The Biology of Plant- Associations

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Babak Ebrahimi, M.S.P.H.

Graduate Program in Entomology

The Ohio State University

2013

Dissertation Committee:

Professor Woodbridge A. Foster, Advisor

Professor P. Larry Phelan

Professor David L. Denlinger

Professor Peter M. Piermarini Copyright by

Babak Ebrahimi

2013 Abstract

Mosquitoes need plant sugar to maximize their activities. While males cannot survive without sugar, females performance increases in the presence of sugar.

Mosquitoes use phytochemicals that are emitted by plants to find the source of sugar. To study mosquitoes’ behavior toward the phytochemicals, olfactometers have been used by researchers. Chapter 2 gives a description of three types of olfactometers that were designed and developed in the vector behavior laboratory at The Ohio State University, to evaluate attractiveness of phytochemicals.

Milkweed and goldenrod flowers are attractive sources of sugar for Aedes vexans,

Culex pipiens, and Cx. restuans in the Midwestern United States. In Chapter 3, attraction of the flowers of these two plants, their extracts, and the synthetic blend of milkweed were compared to unbaited and honey-baited traps. Flower-baited traps were more attractive to Ae. vexans and Culex spp. than traps with other baits. In Chapter 4, the possibility that Anopheles gambiae, a major African malaria vector, can learn plant odors and single phytochemicals, was explored in a mesocosm and an olfactometer. Results showed that associative olfactory learning depends on release rates of compounds during testing and conditioning. They also showed that when mosquitoes associated a compound with a sugar reward during conditioning, they were still highly attracted to it, even in the presence of a novel compound. The effect of sugar availability on vectorial capacity, ii including survival, biting rate, and fecundity was assessed in Chapter 5. Vectorial capacity was increased in a sugar-rich environment, unlike an earlier study, probably because mosquitoes had access to human blood for long period of time, allowing a higher biting rate. Fecundity was slightly decreased in the sugar-poor environment, probably because of a lower insemination rate in females, owing to rapid male die-off, and the females’ ability to use blood to provide energy. In addition to sugar availability, reproductive fitness of An. gambiae can be affected by rain. The experiments described in

Chapter 6 showed that water agitation stimulates egg hatching, a first for this genus of mosquito. Eggs stored as long as a month hatched upon agitation. This characteristic has led to a new method of handling and colony maintenance in this species. Mosquitoes and other rely on many aspects of natural light in their lives, but artificial light generated by incandescent and fluorescent lamps and used in laboratory experiments contains a flicker component, a byproduct of alternating current. can detect this flicker, and it may affect their responses to light-based behavior and development. A newly developed flicker-free dimmer for LEDs is described in Chapter 7. A low-pass filter has been placed between an Arduino microcontroller and a strip of LEDs to remove flicker. The microcontroller controls dimming duration.

iii Dedication

To Daryoush, Maziyar, and Mehrnoosh Ebrahimi, Pourandokht Raeisi

and to my best friend and love, Leila Farivar

iv Acknowledgments

Many people paved my path through my Ph.D. program, by their intellectual, emotional, and financial support. I thank my advisor, Woodbridge Foster, who patiently taught, inspired, trained, encouraged, and assisted me through all steps. My appreciation to him goes beyond words; he opened my eyes to another field of science, and to another philosophy. I appreciate Larry Phelan for teaching me the principles of chemical ecology and behavior, and for his valuable contribution and suggestions in Chapters 2, 3, and 4. I was fortunate to have David Denlinger in my committee for all of his support and suggestions, particularly for Chapters 5 and 6. I am also grateful to have had Glen

Needham in my committee for his support for providing lab material for a part the study in Chapter 4. I also thank Peter Piermarini for his suggestions on all chapters of this dissertation. Many of my friends and colleagues contributed to my research and helped me since I have started my Ph.D. Program. I was fortunate to work alongside Bryan

Jackson, whose constructive discussions and support helped me during the past 3 years.

He also contributed to Chapter 5, for which I am grateful. I also thank Chris Stone particularly for his contribution in constructing the mesocosms and also for his suggestions in Chapter 5. It was a pleasure to work alongside Philip Otienoburu, who helped me in Chapters 2 and 3. I am also grateful to have collaborated with Reza Farivar v in Chapter 7, which would have been impossible to fulfill without his ideas and support. I am also thankful for the encouragement and support that I received from my parents,

Daryoush Ebrahimi and Pourandokht Raeisi, my brother, Maziyar Ebrahimi, and my sister, Mehrnoush Ebrahimi. Finally, I thank my best friend and my wife, Leila Farivar, whom I have been fortunate to have for her support, inspiration, and for her comments on statistical analyses of this dissertation.

vi Vita

June 1993 ...... Adab High School

February 1999 ...... B.S. Plant Protection, Isfahan

University of Technology

February 2003 ...... M.S.P.H. Medical Entomology

and Vector Control, Tehran

University of Medical Sciences

2004 – 2006 ...... C.E.O., Isfahan Adrian Avin Co.

2007-2013 ...... Graduate Teaching /

Research Associate,

The Ohio State University

Publications

Otienoburu P.E., Ebrahimi B., Phelan P.L., Foster W.A. 2012. Analysis and Optimization of a Synthetic Milkweed Floral Attractant for Mosquitoes. Journal of Chemical Ecology.

38(7): 873-881.

Nikbakhtzadeh M.R., Hemp C., Ebrahimi B. 2007. Further Evidence on the Role of

Cantharidin in the Mating Behaviour of Blister Beetles (Coleoptera: Meloidae).

Integrative Biosciences. 11: 141-146. vii Nikbakhtzadeh M.R., Ebrahimi B. 2007. Detection of Cantharidin Related Compounds in

Mylabris impressa (Coleoptera: Meloidae). Journal of Venomous Animals & Toxins.

13(3): 686-693.

Yaghoobi-Ershadi M.R., Akhavan A.A., Zahraie-Ramazani A.V., Abai M.R., Ebrahimi B.,

Vafaie-Nezhad R., Hanafi-Bojd A.A. and Jafari R. 2004. Epidemiological study in a new focus of cutaneous leishmaniasis in the Islamic Republic of Iran. Letter to the editor,

Eastern Mediterranean Health Journal. 10(4/5): 688.

Yaghoobi-Ershadi M.R., Akhavan A.A., Zahraie-Ramazani A.V., Abai M.R., Ebrahimi B.,

Vafaie-Nezhad R., Hanafi-Bojd A.A. and Jafari R. 2003. Epidemiological study in a new focus of cutaneous leishmaniasis in the Islamic Republic of Iran. Eastern Mediterranean

Health Journal. 9(4): 816-826.

Alempoor-Salemi J., Shayeghi M., Zeraati H., Akbarzadeh K., Basseri H., Ebrahimi B.,

Rafinejad J. 2003. Some aspects of head lice infestation in Iranshahr area (southeast of

Iran). Iranian Journal of Public Health. 32(3): 60-63.

Fields of Study

Major Field: Entomology

viii Table of Contents

Abstract...... ii Dedication...... iv Acknowledgments...... v Vita...... vii List of Tables...... xi List of Figures...... xiv

Chapters 1. Introduction: …...... 1 Effect of Sugar on Mosquitoes...... 2 Location of Host Plants by Volatiles...... 5 Practical Use of Phytochemicals...... 7 Potential Influence of Learning on Attraction to Plants...... 11 How Plant Communities Can Alter Vectorial Capacity...... 12 Objectives...... 14

2. Olfactometer Design for Aedes vexans and Anopheles gambiae Abstract...... 16 Introduction...... 17 Materials and Methods...... 19 Results...... 31 Discussion...... 34

3. Innate Attraction to Plant Volatile Abstract...... 44 Introduction...... 45 Materials and Methods...... 46 Results...... 53 Discussion...... 55

4. Learning: Association of Volatiles with Sugar Abstract...... 67 Introduction...... 68

ix Materials and Methods...... 71 Results...... 84 Discussion...... 90 Conclusion...... 98

5. Effect of Plant Community on Vectorial Capacity Abstract...... 120 Introduction...... 121 Materials and Methods...... 123 Results...... 132 Discussion...... 134

6. Delayed Egg Hatching of Anopheles gambiae (Diptera: Culicidae), Pending Water Agitation Abstract...... 150 Introduction...... 151 Materials and Methods...... 154 Results...... 162 Discussion...... 166

7. A non-flickering and programmable dimmer for LED lights Abstract...... 188 Introduction...... 189 Materials and Methods...... 191 Results and Discussion...... 196

Bibliography Chapter 1...... 205 Chapter 2...... 215 Chapter 3...... 220 Chapter 4...... 224 Chapter 5...... 229 Chapter 6...... 233 Chapter 7...... 240

x List of Tables

Table 2.1: Comparison of discrimination indices (percentage of mosquitoes in one arm / mosquitoes in both arms) in Jepson-Healy olfactometer, constructed from two different materials. The comparisons were made according to the type of experiment, i.e., Honey vs. Blank, Honey vs. Honey, and Blank vs. Blank arms (Honey = 2.5 g), with H0: Discrimination Index = 0.5, and H1: Discrimination Index ≠ 0.5. Significant P-values indicate that mosquito response in one arm was higher than the other arm...... 37 Table 2.2: Percentage of total responses to both arms (mean percentage ± SE) in each type of experiment, comparing olfactometer materials:. polypropylene vs. polyethylene...... 38 Table 4.1: Mean (± SE) response latency of males Anopheles gambiae to 12 chemicals diluted in mineral oil. Among 4 concentrations (1, 10, 100, and 1000 nl/ml) for each chemical, the one with shortest mean response time (sec) of mosquitoes was reported, along with proportion of mosquitoes that probed or walked, flew, or did not respond. Mann-Whitney U test was used for comparison between mean of each chemical and mineral oil (40 sec)...... 102 Table 4.2: Estimated regression line of release rate (mg/hr) by concentration (%v/v) for phenylacetaldehyde, linalool oxide, and Z-β-ocimene. Mineral oil was used as the diluent in a 3.81 by 5 cm low-density polyethylene sachets (51µm thickness) at 25 °C an 70% RH...... 103 Table 4.3: Comparison of dose response of Anopheles gambiae males and females to phenylacetaldehyde, linalool oxide, and Z-β-ocimene at three concentrations (µl / ml) in sachet. The mean percent discrimination out of total responding (treatment arm / sum of

xi two arms), and total responses out of total released mosquitoes, of the three concentrations were compared to each other...... 104 Table 4.4: Factors that may influence associative learning in Anopheles gambiae. Prior to conditioning and testing, the tested subject should be hungry. During conditioning in this study, mosquitoes were exposed to unconditioned and conditioned stimuli simultaneously, for 15 min. Exposure time of testing was 12 hr...... 105 Table 5.1. Plant compositions in high sugar or low sugar mesocosms. Number of used plants in each mesocosm is before each species name...... 139 Table 5.2. Predicted mean age at death of females in each replicate...... 140 Table 5.3. Mean daily biting rates per female in sugar rich and sugar-poor mesocosms140 Table 5.4. Vectorial capacity, calculated based on C = (m ∙ a 2 ∙ p n) / (-log(p)) when m = 5, n = 12, p = predicted mean age...... 140 Table 5.5. Mean daily egg output per female for each replicate after 20 days of blood- feeding...... 141 Table 5.6. Mean wing size of surviving females in sugar rich mesocosms, collected on day 21, for each replicate. Mean wing sizes of 4 replicates were tested using one-way ANOVA with Tukey post hoc...... 141 Table 6.1. Details of the effect of seven consecutive daily agitations on mean proportion (%) hatching of Anopheles gambiae eggs (1egg / 1.5 mL in well). Three data sets are compared: 1) daily spontaneous hatch rates of control (non-agitated) groups on successive days at the same times as the treatment (5 min/d agitation) group counts, and 2) daily stimulated hatch rates of treatment (agitated) groups a) immediately after (<10 min) each agitation, and b) during the 24-h period before the next one. Graphic summary of data with daily times combined (i) is presented in Fig. 1. (n = 15)...... 175 Table 6.2. Effect of grouped eggs on proportion of Anopheles gambiae eggs that hatched (% mean ± SE). Comparison of hatching rates of solitary eggs (1egg / 1.5 mL in well)a and grouped eggs (10 eggs / 5mL in test tube)a before the first agitation of pharate first- instar larvae (DH0 , 2 d post-oviposition) and Total Hatch by Agitation after 1 wk consecutive daily agitations (THA). Temperature 25.5°C...... 179 xii Table 6.3. Details of effect of storage duration and temperature on proportion of Anopheles gambiae eggs hatching (10 eggs / 5mL in test tube). Comparison of mean hatch rates (± SE, n = 5 a) after storage for five durations (2-29 d) at 15.5 and 25.5°C, before the first agitation (DH0) and again <10 min after the first agitation of pharate first- instar larvae, and of Total Hatch (TH) after 6 consecutive days of agitation. Post-storage (6-d agitation period) temperature 25.5°C...... 181

Table 6.4. Regression models describing relationship between Daily Hatch (DHi) of Anopheles gambiae eggs (10 eggs / 5mL in test tube) after 5 min agitation per day (0 ≤ i ≤ 6), storage duration (2 ≤ age ≤ 29), and temperature (temp = 15.5 or 25.5). Daily Hatch

(DHi) ranged between 0 and 1...... 183 Table 7.1 Codes to control LED lights. Parts of the codes that can be changed are boldfaced...... 199

xiii List of Figures

Fig. 2.1: A drawing of the Y-shape olfactometer, connected to a mosquito- release cage...... 39

Fig. 2.2: Parts of assembled dual-port olfactometer. Direction of airflow is left to right, from trapping jars towards mosquito-release cage...... 39

Fig. 2.3: Left: front view of a funnel port installed on a jar. Right: side view of a baffle port...... 40

Fig. 2.4: Three Jepson-Healy olfactometers made of polyethylene plastic bags...... 40

Fig. 2.5: Mean response percentage of Aedes vexans to 5 g honey and control in Y-shape and dual-port olfactometers, after 12 hr...... 41

Fig. 2.6: Mean response percentage of Anopheles gambiae adults (both sexes) to two arms, baited with honey (2.5 g) or left blank, in olfactometers made with polyethylene or polypropylene bags...... 42

Fig. 2.7: Mean percent responding of released 2-d-old Anopheles gambiae mosquitoes to the arms of Jepson-Healy olfactometer baited with males and females of the same species vs. no mosquitoes (number of replicates = 9)...... 43

Fig. 3.1: Percentage of Aedes vexans (both sexes) attracted to one bud of milkweed flower (4g) in a dual choice olfactometer...... 60

xiv Fig. 3.2: Percentage of Aedes vexans (both sexes) attracted to goldenrod flower (4 g) in a dual choice olfactometer...... 60

Fig. 3.3: Percentage of Aedes vexans (both sexes) attracted to one floret of milkweed flower and honey in a dual choice olfactometer...... 61

Fig. 3.4: Percentage of Aedes vexans (both sexes) attracted to goldenrod flower and honey in a dual choice olfactometer ...... 61

Fig. 3.5: Number of mosquitoes (Culex spp. and Aedes vexans) per night at each trap

(site), September 5, 2008. All traps were unbaited and served as blank suction traps...... 62

Fig 3.6: Total number of mosquitoes in goldenrod flower-baited and control-paired

(blank) traps, collected from each location...... 62

Fig. 3.7: Mean number of Aedes vexans and Culex spp. from unbaited CFG traps paired the traps baited with goldenrod flower...... 63

Fig. 3.8: Mean number of Aedes vexans and Culex spp. per night in blank CFG traps, paired with traps baited with the extract of goldenrod flower...... 63

Fig. 3.9: Mean number of male and female Culex spp. and Aedes vexans per night per location. All traps were unbaited and served as blank suction traps...... 64

Fig 3.10: Mean number of male and female Aedes vexans and Culex spp. collected from unbaited traps, and traps baited with milkweed flower, milkweed blend, and honey...... 64

Fig. 3.11: Mean number of mosquitoes collected from unbaited traps that were setup at 1,

2, and 3 m away from a CFG trap that dispensed the synthetic milkweed blend at 1000 times of the original concentration...... 65

xv Fig. 3.12: Mean number of mosquitoes, collected from unbaited traps, and traps that were baited with milkweed synthetic blend at higher concentrations (x100 and x1000) compared to what developed in olfactometer for Culex spp...... 66

Fig. 4.1: Comparison of mean percent responding of naïve and conditioned Anopheles gambiae adults in an olfactometer baited with plant volatiles (Senna didymobotrya) vs. blank. 24 hr prior to the olfactometer test, mosquitoes were fed 2% sugar solution alone

(naïve control), and with the sugar solution while being exposed to the headspace of S. didymobotrya for 15 min (conditioned treatment)...... 106

Fig 4.2: Survival of Anopheles gambiae females and males on water (0), and after 12 hr feeding on three concentrations of sugar solutions (2, 50, and 95%), and on sugar cubes

(100%). Day 0 represents 12 hr after removing sugar (2 d post-emergence)...... 107

Fig 4.3: Toxicity range. Survival of Anopheles gambiae females and males after 15 min feeding on 2% sugar solution (control) containing 2.5 µl phenylacetaldehyde (PHE), 5 µl linalool oxide (LO), or 5, 10, and 50 µl Z-β-ocimene (ZBO) in 4 ml of solution...... 108

Fig. 4.4: Weight loss (mg) of three concentrations (1, 10, and 100 %v/v) of phenylacetaldehyde (PHE), linalool oxide (LO), and Z-β-ocimene (ZBO) diluted with 1 ml mineral oil in 3.81 by 5 cm low-density polyethylene sachets...... 109

Fig 4.5: Estimated regression lines for release rates (mg/hr) of phenylacetaldehyde

(PHE), linalool oxide (LO), and Z-β-ocimene (ZBO) by concentration (%v/v). Mineral oil was used to dilute the compounds in 3.81 by 5 cm low-density polyethylene sachets

(51 µm thickness) at 25 °C and 70% RH...... 110

xvi Fig. 4.6: Dose-responses of males and females of Anopheles gambiae to phenylacetaldehyde. Mean percent responding of released adults to each arm of olfactometers baited with three concentrations (0.1, 1, and 10 µl/ml) vs. blank...... 111

Fig. 4.7: Dose-responses of males and females of Anopheles gambiae to Z-β-ocimene.

Mean percent responding of released adults to each arm of olfactometers baited with three concentrations (1, 10, and 100 µl/ml) vs. blank...... 112

Fig. 4.8: Dose-responses of males and females of Anopheles gambiae to linalool oxide.

Mean percent responding of released adults to each arm of olfactometers baited with three concentrations (1, 10, and 100 µl/ml) vs. blank...... 113

Fig. 4.9: The mean percent responses of released Anopheles gambiae to each arm of olfactometers, baited with 10 µl phenylacetaldehyde (PHE) vs. 10 µl linalool oxide (LO).

Adults were conditioned by 2% sugar solution alone (naïve mosquitoes), and mixtures of

2.5 µl linalool oxide or phenylacetaldehyde in 4 ml 2% sugar solution...... 114

Fig. 4.10: The mean percent responses of released Anopheles gambiae to each arm of olfactometers, baited with 1 µl phenylacetaldehyde (PHE) vs. 100 µl Z-β-ocimene

(ZBO). Adults had been conditioned with 2% sugar solution alone (naïve mosquitoes), and mixtures of 10 µl ZBO or 2.5 µl PHE in 4 ml 2% sugar solution...... 115

Fig. 4.11: The mean percent responses of released Anopheles gambiae to each arm of olfactometers, baited with sachets of 1 µl phenylacetaldehyde (PHE) vs. 100 µl Z-β- ocimene (ZBO). Adults had been fed 2% sugar solution while exposed to either 10 µl

ZBO or 2.5 µl PHE placed on the germination papers, or no volatile (naïve)...... 116

xvii Fig. 4.12: The mean percent responses of released Anopheles gambiae to each arm of olfactometers, baited with 1 or 30 µl phenylacetaldehyde (PHE) in sachets vs. blank.

