyllene (Sigma 98%)) were obtained from the Gries-laboratory (Simon Fraser University, Burnaby, BC, Canada). A stock solution was prepared by diluting 1 g (weighed to the nearest 0.1 mg) of each authentic standard into 50 ml of HPLC grade npentane (total of four working stocks). An aliquot (100 µl) of each working stock was pipetted into a 10-ml volumetric flask and diluted to the mark with HPLC grade npentane. The flask was then shaken and sonicated for 5 min to facilitate complete dissolution. A subsample (0.5 ml) was transferred to a septum-capped GC vial for calibration. Four different volumes of reagent standards were injected into the zNose! using the zNose ! 3500 sample injector (0.1 µl, 0.2 µl, 0.4 µl and 0.8 µl- five times per volume). Calibration curves were developed for each chemical. The system was completely baked, cleaned and calibrated for each experiment. The zNose! was programmed as follows: The inlet temperature was 200°C, the valve temperature was 165°C, and the initial column temperature was 40°C. During analysis 96  the column temperature was ramped at the rate of 10°C per second to reach a final column temperature of 200°C. The SAW sensor was operated at 50°C and the trap was operated at 250°C. Helium flow was set at 3.00 ccm. The sampling was set for 10 seconds (Sample flow 20 ccm) after which the system switched to 20 sec of data acquisition mode After each data sampling period the system needed a 30-sec baking period, in which the sensor was heated briefly to 150°C and after which the temperature conditions of the inlet, column, and sensor were reset to the initial conditions. In between each sample measurement at least one blank (one ml HPLC grade n-pentane in a 40ml GC vial with silicon septa) was run to ensure cleaning of the system and a stable baseline.  4.2.9 Data analyses 4.2.9.1 Chemical baselines before infestation Chemical levels at one hour before infestation in control and treatment groups were analyzed by ANOVA (SPSS ver.16).  4.2.9.2 Comparing level of chemicals in control and treatment groups A generalized estimating equations (GEE) technique was used to analyze the data (Using R ver. 2.9.0) 11(Ballinger 2004). Time was considered as a factor (with 3 levels) as opposed to a continuous variable. This allowed more flexibility for my model (I didn't assume a monotone linear effect).  11  GEE analysis has been conducted by the statistical consulting and research laboratory (SCARL) at the Department of Statistics, The University of British Columbia.  97  ="8#E%&42'&# 4.3.1 Identifying indicator chemicals As explained in the previous chapter, I selected and identified four chemicals as indicators of T. ni infestation (Fig 3.4): (Z)-3-hexenyl acetate, (E)-!-ocimene, limonene and !-caryophyllene. All of these chemicals have been previously reported in volatile blends of tomato plants ( Buttery et al. 1987; Baldwin et al. 2008; Mayer et al. 2008; Mathieu et al. 2009). The following model was fitted to the data to explain chemicals changes over time with consideration of the treatment that plants received: "Chemical = # 0 + #1 (treatment1) + # 2 (treatment2) + # 3 (treatment3) + # 4 (time1) + # 5 (time2) + error  !  For everything being held constant "1 for example, represents the change in the chemical level with a unit change in treatment. We want to see which beta is significantly different ! from zero. None of the indicator volatiles were detected in empty bags. No significant  differences among volatile levels were found at the baseline. Please note that in greenhouse experiments, time was not a continuous variable. The line in the graphs is simply an aid for following the direction of changes in chemical volatiles. The level of (Z)-3-hexenyl acetate emission (Fig 4.4) had a direct and statistically significant correlation with pest density and time (Table 4.1). Higher densities of pests induced greater emissions of (Z)-3-hexenyl acetate and the level increased over time.  98  4.3.2 Experiment one: effect of infestation level on volatile emission  (Z)-3-hexenyl acetate  limonene  (E)- !-ocimene  !-caryophellene  FIG 4.4 VOLATILE EMISSION LEVEL (NANOGRAM) IN CONTROL AND TREATMENT TOMATO PLANTS OVER TIME (HOURS AFTER PLACEMENT OF LARVAE) IN RESPONSE TO DIFFERENT CABBAGE LOOPER INFESTATION LEVEL (MEAN±SD)  99  TABLE 4.1. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER INFESTATION LEVEL ON THE (Z)-3-HEXENYL A CETATE EMISSION FROM TOMATO PLANTS OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  28.562  1.803  251  p<0.001  Treatment 1 (one larva)  60.240  1.170  2651  p<0.001  Treatment 2 (three larvae)  111.630  1.017  12037  p<0.001  Treatment 3 (nine larvae)  233.928  0.924  64062  p<0.001  Time1 (6-12)  40.513  2.110  369  p<0.001  Time2 (12-24)  67.244  3.394  392  p<0.001  In the case of (E)-!-ocimene, the emission level in treatments with one or three pests was lower than that in the control plants for up to 12 hours after infestation (Fig 4.4). Six hours after infestation, the level of (E)-!-ocimene in treatments with one larva was 94.27± 10.60 ng below the baseline and in treatments with three larvae the level was 34.925±5.43 ng below the baseline. At the same time, the level of (E)-!-ocimene in control plants after 6 hours was 0.457±5.35 below the baseline. In contrast, the level of (E)-!-ocimene in treatments with nine larvae was higher than in control plants and increased with time. Despite the differences in emission pattern, the volatile levels in treatment and control groups were significantly different at each time point (Table 4.2).  100  TABLE 4.2. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER INFESTATION LEVEL ON THE (E)-!-OCIMENE EMISSION FROM TOMATO PLANTS OVER TIME  Estimate  Std Error  Wald test  Sig  -193.15  12.19  251  p<0.001  Treatment 1 (one larva)  68.78  2.46  781  p<0.001  Treatment 2 (three larvae)  191.87  3.46  3069  p<0.001  Treatment 3 (nine larvae)  401.11  5.32  5683  p<0.001  Time1 (6-12)  109.77  9.86  124  p<0.001  Time2 (12-24)  457.36  23.71  372  p<0.001  Intercept  The limonene emission pattern was quite similar to that for (E)-!-ocimene with a significant decrease in emission level in treatments with one or three larvae and an increase in the emission level in treatments with nine larvae. The difference was that it took about 24 hours in treatments with one larva for the emission level to pass the control emission level while in treatments with three larvae, the emission level passed the control level before 12 hours of infestation (Fig 4.4). The emission levels were significantly different in treatment and control groups at each time point (Table 4.3).  101  TABLE 4.3. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER INFESTATION LEVEL ON THE LIMONENE EMISSION FROM TOMATO PLANTS OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  -61.51  3.26  357  p<0.001  Treatment 1 (one larva)  -65.42  1.52  1848  p<0.001  Treatment 2 (three larvae)  15.11  1.22  154  p<0.001  Treatment 3 (nine larvae)  176.17  1.19  21925  p<0.001  Time1 (6-12)  64.29  3.55  329  p<0.001  Time2 (12-24)  123.76  6.05  418  p<0.001  There was a similar pattern in the emission of !-caryophyellene (Fig 4.6). Six hours after infestation, the emission rate was lower in treatments with one larva (572.23 ± 4.70 ng below baseline) and three larvae (202.88 ± 5.42 ng below baseline) compared to the control group (4.78 ± 3.02ng above baseline). Twelve hours after infestation, the emission level in treatments with three larvae was greater than in the control groups (3135.72 ± 1018.80 ng) but in treatments with one larva it was still below the control level (358.60 ± 38.77 ng below baseline). At the same time, in the treatment group that received nine larvae, the emission level was higher than the control group at all times. Despite the unusual variation in volatile emission in the treatment with nine larvae at 6 hours after infestation (2495.30 ± 2024.92 ng), the volatile levels were significantly different in all treatment and control groups at all time points (Table 4.4). This variation never occurred in other experiments.  102  TABLE 4.4. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER INFESTATION LEVEL ON THE !-CARYOPHELLENE EMISSION FROM TOMATO PLANTS OVER TIME.  Estimate  Std Error  Wald test  Sig  -1788.94  148.22  146  p<0.001  -36.02  2.93  151  p<0.001  Treatment 2 (three larvae)  2335.28  60.79  1476  p<0.001  Treatment 3 (nine larvae)  6581.37  111.53  3482  p<0.001  Time1 (6-12)  2435.45  228.81  113  p<0.001  Time2 (12-24)  2848.58  206.95  189  p<0.001  Intercept Treatment 1 (one larva)  4.3.3 Experiment two: effect of larval distribution within plant For all indicator chemicals, treatment had a statistically significant effect on the emission level of volatiles. In the case of (Z)-3-hexenyl acetate and limonene, infestation in the upper canopy induced a slightly higher emission of the volatiles at each time point (~35 to 50 ng) but in the case of (E)-!-ocimene and !-caryophellene, despite the statistical significance, the difference was very small (~5 to 10 ng) (Fig 4.5).  103  (Z)-3- hexenyl acetate  limonene  (E)- !-ocimene  !-caryophyllene  FIG 4.5 VOLATILE EMISSION LEVEL (NANOGRAM) IN CONTROL AND TREATMENT TOMATO PLANTS OVER TIME (HOURS AFTER PLACEMENT OF LARVAE) IN RESPONSE TO DIFFERENT CABBAGE LOOPER DISTRIBUTION WITHIN THE PLANT CANOPY (MEAN±SD)  104  TABLE 4.5. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER DISTRIBUTION WITHIN THE TOMATO PLANT CANOPY ON THE EMISSION LEVEL OF (Z)-3-HEXENYL ACETATE OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  41.690  3.898  114.4  p<0.001  Treatment 1 Upper canopy  242.978  1.118  47251.8  p<0.001  Treatment 2 Lower canopy  212.342  1.304  26529.3  p<0.001  51.3  4.649  121.8  p<0.001  85.750  7.717  123.5  p<0.001  Time 1 (6-12) Time 2 (12-24)  In the case of (Z)-3-hexenyl acetate, the coefficient estimates in treatment 1 and treatment 2 do not overlap (242.978 ± 1.118 in upper canopy and 212.342 ± 1.304 in lower canopy).  TABLE 4.6. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER DISTRIBUTION WITHIN THE TOMATO PLANT CANOPY ON THE EMISSION LEVEL OF (E)-!-OCIMENE OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  -272.77  26.04  110  p<0.001  Treatment 1 Upper canopy  440.23  2.24  32977  p<0.001  Treatment 2 Lower canopy  414.64  2.31  32205  p<0.001  Time 1 (6-12)  240.47  21.99  120  p<0.001  Time 2 (12-24)  575.57  52.37  121  p<0.001  In the case of (E)-!-ocimene, although the level of chemicals are very close to each other in both treatment groups, their coefficients still do not overlap (440.23 ± 2.24 in upper 105  canopy and 414.64 ± 2.31 in lower canopy) and both treatments’ coefficients were significantly different than zero therefore; they both had significant effect on volatile emission. Please note that in these experiments, time is a variable itself and it is not continuous. The lines in graphs are only visual aids so one can follow the pattern of change in emission level of volatiles.  TABLE 4.7. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER DISTRIBUTION WITHIN THE TOMATO PLANT CANOPY ON THE EMISSION LEVEL OF LIMONENE OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  -81.81  7.18  130  p<0.001  Treatment 1 Upper canopy  187.87  1.27  21982  p<0.001  Treatment 2 Lower canopy  162.67  1.37  14192  p<0.001  Time 1 (6-12)  99.72  8.92  125  p<0.001  Time 2 (12-24)  157.17  13.93  127  p<0.001  Coefficients in limonene also do not overlap and both treatments had significant effects on the volatile emission.  106  TABLE 4.8. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER DISTRIBUTION WITHIN THE TOMATO PLANT CANOPY ON THE EMISSION LEVEL OF !-CARYOPHYLLENE OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  -3931.20  358.62  120  p<0.001  Treatment 1 Upper canopy  5903.41  1.19  24470091  p<0.001  Treatment 2 Lower canopy  5880.14  1.32  19927975  p<0.001  Time1 (6-12)  5844.92  533.41  120  p<0.001  Time2 (12-24)  5676.07  517.50  120  p<0.001  In the case of !-caryophellene, although emission level is very close in upper and lower canopy, but statistically, the coefficients do not overlap (5903.41 ± 1.19 in upper canopy and 5880.14 ± 1.32) and both treatments had significant effect on the emission of !caryophellene.  107  4.3.3 Experiment three: effect of feeding duration by larvae (Z)-3-hexenyl acetate  limonene  (E)- !-ocimene  !-caryophyllene  FIG 4.6 VOLATILE EMISSION LEVEL (NANOGRAM) IN CONTROL AND TREATMENT TOMATO PLANTS OVER TIME (HOURS AFTER PLACEMENT OF LARVAE) IN RESPONSE TO DIFFERENT FEEDING DURATION OF CABBAGE LOOPER LARVAE (MEAN±SD)  Removing larvae at different times had significant effects on the level of volatile emission for all indicator chemicals (Fig 4.6). In all cases, the level of volatile emissions 108  dropped after the larvae were removed but it never reached the level of intact plants. In the cases of (E)-!-ocimene and !-caryophyllene it was clear that there was no significant difference between treatments before pests were removed from one of the treated groups (Table 4.10 and 4.12)  TABLE 4.9. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER FEEDING DURATION ON THE EMISSION LEVEL OF (Z)-3-HEXENYL ACETATE FROM TOMATO PLANTS OVER TIME  Estimate  Std Error  Wald test  Sig  7.4  15.9148  2.160e-01  p<0.001  Treatment 1 (removed-6hr)  305.8579  0.9169  1.113e+05  p<0.001  Treatment 2 (removed-12hr)  627.4453  0.9445  4.413e+05  p<0.001  Treatment 3 (removed-24hr)  666.9968  0.7812  7.291e+0.5  p<0.001  Time1 (6-12)  112.3625  29.6081  1.440e+01  p<0.001  Time2 (12-24)  -167.1500  15.1727  1.214e+02  p<0.001  Intercept  For all indicator chemicals, all treatment coefficients are significantly different than zero. Among treatments, treatment three has a greater effect on the emission of volatiles.  109  TABLE 4.10. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER FEEDING DURATION ON THE EMISSION LEVEL OF (E)-"-OCIMENE FROM TOMATO PLANTS OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  127.14  40.67  9.78e+00  p<0.001  Treatment 1 (removed-6hr)  1948.30  6.38  9.34e+04  p<0.001  Treatment 2 (removed-12hr)  2689.12  7.44  1.31e+05  p<0.001  Treatment 3 (removed-24hr)  2792.26  4.29  4.24e+05  p<0.001  Time1 (6-12)  -54.76  46.86  1.37e+00  0.2425  Time2 (12-24)  -318.11  71.51  1.98e+01  p<0.001  TABLE 4.11. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER FEEDING DURATION ON THE EMISSION LEVEL OF LIMONENE FROM TOMATO PLANTS OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  -105.14  13.47  60.9  p<0.001  Treatment 1 (removed-6hr)  328.18  2.23  21586.2  p<0.001  Treatment 2 (removed-12hr)  543.82  2.80  37781.9  p<0.001  Treatment 3 (removed-24hr)  602.38  1.97  93585.8  p<0.001  Time1 (6-12)  156.36  18.46  71.7  p<0.001  Time2 (12-24)  176.22  22.91  59.2  p<0.001  110  TABLE 4.12. GEE REGRESSION COEFFICIENTS- EFFECT OF CABBAGE LOOPER FEEDING DURATION ON THE EMISSION LEVEL OF !-CARYOPHYLLENE FROM TOMATO PLANTS OVER TIME  Estimate  Std Error  Wald test  Sig  Intercept  1398.87  418.91  1.12e+01  p<0.001  Treatment 1 (removed-6hr)  7883.70  2.47  1.02e+07  p<0.001  Treatment 2 (removed-12hr)  15110.76  2.90  2.72e+07  p<0.001  Treatment 3 (removed-24hr)  17740.21  1.42  1.55e+08  p<0.001  Time1 (6-12)  -531.76  586.07  8.20e-01  0.36423  Time2 (12-24)  -3764.50  705.76  2.84e+01  p<0.001  111  ="=#F+&54&&+)*# In this chapter I explored the possibility of using HIPV-based indicators for detecting infestations by T. ni larvae and obtaining additional information about the pest population and severity of damage. I looked at the effect of pest infestation level, distribution of pests within the plant canopy and feeding duration on the emission of indicator volatiles. I will discuss the results by looking at each individual indicator compound. The first compound is (Z)-3-hexenyl acetate which is a green leaf volatile. Green leaf volatiles (GLVs) are C6 aldehydes, alcohols, and their esters formed through the hydroperoxide lyase pathway of oxylipin metabolism. Plants start to form GLVs after disruption of their tissues and after suffering biotic or abiotic stresses (Maes and Debergh 2003; Matsui 2006). GLVs have been shown to significantly activate jasmonic acid production in exposed corn (Zea mays) seedlings, thereby priming these plants specifically against subsequent herbivore attack(Engelberth et al. 2007). In addition to the major ecological functions of GLV biosynthesis which is related to resistance against both herbivores and pathogens, the genetic modification of GLV biosynthesis could be a unique approach for improving plant resistance against such biotic stresses (Shiojiri et al. 2006). In the density experiment, I observed a direct link between numbers of pests and the emission level of (Z)-3-hexenyl acetate. It seems that higher feeding pressure by T. ni translated into higher damage to the cell membranes and consequently led to greater amounts of (Z)-3-hexenyl acetate (Fig.4.4). This effect became clearer in the feeding duration trial where there was a sharp drop in the emission level of (Z)-3-hexenyl acetate in treatments where pests were removed after 6 hours (Fig 4.6), going from 523.75 ± 3.78ng at 6 hours after infestation to 172.70 ± 4.47ng at 24 hours (12 hours after I  112  removed the pests from the plants). In experiment two (larvae distribution within the plant canopy), I observed a slight difference in the level of (Z)-3-hexenyl acetate in plants that received damage in the upper canopy compared to ones that were infested in the lower part. The result might suggest a slightly higher concentration of (Z)-3-hexenyl acetate in the young leaves but the plants I used were relatively short (about a meter tall) compared to those in commercial greenhouses and the lower canopy of the plant was not significantly different from the upper canopy in these young plants. Overall, (Z)-3hexenyl acetate alone might not be a reliable indicator for specific pest damage. However, in combination with other compounds it might be a good indicator of pest presence. For example (Deng et al. 2005) used three C6-aldehydes, hexanal, (Z)-3hexenal, and (E)-2-hexenal to study defensive responses of tomato plants to Helicoverpa armigera. The second indicator chemical is (E)-!-ocimene which is an acyclic monoterpene. Terpenes are the most abundant and varied class of HIPVs. For instance, it has been shown that in response to the beet armyworm Spodoptera exigua (Hubner), the volatile (E)-!-ocimene was released from leaves of both undamaged and insect-damaged Medicago truncatula plants, but at higher levels in insect-damaged plants. The higher emission rate of the volatile (E)-!