Adults had been fed 2% sugar solution while exposed to the volatiles of 2.5 µl PHE placed on the germination paper or no volatile (naïve)...... 117

Fig. 4.13: The mean percent responses of released Anopheles gambiae females to each arm of olfactometers, baited with 100 µl Z-β-ocimene (ZBO) in sachet vs. blank. Adults had been fed 2% sugar solution while exposed to the volatiles of either 10 or 100 µl ZBO released from sachets, or no volatile (naïve)...... 118

Fig. 4.14: The mean percent responses of released Anopheles gambiae males to each arm of olfactometers, baited with 100 µl Z-β-ocimene (ZBO) in sachet vs. blank. Adults had been fed 2% sugar solution while exposed to the volatiles of either 10 or 100 µl ZBO released from sachets, or no volatile (naïve)...... 119

Fig. 5.1. (A) Schematic drawing of a mesocosm as it appeared during experimentation identifying the following: (a) metal-halide grow light, (b) humidifier exhaust pipe, (c) resting site, (d) oviposition site, and (e) temperature and humidity sensor. (B) A panoramic view from the antechamber looking into the mesocosm during an experiment...... 142

Fig. 5.2. Anopheles gambiae males and females were held in a mesocosm overnight where they had access to six plants. This was replicated four times. (A) Proportion of males and females positive for fructose. (B) Mean amount of fructose (µg) ±SD for both males and females...... 143

xviii Fig. 5.3. Cumulative percent mortality of male and female An. gambiae and daily biting rate per surviving female over a 21-d period in a mesocosm when provided with 24-h access to nectariferous plants and 1-h nightly access to a human host...... 144

Fig. 5.4. Kaplan-Meier survivorship curves for male and female An. gambiae when exposed to sugar-rich or sugar-poor plants. Each line represents one replicate...... 145

Fig. 5.5. Daily biting rates of mosquitoes in sugar-rich and sugar-poor mesocosms

(circles) of four replicates (colors), and the associated 3-d moving averages (lines). Biting rates were calculated by dividing the number of observed bites per night by the number of females predicted to be present (initial number of females times the predicted survivorship)...... 146

Fig. 5.6. Distribution of biting rates among replicates. In replicate 2, the biting in sugar rich mesocosm was “marginally” higher than in sugar-poor...... 147

Fig. 5.7. Insemination rates (3 days cumulative) of dead females collected daily. Blue line: sugar-poor mesocosm; Red line: sugar-rich mesocosm...... 148

Fig. 5.8. Mean wing size of dead females and male collected daily from a sugar rich environment...... 149

Fig. 6.1. Effect of seven consecutive daily agitations on mean daily hatch (% DHi).

Comparison of DHi of non-agitated and agitated (5min/d) Anopheles gambiae eggs

(1egg / 1.5 mL in well) from i = 0 (1 d post-oviposition) until i = 6 (7 d post- oviposition). Temperature 25.5°C. Details presented in Table 6.1...... 185

Fig 6.2. Effect of storage duration and temperature on mean daily hatch (% DHi) of

Anopheles gambiae eggs (10 eggs / 5mL in test tube) agitated 5 min/d on six consecutive xix days, from i = 0 (before the first agitation) until i = 6 (after the sixth agitation).

Comparisons after each of five different storage durations (2-29 d) at either 15.5° or

25.5°C. Post-storage (6-d agitation period) temperature 25.5°C...... 186

Fig 6.3. Effect of storage duration and temperature on mean (± SE) Total Hatch by

Agitation (% THA) of Anopheles gambiae eggs (10 eggs / 5mL in test tube) after six consecutive days of agitation (5 min/day) following removal from storage. Comparisons of THA after each of five different storage durations (2-29 d) at either 15.5° or 25.5°C.

Post-storage (6-d agitation period) temperature 25.5°C...... 187

Figure 7.1. Dimming AC current by turning off part of current in each half of sinusoidal

AC wave. The proportion that is 'off' relative to 'on' increases, the current decreases, and the flicker frequency increases...... 202

Figure 7.2. Block diagram of how electric signal is manipulated to create final output, a flicker-free light...... 202

Figure 7.3. Simplified view of additional circuitry in the system...... 202

Figure 7.4. Transition diagram of RC filter...... 203

Figure 7.5: Circuitry of the LED strip light...... 203

Figure 7.6. Electrical relationship between lux and the PWM duty cycle, showing non- linearity...... 204

Figure 7.7. Correlation between lux and requested PWM value after adding exponential function to the codes...... 204

xx Chapter 1: Introduction

This dissertation is a contribution to several topics central to our knowledge of the effects of plants on the behavior and reproductive performance of adult mosquitoes. This includes studies of 1) attraction of mosquitoes to the volatile organic compounds (VOCs) of potential host plants, 2) their ability to learn to associate those VOCs with sugar- producing plants, and 3) the effect of those plants on mosquito fecundity and coefficients of vectorial capacity, sensu Garett-Jones (1964). But it also includes tangential developments and studies ancillary to the main theme: 1) tests of new olfactometer designs for analyzing olfactometer results, 2) an examination of a stimulus that initiates egg hatching, and 3) a new device for eliminating light flicker in laboratory studies of mosquitoes and other insects. Because mosquitoes are so important to the health of humans and animals, anything that can be learned about the largely unexplored plant- mosquito relationship has the potential for improving human health worldwide.

Eggs that are laid by mosquitoes on water or damp soil hatch into first instar larvae when flooded or agitated in water. The larvae go through four stages of development before changing into pupae. Females and males usually feed on natural sugar resources shortly after emergence, to increase their energy reserves for mating and host-seeking. All females, except for autogenous ones (i.e., blood meal for initial egg development is not required), need one or more blood-meals to produce eggs. Females 1 can uptake disease-causing pathogens by feeding on infected hosts, and transfer those pathogens to uninfected individuals. Infective bites of mosquitoes cause more than 500 million clinical diseases worldwide each year. The importance of the mosquito species investigated in this dissertation, Anopheles gambiae, Aedes vexans, and Culex spp. is underlined by recent statistics: Malaria has caused more than 1 million deaths each year

(Snow et al. 2005), and has resulted in direct and indirect economic problems in the world (Sachs and Malaney 2002). In 2012, CDC reported 5,674 human cases of the West

Nile Virus (WNV) disease, with 286 deaths just in the U.S. The following topics provide an abbreviated general background of the plant-mosquito relationship and a statement of objectives. Further details are presented in the individual chapters.

Effect of Sugar on Mosquitoes

Only female mosquitoes take blood meals, but both sexes feed on plants sugar

(Yee et al. 1992, Foster and Hancock 1994, Gary and Foster 2004). However females of the genus Toxorhynchites feed exclusively on sugar. Sugar-feeding is the only source of nutrition for males (Yuval et al. 1994), and it extends females longevity as shown for

Aedes aegypti (Thorsteinson and Brust 1962, Nayar and Sauerman 1975), Culex pipiens quinquefasciatus (Patterson et al. 1969), Aedes taeniorhynchus, Ae. sollicitans,

Anopheles quadrimaculatus, Psorophora columbia (Nayar and Sauerman 1975), and An. gambiae (Straif and Beier 1996, Gary and Foster 2001, 2004, Impoinvil et al. 2004).

Unlike blood-feeding behavior, which is usually restricted to limited numbers of hosts, sugar-feeding can occur on a wide range of sugar-providing plants, though preferences 2 are evident (Foster 1995). Natural sources of sugar for mosquitoes are floral and extra- floral nectars (Patterson et al. 1969, Foster and Hancock 1994, Burkett et al. 1999,

Gouagna et al. 2010), honeydew (Gary and Foster 2004), and ripening fruits (Joseph

1970). Müller et al. (2010a) showed that An. gambiae was attracted to sugar baits in the first half of the night, whereas blood-host was sought in the second half of the night.

However, Culex nigripalpus feeds on sugar before taking a blood-meal (Hancock and

Foster 1997). Nonetheless, in some species the two behaviors may occur simultaneously or overlap to some extent (Yee et al. 1992).

Presence of fructose in mosquitoes can be detected using Van Handel's cold- anthrone test (Van Handel 1965, 1972). Because this sugar does not occur as a metabolite in mosquitoes but is common in plants, its presence in the body is an indication of the presence of an undigested sugar meal from plants and plant-derived sources. Ingested sugar may either provide energy for flight or be converted to glycogen and lipid for storage in fat body and flight muscles. Because sugar-feeding is more crucial in zoophilic species than in anthropophilic species, at least partly because a greater portion of a human blood meal is converted into energy reserves, the zoophilic species usually take sugar every 3-4 days even if blood is accessible (Nayar and Sauerman 1975, Wittie 2003,

Fernandes and Briegel 2005). Another reason may be that An. gambiae and Aedes aegypti, two anthropophilic species, are also endophilic. Ae. aegypti mates around the blood host, in West Africa An. gambiae sometimes mates indoors (Dao et al. 2008), and both species often oviposit very close to houses. Thus frequent sugar-feeding may not be necessary. The importance of sugar-feeding to specialist and generalist blood-feeders was 3 demonstrated by comparing the flight ranges of An. gambiae and a generalist blood feeder, An. atroparvus, showing that An. atroparvus flew a longer distance, when 10% sugar solution was presented ad lib. Also, in An. atroparvus sugar-fed females flew a longer distance than the blood-fed ones (Kaufmann and Briegel 2004). Infected vectors that can for a longer distance, may contribute to the geographic distribution of diseases in a larger area.

Laboratory studies have shown that sugar-feeding decreases the biting frequency of An. gambiae (Straif and Beier 1996, Gary and Foster 2001). Also, sugar deprivation of females can either increase their response to host stimuli or decrease their susceptibility to repellents (e.g., Bowen and Romo 1995, Gary and Foster 2001, Braks et al. 2006).

However, physiological factors (e.g., body size, age, energetic reserve, etc.) may affect the frequency and volume of sugar taken by mosquitoes. It is likely that mosquitoes with higher energetic reserve take the first sugar-meal later than those with a low energetic reserve.

Ma and Roitberg (2008) reported a higher reproductive success in An. gambiae females when sugar was available. Maybe because females that fed on sugar had a higher chance of mating with the males that survived longer (Stone et al. 2009). These studies indicate that plant-derived sugar may be important to vectors of disease under natural conditions and may be an aspect of their vulnerability.

4 Location of Host Plants by Volatiles

Floral scents are complex mixtures of small volatile organic compounds (VOCs) with 5-20 carbons that can be produced by almost any floral tissue (Raguso 2001,

Dudareva and Pichersky 2006). Plants emit these volatiles to attract pollinators and reward them by providing sugar (Knudsen and Tollsten 1993, Jürgens et al. 2002).

Semiochemicals are produced and released at various rates depending on environmental conditions such as temperature and diel cycle (Metcalf and Kogan, 1987). In a study, the intensity and composition of its floral scents of Silene otites, often visited by moths, pollinators, and Cx. pipiens pipiens, changed during different times of the day (Dötterl et al. 2012). A higher emission of plant compounds can also signal the sugar abundance to pollinators. Guerenstein et al. (2004) demonstrated a direct correlation between the emission of CO2 from flowers of Datura wrightii and the sugar production. Insects that are innately attracted to the compounds can take sugar, and those with learning capability can associate the compound with sugar (i.e., reward). Wind or airflow is essential to deliver volatiles to the odor receptors of insects (Cardé 1996), and mosquitoes fly upwind towards the source of odor (Geier et al. 1999).

In the first record of mosquito attraction to plants, anophelines were found in the fruit-baited traps (Bates 1949). Honey has been used in previous studies as a attractant that can represent phytochemicals (e.g, De Meillon et al. 1967, Foster and Takken 2004,

DeGennaro et al. 2013). Increasing the release rate of honey VOCs in a dual-choice wind tunnel, resulted in a higher response of Cx. nigripalpuls to the honey (Hancock and

Foster 1997). However, in a mesocosm, An. gambiae showed an opportunistic behavior 5 depending on accessibility and quality of the sugar resources in mesocosm trials (Stone et al. 2012). Mosquitoes appear to capitalize on this relationship by using plant odors to locate sources of flower nectar, though few species pollinate plants. Extrafloral nectaries also may provide olfactory cues, but these scarcely have been investigated. Some field studies have suggested that mosquitoes fed on a very select number of plants (Abdel-

Malek and Baldwin 1961, Abdel-Malek 1964, Müller and Schlein 2006). Studies of mosquito responses to flower volatiles include attraction of Ae. aegypti to flowers of clovers, mustard, buckwheat, odors of rose perfume (Thorsteinson and Brust 1962), and extracts of Canada goldenrod and common milkweed (Vargo and Foster 1982). Healy and

Jepson (1988) demonstrated that natural flowers and extract of yarrow flowers attracted

Anopheles arabiensis. Also, in olfactometer bioassays, extracts of common milkweed

(Mauer and Rowley 1999), flowers of Silene otites (Jhumur et al. 2008), and natural flowers, extract and synthetic blend of common milkweed flowers (Otienoburu et al.

2012) were attractive to Cx. pipiens pipiens. Recent studies have demonstrated the attraction of An. gambiae to 13 plant species in Kenya (Manda et al. 2007a), 5 plants in

Burkina Faso (Gouagna et al. 2010), and Parthenium hysterophorus (Nyasembe 2012). It is interesting that the innate attraction of An. gambiae to P. hysterophorus is among the highest (Manda et al. 2007b), but the mosquito had the lowest survivorship because the plant does not produce any identifiable amount of monosacharide (Manda et al. 2007a).

There are two possible reasons for the attraction: a) the plant has other benefits to the mosquito and b) the attraction was based on the innate response and the mosquito can learn to visit other plants that provide sugar. These studies have demonstrated that odor 6 facilitates mosquitoes in their search for sugar. Apparently, mosquitoes prefer certain plants unless sugar availability is restricted to other plant species later in the season

(Foster 2008). The potential application of species-specific phytochemicals is to use in surveillance of those species. The opportunities for using these attractants to study and control mosquitoes are only beginning to be realized.

Practical Use of Phytochemicals

Attractants can be used in mosquito traps to detect early virus activity in captured mosquitoes and to monitor their populations. Mosquito surveillance workers have relied on light traps to collect mosquitoes, but those also catch many other insects, and their use is laborious and expensive because of the need for a source of electricity. Light has been supplemented or replaced by the deployment of and human kairomones such as carbon dioxide and octenol, which act as cues for females to locate blood meals, but these attract only the empty, blood-hungry portion of the female population. Gravid traps are cumbersome and collect only gravid females of the few species that are attracted to the odor of oviposition sites (Foster 1995, 2008). Phytochemical attractants have the potential to provide a widely applicable alternative. Males often emerge first, and they take only nectars from plants, so phytochemicals may be the best tool for locating sources of mosquito production and estimating adult populations. Female mosquitoes typically search for sugar prior to taking the first blood meal, but they also sugar-feed before subsequent blood meals and periodically during mid-gonotrophic cycle. Furthermore, females in prehibernation reproductive diapause take sugar frequently but do not take 7 blood meals and therefore cannot be trapped using human kairomones or gravid traps.

Preliminary evidence shows that phytochemicals are able to attract substantial numbers of mosquitoes in the field (Müller et al. 2010). Because the production of natural VOCs is restricted by time of the day and season, synthetic VOCs can be applied when the natural plants are not available. These compounds are often used in low amount in traps which may reduce the cost. The synthetic phytochemicals are created based on the headspace or extract of plants that are attractive to mosquitoes (e.g., Jhumur et al. 2008, Otienoburu et al. 2012). Mauer and Rowley (1999) found that milkweed flower is attractive to Cx. pipiens, but the synthetic blend was not attractive in olfactometer maybe because they created the blend based on the most abundant compounds. Specific formulations of

VOCs that have been shown to be attractive in the laboratory have yet to be thoroughly tested under field conditions, so the topic remains an active area of investigation.

It is interesting to study the attraction of mosquitoes to both phytochemicals and kairomones, because some mosquito species respond to both chemicals at the same time.

Thus, the combination of both attractants may either increase the catch of mosquitoes in the field traps, or result in an inhibitory effects the two attractants.

The most common methods of controlling mosquitoes are insecticide-treated bed nets (ITN) and indoor residual spraying (IRS). Nonetheless, Shaukat et al. (2010) suggested that ITN and IRS should be combined with other control methods, to be able to control malaria effectively. Plant scents and sugar can be good candidates to be used in integrated vector management (IVM). Phytochemicals can be used in combination with toxins to attract and kill mosquitoes. In recent years, many studies have shown the 8 efficiency of attractive toxic sugar baits (ATSB) to control mosquitoes (Müller and

Schlein 2006, 2008, Schlein and Müller 2008, Müller et al. 2010a, b, c, Beier et al. 2012).

In these studies, sugar was mixed with the toxins that act through digestive system of insects (e.g., spinosad and boric acid). The attractant was made by mixing rotting fruits, juice, wind or beer, and brown sugar. Then, the mixture was left at room temperature or in the sun for few days, so that yeasts and bacteria fermented the sugar and released volatiles. Field studies showed that the ATSB that was sprayed on vegetation around larval development strongly suppressed or completely eliminated mosquito populations.

ATSB effectively controlled the populations of An. sergentii and Aedes caspius in the oasis where the plants were sprayed with the mixture (Müller and Schlein 2006). In the similar study, the population of Cx. pipiens was controlled in an oasis, in an arid area, when the ATSB was used (Müller et al. 2010c). Also, when the ATSB was sprayed on plants around larval habitats in Mali, An. gambiae abundance drastically dropped, and the gonotrophic cycle was decreased (Müller et al. 2010a), indicating that vectorial capacity can be lowered effectively. However, the effect of using ATSB on other insects, particularly pollinators, has remained to be answered. It is possible to develop and use synthetic attractants in killing stations that mainly catch mosquitoes, or to apply toxins that are harmful to mosquitoes only. The synthetic odors, created based on the headspace profile of plants, have been attractive in laboratory experiments. But more studies are needed to develop synthetic attractants that can compete with natural plants. One may suggest that using the same insecticide and sugar in the ATSB may result in resistant in mosquitoes. Wada-Katsumata et al. (2013) have demonstrated that a strain of German 9 cockroaches has evolved to taste glucose, used in the sugar traps for many years, as a

“bitter” compound. But the mutant cockroaches were still able to taste fructose as a

“sweet” compound. One potential method to avoid similar resistance in mosquitoes is to periodically change the type of sugar and insecticide. The insecticide can also be substituted with a microorganisms that can potentially reduce mosquito infection (Lindh et al. 2006).

Another possible method to control mosquitoes is to remove the sources of sugar from an area (Abdel-Malek 1964). Comparison of mosquitoes' performance in sugar-rich and sugar-poor mesocosms has already shown how restricted access to sugar may affect the population, reproduction, and biting frequency of mosquitoes (Stone et al. 2012).

Also, one study showed that in an oases with blooming trees, females of Anopheles sergentii were larger and survived longer than females in the oases without blooming trees (Gu et al. 2011). However, it should be noted that even if all sugar-rich plants are eliminated, mosquitoes can feed on the honeydew, produced by pests that attack other plants. Also, it is not practical to remove all sources of sugar in areas where the sugar productive plants are widely distributed. It is important to investigate whether mosquitoes can quickly learn to find and obtain sugar from other sources, or if their populations decrease dramatically by removal of sugar sources.