-ocimene, suggests that it plays an active role in indirect insect defenses in M. truncatula (Navia-Gine et al. 2009). The defensive property of (E)-!-ocimene is not limited to plants. It has been shown that (E)-!-ocimene may provide some protection of box elder bugs from predation (Palazzo and Setzer 2009). It has also been shown that glandular trichomes of tomato plants store considerable amounts of mono- and sesquiterpenes (van Schie et al. 2007). In  113  experiment one (different infestation level) I observed an interesting phenomenon. If we assume that the emission of (E)-!-ocimene is only caused by mechanical aspects of feeding via damaging the trichomes, then we should expect to see an increasing pattern of emission in response to number of feeding larvae as I observed for (Z)-3-hexenyl acetate. However, as seen in Figure 4.4, in treatments infested with one or three larvae, there was a significant decrease in the amount of (E)-!-ocimene within the first 12 hours after infestation, 94.27 ± 10.60 ng below the baseline in treatments with one larva and 34.92 ± 5.43ng below the baseline in treatments with three larvae. This was not the case in treatments with nine larvae. In these treatments I observed an increase in the level of (E)!-ocimene over time. This pattern was repeated with other mono- and sesquiterpene indicator compounds (Fig 4.4). Martel and Malcolm (2004) investigated the density effects of the oleander aphid, Aphis nerii, on cardenolide expression in two milkweed species, Asclepias curassavica and A. incarnata. They found that cardenolide concentration (microgram/g) of A. curassavica in both aphid-treated leaves and opposite, herbivore-free leaves decreased initially in comparison with aphid-free controls, and then increased significantly with A. nerii density. In addition, they found significantly higher cardenolide concentration in aphid-treated leaves compared to herbivore-free leaves. I have two theories that might help to explain this observation. Maffei et al. (2007a, 2007b) explained early events that occur in plants when they first encounter herbivores involving rapid signaling via calcium channels. There might be a threshold that needs to be met before defensive responses are activated which would explain the delay in the emission of mono- and sesquiterpenes in treatments with 1-3 feeding larvae. Such delays have been reported before (Kant et al. 2004). In contrast, when the damage is greater,  114  these thresholds would be passed more quickly so there is no delay in the emission of the volatiles. Some greenhouse practices that can cause severe damage were also reported to cause a rapid increase in the emission of these volatiles (Jansen et al. 2009). This might explain the delay in the emission of volatiles but I also noticed a decrease in the amounts of volatiles below the baseline. I do not have any evidence to support the following claim so it is purely speculative. Like other compounds in plants, production of these volatiles is energy intensive. One of the known functions of these volatiles is their indirect defensive function (Arimura et al. 2005). We know that some of these volatiles will be produced de novo or their level of production will increase significantly even in undamaged parts of the plant in response to herbivory ( Turlings and Tumlinson 1992; Turlings et al. 1995; Dicke 1999; Van Poecke et al. 2001; Mattiacci et al. 2001; Neveu et al. 2002; Rose and Tumlinson 2005). The indirect defense via emitting HIPVs is not the only form of defensive mechanism in plants. I have no evidence to prove this claim but I think when the level of threat is low (i.e. one larva), emitting HIPVs might be considered an expensive option so by decreasing the emission level, the plant might invest the excessive energy into another form of defense. However, if the threat persists, then they may shift their defensive mechanism to plan B (emission of HIPVs -indirect defense). This is just a hypothetical explanation. Hare et al. (2003) showed that the production of glandular trichomes for herbivore resistance by Datura wrightii is costly because plants that produce glandular trichomes, a factor conferring resistance to some insect herbivores, also produced 45% fewer seeds than plants that produce non-glandular trichomes when grown in a common garden. Heil and Baldwin (2002) explained that  115  studying fitness costs in plants is relatively difficult due to the lack of precise methods and measurement instruments but this can be an exciting area for further research. In my second experiment with larvae on different parts of the canopy, I observed statistically significant differences in the emission levels of (E)-!-ocimene (95% CI do not overlap); however, as explained before this difference is relatively small due to the fact the plants in the experiment were young and small. In the third experiment (feeding duration), I observed a decrease in the amount of (E)-!-ocimene after the pests were removed but it didn’t reach the level of clean plants within the period of observation. This decrease in the amount of volatiles might be a response to the elimination of mechanical damage caused by pests. It has been shown that continuous mechanical wounding can trigger the same responses as herbivores (Mithofer et al. 2005). It might be useful to follow the emission changes for a longer period of time after removing the pest to determine how long the effect would last. The third indicator compound is limonene which is a cyclic monoterpene. Limonene is also known to have defensive properties against both pests and diseases either directly (Byun-McKay et al. 2006; Ben-Yehoshua et al. 2008) or indirectly (Verheggen et al. 2008). I observed the same pattern in the first experiment (different infestation level) with a longer delay period for limonene emission in the treatments with one larva (Fig 4.4). The second experiment (distribution of larvae within plant canopy) showed a significant effect of treatment on the emission level with greater effect on the upper canopy which might suggest higher concentration/production of limonene in younger leaves. In the feeding duration experiment, the level of emission remains relatively steady with a small slope after removing the pests unlike the !-caryophyellene level which drops sharply  116  after pests are removed. Further studies are required to determine how long plants continue to emit volatiles after pest removal. The last indicator chemical is !-caryophyellene which is a sesquiterpene. !-Caryophyllene has been reported as an herbivore-induced plant volatile both above ground (Abel et al. 2009) and belowground (Hiltpold and Turlings 2008). It has been reported as an abundant sesquiterpene in the blend of tomato volatiles (Vercammen et al. 2001). I only observed a delay period in emission in treatments infested with one larva in the first experiment. In the second experiment, the difference was very small yet both treatments had a statistically significant effect. In the third experiment, a sharp drop in emission was observed after removing the larvae. This might be because of the relatively higher cost of production of this compound compared to a simple monoterpene like limonene. My results indicate the suitability of these volatile chemicals as indicators of pest presence. We can also obtain additional information about the infestation. However, we must consider all the factors that can cause variation in the emission level of these volatiles. Relying on individual volatiles as indicators might be misleading. Therefore, one should look at the pattern of multiple indicators. Further experimentation in commercial greenhouses is necessary to determine the suitability of these volatiles as pest presence indicators and better understand the effect of biotic and abiotic factors on their emission level. My results in this chapter add more insight into the concept but this topic is still at the preliminary stages and more work must be done before we can fully understand all aspects of this system.  117  ="?#E%6%,%*5%&# Abel C, Clauss M, Schaub A, Gershenzon J, Tholl D. 2009. Floral and insect-induced volatile formation in Arabidopsis lyrata ssp. Petraea, a perennial, outcrossing relative of A. thaliana. Planta 230:1-11. Agrawal AA. 2004. Plant defense and density dependence in the population growth of herbivores. Am Nat 164:113-120. Anderson P, Agrell J. 2005. Within-plant variation in induced defence in developing leaves of cotton plants. Oecologia 144:427-434. Arimura G, Kost C, Boland W. 2005. Herbivore-induced, indirect plant defences. Biochim Biophys Acta 1734:91-111. Baldwin EA, Goodner K, Plotto A. 2008. Interaction of volatiles, sugars, and acids on perception of tomato aroma and flavor descriptors. J Food Sci 73:S294-307. Ballinger GA. 2004. Using generalized estimatin equations for longitudinal data analysis. Organizational Res Meth 7:127-150. Ben-Yehoshua S, Rodov V, Nafussi B, Feng X, Yen J, Koltai T, Nelkenbaum U. 2008. Involvement of limonene hydroperoxides formed after oil gland injury in the induction of defense response against penicillium digitatum in lemon fruit. J Agric Food Chem 56:1889-1895. Buttery RG, Liang LC, Light DM, 1987, tomato leaf volaile aroma component. J Agric Food Chem 35: 1039-1042 Byun-McKay A, Godard KA, Toudefallah M, Martin DM, Alfaro R, King J, Bohlmann J, Plant AL. 2006. Wound-induced terpene synthase gene expression in sitka spruce  118  that exhibit resistance or susceptibility to attack by the white pine weevil. Plant Physiol 140:1009-1021. Deng C, Li N, Zhu W, Qian J, Yang X, Zhang X. 2005. Rapid determination of c6aldehydes in tomato plant emission by gas chromatography-mass spectrometry and solid-phase microextraction with on-fiber derivatization. J Sep Sci 28:172176. Dicke M. 1999. Specificity of herbivore-induced plant defences. Novartis Found Symp 223:43-54; discussion 54-9, 160-5. Engelberth J, Seidl-Adams I, Schultz JC, Tumlinson JH. 2007. Insect elicitors and exposure to green leafy volatiles differentially upregulate major octadecanoids and transcripts of 12-oxo phytodienoic acid reductases in Zea mays. Mol Plant Microbe Interact 20:707-716. Hare JD, Elle E, van Dam NM. 2003. Costs of glandular trichomes in Datura wrightii: A three-year study. Evolution 57:793-805. Heil M, Baldwin IT. 2002. Fitness costs of induced resistance: Emerging experimental support for a slippery concept. Trends Plant Sci 7:61-67. Heil M, Silva Bueno JC. 2007. Within-plant signaling by volatiles leads to induction and priming of an indirect plant defense in nature. Proc Natl Acad Sci U S A 104:5467-5472. Hiltpold I, Turlings TC. 2008. Belowground chemical signaling in maize: When simplicity rhymes with efficiency. J Chem Ecol 34:628-635. Jansen RMC, Hofstee JW, Wildt J, Verstappen FWA, Bouwmeeter HJ, Posthumus MA. 2009. Health monitoring of plants by their emitted volatiles: Trichome damage  119  and cell memberane damage are detactable at greenhouse scale. Ann Appl Biol 154:441-452. Kant MR, Ament K, Sabelis MW, Haring MA, Schuurink RC. 2004. Differential timing of spider mite-induced direct and indirect defenses in tomato plants. Plant Physiol 135:483-495. Maes K, Debergh PC. 2003. Volatile emitted from in vitro grown tomato shoots during abiotic and biotic stress. Plant Cell, Tissue Organ Culture 75:73-78. Maffei ME, Mithofer A, Boland W. 2007a. Before gene expression: Early events in plant-insect interaction. Trends Plant Sci 12:310-316. Maffei ME, Mithofer A, Boland W. 2007b. Insects feeding on plants: Rapid signals and responses preceding the induction of phytochemical release. Phytochem 68:29462959. Martel JW, Malcolm SB. 2004. Density-dependent reduction and induction of milkweed cardenolides by a sucking insect herbivore. J Chem Ecol 30:545-561. Mathieu S, Cin VD, Fei Z, Li H, Bliss P, Taylor MG, Klee HJ, Tieman DM. 2009. Flavour compounds in tomato fruits: Identification of loci and potential pathways affecting volatile composition. J Exp Bot 60:325-337. Matsui K. 2006. Green leaf volatile: Hydroperoxide lyase pathway of oxylipin metabolism. Current Opin Plant Biol 9:274-280. Mattiacci L, Rocca BA, Scascighini N, D'Alessandro M, Hern A, Dorn S. 2001. Systemically induced plant volatiles emitted at the time of "Danger". J Chem Ecol 27:2233-2252.  120  Mayer F, Takeoka GR, Buttery RG, Whitehand LC, Naim M, Rabinowitch HD. 2008. Studies on the aroma of five fresh tomato cultivars and the precursors of cis- and trans-4,5-epoxy-(E)-2-decenals and methional. J Agric Food Chem 56:37493757. Mithofer A, Wanner G, Boland W. 2005. Effects of feeding Spodoptera littoralis on lima bean leaves. Ii. Continuous mechanical wounding resembling insect feeding is sufficient to elicit herbivory-related volatile emission. Plant Physiol 137:11601168. Navia-Gine WG, Yuan JS, Mauromoustakos A, Murphy JB, Chen F, Korth KL. 2009. Medicago truncatula (E)-beta-ocimene synthase is induced by insect herbivory with corresponding increases in emission of volatile ocimene. Plant Physiol Biochem 47:416-425. Neveu N, Grandgirard J, Nenon JP, Cortesero AM. 2002. Systemic release of herbivoreinduced plant volatiles by turnips infested by concealed root-feeding larvae Delia radicum l. J Chem Ecol 28:1717-1732. Palazzo MC, Setzer WN. 2009. Monoterpene hydrocarbons may serve as antipredation defensive compounds in Boisea trivittata, the boxelder bug. Nat Prod Commun 4:457-459. Pavia H, Toth G, Aberg P. 2002. Optimal defense theory: Elasticity analysis as a tool to predict intraplant variation in defenses. Ecology 83:891-897. Rose US, Tumlinson JH. 2005. Systemic induction of volatile release in cotton: How specific is the signal to herbivory? Planta 222:327-335.  121  Shiojiri K, Kishimoto K, Ozawa R, Kugimiya S, Urashimo S, Arimura G, Horiuchi J, Nishioka T, Matsui K, Takabayashi J. 2006. Changing green leaf volatile biosynthesis in plants: An approach for improving plant resistance against both herbivores and pathogens. Proc Natl Acad Sci U S A 103:16672-16676. Turlings TC, Loughrin JH, McCall PJ, Rose US, Lewis WJ, Tumlinson JH. 1995. How caterpillar-damaged plants protect themselves by attracting parasitic wasps. Proc Natl Acad Sci U S A 92:4169-4174. Turlings TC, Tumlinson JH. 1992. Systemic release of chemical signals by herbivoreinjured corn. Proc Natl Acad Sci U S A 89:8399-8402. Van Poecke RM, Posthumus MA, Dicke M. 2001. Herbivore-induced volatile production by Arabidopsis thaliana leads to attraction of the parasitoid Cotesia rubecula: Chemical, behavioral, and gene-expression analysis. J Chem Ecol 27:1911-1928. van Schie CC, Haring MA, Schuurink RC. 2007. Tomato linalool synthase is induced in trichomes by jasmonic acid. Plant Mol Biol 64:251-263. Vercammen J, Pham-Tuan H, Sandra P. 2001. Automated dynamic sampling system for the on-line monitoring of biogenic emissions from living organisms. J Chromatogr A 930:39-51. Verheggen FJ, Arnaud L, Bartram S, Gohy M, Haubruge E. 2008. Aphid and plant volatiles induce oviposition in an aphidophagous hoverfly. J Chem Ecol 34:301307.  122  CHAPTER FIVE: EFFECT OF PEST INFESTATION, CROP MAINTENANCE PRACTICES AND ENVIRONMENTAL FACTORS ON VOLATILE EMISSION RATE OF TOMATO PLANTS IN A COMMERCIAL GREENHOUSE12.  ?"!#0*',)/45'+)*# In previous chapters, I explained the importance of the greenhouse vegetable industry as a growing segment of Canadian agriculture. I mentioned that due to the current global economic crisis and its significant effects on the greenhouse tomato market, it is very important to concentrate on protecting our crops and reducing the crop loss caused by pests and diseases. I briefly introduced pest monitoring as an important pillar of a successful IPM program and discussed the possibility of using herbivore-induced plant volatiles (HIPVs) as an indicator of pest presence for pest monitoring. I introduced an ultra fast gas chromatograph (zNose!) that can be used for monitoring changes in the HIPVs in the field. I showed that we could discriminate clean and infested tomato plants that were infested with T. ni based on their volatile emission pattern both in the laboratory and in a research greenhouse at early stages of damage. I then took this concept to a different level and investigated the possibility of obtaining additional information about the pest population by monitoring HIPV emissions. I showed that it is possible to obtain some information about T. ni infestation level, distribution of T. ni within the tomato plant canopy and feeding duration by monitoring HIPVs.  12  A version of this chapter has been submitted for publication. Miresmailli S., Gries G.J., Gries R. M., Zamar, R.H. and Isman M.B. 2009. Effect of pest infestation, crop maintenance practices and environmental factors on volatile emission rate of tomato plants in a commercial greenhouse  123  I was always aware of the fact that what I would find in the laboratory might not necessarily apply to the field where there is greater variability. There are several factors that can affect the accuracy of results when it comes to studying HIPVs in the field (Jakobsen 1997; Gouinguene and Turlings 2002). Some examples of such factors to consider are light intensity (Loreto et al. 2006), relative humidity (Vallat et al. 2005), temperature (Maleknia et al. 2009), and photoperiod (Picone et al. 2004). Therefore, I decided to test this concept inside a commercial greenhouse. Recent improvements in analytical tools enable researchers to collect more accurate data about HIPVs and plant systems within their growing environment in short periods of time (Kunert et al. 2002; Pasini et al. 2004; Lu et al. 2006; Oh et al. 2008; Zhong et al. 2009). I used the zNose! to collect samples in the field (i.e. a commercial greenhouse). I also recorded environmental factors and took note of some greenhouse maintenance practices (i.e. removing shoots or picking fruits) during the sampling period. I had two objectives in this study. First I wanted to know if the concept of detecting pest infestation via monitoring plant volatile emission is applicable in a commercial greenhouse and second, I wanted to study the effect of environmental and maintenance related factors on the emission of volatiles from tomato plants and compare them with T. ni infestation.  124  ?"1#(.'%,+.2&#.*/#(%'@)/&# 5.2.1 Commercial greenhouse I collected all my samples from Houweling’s Hot House (Delta, BC). Samples were collected during the months of June, July and August of 2008.  5.2.2 Model pest I chose the cabbage looper (Trichoplusia ni, Noctuidae) as my model insect. During the sampling process, I fully scouted each sample plant and noted the presence of actual pests or their damage symptoms.  