The use of toxic plants that can kill either mosquitoes or the pathogen inside them has not been fully explored. If planted around human habitations, they can potentially decrease the number of infected human cases. However, the plants should be safe for human, animals, and pollinators; otherwise, they may pose a higher risk than insecticides. 10 Potential Influence of Learning on Attraction to Plants

The effectiveness of phytochemicals as attractants for surveillance and control of mosquitoes depends, to some extent, on their innate responsiveness to them. The same applies to the principle behind the prospects for suppressing their populations by removing their preferred sugar sources. If attraction can be modified by a mosquito’s experience, the specific composition of the plant community may determine its response to particular phytochemicals, rather than an innate preference for them, and it may be less susceptible to synthetic attractants used in traps or ATSB applications and to seasonal, geographical, or artificially induced differences among communities. Learning and memory assist individual organisms to find a location, food, mate or a breeding site.

Learning is defined as “a change in behavior due to an experience that increases the fitness of an individual because it recognizes and remembers the best resources” (McCall and Kelly 2002). Therefore, one might expect that mosquitoes will be flexible in their plant-host preferences. Previous studies have shown mosquitoes' ability to return to the location where a blood meal was obtained during the previous gonotrophic cycle (McCall and Kelly 2001), to oviposit on a larval development site where it had earlier emerged

(McCall and Eaton 2001), and to associate plant odors or single-compound with sugar

(Jhumur et al. 2006, Tomberlin et al. 2006 Sanford and Tomberlin 2011). Associative learning using visual cues and blood has also been reported in An. gambiae (Chilaka et al.

2012), indicating that associative learning is not limited to the olfactory system of this nocturnal species. Mosquitoes that can learn to readily locate sugar may have a greater 11 survival, which has the side-effect of increasing their vectorial capacity. Menda et al.

(2013) showed that Ae. aegypti females that were exposed to octenol or dark surface while experienced electric shock, avoided the odor (i.e., octenol) or the dark surface (i.e., the visual cue) for one hour. But the conditioned mosquitoes did not avoid the odor or visual cue after 24 hr, indicating that probably the incident was remembered for a short time. Mosquitoes' ability to learn may complicate the control methods, as laboratory experiments showed that pre-exposure of Ae. aegypti females to DEET can result in habituation to the compound (Stanczyk et al. 2013). The results to date indicate that that the ability of vectors to learn does exist and can have major epidemiological consequences, but this topic has barely been explored.

How Plant Communities Can Alter Vectorial Capacity

Vectorial capacity is defined as “the total number of new infected hosts that can be derived daily, directly from one original infection in a specific environmental condition” (Garrett-Jones 1964, Dye 1992). It is often used to understand how mosquito populations and their behavior may influence the transmission of a disease. The main components of vectorial capacity are female density per human host, biting frequency, survival rate, and extrinsic incubation period. Among the components, survival rate and biting frequency have the greatest effect on vectorial capacity. An. gambiae survive longer on a 50% sucrose solution and cassava (Manihot esculenta), followed by castor bean (Ricinus communis) with accessible extra-floral nectaries, and lantana (Lantana camara) with flowers (Gary and Foster 2005). In the laboratory, sugar-feeding increases 12 longevity and decreases the biting frequency of mosquitoes, which affect vectorial capacity in opposing directions (Straif and Beier 1996, Gary and Foster 2001). Recent studies in the field have shown that the availability of plants providing large amounts of sugar results in a profoundly higher vectorial capacity of the local mosquito population

(Gu et al. 2011, Beier 1996). Another study, conducted in a large mesocosm, found just the opposite (Stone et al. 2012). Therefore, the outcome of the balance between the influences of sugar on biting frequency and survival remain uncertain. The other factor that may complicate measuring vectorial capacity is the body size of mosquitoes. Small females of An. gambiae (approximately 3 mm wing length) need an additional blood meal (Feinsod and Spielman 1980, Lyimo and Takken 1993), but they are not as successful as large females in finding a host. Therefore, smaller An. gambiae females are most likely to take sugar meals earlier and more frequently than larger mosquitoes.

Egg production (e.g., Eischen and Foster 1983, Manda et al. 2007) and insemination rates can be increased when sugar is available. Thus, sugar can indirectly affect the density component of vectorial capacity. Males' survival is directly affected by sugar availability, and males that survive longer have a greater chance to mate and inseminate females. Also, in most species males form a swarm to mate with females, and only males that have enough energy may successfully inseminate females. After swarm, males of An. freeborni became so depleted that they needed to feed on sugar (Yuval et al.

1994). However, more reserve may increase the males' survival even if sugar is absent. It should be noted that the space in which the studies are conducted can affect insemination rates (Gary et al. 2009) and survival. Understanding how this plays out under realistic 13 conditions is essential to any prospects for decreasing pathogen transmission by removing specific plant hosts or otherwise altering the plant community.

Objectives

The objectives of this dissertation can be couched in the following questions and the methods used to answer them:

1) What type of olfactometer, used to evaluate mosquito attraction to phytochemicals, gives the greatest response and discrimination, when used with the mosquitoes Aedes vexans and Anopheles gambiae? Several designs were built and tested in the laboratory, using odors known from previous studies to be attractive to mosquitoes.

2) Which sources of odor are most attractive to Aedes vexans and Culex spp. mosquitoes in the laboratory and field? Whole flowers, flower extracts, synthetic blends of flowers, and honey were all used as test attractants in olfactometers and in battery- powered field traps.

3) Can An. gambiae learn to associate the odors of plants with the likelihood of obtaining sugar? Its potential for associative learning between phytochemicals and sugar was evaluated by applying several conditioning methods with both whole plants and individual compounds and testing the results with an olfactometer.

4) Does the sugar-productivity of a plant community raise or lower the vectorial capacity and fecundity of an An. gambiae population? High-sugar and low-sugar mesocosms within large greenhouse rooms were used to measure biting frequency, survival, insemination rate, and fecundity of mosquito cohorts. 14 5) Why do An. gambiae eggs fail to hatch if they remain undisturbed in their oviposition sites? Laboratory experiments tested the effects of repeated mechanical stimulation on hatching and the viability of eggs left undisturbed for various durations.

6) Can a laboratory light source be designed and built that simulates natural light by eliminating all flicker? An electronic device was created that both eliminates flicker and can control the duration of the diel light cycle and its crepuscular periods.

15 Chapter 2: Olfactometer Design for Aedes vexans and Anopheles gambiae*

Abstract

To design and test a good olfactometer for Aedes vexans, a short Y-shape and a long dual- port type wind-tunnel olfactometer were tested, with honey as the bait in one arm versus an unbaited arm (blank). All parts of the Y-shape olfactometer and flight chamber of the dual-port olfactometer were made of acrylic plastic. Arms and funnel traps in the dual- port olfactometer were made of glass. To assess the highest difference between responses of two arms, the discrimination index was calculated by dividing the number of mosquitoes in one arm, often the honey arm, by the number of mosquitoes in both arms.

Overall, the best discrimination was observed in a dual-port olfactometer, but the total responses, i.e., the number of mosquitoes in both arms divided by the total number of released mosquitoes, did not differ between the two olfactometers. The highest total responses of Anopheles gambiae were demonstrated in a polypropylene olfactometer based on a design by Jepson and Healy. The discrimination indices were significant when the polyethylene- and polypropylene-made olfactometers were setup with honey vs. blank. The Jepson-Healy olfactometer shared a similar design with this dual-port

* B. Ebrahimi, P. E. Otienoburu, P. L. Phelan, W. A. Foster. This chapter is not intended for publication. 16 olfactometer, but the new one was made of disposable polyethylene plastic bags, particularly because of its lower cost.

Keywords: Aedes vexans, Anopheles gambiae, olfactometer

Introduction

Olfactometers have been designed and used to study the role of odors in the behavior of mosquitoes and in discovering a variety of zoochemicals, phytochemicals, and repellents (Mayer and James 1970, Carlson et al. 1973, Takken 1991, Dogan and

Rossignol 1999, Mauer and Rowley 1999, Geier et al. 1999, Bosch et al. 2000, Allan et al. 2006, Jhumur et al. 2007, Otienoburu et al. 2012). There are many advantages in using olfactometers, such as their use at any time of year, the small space they occupy, their easy setup and low cost, their use in recording behavior for detailed analyses (Geier and

Boeckh 1999).

Two main types of olfactometers have been used in bioassay experiments with mosquitoes: Y-shape and dual-port wind-tunnel olfactometers. These olfactometers are usually “dynamic” (i.e., they are wind tunnels), because air is pumped or blown into their arms and then taken out from back of release chambers.

17 The oldest Y-shape olfactometer was designed by McIndoo (1926). The most common Y-shape olfactometer adapted for mosquitoes was designed by Geier and

Boeckh (1999), in which mosquitoes are released into a single 13-cm-diameter flight tunnel, fly 50 cm upwind in the tunnel, then enter either of the two arms of the “Y” configuration. One of the earliest dual-port wind-tunnel olfactometer for mosquitoes

(referred to as dual-port in this paper) was designed and used by Schreck et al. (1967).

Since then, several variations of these have been used , including a dual-port modification of Geier’s original Y-shape design (Geier et al. 1999). In both types, usually one arm is baited whereas the other arm is left blank. Mosquitoes fly upwind, toward the air that comes from both arms, then at the juncture they choose between the air in the clean and baited arms. In some experiments, the two arms are baited with different substances, so that preferences between two odors can be measured simultaneously.

The main objective of this study was to compare the Y-shape and dual-port olfactometers for Aedes vexans. Also, the response of Anopheles gambiae in polyethylene and polypropylene bags, used in a dual-port Jepson-Healy olfactometer, were compared.

The goal was to choose an olfactometer for each species, with both a high response to the baited arm and a high discrimination between baited and blank (control) arms. Finally, the response of An. gambiae to trapped adults of the same species was tested in a Jepson-

Healy olfactometer.

18 Materials and Methods

Rearing and Maintenance of Mosquitoes

Anopheles gambiae:

The Mbita strain of An. gambiae s.s. (S form) was originally colonized and identified from locally collected material in 2001 by the staff of the International Centre of Insect Physiology and Ecology (ICIPE) at Mbita Point, Suba District, Nyanza, Kenya.

A colony of this strain has been maintained at The Ohio State University for several years

(Biosafety protocol No. 2005R0020). All stages were held in a rearing room at 25.5 ±

1°C, 70 ± 5% RH, and L:D about 12:12 h with 45-min crepuscular transitions between photophase and scotophase (2000 and 0730 hours). Larvae were reared in shallow aluminum pans in aged tap water, as described by Gary and Foster (2001). On the eighth and ninth days of feeding, pupae were removed with a clean plastic pipette and transferred to a plastic container (9.0 cm diameter by 1.2 cm height) filled with 50-75 mL water. These containers were placed in a cage (16 by 21 by 27 cm). Emerged adult mosquitoes had access to water and 10% sucrose solution ad libitum. To obtain eggs, one of the authors (B.E.) allowed 50-70 females to feed on his arm (Biomedical IRB protocol

No. 200440193, FWA No. 00006378). The blood-feeding duration was about 15 min.

Gravid mosquitoes laid eggs 3 days after taking blood. Post embryonation, >2 days after oviposition, eggs were agitated to induce hatching as needed (see Ebrahimi et al. 2013) for olfactory experiments.

19 For olfactory experiments, 200 pupae were collected and placed in a cage as described above. Approximately 75% of the pupae would emerge the following morning.

Adults had access to water only.

Aedes vexans:

Attempts to rear Ae. vexans collected from field in Ohio was not successful, so the

Niagara River strain of Ae. vexans, collected in Ontario, Canada, and cage-adapted and provided by R. Kuhn of the University of Mainz, Germany, was maintained and used at

The Ohio State University (Biosafety protocol No. 2005R0020). All stages were held at

26 ± 1°C, 70 ± 5% RH, and L:D about 15:9 h with 45-min crepuscular transitions between photophase and scotophase (2130 and 0600 hours). Adults fed on human blood

(W.A.F) weekly. Water and a 10% sucrose solution were provided in separate tilted 10- mL vials with protruding wicks. The oviposition site was a container filled with damp cotton rolls beneath 2-4 layers of damp cheesecloth, with fresh moss on top

(approximately 2 cm depth). The oviposition container was always kept damp with sprayed water. Eggs were hatched by submerging some of the materials (e.g. cotton rolls, cheesecloth, or moss) in aged tap water containing a small amount (<10mg) of a yeast- lactalbumin-rat chow mixture. Because it was not possible to estimate number of eggs in the oviposition materials, various numbers of first-instar larvae were obtained upon each submersion. Two-hundred first-instar larvae were transferred to an aluminum pan containing 450 mL of aged tap water. Larvae were fed daily with increments of 60, 120,

20 120, 120, 180, and 180 mg powdered Tetramin®. Collected pupae were transferred to a container and put in a cage (as described above). Adults had access to water at all times, and they were tested in the olfactometers 2 d post-emergence.

Adult preparation for olfactometer experiments

In most previous olfactometer studies, mosquitoes were aspirated from a cage and transferred to a container or chamber connected to an olfactometer. They were released after 30 min to 3 hr of acclimation (e.g., Hern and Dorn 2001). To minimize behavior- modifying agitation prior to olfactometer tests (cf. Bateson et al. 2011), pupae were allowed to emerge in a cage (16 by 21 by 27 cm), then the container with unemerged pupae was slowly removed from the cage. The number of unemerged pupae were counted to determine the number of emerged adults. Unemerged pupae were added to pupae that were transferred daily to a new cage. Because males often emerged 1 d before females, adding unemerged pupae (presumed to be female) kept the sex ratio at close to 50:50.

One to two hr before onset of an olfactometer test, each cage of adult mosquitoes was transferred from the rearing room to the bioassay laboratory, where the olfactometer was ready on a bench top and the light regime was the same. Cages had a 5-inch (12.7- cm) diam. sleeved opening in front for access into the cage and a 4-inch (10.2-cm) diam. screen in back for ventilation. Before a test, the sleeved opening of the cage was connected to the flight chamber, and a 4-inch- (10.2-cm-) diam. exhaust duct was connected to the screen in the back of cage, to direct effluent air to a hood and out of the laboratory. Therefore, the mosquito cage served as a release chamber. 21 Mosquitoes were released into the flight chamber approximately 2 hr prior to scotophase, by connecting the cage and flight chamber. A network of white Christmas mini-lights was suspended 1.5 – 1.7 m above the olfactometer. To reduce and spread the light, a 1.5 by 3-m black cotton cloth was suspended horizontally 20 cm below the lights.

To shield mosquitoes from visual distractions on all sides, a curtain of black cotton cloth surrounded the olfactometer. This horizontal cloth was suspended down to the side of the bench top. Tests ran for 12 hr, overnight, to increase total catch. This long period was allowed, because previous video-recordings had demonstrated that the mosquitoes responded to volatiles during several phases of the diel cycle in olfactometer tests (e.g.,

Healy and Jepson 1988, Jepson and Healy 1988, Otienoburu et al. 2012), and had indicated that, during this 12-hr test, habituation to odor did not occur at the release rate used. The trapped mosquitoes in the arms, remaining in the flight chamber, or still in the release cage were counted at the end of each test.

Olfactometer Designs

Y-tube olfactometer for Ae. vexans

The Y-tube olfactometer was constructed according to the design of Geier and

Boeckh (1999) with some modifications. The flight tunnel was replaced by a dome-shape plastic flight chamber that fitted the opening of the cage (Fig. 2.1). The angle between the two arms of the olfactometer was 90°. The two arms were 7 cm apart at their closest distance in the dome-shape flight chamber. Each arm was 20 cm long by 7 cm diameter.

Two clean and clear acetate sheets, forming funnels with a 13-cm-wide opening and a 22 narrower upwind end of 3 mm diam, were placed inside each arm, so that the distance between the wide opening and the flight chamber was 5 cm. The wide ends opened into the flight chamber and the narrower ends pointed into a cap that fitted on the arm. There was a 7.5-mm-diam. hole at the center of each cap for air-tube connections. To release mosquitoes, the opening of the dome was gently passed through the stockinette until the dome was fitted snugly into the opening of the cage.

Dual-port olfactometer for Ae. vexans

The other type of olfactometer, redesigned in this study, was a wind tunnel with dual-ports, adapted from Hancock and Foster (1993). In the new design, a 30 x 40-cm- wide, 90-cm-long wind tunnel made of clear acrylic was used (Fig. 2.2). A cage of adult mosquitoes was placed at one end of the wind-tunnel (flight chamber), where there was a

21-cm-diam. hole at the center, approximately 4 cm above the bottom of the flight chamber. A connector with a sliding gate was placed between the cage of mosquitoes

(release chamber) and the flight chamber. Two cylindrical glass jars (15 cm long by 7 cm diam.) were placed in two holes (7.2 cm diam.) at the upwind end of the flight chamber.

One borosilicate glass funnel, with the wide end opening into the flight chamber, was placed on the opening of each jar and held in place with Parafilm ® (Fig. 2.3). The narrower end of the funnel (3 mm diam.) was pointing into the jar. When jars and funnels were in place, connected to the flight chamber, the distance between the small openings of the two trapping arms was 21 cm. Purified and humidified air was introduced into each jar through a 7.5 mm hole in the upwind end of each jar. This trapping system was 23 designed so that mosquitoes would fly upwind and enter the jar through the hole in the funnel but could not fly back to the flight chamber.

The amount of air that was blown into the olfactometer could not keep the humidity in the flight chamber at an optimum level (70 ± 10% RH). To maintain that level, a layer of dampened cotton wool, covered by a black cotton cloth, was placed on the floor of the flight chamber.

Jepson-Healy olfactometer for An. gambiae

Among the disadvantages of commonly used olfactometers are high cost of construction, need for laborious maintenance, and possibility of chemical contamination between replicates. Jepson and Healy (1988) designed and used an olfactometer made almost entirely of disposable polyethylene plastic tubing, to test attraction of Aedes aegypti and Anopheles arabiensis to plant volatiles (Healy and Jepson 1988, Jepson and

Healy 1988). In the present study, a modified Jepson-Healy olfactometer was designed and tested for An. gambiae. It followed the dual-port olfactometer (see above) design, but with a shorter wind-tunnel (i.e., flight chamber), with baffle ports, and with materials used by Jepson and Healy (Fig. 2.4). The baffle ports were used because Verhulst et al.

(2008) had already assessed their performance for An. gambiae and found them to be more effective. Jepson and Healy had placed a thin wire framework within the polyethylene tubing to make a flight chamber. However, in this new design, a 32 by 32 by

20 cm framework was placed outside of a disposable plastic bag, which served as a flight chamber, with 1-3 small ACCO binders clips on each wire of the framework to maintain 24 the box-shape construction of the bag. Therefore, it was unnecessary to wash and heat- treat wires between replicates, and the framework could not puncture the bag during set- up. Both polyethylene (Uline ®, Milwaukee, WI) and polypropylene (Ovenbag, Reynolds

®) bags were tested, to select the material that would result in the highest response and discrimination between treatment and control traps. The bags were 51 by 76 cm with one open end (51 cm). Approximately 5 cm of the opening of the bag was held by a rubber band around wider end of a 4-to-5-inch (10.2- to 15.2-cm) diam. galvanized connector.

When the olfactometer was ready to run, the 4'' end of the connector was pushed into the stockinette that covered the opening of the releasing cage of mosquitoes, so that mosquitoes could easily fly into the flight-chamber bag.