5.2.3 Indicator chemicals The same indicator volatile chemicals that were collected and identified by GC-MS in the laboratory (as explained in Chapter Three) were used for this study: (Z)-3-hexenyl acetate, (E)-!-ocimene, limonene and !-caryophyllene. The zNose! was tuned and calibrated (C4-C24 as explained in Chapter Two) prior to each set of experiments with authentic standards.  5.2.4 zNose! program properties The zNose! was programmed as follows: The inlet temperature was 200°C, the valve temperature was 165°C, and the initial column temperature was 40°C. During analysis the column temperature was ramped at the rate of 10°C per second to reach a final column temperature of 200°C. The SAW sensor was operated at 40°C and the trap was operated at 250°C. Helium flow was set at 3.00 ccm. The sampling was set for 60  125  seconds (Sample flow 20 ccm) after which the system switched to 20 seconds of data acquisition mode. After each data sampling period the system needed a 30-second baking period, in which the sensor was heated briefly to 150°C and after which the temperature conditions of the inlet, column, and sensor were reset to the initial conditions. In order to create consistent sampling patterns, I used a stainless steel funnel (15 cm diameter, 20 cm long) that screwed over the zNose! inlet ensuring a consistent distance from the plants during sampling. I cleaned and baked the funnel at 250°C for two hours prior to each sampling period.  5.2.5 Recording environmental condition Temperature, relative humidity, light intensity and airflow were recorded using a Fisher Scientific Enviro-Meter!.  5.2.6 Selecting sampling sites I was not permitted to conduct an experiment inside this commercial greenhouse. I was also not permitted to infest plants or modify any operational practices. Therefore, this study is a survey instead of a designed or controlled experiment. As part of my research contract, I had access to one section of the greenhouse once a week from 12:00 to 15:00 pm with some gaps due to conflicts with the greenhouse maintenance schedule and my sampling plan. In order to create an unbiased representative sampling strategy, before entering the greenhouse, I randomly selected 30 spots within the working section, using the Google Earth! software (Google Inc. Mountain View, CA) (Fig 5.1).  126  FIG 5.1 SAMPLING SITES IN THE HOUWELING’S HOT HOUSE ,DELTA, BC.  127  Sample  GPS Coordination  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30  49° 3'14.09"N-123° 2'44.70"W 49° 3'11.76"N-123° 2'44.05"W 49° 3'10.41"N-123° 2'43.07"W 49° 3'13.30"N-123° 2'43.06"W 49° 3'14.16"N-123° 2'41.30"W 49° 3'11.20"N-123° 2'41.31"W 49° 3'13.03"N-123° 2'40.51"W 49° 3'10.57"N-123° 2'38.97"W 49° 3'13.82"N-123° 2'39.29"W 49° 3'12.20"N-123° 2'38.50"W 49° 3'10.14"N-123° 2'37.58"W 49° 3'13.57"N-123° 2'37.85"W 49° 3'11.48"N-123° 2'35.89"W 49° 3'14.03"N-123° 2'36.39"W 49° 3'13.27"N-123° 2'34.37"W 49° 3'11.03"N-123° 2'33.50"W 49° 3'14.22"N-123° 2'32.78"W 49° 3'13.19"N-123° 2'31.35"W 49° 3'11.91"N-123° 2'30.41"W 49° 3'14.51"N-123° 2'29.73"W 49° 3'10.33"N-123° 2'29.70"W 49° 3'13.25"N-123° 2'28.69"W 49° 3'11.34"N-123° 2'28.26"W 49° 3'12.20"N-123° 2'27.43"W 49° 3'10.46"N-123° 2'26.26"W 49° 3'14.02"N-123° 2'26.60"W 49° 3'13.44"N-123° 2'24.73"W 49° 3'11.40"N-123° 2'24.64"W 49° 3'11.50"N-123° 2'22.65"W 49° 3'14.18"N-123° 2'22.99"W  FIG 5.2 GPS COORDINATION OF SAMPLING SITES- SAMPLING TEAM  I then extracted the GPS coordinates from the software and found the sampling sites inside the greenhouse using a Garmin GPSmap 60CSx (Garmin, Olathe, KS). Because the grower used the V-system for growing his crops and moved crops gradually as they grew, I marked each sample plant with a piece of ribbon to ensure that I was sampling the same plant over time. I recorded some of the practices that were done on the sampled plants (i.e. removing shoots, picking fruits or opening windows on top of the plants) at  128  the time of sampling and I recorded the environmental conditions at each sampling site at the same time.  5.2.7 Data analyses A generalized estimating equations (GEE) regression technique was used to analyze the data (Using R ver. 2.9.0) 13(Ballinger 2004). Pest infestation (presence of pests or damage symptoms), maintenance practices (open window, removing shoots or picking fruits), environmental factors (temperature, RH%, light intensity and airflow) and time (with seven levels) were each considered as a factor.  13  GEE analysis has been conducted by the statistical consulting and research laboratory (SCARL) at the Department of Statistics, The University of British Columbia.  129  ?"8#E%&42'&# I only encountered pests once in the form of a small infestation of T. ni on one of my sample plants. On other occasions I found damage symptoms of pests on the foliage of sample plants but I could not locate pests themselves. I did not find any disease infestation at my sampling sites and I only found one spot with signs of spider mite damage, but the level of damage was very low due to an implemented biological control program. Among all four indicator chemicals, I observed the highest emission of !-caryophyllene when I had observed the actual pest or signs of damage (Fig 5.3 and Fig 5.4). The level of !-caryophyllene increased from ~ 3,000 to 4,000 ng (150 to 200 ppb) in clean plants to ~10,000 ng (500 ppb) in damaged plants without T. ni larvae and ~ 60,000 ng (3000 ppb) in damaged plants with T. ni larvae. Emission of other indicator chemicals in response to T. ni also differed but less pronounced.  130  Limonene  FIG 5.3 VARIATION OF INDICATOR CHEMICALS AND ENVIRONMENTAL FACTORS AND RECORDS OF PEST INFESTATION AND MAINTENANCE PRACTICES AT SAMPLE SITE NUMBER TWO – DAMAGE SYMPTOMS, NO PEST AT WEEK 8  131  Limonene  FIG 5.4 VARIATION OF INDICATOR CHEMICALS AND ENVIRONMENTAL FACTORS AND RECORDS OF PEST INFESTATION AND MAINTENANCE PRACTICES AT SAMPLE SITE NUMBER FOUR- PEST PRESENCE AT WEEK EIGHT  132  TABLE 5.1. GEE REGRESSION COEFFICIENTS- (Z)-3-HEXENYL ACETATE Estimate  Std Error  Wald test  Sig  Intercept  408.1501  93.2760  19.15  p<0.001  Pest presence  41.9823  13.8600  9.17  p<0.01  Pest damage symptoms  4.8714  8.7314  0.31  0.5769  Open window  50.1083  44.1009  1.29  0.2559  Removing shoots  658.0727  23.5872  778.38  p<0.001  Picking fruits  -304.4295  16.2413  351.34  p<0.001  Temperature  -0.7988  1.5553  0.26  0.6075  Relative humidity  -1.7296  1.5446  1.25  0.2628  Light intensity  0.0229  0.0823  0.08  0.7811  Air flow  -2.0955  3.7390  0.31  0.5752  Time (Week 1-Week 2)  -77.3690  14.7965  27.34  p<0.001  Time (Week 2-Week 3)  -90.6191  14.2321  40.54  p<0.001  Time (Week 3-Week 4)  -39.3291  12.7390  9.53  p<0.001  Time (Week 4-Week 5)  -62.3358  11.1357  31.34  p<0.001  Time (Week 5-Week 6)  21.1675  14.3251  2.18  0.1395  Time (Week 6-Week 7)  -79.9492  11.4569  48.70  p<0.001  Time (Week 7-Week 8)  -158.2364  14.9140  112.57  p<0.001  While pest presence had a significant effect on the emission level of (Z)-3-hexenyl acetate, no significant difference were found in damaged plants without the pest. Opening windows and environmental factors also had no significant effect on emission level.  133  TABLE 5.2. GEE REGRESSION COEFFICIENTS- (E)-!-OCIMENE Estimate  Std Error  Wald test  Sig  Intercept  1371.752  784.186  3.06  p<0.05  Pest presence  3524.058  112.122  987.88  p<0.001  Pest damage symptoms  986.089  68.117  201.99  p<0.001  Open window  -239.640  378.154  0.40  0.53  Removing shoots  119.818  83.635  2.05  0.15  Picking fruits  -295.688  47.253  39.16  p<0.001  Temperature  -12.995  15.076  0.74  0.39  Relative humidity  -11.859  11.222  1.12  0.29  Light intensity  -0.138  0.215  0.41  0.52  Air flow  32.121  31.182  1.06  0.30  Time (Week 1-Week 2)  86.261  53.595  2.59  0.11  Time (Week 2-Week 3)  1560.005  30.102  2685.80  p<0.001  Time (Week 3-Week 4)  1626.192  54.465  891.46  p<0.001  Time (Week 4-Week 5)  1628.971  51.159  1013.86  p<0.001  Time (Week 5-Week 6)  1835.454  64.405  812.18  p<0.001  Time (Week 6-Week 7)  1706.823  65.817  672.51  p<0.001  Time (Week 7-Week 8)  1371.301  69.834  385.59  p<0.001  In the case of (E)-!-ocimene, both damage and presence of pests had a significant effect and among practices, only fruit picking had a significant effect on the its emission level. Environmental factors had no significant effect on the emission levels.  134  TABLE 5.3. GEE REGRESSION COEFFICIENTS- LIMONENE Estimate  Std Error  Wald test  Sig  Intercept  394.3992  376.6762  1.10  Pest presence  2265.7977  68.8989  1081.48  p<0.001  Pest damage symptoms  245.6056  33.8346  52.69  p<0.001  Open window  -354.8187  218.4750  2.64  0.104  Removing shoots  849.8344  96.2715  77.92  p<0.001  Picking fruits  -286.1433  34.1700  70.13  p<0.001  Temperature  -10.6775  4.8066  4.93  p<0.05  Relative humidity  4.3043  7.1018  0.37  0.544  Light intensity  -0.0938  0.1289  0.53  0.467  Air flow  36.6601  19.3304  3.06  0.058  Time (Week 1-Week 2)  393.6367  38.0709  106.91  p<0.001  Time (Week 2-Week 3)  858.1565  20.9580  1676.61  p<0.001  Time (Week 3-Week 4)  861.0540  36.4886  556.86  p<0.001  Time (Week 4-Week 5)  868.0864  31.5279  758.12  p<0.001  Time (Week 5-Week 6)  976.8227  39.8603  600.55  p<0.001  Time (Week 6-Week 7)  904.4213  43.5616  431.06  p<0.001  Time (Week 7-Week 8)  859.3140  37.7261  518.82  p<0.001  0.295  In the case of limonene, presence and damage signs of the pest, fruit picking and removing the shoots had significant effects on the emission level. Temperature among the environmental factors was also found to have a significant effect on the emission level.  