Each olfactometer had two baffle ports, each made of 5.5-cm-long by 4-cm-diam. aluminum EMT Screw Connector Concrete (Fig. 2.3). Three screws had been drilled into a locknut to hold a 5.5-cm diam. metal screen (New York Wire ® 10205) perpendicular to the length of the connector. A 3-mm gap between the screen and the edge of the aluminum connector was adjusted by turning the locknut. To connect the flight chamber to the two arms, two 4-cm-long openings were made along the sealed end of the bag, approximately 40 cm apart. Then the open end of the port was inserted half its length into the 4-cm openings of the bag, and it was secured with rubber bands. Each trapping arm was a 25- cm-wide by 35-cm-long plastic tube, made of the same material as the flight- chamber bag. To make a trapping bag, both ends of the plastic tube, except for 4 cm at the center of one wide side, were sealed. The screen end of the aluminum port was inserted half its length into the 4-cm opening of the trapping bag, and it was secured with rubber 25 bands. The distance between the center of the two ports was 31 cm. Air was directed into the trapping bag through Tygon tubes, which were connected to the bags by 1.1-cm-long by 4.8-mm-diam. brass connectors.

After the mosquitoes were counted at the end of each olfactometer test, all bags were discarded, and metal components (e.g., aluminum connectors, screens, brass connectors, and 4'' to 5'' galvanized connectors) were washed with scalding water and heated at 120°C in an oven for 1-2 hr. Then they were allowed to cool to room temperature before being used in the next test.

Air supply

In all dynamic olfactometer tests in this study, air was supplied by a GAST diaphragm-type compressor (Model DOA-P704-AA), Benton Harbor, MI. A Tygon ® tube (AAB00007) carried air from the pump to two cylinders of activated carbon, which purified air, and then into a bubbling system (26 ± 1°C deionized water) to increase humidity to 60-70% RH and to keep the temperature at 26 °C. Clean metal air manifolds or glass Y-shaped connectors were used to split the air. Smaller Tygon tubes carried air into the air connectors of the olfactometer traps. Flowmeters were connected to the tubes to measure the airflow. If it was necessary to adjust the airflow, tube clamps were used on

Tygon tubes, or adjustable manifolds were installed. Then the airflow of each tube was adjusted before connecting it to an olfactometer arm. Airflows were maintained at 24 mL/s and 12 mL/s for Aedes vexans and Anopheles gambiae respectively.

26 Bioassay Experiments

All components of the olfactometers were handled with clean gloves. Metal and glass parts were washed and cleaned with hot water and heated in an oven at 120°C for 1-

2 hr to dry. Activated carbon was replaced every 6 mo, when all Tygon tubes were replaced. Except for the flight chamber and arms of the Jepson-Healy olfactometer, which were used and disposed after each olfactometer test, the acrylic plastic components of the dual-port and Y-shape olfactometers were washed with scalding water and wiped with clean, dry paper towels. At the end of each olfactometer test, the number of mosquitoes in the arms, flight chamber, and cage were counted. The response, which was the proportion trapped in each arm, was calculated by dividing the number of mosquitoes in the arm by the total number of released mosquitoes. For each test, the discrimination index was calculated by dividing the number of responding mosquitoes in the treatment arm, which was typically honey, by the total response in both arms (i.e., treatment and control) as follows:

Discrimination Index = Treatment / (Treatment + Control)

An outcome of 1 indicates that all mosquitoes chose the treatment – an ideal response, whereas an outcome of 0 shows the poorest performance.

The attractant used for the bioassays was clover honey, Great Value ® (Hillsboro, KS), purchased from WalMart. Previous studies have used honey as a standard attractant, to test mosquitoes in olfactometers (Hancock and Foster 1993, 1997, Foster and Takken

2004, DeGennaro et al. 2013).

27 Comparison of Y-shape and Dual-port olfactometers for Aedes vexans

At the time of designing this experiment, only Y-shape and dual-port olfactometers were constructed. The main goal was to compare the efficiency of these olfactometers for Ae. vexans.

First, each olfactometer was tested by releasing mosquitoes as described above, without adding any attractant to either arm of the olfactometer to make sure that purified/humidified air does not attract mosquitoes. No mosquito was captured in either of the arms during 12-hr olfactometer tests. Then, the interior wall of one of the arms, except for the top 2 cm, was coated with 5 g clover honey, and the funnels were held in place using Parafilm. The second arms of each olfactometer was blank, i.e., it was not treated with anything (control). To determine which type of olfactometer gave the best discrimination, indices calculated for the Y-shape and dual-port olfactometers were compared to each other. Respectively, four and six replicates were conducted for the Y- shape and dual-port olfactometers.

Comparison of polyethylene- vs. polypropylene-made olfactometers for Anopheles gambiae

Two plastic materials were tested with An. gambiae to assess which one allowed both the highest total response and the best discrimination between baited and blank arms. The

Jepson-Healy olfactometers were made with one or the other of these two materials, then the following types of experiments were conducted:

a) 2.5 g honey in one arm and nothing in the other (blank) 28 b) 2.5 g honey in both arms

c) no bait in either of the arms (blank)

Each test of150 ± 20 adults (combined sexes) was run 7 – 10 times.

Response of An. gambiae to the trapped adults of same species

In general, a control arm, usually an arm without any material (blank), has been used in olfactometer tests, to take into account the effect of other factors (e.g., anemotaxis to the airflow). If individuals of a species interact with each other (e.g., using social or sex pheromones), olfactometer tests are run with one insect at a time. So far, there have been no confirmed reports of sex pheromones or social pheromones for An. gambiae or other mosquitoes. Therefore, adults of this species have been released in large numbers in all olfactometry studies, on the assumption that individuals behave independently of one another. Nonetheless, this experiment was designed to ensure that An. gambiae males or females already trapped in one arm have no effect on others still in the release cage or flight chamber.

Adults were kept on water for 2 d post-emergence (2-d-old). Then 50–135 2-d-old adults (equal sex ratio) were transferred into one of two clean cylindrical screens (10 cm diam. by 20 cm length) through a 2-cm-diam. hole on one side of the cylinder. To provide water for mosquitoes, three cotton wicks were soaked with water and put inside the cylinder. The hole was blocked using a piece of clean aluminum foil, then the cylinder was placed inside one of the arms of the olfactometer. The second cylindrical screen, with three dampened wicks and aluminum foil, was put in the second arm of the olfactometer. 29 Thus, released mosquitoes had a choice of the “mosquito arm” or the “blank arm” in the

Jepson-Healy olfactometer. The trapped mosquitoes in each arm were counted after 12 hr.

This experiment was repeated 9 times with alternating positions of the “mosquito arm”.

Statistical Analyses

The number of mosquitoes released in each olfactometer test was 150 ± 20, and each test was repeated 3-10 times. Therefore, between 400 and 1500 mosquitoes were tested for each objective. Chi-square goodness-of-fit tests were used to compare the number of responding males and females in each arm. Expected values were calculated with the assumption that the total number of responses, by either sex, were equally divided between the two arms.

The discrimination index was calculated for each replicate, by dividing number of mosquitoes in the baited arm by the total number of mosquitoes in both arms. Angular transformation was applied to the proportion, and the value was tested using a 'one- sample t-test' with the assumption of a 50% response in each arm (Arcsin√0.5 = 0.785), indicating no difference between the two arms (Allan and Kline 1995, Dekker et al.

2001). The above analysis shows how accurately mosquitoes may find the source of a compound in one arm. However, to reveal how effective a chemical compound is in attracting mosquitoes, the total response was reported. To calculate the total response, the total number of mosquitoes in both arms was divided by the total number of released mosquitoes. The total responses between two different olfactometer experiments were compared using an independent-samples t-test. Whenever data were not normal, Mann- 30 Whitney U tests were used instead of an independent-samples t-test. A two-way ANOVA was used to detect interactions between the bag materials and type of experiment (i.e.,

Honey vs. Honey, Honey vs. Blank, and Blank vs. Blank). For all statistical analyses, the criterion for significant differences was α = 0.05 (Sokal and Rohlf 1995, Zar 2010).

Results

Comparison of Y-shape and Dual-port olfactometers for Aedes vexans

Because there were no significant differences between the two sexes in response to the honey and blank arms (P > 0.05) in the Y-shape and dual-port olfactometers, the sexes were pooled for further analyses.

a) Y-shape olfactometer: Fig. 2.5 shows the mean responses of mosquitoes to the

5 g of honey vs. blank arms in the Y-shape olfactometer during a 12-hr test. The mean percentages of response of released mosquitoes to the honey and control arms were

51.7% (SE = 5.2) and 19.9% (SE = 2.4), respectively. Also, the discrimination index showed that 71.9% (SE = 3.2) of the total responding mosquitoes were found in the honey arm, a significant honey bias (t = 6.341, df = 3, P = 0.008).

b) Dual-port olfactometer: The mean responses to 5 g of honey vs. control arms of the dual-port olfactometer were 50.3% (SE = 5.4) and 2.2% (SE = 2.2) respectively, as shown in Fig. 2.5. Out of all responding mosquitoes, 95% (SE = 5) chose honey over the blank (t = 7.13, df = 5, P = 0.001).

31 This discrimination index for the dual-port olfactometer was significantly higher than the index for the Y-shape olfactometer (Mann-Whitney U test = 2; n1 = 4 and n2 =

6; P = 0.038).

To evaluate the relative efficiency of the Y-shape and dual-port olfactometers, the responses to honey, out of the total mosquitoes released, also were compared. The same analysis was applied to responses to the control arms in the two olfactometers. The responses to honey arm were the same in both olfactometers (t = 0.176, df = 8, P = 0.87), whereas fewer adults were trapped in the control arm of the dual-port olfactometer compared to control arm of the Y-shape one (t = 5.38, df = 8, P = 0.001). This was the reason for a significantly higher total response in both arms of the Y-shape olfactometer

(80.1%, SE = 7.6%), compared to the dual-port olfactometer (55.6%, SE = 4.9%) (t =

2.93, df = 8, P = 0.019).

Comparison of polyethylene- vs. polypropylene-made olfactometers for Anopheles gambiae

Both sexes of An. gambiae were combined for further analyses because mosquito responses were not affected by sex. The results to both arms in polyethylene and polypropylene bags are presented in Fig. 2.6. To see whether the response to the two arms differed in each test, the discrimination indices were calculated and analyzed using a one- sample t-test after angular transformation (Table 2.1). The indices for the polyethylene- and polypropylene-made olfactometers were 74.6% and 73.1%, respectively. The statistical analyses showed that, regardless of the bag material, when only one arm was 32 baited with honey, mosquitoes were significantly attractive the honey-arm. A two-way

ANOVA did not detect any interactions between type of experiment and bag material on the results of total responses and discrimination indices (P > 0.5). The discrimination indices of the three types of experiments varied significantly, with the highest mean discrimination indices when only one arm was baited with honey (F = 6.1; df = 2, 36; P =

0.005). The mean total responses of all three types of experiments was higher when polypropylene bag was used (F = 5.69; df = 1, 36; P = 0.02). Detailed analyses revealed that the total response to both blank arms was higher in the polypropylene bag than in the polyethylene bag (Table 2.2).

Response of An. gambiae to trapped adults of same species

Responses of mosquitoes to each arm were not affected by sex; therefore, all sexes in each arm were pooled together for further analyses. As shown in Fig. 2.7, the mean percent response of released mosquitoes to the arm holding mosquitoes was 11.1%

(SE = 3.4), not statistically different from 7.4% (SE = 2.9) trapped in the control arm.

Also, 61.1% (SE = 9.3) of all responding mosquitoes were trapped in the arm that was baited with mosquitoes, an insignificant difference (one-sample t-test = 1.298, df = 8, P =

0.23), indicating that the presence of mosquitoes in the arm of olfactometer did not influence the non-responding mosquitoes.

33 Discussion

Comparison of Y-shape and Dual-port olfactometers for Aedes vexans

Discrimination between honey and blank arms was higher in the dual-port olfactometer. The total response in the Y-shape olfactometer was higher, probably because of the very short flight chamber and the short distance between the two arms.

The preliminary tests with the Hancock-Foster olfactometer (Hancock and Foster 1993,

1997), using Ae. vexans and Culex pipiens, showed a lower discrimination between honey and blank than the same comparison in the dual-port olfactometer of this study.

Both olfactometers were approximately the same length; however, the distance between the two arms in this olfactometer was greater. The distance between the two arms was probably one of the reasons that other researchers changed the Y-shape design to a dual- port olfactometer for Aedes aegypti (Geier and Boeckh 1999, Geier et al. 1999), allowing mosquitoes to hover between two parallel airstreams emerging from the two arms. In some studies, the arms were place at opposite sides or ends of a long wind tunnel (Dogan and Rossignol 1999), sometimes with counter-direction airflows that were exhausted from middle of the flight chamber (Torto et al. 2010, Nyasembe et al. 2012). In this study, a dual-port olfactometer with parallel airstreams was designed for further studies with

Ae. vexans and Cx. pipiens (Otienoburu et al. 2012), mainly because of better discrimination.

34 Comparison of polyethylene- vs. polypropylene-made olfactometers for Anopheles gambiae

Mosquitoes in both polyethylene- and polypropylene-made olfactometers were able to successfully discriminate the honey arm from the blank one. The mean total response was higher in the polypropylene-made olfactometer when one arm was baited with honey. A low total response was expected when both arms were blank. However, the total response of the blank-vs.-blank experiment was higher in the polypropylene bag than in the polyethylene one, indicating a higher mosquito activity in the polypropylene bags. Polyethylene plastic bags, therefore, are recommended for use in Jepson-Healy olfactometers, particularly because the cost of making an olfactometer with polyethylene was approximately one-third of making the same-size olfactometer with polypropylene bags.

Response of An. gambiae to the trapped adults of same species

Results of this test showed that the mosquitoes are neither attracted to, nor repelled by, each other at distance. It has been reported that the behavior of Aedes sierrensis (Ahmadi and McClelland 1985) and Aedes aegypti (Cabrera and Jaffe 2007) is affected by pheromones. However, this experiment showed that it is not the case for An. gambiae, and the results of testing mosquitoes in large numbers are probably similar to those of testing individual mosquitoes. Thus, mass-release of mosquitoes in olfactometers, which is typical of olfactory studies using them (e.g., Geier et al. 1999,

Foster and Takken 2004) does not compromise or confuse the results. 35 Acknowledgements

The authors are most grateful to Roland Kuhn for the use of his Niagara River strain of

Aedes vexans and they thank Philip E. Otienoburu and Christopher M. Stone for their suggestions and help in designing olfactometers. We also thank Leila Farivar, Department of Economics, The Ohio State University, for her thoughtful discussions on statistical analyses, and Robert L. Aldridge, Ashley N. Jackson, and Benjamin D. Barker for maintaining mosquito colonies and providing adults for experiments. This work was supported by NIH grants R01-AI064506 and R01-AI077722 from the National Institute of Allergy & Infectious Diseases (NIAID) to W.A.F. Its content is solely the responsibility of the authors and does not represent the official views of the NIAID for the National Institute of Health.

36 Table 2.1: Comparison of discrimination indices (percentage of mosquitoes in one arm / mosquitoes in both arms) in Jepson-

Healy olfactometer, constructed from two different materials. The comparisons were made according to the type of experiment,

i.e., Honey vs. Blank, Honey vs. Honey, and Blank vs. Blank arms (Honey = 2.5 g), with H0: Discrimination Index = 0.5, and H1:

Discrimination Index ≠ 0.5. Significant P-values indicate that mosquito response in one arm was higher than the other arm.

Polypropylene Polyethylene

3 Blank vs. Blank vs. 7 Honey vs. Blank Honey vs. Honey Honey vs. Blank Honey vs. Honey Blank Blank Discrimination Index 73.1 ± 8.1 a 43.2 ± 9.7 42.7 ± 12.4 74.6 ± 8.7 a 36.9 ± 16.8 45.2 ± 14.9 (% ± SE) P-value of one- 0.022 0.587 0.565 0.018 0.487 0.788 sample t-test

a Discrimination index in Honey vs. Blank experiment was calculated by dividing the number of mosquitoes in the Honey arm by

those in the Blank arm. Table 2.2: Percentage of total responses to both arms (mean percentage ± SE) in each type of experiment, comparing olfactometer materials:. polypropylene vs. polyethylene.

Polypropylene Polyethylene t-test df P-value Experiment Type

Honeya vs. Blank 56.5 ± 6.9 32.6 ± 5.9 2.66 18 0.016

Honeya vs. Honeya 47.4 ± 8.7 46.8 ± 14.8 0.1 10 0.92

Blank vs. Blank 42.8 ± 10 8.5 ± 2.9 2.44 9 0.037

a Honey = 2.5 g

38 Fig. 2.1: A drawing of the Y-shape olfactometer, connected to a mosquito- release cage.

Fig. 2.2: Parts of assembled dual-port olfactometer. Direction of airflow is left to right, from trapping jars towards mosquito-release cage.

39 Fig. 2.3: Left: front view of a funnel port installed on a jar. Right: side view of a baffle port.

Fig. 2.4: Three Jepson-Healy olfactometers made of polyethylene plastic bags.

40 Honey Control

Dual-port Dual-port

Y-shape Y-shape

60% 50% 40% 30% 20% 10% 0% 10% 20% 30%

Mean response (%)

Fig. 2.5: Mean response percentage of Aedes vexans to 5 g honey and control in Y-shape and dual-port olfactometers, after 12 hr.

41

Polyethylene

Blank Blank

Honey Honey

Honey HoneyBlank

50% 40% 30% 20% 10% 0% 10% 20% 30% 40%

Blank Blank

Honey Honey

Honey HoneyBlank Polypropylene

50% 40% 30% 20% 10% 0% 10% 20% 30% 40%

Fig. 2.6: Mean response percentage of Anopheles gambiae adults (both sexes) to two arms, baited with honey (2.5 g) or left blank, in olfactometers made with polyethylene or polypropylene bags.

42 30

25 ) % (

d e

s 20 a e l e r

l a t o t 15 m o r f

e s n o

p 10 s e r

n a e M 5

0

With mosquitoes No mosquitoes

Fig. 2.7: Mean percent responding of released 2-d-old Anopheles gambiae mosquitoes to the arms of Jepson-Healy olfactometer baited with males and females of the same species vs. no mosquitoes (number of replicates = 9).

43

Chapter 3: Innate Attraction to Plant Volatiles *

Abstract

Phytochemicals can be used for surveillance and control of mosquitoes. Flowers of

Canada goldenrod (Solidago canadensis) and common milkweed (Asclepias syriaca) were attractive to Aedes vexans compared to a blank arm in olfactometer tests, however honey was more attractive than the flowers. Field experiments were designed based on uneven distributions and densities of Ae. vexans and Culex spp. to test attractiveness of fresh flower cuttings, a flower extract, a milkweed minimal synthetic blend, and honey in

CFG traps. Both sexes of Ae. vexans and Culex spp. males were attracted to goldenrod flowers, however goldenrod extract was attractive only to males of Culex spp. Both sexes of Ae. vexans were caught more often in CFG traps baited with milkweed flowers, but there was no difference in the number of mosquitoes that were collected from traps baited with either honey, milkweed synthetic blend or left blank. The blend concentration and the distance from release source did not have any effect on attraction efficiency.

Keywords: olfaction, synthetic attractants, Aedes vexans, plant volatiles

* B. Ebrahimi, P. E. Otienoburu, P. L. Phelan, W. A. Foster. In preparation 44 Introduction

Many studies have been conducted on trapping hungry females using carbon dioxide and vertebrate kairomones to attracts female mosquitoes (Becker et al. 1995). In recent years, developing plant-based traps for mosquitoes has received some attention (Kline et al.

1990, Mauer and Rowley 1999, Müller et al. 2010). Floral scents are complex mixtures of small volatile compounds (C5-C20) that can be produced by almost any floral tissue

(Dudareva and Pichersky 2000, Raguso 2001). Some species such as nocturnally pollinated plants emit strong scents, which serve as synomones, to attract pollinators

(Knudsen and Tollsten 1993, Jürgens et al. 2002). But mosquitoes are considered nectar thieves that use plants volatiles as kairomones. Phytochemical attractants can be used in traps for early-warning systems because they can attract newly emerged female and male mosquitoes (Foster 2008). Also, they can be used to develop attractive toxic sugar baits to control mosquitoes in the field (Müller and Schlein 2006, 2008, Schlein and Müller 2008,

Müller et al. 2010).

Aedes vexans is one of the most important biting mosquitoes in the metropolitan areas of the upper Midwest U.S. It breeds in surrounding agricultural land, temporary wetlands, and roadside ditches from which it immigrates into suburban and urban areas. It is also considered as a bridge vector for West Nile Virus. However, there is no knowledge of volatile compounds of the plants that attract Ae. vexans (Yee et al. 1992). The objective of this study was to compare the attractiveness of two extracts that were developed based on

45 volatile profiles of the flowers of Canada goldenrod (Solidago canadensis) and common milkweed (Asclepias syriaca) (Otienoburu 2011), and a synthetic blends of common milkweed (Otienoburu et al. 2012) with their fresh whole flowers and with honey in olfactometer designed for Ae. vexans, and in field traps for various mosquito species, including Ae. vexans.

Materials and Methods

Mosquitoes

Laboratory experiments were conducted with an Ae. vexans colony that was reared and maintained at The Ohio State University. Details of rearing this species have been described in Chapter 2.

Olfactometer Experiments

To assess attractiveness of fresh flowers, synthetic blends of their volatiles, and honey, behavioral assays were conducted in a dual-port olfactometer described in Chapter 2.

Two hundred Ae. vexans pupae were counted and allowed to emerge into a small mosquito cage supplied with water on soaked cotton wicks. The following morning, unemerged pupae were removed from the cage. Bioassay experiments were started 2 d post emergence by connecting a cage of adults to a dual-port olfactometer. The treatment position of the olfactometer arms was altenated, and the olfactometer was thoroughly

46 cleaned in between replicates. The olfactometer experiments were run for a 12-hr period that included the entire scotophase.

Olfactometer Bioassays:

To determine if mosquitoes are attracted to milkweed and goldenrod flowers, 4 g of each flower were tested separately against a blank in the dual-port olfactometer (3 replicates).

Also, mosquito attraction toward 2.5 g honey coated inside one of the arms was compared with one floret (0.08 g) of a milkweed blossom in the dual-port olfactometer (7 replicates). In a similar experiment, mosquito preference towards 5 g honey vs. 2 g goldenrod flowers was tested in the olfactometer.

Field Experiments

Study Site: The study site was a meadow lying east of a 13.7-ha woodlot at Don Scott

Field in Columbus, Ohio, at N 40 5' 8'', W 83 5' 2'', described by Haramis and Foster

(1983, 1990). The main woodlot was surrounded by human dwellings in west and northwest, a fenced pasture in south. The rest of the area is a farm land for cultivating maize or soybeans. There was an animal facility about 500 m southwest of the study area, housing herds of cattle that were allowed to graze on the pasture. A drainage ditch runs along two-thirds of the east and half of the northern edge of the woodlot. The main tree species were beech (Fargus grandifolia), red maple (Acer rubrum), and red oak (Quercus rubra) (Haramis and Foster 1983). The early season flowering shrubs at the woodlot's

47 edge were garlic mustard (Alliaria petiolata), trillium (Trillium), Japanese honeysuckle

(Lonicera japonica), daisy fleabane (Erigeron annuus), dogbane (Apocynum medium), and snakeroot (Eupatorium rugosum). The main flowering shrubs during the peak of adult mosquitoes in mid and late season were common milkweed (Asclepias syriaca),

Queen Anne's lace (Daucus carota), and Canada goldenrod (Solidago canadensis).

CFG Trap: This study employed Counterflow Geometry traps (CFG) (Kline 1999), also known as Mosquito Magnet X (MMX), from American Biophysics Co. (ABC) (North

Kingstown, RI, U.S.A.). Previous studies had shown the greater efficiency of this trap compared to other commercially available traps (Mboera et al. 2000, Kline 2002,

Cooperband and Cardé 2006, Brown et al. 2008, Schmied et al. 2008). Because of the design of CFG trap, mosquitoes can be drawn into the trap, nearby a tube that dispenses volatiles. The traps were always handled by powder-free nitrile gloves and were cleaned thoroughly with alcohol prior to set-up in the field. A strip of dry netting (ca. 20 by 20 cm), impregnated with Deltamethrin (5%), which has a very low vapor pressure, was placed in the collecting compartment of the trap to kill the mosquitoes for convenient collection and to kill spiders that entered the trap. Rechargeable batteries (12 DCV) were able to power the fans for about 72 hr; nonetheless, they were changed every 2 d. To diminish damage from direct sunlight and rain, the batteries were covered with a plastic bag and were hidden under plants if possible.

48 Baits: CFG traps were baited with either fresh flower cuttings, a flower extract, a milkweed minimal synthetic blend, honey, a solvent, or nothing. About 10 g of honey was used to coat a coiled paper that was hung below the intake of the trap. Extracts of flowers were prepared by submerging fresh-collected florets of milkweed or goldenrod in

HPLC grade n-pentane (Fisher Scientific), in a 1:8 ratio (w/v), for 24 hr. Then, the mixture was poured into a filter paper to separate flower materials from the extract. The milkweed minimal synthetic blend was based on 3 main components found by GC-MS in the extract of milkweed flower: Phenylacetaldehyde (3.4 µg), Benzaldehyde (6.9 µg), and

E-2-Nonenal (2.8 µg) in 1 ml solvent (Otienoburu et al. 2012). 600 µl of the pure blend was diluted in 1 ml mineral oil and put in a polyethylene Eppendorf tube to reduce the release rate of the blend.

Trap Setup: Field experiments were conducted between June and October of 2008 and

2009. CFG traps were baited with either fresh flower cuttings, a flower extract, a synthetic blend, honey, a solvent, or nothing. A clean metal wire was used to hold a container of extract, blend, and solvent ca. 5 cm below the trap outlet. Fresh flower cuttings (5 g) were suspended at the stem from the bottom of outlet. Honey (10 g) was placed either in a cup or coated on a coiled filter paper and suspended about 5 cm below the trap's outlet. Traps were arranged in a line in a shrubby ecotone on the eastern side

(generally leeward) of the main woodlot bordering the farm clearing.

49 The traps were hung from metal pipes embedded at an angle in soil beneath the shrubbery about 1 m outside the edge of the woodlot. The traps were suspended 1 ± 0.2 m above the ground. An area of 0.5 m radius around each trap was cleared of shrubs to facilitate mosquito movement around the trap and to provide room for servicing the traps and batteries. The distance between traps depended on the objective of the experiment. When the traps were not running, they were cleaned with 70% alcohol and clean paper towel, the batteries were disconnected, and the inlets were blocked by a cap. Mosquitoes were collected from traps every 3 d. Before the traps were opened, they were inspected for live mosquitoes inside the trapping compartment. If there were any, the trap was opened and the living mosquitoes were gently held by the impregnated net for few minutes to die.

Then the trap was turned upside-down over a large funnel made of white glossy cardboard and the mosquitoes collected in small plastic containers. The 3-night tests were repeated 3 times (3 nights x 3 times = 9 nights total).

Evaluation of Mosquito Distribution: In the first setup (2008), 16 CFG traps were placed 10 m apart and baited randomly with the fresh goldenrod flowerhead, a synthetic blend of goldenrod, or nothing. To assess whether mosquito populations were evenly distributed alongside the woodlot, unbaited traps were run for three nights.

Attraction to Goldenrod flower and extract: Eight locations were selected on the edge of the woodlot, 9 m apart, with two CFG traps at each location that were 1.25 ± 0.25 m

50 away from each other. Each pair of traps was connected to a set of two parallel-connected batteries, so that both traps had the same power. The traps were run for three nights, with one unbaited trap and one baited with a goldenrod flowerhead. Every afternoon, the flower baits were replaced with fresh cuttings. This experiment was replicated once.

To assess attraction to the goldenrod extract, the traps were setup as stated above but one of the traps at each location was baited with the extract.

Attraction to milkweed flowers, synthetic blend, and honey: In 2009, six locations were selected on the edge of the woodlot, and four CFG traps were setup 1.25 ± 0.25 m apart from each other at each location. The last trap at one location was 5.5 m away from the first trap of the closest nearby location. All four traps were connected to a set of three parallel-connected batteries. One of the traps at each location was left unbaited, while the other three were baited with either honey, the milkweed minimal synthetic blend, or a flowerhead of milkweed. To evaluate trap-night population fluctuations in general mosquito flight activity, all traps were run unbaited for three nights, and the mean number of mosquitoes per night was calculated.

Effect of blend concentration: To test the effect of release rate of the milkweed blend, two experiments were conducted. In the first one, we compared the total number of captured mosquitoes at 1, 2, and 3 m away from the source of the blend to check for arrestment of mosquitoes at some distance from the baited trap. Only the first CFG trap at

51 each location (block) contained the milkweed minimal blend (B3), at 1000 times of the original concentration that was used in previous field experiments (see above). The other three traps at each locations were unbaited. All were connected to batteries, but the trap holding the blend was capped so that the mosquitoes could not enter. The distance between locations was 6 m. In the second experiment, two different concentrations of the milkweed blend (B3x1000 and B3x100) were tested along with two blanks as controls in each block of four traps. The location of the different traps within the block was not changed. The total number of mosquitoes in each trap was counted and analyzed using a two-way ANOVA.

Statistical analysis

The proportion of responses to the two arms of the olfactometer were analyzed using a one-sample t-test. The percentages of the two species (Ae. vexans and Culex spp.) were calculated by dividing the number of each species and each sex by the total number of mosquitoes trapped in each arm of the olfactometer. Field data were analyzed with paired t-tests (for paired groups) or two-way ANOVA for randomized block design (for comparison of more than two groups), to compare the mean numbers collected in traps with different baits.

52 Results

Olfactometer Bioassays

Although the total responses to 4 g of fresh flowers of milkweed and goldenrod were low

(11.61 and 16.15% respectively), there was a difference between the two arms, showing that the milkweed and goldenrod flowers were attractive to mosquitoes (Figs. 3.1 and

3.2). The percentage of Ae. vexans (both sexes) that were attracted to one floret of a milkweed flower and 2.5 g of honey in a dual choice olfactometer are presented in Fig.

3.3. They demonstrated that this amount of honey is more attractive to the mosquitoes (P

<0.05), with 20.9% of the released mosquitoes trapped in the honey arm and 11.2% trapped in the arm with a 0.08-g milkweed floret arm. The percent response to the 5-g- honey arm was 21.74% vs. no mosquitoes in the arm that was baited with 2 g of goldenrod (Fig. 3.4) (P < 0.05).

Field Experiments

Evaluation of Mosquito Distribution:

The Primary field collections using attractants (flower and synthetic blend of goldenrod) indicated that there was no difference between baited and unbaited traps. We observed that some of the traps caught more mosquitoes than others, regardless of the type of bait that was used. To assess whether the mosquito population was evenly distributed alongside the edge of the woods, traps were run unbaited. Because there was no

53 difference in number of collected males and females, they were pooled together for further analyses. Fig. 3.5 shows the number of mosquitoes per night for each of 16 unbaited CFG trap. A total of 287 mosquitoes (277 Culex spp. and 10 Ae. vexans) were collected in 3 nights (data from sites 1, 2 and 12 were excluded because of malfunctions of the traps at those locations). The graph shows that the population was not evenly distributed, and a t-test showed that the first 8 CFG traps had more Culex spp. than the last eight traps (P = 0.02). The highest numbers of mosquitoes were collected from shady areas, where the shrubby vegetation was more dense.

Attraction to Goldenrod flower and extract: It was interesting that few unbaited traps caught more mosquitoes than some of the baited traps (Fig. 3.6), when the traps were paired. The paired traps let us compare unbaited and baited traps despite population fluctuations. Fig. 3.7 shows the number of trapped mosquitoes, according to species and sex per night at each location. A paired t-test showed that the CFG traps baited with a

Goldenrod flowerhead trapped significantly more Culex spp. and Ae. vexans (P = 0.01).

When the mean numbers of mosquitoes per night for each species were separated by sex

(Fig. 3.7), males and females of Ae. vexans and males of Culex spp. preferred goldenrod flowers (P< 0.01), however there was no significant difference in number of Culex spp. females that were attracted to the flower vs. the blank (P = 0.09). The extract of goldenrod flower was attractive only to Culex spp. males (P = 0.03) (Fig. 3.8).

54 Attraction to milkweed flowers, synthetic blend, and honey: Mean numbers of each sex of Culex spp. and Ae. vexans per night at each trap were counted for all 24 unbaited traps and analyzed using one-way ANOVA. Although Ae. vexans mosquitoes were distributed evenly in the six locations (Fig. 3.9), more males and females of Culex spp. were caught in locations b and c (males: P = 0.001; females: P = 0.04). Therefore, all experiments were carried out using this randomized block design.

Mean numbers of mosquitoes per night at each baited and unbaited trap are presented in the Fig. 3.10. Only males of Aedes vexans showed high attraction toward Milkweed flower (P = 0.03). Although females of Aedes vexans were caught in Milkweed flower- baited traps slightly more often than the other three treatments, the difference was not significant (P = 0.13).

Effect of blend concentration: Figs. 3.11 and 3.12 show the number of mosquitoes in each trap per night. The statistical analyses showed that neither concentration nor distance had any effect on the number of trapped mosquitoes (P >0.05).

Discussion

Olfactometer Bioassays

The results showed that in olfactometer tests, Ae. vexans is attracted to flowers of milkweed and goldenrod. However, both sexes prefer honey over milkweed or goldenrod flowers. It can be concluded that a single milkweed floret is attractive to mosquitoes but

55 not as much as is honey. Previous laboratory studies have shown attraction of other species of mosquitoes to honey (Thorsteinson and Brust 1962, de Meillon et al. 1967,

Wensler 1972, Hancock and Foster 1993, Foster and Takken 2004). Also, Vargo and

Foster (1982) reported that Aedes aegypti females were more attracted to honey extract in the early stages of food deprivation whereas they chose a milkweed extract in late stages.

Our study is the first to demonstrate the attraction of Ae. vexans to honey and the flowers of milkweed and goldenrod in olfactometer tests. Because Ae. vexans adults cannot survive more than two days on water, we were not able to assess their responses based on food-deprivation stages. One may argue that the response proportions might have been affected by the amounts of honey and flowers that were used in olfactometer tests.

Although we cannot reject this hypothesis for Ae. vexans, the results of similar olfactometer tests with Anopheles gambiae in our laboratory, do not support it. Total responses and response proportions of An. gambiae in dual-port olfactometers that were baited with 150 µl Parthenium synthetic blend vs. different amounts of honey (0.7, 1.5, and 3 g) were not changed (M.R. Nikbakhtzadeh, personal communication). Nonetheless, increasing the amount of honey by > 10 times may affect mosquito behavior.

Field Experiments

In the field studies, we reported two predominant mosquito species, Ae. vexans and Culex spp., that were collected from CFG traps. In the preliminary field studies, we were not able to discriminate between flower-baited and blank traps that were setup 10 m apart

56 because of population fluctuation of mosquitoes. However, when the blank and baited traps were paired, 1 m apart, we were able to demonstrate a higher attraction of both sexes of Ae. vexans and males of Culex spp. to the CFG traps baited with the goldenrod flowers. The results suggest that using phytochemicals to trap mosquitoes may be effective at short-range. There are two explanations for the short-range attraction of phytochemicals: 1) sugar sources are not as scarce as blood sources, and 2) mosquitoes have developed odor receptors that can detect a wide range of phytochemicals. It should be noted that the study was carried out between mid September and early October, when goldenrod was the most common blooming plant in the area, thus eliminating all goldenrod plants from the area was impossible. The competition problem with natural plant hosts has been addressed by (Foster 2008). Considering the longer effective range of zoochemicals (Gillies and Wilkes 1969, 1970, 1972, Okumu et al. 2010), more studies are needed to evaluate the effectiveness of traps baited with phytochemicals vs. zoochemicals.

Test results from the field were contrary to the olfactometer tests; wild Ae. vexans was more attracted to milkweed and goldenrod flowers than to honey. In all field experiments with three baited traps, flowers of milkweed and goldenrod were more attractive baits than the others. Sugar feeding of Ae. vexans has been observed in the field studies

(Grimstad and DeFoliart 1975, Yee et al. 1992). But it is interesting that honey was not attractive, because (Kline et al. 1990) found a honey extract attractive to Aedes taeniorhynchus and Culex nigripalpus in the field. While an extract of honey was used by

57 Kline et al. (1990), we used pure honey that was coated on a coiled filter paper. Thus, it is likely that more volatiles were released when the honey extract was applied in CDC traps. It should be noted that the strain of Ae. vexans that was used for olfactometer experiments had been laboratory-adapted and then maintained in laboratory for a long time, so the odor receptors of the lab strain may have changed.

The extract of goldenrod flowers was attractive only to males of Culex spp., and neither of the synthetic blends was able to attract mosquitoes in the field. Also, Mauer and

Rowley (1999) were not able to make an attractive synthetic blend for Culex pipiens, but

Jhumur et al. (2008) were successful. More studies are needed to assess whether the process of making flower-based extract changes the quantity and profile of the chemicals that are naturally attractive to mosquitoes.

Increasing the blend concentration by 1000 times did not affect the blend attractiveness to any of the field mosquitoes and no arrestment was detected by placement of blank CFG traps away from the baited trap, indicating that the blend concentration was not the factor for low trap catch, but maybe the blend did not have the main attractive compounds, or the ratio of existing compounds were different from those that are released by flowers. In a similar study in our mesocosms, increasing the blend concentration did not boost the collection of An. gambiae in the CFG traps (M.R. Nikbakhtzadeh, personal communication).

58 Acknowledgments

We would like to thank Robert Aldridge for his help in rearing mosquitoes used in this research, Bryan T. Jackson for his help in setting up the field component of this research, and Roland Kuhn for supplying his laboratory-adapted Niagara River strain of Aedes vexans. This study was supported, in part, by NIH grant R01-AI064506 from the National

Institute of Allergy & Infectious Diseases (NIAID) to W.A.F. Its content is solely the responsibility of the authors and does not represent the official views of the NIAID for the National Institute of Health.

59 14 11.61 12 10 8

% 6 4 2 0.00 0 Milkweed Flower (1 bud) Blank Choices

Fig. 3.1: Percentage of Aedes vexans (both sexes) attracted to one bud of milkweed flower (4g) in a dual choice olfactometer.

16 13.98 14

12

10

8 % 6

4 2.17 2

0 Goldenrod Flower Blank

Fig. 3.2: Percentage of Aedes vexans (both sexes) attracted to goldenrod flower (4 g) in a dual choice olfactometer.

60 25

20.86 20

15 11.21 % 10

5

0 Honey (2.5g) 1 Milkweed Floret (0.08g) Choices

Fig. 3.3: Percentage of Aedes vexans (both sexes) attracted to one floret of milkweed flower and honey in a dual choice olfactometer.

25 21.74

20

15 % 10

5

0.00 0 Honey (5gr) Goldenrod Flower (2gr)

Fig. 3.4: Percentage of Aedes vexans (both sexes) attracted to goldenrod flower and honey in a dual choice olfactometer.

61 30

25 t

h 20 g i N

/

15 r Culex sp. e b Aedes vexans

m 10 u N 5

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Sites

Fig. 3.5: Number of mosquitoes (Culex spp. and Aedes vexans) per night at each trap

(site), September 5, 2008. All traps were unbaited and served as blank suction traps.