135  TABLE 5.4. GEE REGRESSION COEFFICIENTS- !-CARYOPHYELLENE Estimate  Std Error  Wald test  Sig  Intercept  5.43e+03  1.24e+03  1.92e+01  p<0.001  Pest presence  5.60e+04  1.65e+02  1.16e+05  p<0.001  Pest damage symptoms  8.71e+03  1.32e+03  4.36e+03  p<0.001  Open window  2.38e+02  6.20e+02  1.50e-01  0.70149  Removing shoots  6.02e+02  1.51e+02  1.59e+01  p<0.001  Picking fruits  -9.62e+02  1.42e+02  4.61e+01  p<0.001  Temperature  -2.62e+01  2.47e+01  1.13e+00  0.28878  Relative humidity  -7.08e+01  1.89e+01  1.41e+01  p<0.001  Light intensity  9.99e-02  5.27e-01  4.00e-02  0.84981  Air flow  3.45e+01  4.58e+01  5.70e-01  0.45168  Time (Week 1-Week 2)  -2.96e+02  1.51e+02  3.87e+00  p<0.05  Time (Week 2-Week 3)  1.68e+03  1.22e+02  1.88e+02  p<0.001  Time (Week 3-Week 4)  1.26e+03  1.58e+02  6.40e+01  p<0.001  Time (Week 4-Week 5)  1.23e+03  1.38e+02  7.83e+01  p<0.001  Time (Week 5-Week 6)  1.29e+03  1.66e+02  6.00e+1  p<0.001  Time (Week 6-Week 7)  1.21e+03  1.52e+02  6.32e+01  p<0.001  Time (Week 7-Week 8)  -1.90e+02  1.74e+02  1.19e+00  0.27450  Both pest presence and damage symptoms were found to have a significant effect on the level of !-caryophellene. Like other compounds, opening the windows above the crops did not have a significant effect but removing shoots and picking the fruits were found to  136  have a significant effect on the emission level. Among the environmental factors, relative humidity was found to have a significant effect on the emission level. For all indicator chemicals, time was found to have a significant effect, indicating that as plants grow, their emission levels change. The changes in all 30 sampling sites are reported in Appendix Two.  137  ?"=#F+&54&&+)*# The main purpose of this study was to test the concept of using HIPVs as pest infestation indicators inside commercial greenhouses. Due to grower-imposed restrictions, I could not design a statistically sound experiment. However, my results support the concept of using HIPVs as possible indicators of pest infestation in commercial greenhouses. I randomly selected my sampling sites and used a funnel attachment for the zNose! to ensure standardized sampling. I used the same method for sampling plants starting from the top of the canopy and gently moving down to lower parts of the canopy. I considered several factors in this study. The GEE analyses enabled me to account for all of these factors together therefore, the significance levels in my results mean that after considering all other factors that play a role in the study, there is still a significant effect caused by this particular factor. I will go through all these factors one by one and discuss their effects on each of the indicator chemicals:  1- Pest presence: I encountered pests only once during the sampling period. However, their effect was quite significant for all indicator chemicals. Specifically, emission of !caryophellene which increased almost 20 times in response to pest presence. This was convincing for the grower who happened to accompany me when we found the infested area to consider this concept as a potential alternative for conventional pest scouting.  2- Damage symptoms: except for (Z)-3-hexenyl acetate, damage symptoms had significant effects on emission level of indicator chemicals. However, interpreting these results is difficult because the exact time of infestation remains unknown. In Chapter  138  Four, I show that damage duration does have a significant effect on emission level of indicator chemicals. I observed a sharp drop in the emission of (Z)-3-hexenyl acetate after I removed T.ni larvae (Fig. 4. 6). Similar patterns were observed for other chemicals but it seemed to take longer before their emission returned to background levels. This might explain the fact that in this study damage signs were not closely linked to emission levels of (Z)-3-hexenyl acetate. Perhaps enough time had elapsed since the damage was initiated. !-caryophellene however, had elevated emission levels despite the absence of pests confirming previous observation (Fig 4.6). The level of !-caryophellene might drop after the pest left the plant but it wouldn’t immediately return to background levels.  3- Greenhouse maintenance practices: The effect of window opening was a key concern for all growers. They wanted to know whether indicator chemicals would still be detectable when windows are open. My results clearly indicate that opening windows and screens do not have a significant effect on the emission or detection of volatiles. Removing shoots and picking fruits on the other hand were found to have significant effects on the emission of almost all indicator chemicals. It has been shown that some greenhouse practices cause an increase in the emission of these volatiles (Jansen et al. 2009), but the effect of these practices is not comparable to that of pest infestation.  4- Environmental factors: Light intensity and airflow did not have any significant effect on emission level of any indicator chemical. Temperature and relative humidity were found to affect the emission of limonene and !-caryophellene respectively.  139  Environmental factors have been shown to affect the release of HIPVs (Jakobsen 1997; Gouinguene and Turlings 2002) but their impact in my study was moderate.  5- Time: For almost all indicator chemicals, time was an effective and significant factor. Temporal variation in the emission levels of volatiles is well known (Dufay et al. 2004). There are several reports of rhythmic emission of volatiles at different times of the day (Dudareva et al. 2003; Hendel-Rahmanim et al. 2007; Waelti et al. 2008). However, even in light of all these variations, pest presence was still the most significant factor.  The results of my study support the concept of using HIPVs as potential pest infestation indicators inside greenhouses. This subject is in its infancy and further studies are required to understand all aspect of this concept. My results do not allow firm conclusions but warrant future investigations.  140  ?"?#E%6%,%*5%&# Ballinger GA. 2004. Using generalized estimatin equations for longitudinal data analysis. Organizational Res Meth 7:127-150. Dudareva N, Martin D, Kish CM, Kolosova N, Gorenstein N, Faldt J, Miller B, Bohlmann J. 2003. (E)-beta-ocimene and myrcene synthase genes of floral scent biosynthesis in snapdragon: Function and expression of three terpene synthase genes of a new terpene synthase subfamily. Plant Cell 15:1227-1241. Dufay M, Hossaert-McKey M, Anstett MC. 2004. Temporal and sexual variation of leafproduced pollinator-attracting odours in the dwarf palm. Oecologia 139:392-398. Gouinguene SP, Turlings TC. 2002. The effects of abiotic factors on induced volatile emissions in corn plants. Plant Physiol 129:1296-1307. Hendel-Rahmanim K, Masci T, Vainstein A, Weiss D. 2007. Diurnal regulation of scent emission in rose flowers. Planta 226:1491-1499. Jakobsen HB. 1997. The preisolation phase of in situ headspace analysis: Methods and prespectives. In: H.F. L, Jackson JF, editors. Plant volatile analysis. Berlin: Springer-Verlag. p 1-22. Jansen RMC, Hofstee JW, Wildt J, Verstappen FWA, Bouwmeeter HJ, Posthumus MA. 2009. Health monitoring of plants by their emitted volatiles: Trichome damage and cell memberane damage are detactable at greenhouse scale. Ann Appl Biol 154:441-452. Kunert M, Biedermann A, Koch T, Boland W. 2002. Ultra fast sampling and analysis of plant volatiles by a hand-held miniaturized gc with pre-concentration unit: Kinetic and quantitative aspects of plant volatile production. J Separation Sci 25:677-684.  141  Loreto F, Barta C, Brilli F, Nogues I. 2006. On the induction of volatile organic compound emissions by plants as consequence of wounding or fluctuations of light and temperature. Plant Cell Environ 29:1820-1828. Lu CJ, Jin C, Zellers ET. 2006. Chamber evaluation of a portable GC with tunable retention and microsensor-array detection for indoor air quality monitoring. J Environ Monit :270-278. Maleknia SD, Vail TM, Cody RB, Sparkman DO, Bell TL, Adams MA. 2009. Temperature-dependent release of volatile organic compounds of eucalypts by direct analysis in real time (dart) mass spectrometry. Rapid Commun Mass Spectrom 23: 2241-2246. Oh SY, Ko JW, Jeong SY, Hong J. 2008. Application and exploration of fast gas chromatography-surface acoustic wave sensor to the analysis of thymus species. J Chromatogr A 1205(1-2):117-127. Pasini P, Powar N, Gutierrez-Osuna R, Daunert S, Roda A. 2004. Use of a gas-sensor array for detecting volatile organic compounds (voc) in chemically induced cells. Anal Bioanal Chem 378:76-83. Picone JM, Clery RA, Watanabe N, MacTavish HS, Turnbull CG. 2004. Rhythmic emission of floral volatiles from Rosa damascena semperflorens cv. 'quatre saisons'. Planta 219:468-478. Vallat A, Gu H, Dorn S. 2005. How rainfall, relative humidity and temperature influence volatile emissions from apple trees in situ. Phytochem 66:1540-1550. Waelti MO, Muhlemann JK, Widmer A, Schiestl FP. 2008. Floral odour and reproductive isolation in two species of silene. J Evol Biol 21:111-121.  142  Zhong Q, Steinecker WH, Zellers ET. 2009. Characterization of a high-performance portable gc with a chemiresistor array detector. Analyst 134:283-193.  143  CHAPTER SIX: SUMMARY AND DISCUSSION When I first started this project in January 2006, I knew that I was not the first person with the idea of using plant-based indicators for monitoring their health (De Moraes et al. 2004; Holopainen 2004; Janssen et al. 2004; Baratto et al. 2005) and I soon learned that I won’t be the last person wanting to study this interesting field (Dekock et al. 2006a; Dekock et al. 2006b; Jansen et al. 2009). A movement has formed in recent years around the core idea of improving the current pest monitoring and crop protection systems. Finding other means of assessing crop health was one of the important aspects of this movement. Many studies have been conducted in recent years to find different types of indicators that can guide us in locating the problem. For example, Oerke et al. (2006) developed a method based on digital infrared thermography which permits a non-invasive monitoring and an indirect visualization of downy mildew development caused by Pseudoperonospora cubensis in cucumber plants. Boissard et al. (2008) developed a cognitive vision system for detecting pests in greenhouses while Skaloudova et al. (2006) designed a computer-assisted estimation system of leaf damage caused by spider mites. Finally, Moularat et al. (2008) demonstrated that a VOC fingerprint specific to fungal development could be detected in indoor environments. In this project, I have tried to contribute to this new approach of crop protection and pest monitoring. My ultimate goal was to investigate the possibility of creating a new tool that can increase the efficiency and performance of human scouts based on an alternative source of information. I chose to work with greenhouse tomatoes because greenhouses represent more controlled environments compared to open fields and tomato plants are  144  very well established model plants for studying herbivore-induced plant volatiles. I used the cabbage looper Trichoplusia ni (Hübner) as the model insect. I raised three major questions: 4) Can we discriminate between clean and infested plants based on the emission pattern of indicator chemicals? This question was not novel and many studies have shown differences in volatile emissions of infested plants, although most of these studies were conducted in laboratories within closed chambers. I took this to a different level and tested plants inside a research greenhouse to see if I can measure the same responses. My main objective was to determine a few infestation indicator volatiles and use them for classifying plants into infested and clean groups. 5) What additional information can we obtain from plant volatile emission patterns that can help us make better decisions? This question is an expansion of the first question. The more information we have about our pest(s) the better we can control them. I looked into the volatile emissions of plants that were infested with different densities of pests at different parts of the canopy and that experienced different periods of pest feeding. My objective was to investigate the possibility of obtaining additional information about the pest population and the severity of damage to plants based on emission patterns of indicator chemicals. 6) What are the effects of environmental factors and greenhouse related production practices on the emission of plant indicator chemicals in a commercial greenhouse? This question addressed the important issue of complexity in commercial greenhouses and how it affects the outcomes of my 145  project. It is very difficult to accurately simulate the conditions of a commercial greenhouse in a research facility. Therefore, I went to a commercial greenhouse for this phase of the work. In addition to levels of indicator chemicals, I recorded main environmental factors (temperature, relative humidity, light intensity and air flow) as well as greenhouse related practices (fruit picking, defoliation, opening windows, etc.). Sample plants were scouted and their general health recorded. The objective of this survey was to investigate the effect of all of these factors on the emissions of volatiles and test the suitability of the indicator chemicals for detecting pest infestations. Here I summarize my results for each question and discuss their significance. To enhance the clarity of my conclusions, at the end of each section, I provide a list of statements that I can or cannot claim based on my results. Can we discriminate clean and infested plants based on emission pattern of indicator chemicals? I determined four indicator chemicals from the volatile blends of tomato and showed that they are emitted in greater quantities by infested plants both in the laboratory and in the research greenhouse. I also showed that the difference in the emission of volatiles is significant enough to allow us to discriminate between clean and infested plants only six hours after infestation. All four indicator chemicals have been previously reported from tomato plants. There were many other compounds in the volatile blend of infested tomato plants that could have been used as good indicators. For example I identified !- phellandrene and linalool in some of the infested plants however; I could not detect them in all experiments and due  146  to their extreme variation, I decided not to use them as indicator chemicals. I also did not have sufficient resources to identify all of the compounds in the tomato plant volatile blends so I limited myself to these four indicator chemicals. Based on my results, I trust that these four chemicals are good indicators of T. ni infestations on the particular cultivar of tomato that I used in my study. Because considerable amounts of these chemicals are stored in grandular trichomes (van Schie et al. 2007), the increase in these volatiles may be due to mechanical plant damage. Mithofer et al. (2005) showed that continuous wounding could induce responses in plants relatively similar to that of pest infestation. On the other hand, Delphia et al. (2007) showed that pests with different modes of feeding might induce different responses in plants. The plant responses that I detected may not entirely be due to pest infestation and may result from just the mechanical damage caused by insect feeding. Mechanical damage -if caused by abiotic factors- is normally not continuous but rather sudden and temporary whereas pest feeding on foliage causes damage that persists as long as the pest insect stays on the plant and continues to feed. For T. ni larvae, feeding can continue for a long time. Therefore, the type and magnitude of the plants’ response should be sufficiently different to be used as indicators of pest infestation. I was able to discriminate between infested and clean plants six hours after the onset of an infestation. Early detection of an infestation is the key to a successful control program. My results are promising in that aspect. If we have sensors capable of detecting these differences, we might be able to locate pest problems well before they become problematic. Given the rapid discoveries and developments in electronics and engineering, we might have such sensors within a few machine generations. I do not  147  claim that we can pinpoint the exact location of a pest based on the emission of volatiles. However, we might be able to locate areas with a higher probability of infestation and then guide scouts to those areas. What additional information we can obtain from plant volatile emission patterns that help us make better decisions? The more we know, the better decisions we can make. In integrated pest management, we have threshold levels that need to be met before we take action. We certainly should not use the same treatment that we use for a fully infested plant when we have only a few pests on the plant. However, it is crucial to know the pest status at all times to be able to make the right decision. We must always be one step ahead. I concentrated on three aspects that can provide us with important information: population density, location of pest within the plant canopy and damage duration. Pest population density was the most important aspect. I showed that there is a significant difference in the response of plants to different pest population densities. I also observed some interesting responses at lower densities. Based on my results in the research greenhouse within this particular pest-plant context, I trust that it is possible to obtain information about the severity of damage at early stages of infestation. I do not claim that we can estimate the exact number of pests on plants but we might be able to label the infested area as moderately or highly infested based on the volatile emission levels. My results from the “ pest location within the plant canopy experiment” showed some significant differences. However, because the plants in my experiment were young and thus still short, results cannot readily be extrapolated. Further experiments with large plants are required to confirm my results. Results of