Fig 3.6: Total number of mosquitoes in goldenrod flower-baited and control-paired

(blank) traps, collected from each location. 62

Fig. 3.7: Mean number of Aedes vexans and Culex spp. from unbaited CFG traps paired the traps baited with goldenrod flower.

Fig. 3.8: Mean number of Aedes vexans and Culex spp. per night in blank CFG traps, paired with traps baited with the extract of goldenrod flower. 63

Fig. 3.9: Mean number of male and female Culex spp. and Aedes vexans per night per location. All traps were unbaited and served as blank suction traps

Fig 3.10: Mean number of male and female Aedes vexans and Culex spp. collected from unbaited traps, and traps baited with milkweed flower, milkweed blend, and honey 64 Fig. 3.11: Mean number of mosquitoes collected from unbaited traps that were setup at

1, 2, and 3 m away from a CFG trap that dispensed the synthetic milkweed blend at 1000 times of the original concentration

65 Fig. 3.12: Mean number of mosquitoes, collected from unbaited traps, and traps that were baited with milkweed synthetic blend at higher concentrations (x100 and x1000) compared to what developed in olfactometer for Culex spp.

66 Chapter 4: Learning: Association of Volatiles with Sugar *

Abstract

Anopheles gambiae mosquitoes were tested for their ability to associate whole-plant volatiles with 2% sucrose solution. These mosquitoes were conditioned and tested with plant volatiles of Senna didymobotrya in an olfactometer, but they did not demonstrate associative conditioning. To condition and test mosquitoes with single-compounds, a series of bioassays was conducted. Mosquito survival on various concentrations of sugar solution, after a 15-min feeding period, revealed that a 2% solution was a satisfactory reward in conditioning trials. The response latency of individual mosquitoes to 11 different volatiles were measured and three compounds were selected for further analyses and tests. The toxicity range, release rate, and dose response were determined for the following compounds: phenylacetaldehyde, linalool oxide, and Z-β-ocimene, which were selected for conditioning and testing trials. Male and female mosquitoes were able to successfully associate phenylacetaldehyde with sugar by showing an enhanced response to it in olfactometer bioassays; however, only females were able to demonstrate such behavior with ocimene. It was concluded that the release rate of volatiles during conditioning and testing is crucial in demonstrating learning capability in mosquitoes.

* B. Ebrahimi, P. L. Phelan and W. A. Foster. In preparation 67 Introducing a novel compound alongside a pre-exposed compound did not distract conditioned mosquitoes from responding to the pre-exposed compound.

Keywords: Anopheles gambiae, associative learning, dose response, phytochemicals

Introduction

Learning enables individual organisms to modify their behavior in response to environmental changes, and thus to maximize their reproductive success. Organisms may learn to obtain food, to nurse their offspring, to escape and to defend themselves, to orient and navigate, and to communicate (Papaj and Lewis 1993).

Categories of Learning

Learning has been classified into two categories: nonassociative and associative

(Papaj and Lewis 1993, Abramson 1994). In nonassociative learning, only one stimulus causes a change of response. Two types of nonassociative learning have been described : habituation and sensitization. If repeated exposure to a stimulus without consequence decreases an organism's response to that stimulus, habituation is said to have occurred.

On the other hand, if such repeated exposure increases the frequency or strength of the response, the phenomenon is known as sensitization.

68 Associative learning refers to the behavior modification in cases in which more than two events are associated with each other. The association is usually between a stimulus and a response. Two types of associative learning are classical (Pavlovian) conditioning, and operant (instrumental) conditioning. In classical conditioning, a conditioned stimulus (i.e., neutral stimulus) is paired with an unconditioned stimulus that elicits an innate response known as unconditioned response. Consequently, the conditioned stimulus triggers the unconditioned response. In the literature, the conditioned stimulus, unconditioned stimulus, and unconditioned response are abbreviated as CS, US, and UR, respectively. Depending on the objectives of the studies, researchers often choose one of two main types of US to either encourage or punish the test subject. These two are known as appetitive and aversive (defensive) conditioning. An example of appetitive conditioning is when an odor (CS) and sugar (US) are presented to a honeybee. If an odor (CS) is paired with a shock or an unpalatable meal (US), this is aversive conditioning (Abramson 1990, 1994, Papaj and Lewis 1993).

Learning in Mosquitoes

Mwandawiro et al. (2000) reported a host-preference in Culex vishnui, Cx. gelidus, and Cx. tritaeniorhynchus, when they were pre-exposed to that host. McCall et al. (2001) have shown that An. arabiensis females tended to return to the site where they were first caught, suggesting that they might use spatial memory. Two separate studies have demonstrated that oviposition preferences of Cx. quinquefasciatus (McCall and

69 Eaton 2001) and Ae. aegypti (Kaur et al. 2003) were increased if those mosquitoes were exposed to the odor (e.g., a repellent) during their aquatic stages. Alonso et al. (2003) were unable to demonstrate any type of associative learning (i.e., appetitive or aversive conditioning) in Ae. aegypti, but Jhumur et al. (2006) were able to condition Cx. pipiens pipiens with chemical mixtures and single compounds. It has also been shown that Cx. quinquefasciatus can associate mixtures of volatiles (Tomberlin et al. 2006) or single- compounds (Sanford and Tomberlin 2011) with sugar (US), and the concentration of sugar solution did not have any effect on their learning capability (Sanford and Tomberlin

2011). In the first report of learning in An. gambiae, females were able to associate visual and olfactory cues with blood (US) (Chilaka et al. 2012). In the two most recent studies on Ae. aegypti, Stanczyk et al. (2013) have found that pre-exposure of females to DEET can result in habituation to the compound, and Menda et al. (2013) were able to pair octenol (CS) with an electric shock (US) in females.

As stated by McCall and Kelly (2002), the learning capability in mosquitoes may increase biting rate (i.e., higher host- or site-fidelity) or survival when they successfully avoid a host defense. Also, if mosquitoes can learn how and where to locate their food sources (i.e., sugar and blood), their survival will increase, which in turn can increase their vectorial capacity. Most of the studies have focused on the innate response of mosquitoes to compounds. However, the learning capability of mosquitoes may provide a baseline for studying how to maximize the effectiveness of plant attractants for surveillance and control, and how to affect vectorial capacity by manipulating the plant

70 community. The main goal of this study was to determine whether the host-plant attraction in An. gambiae can be modified by experience with sugar availability and plant volatiles. To achieve the main goal, mosquitoes were conditioned with plant volatiles or single-compound during sugar-feeding, then their response to the pre-exposed volatiles was tested in a modified Jepson-Healy olfactometer, and that response was compared to the response of naïve mosquitoes that were fed with sugar. Also, to determine some of the factors that may affect learning capability, An. gambiae adults were a) conditioned through olfactory and gustatory system and b) conditioned and tested with different qualities and quantities of unconditioned stimuli (i.e., volatiles).

Materials and Methods

Rearing and Maintenance of Anopheles gambiae

For this study, the Mbita strain of An. gambiae s.s was reared and maintained according to the method described in Chapter 2.

Olfactometer design

A modified Jepson-Healy design with baffle ports, as described in Chapter 2, was used for olfactometer bioassays of this study.

Conditioning container

71 The container was 10 cm diameter by 6.5 cm height, with a 2-cm hole on the side.

To facilitate the use of the containers and to avoid mosquito escape from the side hole of the container, a cover was made with a 2.5 by 2.5-cm piece cut from a nitrile glove. A diagonal1.5-cm slit was made in the center of the nitrile piece . Another identical piece was placed on top of the first one, such that the two slits made a cross-shaped (X) slit, then they were taped over the side-hole of the container. Because of the elasticity of the nitrile glove cut-outs, the X slit was always closed unless the tip of an aspirator (1-2 cm diameter) was pushed through the X slit. For air ventilation, an 8-cm hole was made on the lid of the container, and an aluminum screen (New York Wire ® 10205) was glued on top of the lid. To prevent any chance of cross-contamination, each container was designated for use with only one compound, and they were washed between experiments.

For some experiments, a 6.5-cm wide by 23-cm long aluminum screen was used as a separator, and placed between a 25-cm-long germination paper, on the interior wall of the container, and a shorter germination paper (22 cm), so that the distance between the two papers was about 5 mm.

Chemicals

The chemicals used: phenylacetaldehyde (90+%), benzaldehyde (≥99.5%), nonanal (≥95%), ethanol, decanal, geranylacetone, linalool oxide, hexane, nonanoic acid,

1-octen-3-ol (octenol), α-pinene, and Z-β-ocimene, were all purchased from Sigma-

72 Aldrich® (Saint Louis, MO, USA). Mineral oil, heavy USP/FCC, was obtained from

Fisher Chemical.

Sachets

To slowly release chemicals, sachets were prepared and used. Low-density polyethylene (LDPE) tubing, purchased from ULINE ®, Pleasant Prairie, WI., with 1.5''

(3.81 cm) wide and 2 mil (51 µm) thickness, was cut into 6-cm-long smaller tubes . After sealing one end with an impulse poly bag sealer (ULINE ®, model H-190), the sachet was filled with mineral oil and one of the compounds (%v/v). Then, the excess air was removed, and the other end was sealed so that the distance between the two sealed sides was 5 cm. In bioassay experiments, the sachets were placed in a horizontal position with one side of the sachet completely in contact with the floor of the trapping bags of olfactometer.

Whole-plant Conditioning

Olfactometer bioassay

To assess whether mosquitoes can associate the plant scents with sugar, mosquitoes were conditioned and tested in several of olfactometer experiments. About 2 hr before conditioning mosquitoes, two to four branches of a Senna didymobotrya plant were wrapped in a 30.5 by 76.2-cm polyethylene plastic bag. Between 1 and 2 conditioning containers with germination paper (25 cm) were placed in each bag. No

73 container was placed in a bag that was designated for collecting and pumping the plant's headspace into a venturi aspirator. Approximately 15 min prior to conditioning, air was pumped into the bottom of the bag and directed to a venturi aspirator designed for transferring 1-d-old mosquitoes. The aspirator vacuumed individual mosquitoes from one end, and exposed them to the plant volatiles in the middle of the aspirator, before blowing them into the conditioning container. Before transferring the mosquitoes to the conditioning container, the germination paper inside it was impregnated with a 4-ml 2% sucrose solution. After transferring 55-60 mosquitoes, each container was placed back inside one of the bags to be exposed to the plant volatiles. Conditioning ceased after 15 min by releasing the mosquitoes into a cage supplied with water.

For each cage of plant-conditioned mosquitoes, one cage of naïve mosquitoes was prepared by transferring 1-d-old mosquitoes into a container, supplied only with 2% sucrose solution, with a clean aspirator as described above. The air was directed into a sealed polyethylene bag, before being directed to the aspirator. The naïve mosquitoes were not exposed to any plant volatiles until they were tested in an olfactometer at the same time that conditioned mosquitoes were tested in a separate olfactometer.

About 2-3 hr before testing the two groups of mosquitoes in separate olfactometers, the branches of one S. didymobotrya were wrapped in a 50.8 by 76.2-cm polyethylene bag. Tygon tubing was used to direct purified and humidified air into a two- way manifold. One tube was connected between one of the outlets of manifold and the bag that was wrapped around the plant, while another tube directed the clean air into an

74 empty polyethylene bag. The clean air (blank) and the plant headspace tubings were connected to the two arms of each olfactometer. Every time, one of 6 available healthy plants was used for conditioning and one for testing, so mosquitoes of each replicate were conditioned and tested with two plant pots of S. didymobotrya. The number of mosquitoes in each arm was counted after 12 hr. The experiment was replicated 10 times.

Single-compound Conditioning

Response latency of males

To measure their response latency to volatiles, an objective was to rank the response time of 1-d-old males of An. gambiae to some compounds that have been detected from many flowers, as well as in honey (Knudsen and Tollsten 1993, Andersson et al. 2002, Dudareva and Pichersky 2006, Jhumur et al. 2008). Each of the following compounds was diluted in mineral oil to make four concentrations (1, 10 , 100 , and

1000 nl/ml) of the following: phenylacetaldehyde, benzaldehyde, ethanol, nonanal, decanal, geranylacetone, linalool oxide, hexane, nonanoic acid, Z-β-ocimene, octanol, and α-pinene. Pure mineral oil was tested as the control. First, a single male was trapped in a 2-dram vial by placing the opening of the vial on a male resting in a cage. When the male flew into the back of the vial, the opening was covered with a screen. The vial was placed in a horizontal position, with the screen facing the opening of a borosilicate glass funnel, 5 cm away from the narrow opening of the funnel. The stem of the funnel was connected to an air supply (See Chapter 2 for details) with an airflow of 12 mL/s. The

75 mosquitoes were acclimated to the vial and airflow for 5 min. Approximately 1-2 mm of the tip of a clean microcapillary tube was dipped into one of the solutions, and was placed in the airflow about 1 cm away from the screen of the vial. The microcapillary tube was held in the airflow for up to 40 s, or until the mosquitoes responded by either probing, flying, or walking. The time between placing the compound into the airflow and the first response of the mosquitoes was considered as the 'response latency'. Five males were tested for each compound-concentration.

Survival values

To detect a feeding-related behavioral response of mosquitoes, they should be deprived of food to the point that they start searching for food. It is particularly important that shortly after conditioning (the first sugar-feeding), mosquitoes are hungry enough to search for food in olfactometer tests. The time period between the first sugar-feeding and the next deprivation may be estimated by analyzing survival curves. The objective of this study was to compare the survival rates of laboratory-reared mosquitoes that were fed various concentrations of sucrose solution (2, 50, and 95%) and solid sugar cubes (100%) for 12 hr. In the evening, the sugar solutions were poured into vials containing wicks and then presented to mosquitoes in three separate cages, with 120 ± 20 1-d-old males and females. A fourth cage contained four sucrose cubes within a small clean shallow cup.

The following morning, the sugar vials and cubes were replaced with vials of water.

Adults had access to water throughout their lives. The control group was always kept on

76 water only. The numbers of dead males and females were counted daily, and survival curves were plotted for the control, the 2, 50, 95% sugar solutions, and sugar-cube treatment.

Toxicity range

In the first conditioning sets, the mosquitoes had to be fed on 2% sugar solution mixed with one of the compounds. This test was designed to find the highest amount of

Z-β-ocimene that does not kill the mosquitoes, when imbibed in sugar solution. Thus, 4 ml of 2% sucrose solution was mixed with 5, 10, and 50 μl Z-β-ocimene in a 10-ml glass syringe. The mixture was vigorously shaken for 30 s before being poured on a 6.5 by 25 cm germination paper that was already placed against the wall of a conditioning container. Earlier tests showed that 4 ml of sugar solution was enough to dampen the 6.5 by 25 cm germination paper without dripping or forming excess droplets in which the mosquitoes' wings and legs might get trapped.

At the start of the experiments in the afternoons, approximately 1-2 min after the paper was dampened, 45-55 1-d-old mosquitoes (mixed sexes) were transferred to the conditioning container by the venturi aspirator within < 2 min. Then, the containers were placed on the benchtop at 25°C for 15 min, so that mosquitoes could feed on the mixture of chemical and sugar solution. An. gambiae mosquitoes prefer to rest on a vertical surface and they immediately showed feeding behavior (i.e., probing) when their tarsi contacted the sugar solution. They were given 15 min to feed, > 95% of mosquitoes were

77 fully engorged. After this time period, the conditioning container was placed inside a cage, and the mosquitoes were released when the lid of the container was opened. The procedure was repeated so that a total of 100-110 sugar-engorged mosquitoes were in each cage with water vials. The same procedure was followed with a 4 ml sucrose solution (2%) mixed with 2.5 μl phenylacetaldehyde and 5 μl linalool oxide, each for a separate survival study. A control group (only 2% sugar solution only) was also prepared, following the same procedures as the other groups. The survivorship of all groups was followed until all mosquitoes were dead. The highest concentration of phenylacetaldehyde and Z-β-ocimene with survival curves similar to that of control (2% sugar solution) were chosen for conditioning.

Measuring release rates

To correlate the release rate (mg/hr) with the concentration (%v/v) of the chemicals from the sachets, three concentrations (1, 10, and 100 μl/ml) in the sachets were prepared for phenylacetaldehyde, Z-β-ocimene, and linalool oxide. A sachet of 1mL mineral oil (control) also was prepared. Each sachet was placed inside a separate trapping bag of a Jepson-Healy olfactometer, and weighed every 3 hr for 12 hr, using a scale with 0.1 mg precision. This procedure was repeated four times for each compound- concentration.

A graph of weight loss (dwt) by time (t) was drawn for each replicate, and the slope of the steady part of the curve was measured. Because it took time for the volatiles

78 to pass through the polyethylene plastic of a sachet, the initial release rate (curve slope of the first segment) was often different from the release rate of a subsequent trend. In this study, the compounds needed an average of 3 hr to reach a steady release rate. The release rate of each compound-concentration was calculated as mg per hr.

Dose response of mosquitoes

To find the optimum dose of each compound that elicits the highest attraction of mosquitoes in an olfactometer, three concentrations (0.1, 1, and 10 μl/ml for phenylacetaldehyde; and 1, 10 and 100 μl/ml for linalool oxide and Z-β-ocimene) were prepared in sachets, as described above (Measuring release rate). Ten replicates of olfactometer tests were conducted for each concentration-compound. The two arms of each olfactometer were baited with the sachet of containing the compound vs. mineral oil alone (control) in a sachet and the olfactometers were run for 12 hr. In the morning, the trapped mosquitoes in each arm was counted, and the percent of released mosquitoes responding was calculated. Also, the “discrimination index” was calculated for each olfactometer experiment by dividing the number of mosquitoes in the treatment arm by the total number of responding mosquitoes in both arms:

Discrimination Index = Treatment / (Treatment + Control)

For each compound, the mean discrimination indices of the three concentrations were analyzed after angular transformation, using one way ANOVA. If the discrimination

79 indices of the three concentrations were not statistically different, the dose eliciting the highest attraction was selected for learning bioassays.

Learning bioassays

To test associative learning in mosquitoes, and to assess factors that may affect it, a series of bioassay tests was conducted. The main factors that may affect associative learning are listed in table 4.4. For the purposes of this study, the starvation periods before and after conditioning (i.e., 24 hr), the exposure time to unconditioned and conditioned stimuli (i.e., 15 min), and the duration of testing mosquitoes in an olfactometer (i.e., 12 hr) were not changed. In addition, only one type of unconditioned stimulus (i.e., sucrose) at a specific concentration of 2% aqueous solution was given to the mosquitoes at the same time that the conditioned stimulus (CS) was presented. The

2% sugar solution was chosen because 24 hr after feeding with that solution, mosquitoes were hungry enough to search for food (see results and discussion of Survival values).

To condition them , 50-60 1-d-old mosquitoes (mixed sexes) were transferred into a conditioning container that was already prepared with conditioned and unconditioned stimuli according to the objective of the study (see below for details). Then, each container was placed under an exhaust duct to direct volatiles out of the laboratory. The mosquitoes were not agitated during conditioning. After a 15-min conditioning period, they were released into a cage and kept on water only. To achieve a desired number of mosquitoes in a cage (120-150 mixed adults), this procedure was repeated two or three

80 times. Meanwhile, at least one cage of naïve mosquitoes was prepared by providing mosquitoes with 2% sugar solution only.

After a 24-hr starvation period, the mosquitoes were tested in the olfactometers.

The percent responding in each arm (i.e., the number of mosquitoes in one of the arms divided by the total number released) was calculated. In addition, the discrimination index was calculated as described above (see dose response of mosquitoes). Males and females were pooled together only when there was no sex-ratio disparity. Unless stated otherwise, 10 replicates were carried out for each experiment.