the “feeding duration experiment”  148  might be of great importance. False alarms can jeopardize the accuracy of a pest monitoring system. If the sensing system were to mark an area as infested when the insect had long left the plant, resources would be wasted. My results showed that removing insects from the plant significantly decreases plant volatile emission. It might not rapidly return to the level of a clean plant but it will be significantly lower than that of an infested plant if feeding continues. Based on these results I trust that it will be possible for a highly sensitive sensor to discriminate between plants with ongoing or past insect feeding damage. What is the effect of environmental factors and greenhouse related practices on the emission of plant indicator chemicals in commercial greenhouses? I surveyed a commercial greenhouse eight times and recorded changes in plant volatile emission levels, environmental factors and production practices during the sampling period. Although this was a survey rather than a controlled experiment, I obtained valuable information. I learned that airflow does not have a significant effect on plant volatile blend if sampled at close range (20cm). The particular greenhouse where I conducted the survey was one the most modern and intensively maintained greenhouses in BC. I encountered pests only once. However, the response of the infested plant was so substantial that it surprised even the grower. I certainly cannot make any claim based on a single observation in only one location in one greenhouse. However, I trust this observation warrants further studies in many more commercial greenhouses. Gaining the trust of a commercial grower is an important factor in this type of study. Of 13 greenhouses in BC recommended by a grower association, I gained access to only one tomato greenhouse. Further research should take place in several greenhouses in different  149  locations to prevent pseudoreplication as explained by Hurlbert (1984). Moreover, experiments should be designed for implementation of this concept at a commercial level. I trust my results and efforts have set the stage for the continued development of an intelligent pest monitoring system for monitoring plant pest infestations.  FIG 6.1. COMPARING CONVENTIONAL PEST MONITORING WITH A PROPOSED INTELLIGENT PEST MONITORING SYSTEM: PEST SCOUTS VARY IN EXPERIENCE AND PERFORMANCE, POSSIBLY RESULTING IN INCONSISTANCIES. INTELLIGENT PEST MONITORING ARE BASED ON MODERN SENSORY SYSTEMS AND CONSIDER A WIDE RANGE OF VARIABLES IN THE DECISION MAKING PROCESS. FEEDBACK FROM HUMAN EXPERTS WILL REMAIN PART OF THE SYSTEM AND WILL IMPROVE ITS PRECISION AND PERFORMANCE.  To conclude, I want to add a few points about the methods I used in my project. I deployed a relatively new portable GC for most phases of the research. There is a range of different opinions about this particular instrument among researchers. Like any new instrument, the zNose! has some limitations and problems that might be improved over time. But these problems won’t get addressed unless researchers use this instrument and push it to its limits and provide feedback to the manufacturer (Table 6.1). The number of laboratories applying this instrument is growing. Nowadays, GC-MS is a standard instrument in analytical chemistry laboratories worldwide. Chromatography 150  dates back to 1903 and the work of the Russian scientist, Mikhail Semenovich Tswett. The German graduate student Fritz Prior developed solid state gas chromatography in 1947. Archer John Porter Martin, who was awarded the Nobel Prize for his work in developing liquid-liquid (1941) and paper (1944) chromatography, laid the foundation for the development of gas chromatography and later produced liquid-gas chromatography (1950) (Braithwaite and Smith 1996). The use of a mass spectrometer as the detector in gas chromatography was developed during the 1950s by Roland Gohlke and Fred McLafferty (Gohlke 1959; Gohlke 1993). There were many scientists who helped improve GC-MS to its current shape and form. The zNose! is a portable device that enables us to perform analytical experiments in the field. It is relatively fast so we can use it for monitoring rapid changes in the emission of volatiles. It is not perfect but it can provide accurate and useful information.  TABLE 6.1 FEEDBACK TO THE ZN OSE! MANUFACTURER  Recommendations Hardware  The limitation of peak resolution might be solved with a longer column. It might slightly increase the analysis time but  would allow for a greater range of  compounds that can be analyzed by the machine.  Operation  The cooling fan for the sensor is located on top of the inlet. Collecting samples in the field, might be affected by the fan especially when the operator is analyzing a compound in low concentrations. An attachment for the inlet (like a funnel) can prevent this problem.  Software  The method panel and setting panel do not synchronize automatically. The operator must make changes twice to be effective. Data logging is relatively difficult. It would be very helpful if the operator could extract the information and record it directly into an excel file. It would be more convenient if the battery charge level was displayed as percentage.  151  The concept of using plant volatiles for pest monitoring is still in its infancy and has a long way to go to reach maturity. This project is my humble effort to contribute to a relatively important and novel issue in pest management using the great pool of knowledge accumulated in past decades in chemical ecology. I hope that other researchers and scientists who are interested in the same subject find this tiny piece of information useful.  152  A"!#E%6%,%*5%&# Baratto C, Faglia G, Pardo M, Vezzoli M, Boarino L, Maffei M, Bossi S, Sberveglieri G. 2005. Monitoring plants health in greenhouse for space missions. Sensors and Actuators 108:278-284. Boissard P, Martin V, Moisan S. 2008. A cognitive vision approach to early pest detection in greenhouse crops. Comp Electron Agric 62: 81-93. Braithwaite A, Smith F. 1996. Chromatographic methods. London: Blackie Academic & Professional. De Moraes CM, Schultz JC, Mescher MC, Tumlinson JH. 2004. Induced plant signaling and its implications for environmental sensing. J Toxicol Environ Health A 67:819-834. Dekock J, Aerts J, Berckmans D, Vermeulen K, Steppe K, Lemeur R, Janssen K, Bleyaret P, Westra J, Rieswijk T. 2006a. Crop monitoring by means of an on-line early warning system ACTA Hort 718:183-188. Dekock J, Aerts JM, Berckmans D, Janssen K, Bleyaert P, Vermeulen K, Steppe K, Lemeur R, Westra J, Rieswijk T. October 2006. Crop monitoring by means of an on-line early warning system. III International Symposium on Models for Plant Growth, Environmental Control and Farm Management in Protected Cultivation Wageninen. 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. J Chem Ecol 33:997-1012.  153  Gohlke R. 1959. Time of flight mass spectrometry and gas-liquid chromatography. Analytical Chem 31:535-541. Gohlke R. 1993. Early gas chromatography/mass spectrometry. J American Sco Mass Spect 4:367-371. Holopainen JK. 2004. Multiple functions of inducible plant volatiles. Trends Plant Sci 9:529-33. Hurlbert S. 1984. Pseudoreplication and the design of ecological field experiments. Ecol Monographs 54:187-211. Jansen RMC, Hofstee JW, Wildt J, Verstappen FWA, Bouwmeeter HJ, Posthumus MA. 2009. Health monitoring of plants by their emitted volatiles: Trichome damage and cell memberane damage are detactable at greenhouse scale. Ann Appl Biol 154:441-452. Janssen K, Vermeulen K, Boonen C, Bleyaert P, Lemeur R, Berkcmans D. 2004. Introduction to speaking plant: Let the crop control the greenhouse climate. Commun Agric Appl Biol Sci 69:151-3. Mithofer A, Wanner G, Boland W. 2005. Effects of feeding Spodoptera littoralis on lima bean leaves. Ii. Continuous mechanical wounding resembling insect feeding is sufficient to elicit herbivory-related volatile emission. Plant Physiol 137:11601168. Moularat S, Robine E, Ramalho O, Oturan MA. 2008. Detection of fungal development in a closed environment through the identification of specific voc: Demonstration of a specific voc fingerprint for fungal development. Sci Total Environ 407:13946.  154  Oerke EC, Steiner U, Dehne HW, Lindenthal M. 2006. Thermal imaging of cucumber leaves affected by downy mildew and environmental conditions. J Exp Bot 57:2121-2132. Skaloudova B, Krivan V, Zemek R. 2006. Computer-assisted estimation of leaf damage caused by spider mites. Comput Electron Agric 53:81-91. van Schie CC, Haring MA, Schuurink RC. 2007. Tomato linalool synthase is induced in trichomes by jasmonic acid. Plant Mol Biol 64:251-263.  155  APPENDIX ONE  1,8 Cineole  Citral  Eugenol  Geraniol  MASS SPECTRA OF EXTERNAL STANDARDS AND NIST LIBRARY MATCH- MAJOR COMPOUNDS OF INSECT REPELLENT FORMULATION  156  MASS SPECTRA OF LIMONENE EXTERNAL STANDARD AND VARIAN LIBRARY MATCH  157  (Z)-3-hexenyl acetate  limonene  (E)-!-ocimene  158  !-caryophyllene  MASS SPECTRA OF INDICATOR CHEMICALS  159  APPENDIX TWO  160  161  162  163  164  165  166  167  168  169  170  171  172  173  174  175  176  177  178  179  180  181  182  183  184  185  186  187  188  189  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0067692/manifest

Comment

Related Items