Effect of release rate during testing

Exp. 1: Testing with 10 µl of linalool oxide and phenylacetaldehyde. Following the methods of Jhumur et al. (2006) and Tomberlin et al. (2006), two groups of mosquitoes were conditioned by feeding them on a mix of one of the compounds and 2% sugar solution, for 15 min. Three groups of mosquitoes were conditioned by feeding on i) 2% sugar solution alone (naïve group), ii) the sugar solution mixed with 2.5 µl phenylacetaldehyde (PHE group), and iii) the sugar solution mixed with 2.5 µl linalool oxide (LO group). The compounds were added to 4 ml 2% sugar solution in a 10-ml glass syringe and shaken vigorously for 30 s. Then the mixtures were slowly poured on a 6.5 by 25 cm germination paper, already placed against the wall of the conditioning container. The three groups (naïve, PHE, and LO) were tested separately in separate olfactometers, with a choice of linalool oxide vs. phenylacetaldehyde sachets. To make

81 the sachets, 10 µl of either linalool oxide or phenylacetaldehyde were placed inside separate sachets and sealed, without adding mineral oil.

Exp. 2: Testing with optimum doses of phenylacetaldehyde and Z-β-ocimene. The following three groups of mosquitoes were conditioned by feeding on: i) 2% sugar solution, ii) the sugar solution with 2.5 µl phenylacetaldehyde, and iii) the sugar solution with 10 µl Z-β-ocimene. Each group of mosquitoes was tested in an olfactometer, baited with a choice of 1 µl of either phenylacetaldehyde or 100 µl of Z-β-ocimene. The compounds were prepared in separate sachets and diluted with 1 ml of mineral oil. These were the doses that elicited the highest discrimination response among other tested doses

(see results of dose response).

Associative learning through olfactory system

Exp. 1: Pre-exposed vs. novel compounds. The previous experiment was conducted by feeding the mosquitoes on the mix of sucrose solution and the compound. To see whether the mosquitoes can be conditioned through the olfactory system, they were fed 2% sugar solution while exposed to the volatiles in the conditioning container. To prepare the conditioning container for this experiment, the screen separator was placed between the long (25cm) germination paper, in touch with the inner wall, and a shorter (22 cm) germination paper. The shorter germination paper was dampened with 4 ml of 2% sugar solution while a single compound was added to the longer one. This way, the mosquitoes

82 were able to drink the sugar solution while their olfactory systems were exposed to the volatile. For the naïve group, the same procedure was followed but no compound was added to the outer germination paper. Mosquitoes were tested in the olfactometers baited with a sachet of either phenylacetaldehyde (1 µl /ml) or Z-β-ocimene (100 µl/ml).

Exp. 2: Effect of release rate during testing. To determine whether a higher release rate of a compound during testing has any effect on the response of conditioned mosquitoes in the olfactometer, four groups of mosquitoes were prepared: two groups of mosquitoes were conditioned with phenylacetaldehyde through the olfactory system, as described above, and the other two fed on 2% sugar solution only (naïve group). Then, one naïve and one conditioned group were tested in separate olfactometers with 1 µl/ml phenylacetaldehyde in a sachet vs. blank (sachet of mineral oil). The other naïve and conditioned groups were tested with 30 µl /ml of phenylacetaldehyde in a sachet, compared to a blank.

Exp. 3: Effect of release rate during conditioning. To assess whether the release rate of a compound during conditioning has any effects on learning, 10 and 100 µl/ml Z-β- ocimene were prepared in sachets, which were suspended inside the conditioning containers. A long germination paper (25 cm) inside the container was saturated with 4 ml of a 2% sugar solution. To condition naïve mosquitoes, a sachet filled with 1ml of mineral oil was placed inside the conditioning container. The mosquitoes were tested in

83 the olfactometers, baited with 100 µl/ml Z-β-ocimene vs. a blank. This experiment was replicated 11 times.

Results

Olfactometer bioassay

In both the naïve and the plant-conditioned groups, the mean percent of mosquitoes responding to the blank arm of olfactometer was about 10% (SE = 2.5) (Fig.

4.1). The mean percent of conditioned and naïve mosquitoes responding to the plant arms were, respectively, 24% (SE = 5.3) and 17% (SE = 2.6). Nonetheless, the mean percentages of discrimination of the naïve and conditioned groups were 65.2% (SE= 6.9) and 65.5% (SE = 8.6) respectively (P > 0.05).

Response latency of males

No mosquito responded to the mineral oil, or to any chemical at the lowest concentration (1 nl/ml). For each chemical, the concentration at which the mosquitoes showed the shortest mean response time was selected. Then, the compounds were ranked in four groups, based on the shortest mean response time at the selected concentrations

(Table 4.1). The males demonstrated the slowest mean responses (32.2 s) to ethanol

(1000 nl/ml) and Z-β-ocimene (100 nl/ml), when about 20% of the males probed, walked, or flew towards these chemicals. The fastest mean response was observed when

84 presented with nonanal (100 nl/ml) and decanal (1000 nl/ml), which in both cases was 8.8 s, with 80% of the males responding. Only these two chemical-concentrations induced significantly faster responses than mineral oil (control) (P = 0.03).

Survival values

The survival curves were categorized into three groups (Fig 4.2): (1) All mosquitoes in the control group that were kept only on water died 3 d post-emergence;

(2) the mosquitoes that were kept on a 2% sugar solution and sugar cubes (100%) died 5 d after removing the sugar; and (3) the mosquitoes that were fed on 50% and 95% sugar solution survived longer, thus they had a smoother mortality curve than all other groups.

Approximately 36 hr after replacing the 2% sugar solution or the sugar cubes with water in the cages of mosquitoes, both males and females experienced a drastic drop in survival, suggesting that mosquitoes were likely to be searching for food approximately

12 -24 hr after sugar-feeding. The mosquitoes’ abdomens were completely engorged within 15 min after the onset of feeding on the solutions. At the time of removing the sugar, few mosquitoes (< 5%) were still feeding on the 2% sugar solution, but many of them (10-25%) were probing the sugar cubes.

Toxicity range

Females and males of An. gambiae in the control group died by day 4 of the experiment (Fig. 4.3). The mosquitoes that were fed with a mixture of 50 μl Z-β-ocimene

85 and 4 ml of 2% sugar solution died at a faster rate than all other groups. Therefore 10 μl

Z-β-Ocimene was used for further bioassay experiments. The mosquito survival with linalool oxide was slightly lower than the control on day 1, suggesting that probably the survival would be higher with a lower amount of this compound. On day 2, the survival of males and females that were fed with a mixture of 2% sugar solution and phenylacetaldehyde was slightly higher than the survival rate of the control group; nonetheless, these two groups had similar curves.

Measuring release rates

The weight losses of sachets of phenylacetaldehyde, linalool oxide, and Z-β- ocimene differed. Some chemicals had a faster release rate in the first 3 hr compared to the subsequent 9 hrs of experiment (Fig. 4.4). Therefore for further analyses, the slope

(i.e., release rate) in a period of 9 hr (between hours 3 and 12 of the experiment) were used. Comparing 1 μl dose of all three compounds, phenylacetaldehyde showed the highest release rate. However, at 10 and 100 μl, the release rates of Z-β-ocimene were higher than those of the other two compounds. The regression lines showed high correlations between concentration and release rate (Table 4.2, and Fig. 4.5).

Dose response of mosquitoes

The results of dose responses of the males and females to 0.1, 1, and 10 µl/ml phenylacetaldehyde are presented in Fig 4.6. Total responses among the three

86 concentrations of phenylacetaldehyde differed, with the lowest response (females = 8.4% and males = 5.4%) occurring at the highest concentration (10 µl/ml) (Table 4.3). In fact, the highest mean percent discriminations were observed at 1 µl/ml of phenylacetaldehyde (females: 67.7%, males: 75.1%), with total responses of 15.2% and

14.8% for females and males, respectively.

At the highest tested dose of Z-β-ocimene (100 µl/ml), the total responses of females and males were 13 and 11.7%, respectively. Out of all responding mosquitoes at this concentration, 62.3% males and 70% females responded to the treatment arm (Fig.

4.7, Table 4.3).

Among 1, 10, and 100 µl/ml concentrations of linalool oxide, mosquitoes showed the highest mean percent discrimination at 10 µl/ml of the compound (Fig 4.8), with the total responses of 18.5 and 14.2% for males and females, respectively. The most attractive doses of phenylacetaldehyde, linalool oxide, and Z-β-ocimene were 1, 10, and

100 µl/ml, respectively.

Effect of release rate on learning

Exp. 1: Testing with unspecified doses of linalool oxide and phenylacetaldehyde.

The results of this bioassay are presented in Fig. 4.9, indicating that naïve mosquitoes may be slightly, yet not significantly, attracted to linalool oxide. But both conditioned groups showed low discrimination indices toward phenylacetaldehyde (P > 0.05),

87 suggesting that mosquitoes were not able to associate the sugar solution with either phenylacetaldehyde or linalool oxide.

Exp. 2: Testing with optimum doses of phenylacetaldehyde and Z-β-ocimene.

In the olfactometer experiments, naïve mosquitoes responded equally to phenylacetaldehyde and Z-β-ocimene (Fig. 4.10). But the mosquitoes that were conditioned with phenylacetaldehyde showed a higher response to the compound, indicating that associative learning had occurred. The mean percent discrimination of the naïve group (47.6%, SE = 8.3) and the phenylacetaldehyde-conditioned group (70.2, SE

= 7.2) were marginally significant (t = 2.07, df = 18, P = 0.053). The mean percent discrimination of the ocimene-conditioned group (47%, SE = 8.1) did not differ from the naïve mosquitoes (P > 0.05).

Learning through olfactory system

Exp. 1: Pre-exposed vs novel compounds. Like the previous results, the naïve mosquitoes did not discriminate between the two arms of the olfactometer baited with phenylacetaldehyde vs. ocimene (Fig. 4.11). Out of the total responses of phenylacetaldehyde-conditioned mosquitoes, 78.3% (SE = 8.5) were attracted to phenylacetaldehyde, which was significantly higher than the discrimination of the naïve group (45.7%, SE = 7.9) (Mann-Whitney = 15, Z = 2.65, P = 0.007). The attraction of

88 ocimene-conditioned group to phenylacetaldehyde was 59.9% (SE = 10.6), but not significantly higher from that of the naïve group (P > 0.05).

Exp. 2: Effect of release rate during testing. The use of a higher concentration (i.e., higher release rate) of phenylacetaldehyde during testing in the olfactometer resulted in similar discrimination indices, indicating that associative learning did not occur (Fig.

4.12) (P > 0.05). However, when the optimum release rate of phenylacetaldehyde (i.e., 1

µl/ml) was used in the olfactometer arm, associative learning was observed. The total responses of released mosquitoes among all four groups ranged between 21.4 and 24.6%

(P > 0.05). Only the conditioned mosquitoes that were tested with 1 µl/ml phenylacetaldehyde had a significantly higher mean discrimination index compared to those of other groups (F = 6.74; df = 3, 36; P = 0.001). They also had a higher discrimination index than the naïve mosquitoes that were tested with the same concentration of phenylacetaldehyde (t = 2.86, df = 18, P = 0.01).

Exp. 3: Effect of release rate during conditioning. Because of the differences in the response of the males and females, they were analyzed separately. The highest mean discrimination index (79.0%, SE = 7.1) was observed in the females that were conditioned with 100 µl/ml ocimene in the sachet (Fig. 4.13), which was significantly higher than the discrimination index of naïve group (51.6%, SE = 8.8) (t = 2.43, df = 20,

P = 0.024). However, the mean discrimination indices of the three groups of males,

89 which ranged between 64.7 and 69.1%, were not significantly different (P >0.05), suggesting that different release rates of ocimene during conditioning did not change the males’ behavior (Fig. 4.14). The total responses did not differ among the three groups of males and females (P >0.05).

Discussion

Olfactometer bioassay

Associative learning with whole-plant volatiles was not observed in An. gambiae, because their response to the headspace of S. didymobotrya was not enhanced after it had been associated with 2% sugar solution. It was unclear whether this result was due to the method, such as the use of conditioning containers and olfactometers, or quality and quantity of the plant volatiles. Probably handling mosquitoes during conditioning affected the mosquitoes’ behavior. Bateson et al. (2011) have shown that honey bees that were shaken for 1 min after conditioning, did not respond to the compound as well as the unshaken group. Nonetheless, the following bioassays with single-compounds showed that associative learning can be demonstrated using olfactometer, so disturbing mosquitoes may only moderately affect their responses. One solution to diminish any effect of agitation and handling, would be to condition and test mosquitoes in a mesocosm.

Previous studies have shown variation in the volatile production among individuals of a plant species (Dudareva and Pichersky 2006). Each group of mosquitoes,

90 in this bioassay, was tested with a plant headspace that was different from the plant used for conditioning the mosquitoes. Therefore, each of the 10 replicates might have been exposed to different headspace quantities or qualities of S. didymobotrya. The high variance particularly in the conditioned group, might be due to use of different whole plants.

Response latency of males

The response latencies of males to 11 volatiles were ranked into four main groups, with the shortest response time to nonanal and decanal, and the longest response time to ethanol and ocimene. Vet et al. (1995) have discussed the importance of response potential, i.e., the potential of evoking an innate response, in naïve parasitoids. Any innate response may be modified by learning, but the chance of changing a strong innate response is less likely, compared to a weak response. In addition to response time that was demonstrated in this study, innate responses can be measured by other methods, such as the number of individuals responding to a stimulus, the walking speed, and the length of a walk. Ocimene was used in further bioassays because it is more commonly found in plant volatiles compared to linalool oxide. Moreover, other studies have found ocimene as one of the main constituents of an attractive minimal synthetic blend based on

Parthenium hysterophorus headspace (M.R. Nikbakhtzadeh, unpublished data).

Jhumur et al. (2006) have reported that naïve Culex pipiens molestus had a higher innate response to phenylacetaldehyde than to veratrole and 2-methoxyphenol. The

91 authors measured innate response based on the proportion of mosquitoes that responded to the chemicals vs. a blank in a two-choice olfactometer. Also, it should be noted that one of their objectives was to mimic the release rate of the three chemicals that are emitted by Silene otites. However, the resulting mosquito responses might have been dose-related, as demonstrated for An. gambiae (see results for details). Also, phenylacetaldehyde has been reported as an attractants for Aedes aegypti (Howse 2003) and Cx. tarsalis (Lothrop et al. 2012). Linalool oxide has been reported as an attractant for mosquitoes (Meijerink et al. 2000, Jhumur et al. 2007, 2008, Nyasembe et al. 2012).

Survival values

Table sugar (sucrose) was selected as the reward for conditioning because Ignell et al. (2010) had reported that Ae. aegypti prefers disaccharides (e.g., sucrose) over monosaccharides. In this study, the lowest survival was observed with the

2% sugar solution and sugar cubes (100%), but those mosquitoes that were fed with 50% and 95% sugar solution had the highest survival. The sugar cubes seemed to work as well as a 2% sugar solution, but mosquitoes fed faster on a sugar solution (>95% were engorged after 15 min). Maybe sugar cubes are a better choice, but mosquitoes should be fed on sugar for a long period of time. In a separate experiment, mosquitoes survived longer on sugar cubes compared to sugar granules or sugar powder. If sugar cubes were accessible throughout the mosquitoes’ lives, males might have lived longer than females.

Probably because the small pores in sugar cubes provide small pools, into which

92 mosquitoes are able to salivate, they can dissolve sugar and ingest it. Nonetheless, one main problem with the use of sugar cubes was that all mosquitoes cannot get the same amount of sugar (reward) from the cubes. Although associative learning in honey bees is positively correlated with sugar concentration (Bitterman et al. 1983), this has not been shown for Cx. quinquefasciatus (Sanford et al. 2012). The preliminary tests with successful methods of conditioning showed that increasing the reward concentration to

5% resulted in a low total response in olfactometers 24 hr after conditioning, probably because mosquitoes were not hungry enough to search for food. Therefore, 2% sugar solution was chosen as the unconditioned stimulus (reward) for mosquitoes in this study.

Toxicity range

The mosquitoes that were fed on 5 µl linalool oxide in sugar solution showed a marginally lower survival rate, compared to the control group. Therefore, for conditioning bioassays a 2.5 µl linalool oxide was used.

As explained above, phenylacetaldehyde and Z-β-ocimene were the two compounds selected for further bioassays. Because one of our objectives was to test for learning by feeding mosquitoes a compound in sugar solution, it was essential to test toxicity of the chemicals. In one of our preliminary experiments, mosquitoes died within

12 hr after feeding on a small amount of nonanal, even though An. gambiae males showed the fastest response to this compound (Table 4.1). Nonanal is known as one of the attractants of mosquitoes (Syed and Leal 2009, Carey et al. 2010).

93 The amount of phenylacetaldehyde that Jhumur et al. (2006) added to their sugar solution was more than 4 times higher than the 2.5 µl that was used in this test. It is interesting that the survival rate of An. gambiae that were fed phenylacetaldehyde in a sugar solution in this study was slightly higher than those that were fed sugar solution only. There are three possible explanations: (1) phenylacetaldehyde stimulated the mosquitoes to feed more, (2) phenylacetaldehyde had some physiological effect on the mosquitoes, and (3) phenylacetaldehyde had some anti-microorganismal effects that help the mosquitoes survive longer. The antibacterial effect of phenylacetaldehyde, isolated from screwworm, has already been documented (Erdmann and Khalil 1986). Also, the long-term effect of chemicals such as the antimicrobial methylparaben on survival of An. gambiae and An. arabiensis have been studied (Benedict et al. 2009), but the connection between this chemical and prolonged life is still unknown.

Measuring release rates

Phenylacetaldehyde, linalool oxide, and Z-β-Ocimene behaved differently in low density polyethylene (LDPE) sachets. Linalool oxide with the highest molecular weight

(170.25) among these compounds, had the lowest release rate. Nonetheless, molecular weight may not have been a good predictor, because the release rate of Z-β-ocimene

(MW = 136.24) was the highest, even though its molecular weight is higher than that of phenylacetaldehyde (MW = 120.15). Maybe other physico-chemical characteristics, such as presence of aromatic rings or lipophilicity, are important to this chemical's membrane

94 permeability. Torr et al. (1997) have studied other factors that can increase the release rate, such as a larger surface area, using thinner sachets, and increased temperature.

However, Mukabana et al. (2012) have reported that the release rates of some chemicals did not change with increasing the thickness of sachets, but it should be noted that they did not use mineral oil to dilute and slow the release of compounds.

Dose response of mosquitoes

Both males and females were found to be selective in responding to doses of phenylacetaldehyde, Z-β-ocimene, and linalool oxide. The results of this experiment and the previous one (i.e., release rate) showed that these two factors are not correlated. The sachet was more permeable to Z-β-ocimene than to phenylacetaldehyde or linalool oxide.

The release rate of 1 µl/ml phenylacetaldehyde, which was found to be more attractive to mosquitoes, was about 0.03 mg/hr, whereas the sachets with an attractive concentration of Z-β-ocimene (100 µl/ml) released approximately 1.68 mg of volatile per hour. The total responses of males and females declined at the high concentration of phenylacetaldehyde, probably due to the saturation of the odor receptors. Nonetheless, the total responses among the three concentrations of Z-β-ocimene and linalool oxide were very similar. Perhaps mosquitoes were able to smell the high concentration of linalool oxide and ocimene, but they chose the blank arm, suggesting a repellent effect.

Effect of release rate on learning

95 The mosquitoes that were conditioned with phenylacetaldehyde showed a higher response to phenylacetaldehyde only if the optimum dose was used in the olfactometer.

But the mosquitoes that were tested with the optimum dose of ocimene, a high dose of phenylacetaldehyde, or non-diluted dose of linalool oxide, were not able to associate the compound with the reward. It was hypothesized that mosquitoes showed no associative learning with Z-β-ocimene, and a weak one with phenylacetaldehyde, because (a) the taste of the compounds induced aversive conditioning, (b) the optimum release rates of the compounds were not used to condition the mosquitoes, or (c) the mosquitoes could associate the reward with specific compounds. In the next series of experiments, hypotheses (a) and (b) were tested.

Conditioning through olfactory system

Exp. 1: Pre-exposed vs novel compounds. The mosquitoes were able to associate 2% sugar solution with phenylacetaldehyde that was emitted from surface of the germination paper, by responding to the compound upon the second exposure. The results suggest that volatile exposure through the olfactory system is enough for An. gambiae mosquitoes to associate the compound with the reward, and feeding on both unconditioned and conditioned stimuli may be an ineffective conditioning method for the mosquitoes, due to its unpalatability. Also, toxicity tests of compounds are unnecessary when mosquitoes can learn by smelling them. Nonetheless, mosquitoes were not able to associate Z-β-ocimene with the reward, suggesting that the compound’s taste was not the only reason that

96 associative learning did not occur with this compound. It should be noted that while the innate responses of the naïve mosquitoes to phenylacetaldehyde and Z-β-ocimene were fairly equal, the mosquitoes were able to associate phenylacetaldehyde with a reward in one trial, and they preferred it even when the novel compound (i.e., Z-β-ocimene) was presented at the same time.

If mosquitoes can show a similarly strong associative learning with blends of phytochemicals and their nectar resources in nature, more advanced odor-based methods of surveillance and control will be needed. Current methods of developing attractants for naïve mosquitoes may not be useful for older and more experienced individuals, because they do not take into account the ability of mosquitoes to learn a novel scent. Similar flexibility in Manduca sexta (Riffell et al. 2008) has shown that while this moth has an innate preference for one plant scent with benzenoids, the main headspace constituents, the moth can explore and learn the other plant species that emit mainly esters.

Exp. 2: Effect of release rate during testing. In this bioassay mosquitoes were conditioned through the olfactory system, but unlike Exp.1, no novel compound was in the second arm of the olfactometers. It was concluded that associative learning in mosquitoes was demonstrated only when the optimum dose of phenylacetaldehyde was used in olfactometer experiments. It is likely that the conditioned mosquitoes were confused by an overly high release rate of phenylacetaldehyde during the testing phase, which may have caused an exhaustion of the odor receptors, due to their constant firing.

97 Exp. 3: Effect of release rate during conditioning. It was hypothesized that one of the reasons that associative learning had not been observed with Z-β-ocimene was because the optimum release rate was not used to condition mosquitoes. This experiment showed that this might be true for females of An. gambiae. Hence, probably the best way to condition mosquitoes is to use an optimum dose of each compound, which can be obtained by dose-response bioassays (see above, for details). This method, in addition, facilitates the process of conditioning mosquitoes. However, it is unclear why males were not able to associate the reward with ocimene. More studies are needed to understand whether the discrepancy between responses of the males and the females is because of different odor receptors, difference in their central nervous systems, or differences in the conditioning method.

Conclusion

Although An. gambiae mosquitoes were unable to associate plant scent (S. didymobotrya) with sugar in olfactometer experiments, they could successfully pair single-compounds (CS) with the reward (US). The unsuccessful trials with the plant volatiles were probably due to differences in volatile dose, which usually varies among individuals of one plant species. To avoid this issue, leaning bioassays were conducted using single compounds.

98 In this study, 24 hr was chosen as the time period between conditioning and testing, because An. gambiae is nocturnal and adults often obtain sugar from plants every night (Gary and Foster 2006). To search for a sugar concentration (i.e., reward) that would keep mosquitoes hungry and motivated enough 24 hr after conditioning, survival rates of 4 concentrations of sucrose were measured. The survival rates with a 2% sugar solution and sugar cubes (100%) were similar. However the 2% sugar solution was selected for conditioning mosquitoes, because they were able to imbibe the solution during a 15-min conditioning period. The toxicity range showed that mosquitoes could be fed 2% sugar solutions containing 2.5 µl phenylacetaldehyde, 2.5 µl linalool oxide, or 10

µl Z-β-ocimene. Phenylacetaldehyde, linalool oxide, and ocimene were selected for measuring the release rates from sachets and to test the dose response relationship of mosquitoes in the olfactometer. The optimum responses were obtained from 1 µl phenylacetaldehyde, 100 µl ocimene, and 10 µl linalool oxide in 1 ml mineral oil. The mosquitoes that were conditioned with 2.5 µl phenylacetaldehyde showed a higher response when tested 24 hr later with 1 µl of the compound vs. blank in the olfactometer.

Only females showed learning capability with ocimene when both conditioning and testing were conducted using 100 µl of the compound. The learning capability was stronger when the mosquitoes were conditioned through the olfactory system. If the conditioned mosquitoes were tested using a novel compound vs. the conditioned compound, mosquitoes responded significantly to the conditioned compound, suggesting that conditioned mosquitoes may not be distracted by novel compounds. This bioassay

99 suggests that in nature, experienced mosquitoes may more frequently visit the plants that provide sugar, even though their innate response to the plants is low.

100 Acknowledgment

The authors thank Krystal Seger, Benjamin D. Barker, and Colin Przybylowicz for their assistance in running parts of bioassay tests; Ashley N. Jackson, Jonathan D. Krammer, and Jacob D. Abramson for maintaining mosquito colonies and providing adults for experiments. We also thank, Joan Leonard, Emily Horn, and George Keeney for access to and help in the OSU Biological Sciences Greenhouse. This work was supported by NIH grant R01-AI077722 from the National Institute of Allergy & Infectious Diseases

(NIAID) to W.A.F. Its content is solely the responsibility of the authors and does not represent the official views of the NIAID for the National Institute of Health.

101 Table 4.1: Mean (± SE) response latency of males Anopheles gambiae to 12 chemicals diluted in mineral oil. Among 4 concentrations (1, 10, 100, and 1000 nl/ml) for each chemical, the one with shortest mean response time (sec) of mosquitoes was reported, along with proportion of mosquitoes that probed or walked, flew, or did not respond.

Mann-Whitney U test was used for comparison between mean of each chemical and mineral oil (40 sec).

Difference from Proportion of each Chemicals Mean ± SE Mineral oil responses (%) (concentration: nl/ml) (sec) Mann- Probing or P Flying None Whitney U Walking nonanal (100) 8.8 ± 7.8 40 40 20 2.5 0.03 decanal (1000) 8.8 ± 7.8 80 0 20 benzaldehyde (10) 16.6 ± 9.55 40 20 40 linalool oxide (1000) 16.6 ± 9.55 5 0.15 40 20 40 octenol (1000) 16.6 ± 9.55 40 20 40 nonanoic acid (100) 24.4 ± 9.55 0 40 60 phenylacetaldehyde (10) 24.4 ± 9.55 0 40 60

α-pinene (10) 24.4 ± 9.55 7.5 0.31 20 20 60 geranylacetone (100) 24.4 ± 9.55 40 0 60 hexane (10) 24.4 ± 9.55 40 0 60 ethanol (1000) 32.2 ± 7.8 0 20 80 10 0.69 Z-β-ocimene (100) 32.2 ± 7.8 20 0 80 102 Table 4.2: Estimated regression line of release rate (mg/hr) by concentration (%v/v) for phenylacetaldehyde, linalool oxide, and Z-β-ocimene. Mineral oil was used as the diluent in a 3.81 by 5 cm low-density polyethylene sachets (51µm thickness) at 25 °C an 70%

RH.

Release rate (mg/hr) = a * Concentration + b

Compound a b r R2 F a

Phenylacetaldehyde 0.01 0.0415 0.994 98.7 771.7

Linalool oxide 0.0088 0.0141 0.997 99.4 1598.4

Z-β-Ocimene 0.0168 0.002 0.995 99.0 1031.6

a for all tests: df = 1, 10; P < 0.0001

103 Table 4.3: Comparison of dose response of Anopheles gambiae males and females to phenylacetaldehyde, linalool oxide, and Z-β-

ocimene at three concentrations (µl / ml) in sachet. The mean percent discrimination out of total responding (treatment arm / sum

of two arms), and total responses out of total released mosquitoes, of the three concentrations were compared to each other.

Phenylacetaldehyde (µl / ml) Z-β-Ocimene (µl / ml) Linalool oxide (µl / ml) Sex 0.1 1 10 1 10 100 1 10 100

1

0

4 37.2 67.7 31.5 44.9 21.9 62.3 31.8 55.6 36 Female Treatment / K-W = 6.3, df = 2, P = 0.043 K-W = 4.99, df = 2, P = 0.08 K-W = 1, df = 2, P = 0.61 Total responses (%) 34.2 75.1 20.6 41.6 25.1 79.0 21 49.8 27 Male K-W = 17.2, df = 2, P < K-W = 6.41, df = 2, P = 0.04 K-W = 2.58, df = 2, P = 0.28 0.0001

Female K-W = 6.7, df = 2, P = 0.035 K-W = 2.8, df = 2, P = 0.247 K-W = 0.03, df = 2, P = 0.98 Analyses of

Total responses Male K-W = 8.4, df = 2, P = 0.015 K-W = 0.86, df = 2, P = 0.651 K-W = 0.55, df = 2, P = 0.758 Table 4.4: Factors that may influence associative learning in Anopheles gambiae. Prior to conditioning and testing, the tested subject should be hungry. During conditioning in this study, mosquitoes were exposed to unconditioned and conditioned stimuli simultaneously, for 15 min. Exposure time of testing was 12 hr.

Starvation Starvation Conditioning Testing period period Conditioned Unconditioned Conditioned Stimulus time Stimulus Stimulus (Volatile) (Reward) (Volatile) 24 hrs Quality Quality 24 hrs Quality Quantity Quantity Quantity 0 Exposure time Exposure time Exposure time (15 min) (15 min) (12 hr)

105 Blank arm Plant arm

Plant 10% 24% s p u o r g

d e n o i t i

d Naïve 10% 17% n o C

25% 20% 15% 10% 5% 0% 5% 10% 15% 20% 25% Mean response proportion

Fig. 4.1: Comparison of mean percent responding of naïve and conditioned Anopheles gambiae adults in an olfactometer baited with plant volatiles (Senna didymobotrya) vs. blank. 24 hr prior to the olfactometer test, mosquitoes were fed 2% sugar solution alone

(naïve control), and with the sugar solution while being exposed to the headspace of S. didymobotrya for 15 min (conditioned treatment).

106 Female 100 90 80 70 ) %

( 60 0

e t

a 2

r 50

l

a 50 v

i 40 v

r 95 u 30 S 100 20 10 0 0 1 2 3 4 5 6 7 8 9

Day

Male 100 90 80 70 ) %

( 60 0

e t

a 2

r 50

l

a 50 v

i 40 v

r 95 u

S 30 100 20 10 0 0 1 2 3 4 5 6 7 8 9

Day

Fig 4.2: Survival of Anopheles gambiae females and males on water (0), and after 12 hr feeding on three concentrations of sugar solutions (2, 50, and 95%), and on sugar cubes

(100%). Day 0 represents 12 hr after removing sugar (2 d post-emergence).

107 100 Female 90 80 70 )

% Control (

60 e

t LO 5ul a

r 50

l PHE 2.5 ul a v

i 40 ZBO 5ul v r

u 30 ZBO 10ul S 20 ZBO 50ul 10 0 0 1 2 3 4 Day

100 Male 90 80 70 )

% Control (

60

e LO 5ul t a

r 50

l PHE 2.5 ul a v

i 40 ZBO 5ul v r

u 30 ZBO 10ul S 20 ZBO 50ul 10 0 0 1 2 3 4 Day

Fig 4.3: Toxicity range. Survival of Anopheles gambiae females and males after 15 min feeding on 2% sugar solution (control) containing 2.5 µl phenylacetaldehyde (PHE), 5 µl linalool oxide (LO), or 5, 10, and 50 µl Z-β-ocimene (ZBO) in 4 ml of solution.

108 25

20 1 PHE 10 PHE )

g 15 100 PHE m (

s 1 ZBO s o l

10 ZBO t

h 10 g i 100 ZBO e

W 1 LO 5 10 LO 100 LO

0 0 3 6 9 12

Time (hr)

Fig. 4.4: Weight loss (mg) of three concentrations (1, 10, and 100 %v/v) of phenylacetaldehyde (PHE), linalool oxide (LO), and Z-β-ocimene (ZBO) diluted with 1 ml mineral oil in 3.81 by 5 cm low-density polyethylene sachets.

109 Fig 4.5: Estimated regression lines for release rates (mg/hr) of phenylacetaldehyde

(PHE), linalool oxide (LO), and Z-β-ocimene (ZBO) by concentration (%v/v). Mineral oil was used to dilute the compounds in 3.81 by 5 cm low-density polyethylene sachets

(51 µm thickness) at 25 °C and 70% RH.

110 Control Treatment

10 6% 2%

Female 1 4% 11%

0.1 8% 5%

15% 10% 5% 0% 5% 10% 15% 10 4% 1% Male 1 6% 9%

0.1 7% 4%

15% 10% 5% 0% 5% 10% 15% Mean response proportion

Fig. 4.6: Dose-responses of males and females of Anopheles gambiae to phenylacetaldehyde. Mean percent responding of released adults to each arm of olfactometers baited with three concentrations (0.1, 1, and 10 µl/ml) vs. blank.

111 Control Treatment

100 3% 10%

Female 10 10% 4%

1 13% 6%

15% 10% 5% 0% 5% 10% 15%

100 2% 9% Male 10 14% 2%

1 8% 5%

15% 10% 5% 0% 5% 10% 15% Mean response proportion

Fig. 4.7: Dose-responses of males and females of Anopheles gambiae to Z-β-ocimene.

Mean percent responding of released adults to each arm of olfactometers baited with three concentrations (1, 10, and 100 µl/ml) vs. blank.

112 Control Treatment

100 8% 5%

Female 10 7% 12%

1 8% 6%

15% 10% 5% 0% 5% 10% 15% 100 12% 5% Male 10 6% 8%

1 10% 6%

15% 10% 5% 0% 5% 10% 15% Mean response proportion

Fig. 4.8: Dose-responses of males and females of Anopheles gambiae to linalool oxide.

Mean percent responding of released adults to each arm of olfactometers baited with three concentrations (1, 10, and 100 µl/ml) vs. blank.

113 LO arm (10ul) PHE arm (10ul)

LO (2.5ul) 7% 18% s p u o r g

d PHE (2.5ul) 22% 29% e n o i t i d n o

C Naïve 26% 9%

30% 20% 10% 0% 10% 20% 30% Mean response proportion

Fig. 4.9: The mean percent responses of released Anopheles gambiae to each arm of olfactometers, baited with 10 µl phenylacetaldehyde (PHE) vs. 10 µl linalool oxide (LO).

Adults were conditioned by 2% sugar solution alone (naïve mosquitoes), and mixtures of

2.5 µl linalool oxide or phenylacetaldehyde in 4 ml 2% sugar solution.

114 ZBO arm (100ul) PHE arm (1ul)

ZBO (10ul) 14% 14% s p u o r g

PHE (2.5ul) 7% 18% d e n o i t i d n

o Naïve 13% 10% C

20% 15% 10% 5% 0% 5% 10% 15% 20% Mean response proportion

Fig. 4.10: The mean percent responses of released Anopheles gambiae to each arm of olfactometers, baited with 1 µl phenylacetaldehyde (PHE) vs. 100 µl Z-β-ocimene

(ZBO). Adults had been conditioned with 2% sugar solution alone (naïve mosquitoes), and mixtures of 10 µl ZBO or 2.5 µl PHE in 4 ml 2% sugar solution.

115 ZBO arm (100ul) PHE arm (1ul)

ZBO (10ul) 7% 10% s p u o r

g PHE (2.5ul) 4% 18%

d e n o i t i d n o Naïve 9% 7% C

20% 15% 10% 5% 0% 5% 10% 15% 20% Mean response proportion

Fig. 4.11: The mean percent responses of released Anopheles gambiae to each arm of olfactometers, baited with sachets of 1 µl phenylacetaldehyde (PHE) vs. 100 µl Z-β- ocimene (ZBO). Adults had been fed 2% sugar solution while exposed to either 10 µl

ZBO or 2.5 µl PHE placed on the germination papers, or no volatile (naïve).

116 Blank arm PHE arm (30ul) Percent Disrimination PHE (2.5ul) 13% 11% 52.1 (8.4)

Naïve 14% 10% 42.9 (7.0) s p u o r g

d e n o

i Blank arm PHE arm (1ul) t i d n o

C PHE (2.5ul) 3% 20% 83.3 (4.0)

Naïve 8% 14% 61.3 (6.6)

25% 20% 15% 10% 5% 0% 5% 10% 15% 20% 25%

Mean response proportion

Fig. 4.12: The mean percent responses of released Anopheles gambiae to each arm of olfactometers, baited with 1 or 30 µl phenylacetaldehyde (PHE) in sachets vs. blank.

Adults had been fed 2% sugar solution while exposed to the volatiles of 2.5 µl PHE placed on the germination paper or no volatile (naïve).

117 Blank ZBO arm (100ul)

ZBO (100 ul) 4% 11% s p u o r g

ZBO (10 ul) 7% 8% d e n o i t i d

n Naïve 5% 7% o C

20% 15% 10% 5% 0% 5% 10% 15% 20% Mean response proportion

Fig. 4.13: The mean percent responses of released Anopheles gambiae females to each arm of olfactometers, baited with 100 µl Z-β-ocimene (ZBO) in sachet vs. blank. Adults had been fed 2% sugar solution while exposed to the volatiles of either 10 or 100 µl ZBO released from sachets, or no volatile (naïve).

118 Blank arm ZBO arm (100ul)

ZBO (100 ul) 3% 6% s p u o r g ZBO (10 ul) 3% 7% d e n o i t i d n o C Naïve 3% 5%

20% 15% 10% 5% 0% 5% 10% 15% 20%

Mean response proportion

Fig. 4.14: The mean percent responses of released Anopheles gambiae males to each arm of olfactometers, baited with 100 µl Z-β-ocimene (ZBO) in sachet vs. blank. Adults had been fed 2% sugar solution while exposed to the volatiles of either 10 or 100 µl ZBO released from sachets, or no volatile (naïve).

119 Chapter 5: Effect of Plant Community on Vectorial Capacity*

Abstract

Here we describe a large-scale mesocosm for use and its effectiveness for experimentation with the African malaria vector Anopheles gambiae s.s. Giles (Diptera:

Culicidae) in temperate climates. A 3-wk survival trial with a single-cohort demonstrated successful mating, blood feeding, oviposition, and long life. Then, four replicates with sugar-rich and sugar-poor plants were conducted to assess survival rate, biting rate, fecundity and insemination rate. The sugar production of plants did not affect on biting rates, fecundity and insemination rate of females. However, survival rates of males and females were greater in the sugar-rich mesocosm. Consequently, vectorial capacity was higher in the sugar-rich mesocosm. Possible methods of controlling mosquito populations using sugar are discussed.

Keywords: Anopheles gambiae, vectorial capacity, large mesocosm, sugar feeding

*B. T. Jackson, B. Ebrahimi, C. M. Stone, and W. A. Foster. In preparation 120 Introduction

Nectariferous plants can provide shelter and nectar meals for vectors, thus vegetation has been considered as one of the predictors of mosquito abundance (Reisen

2010). Sugar is the only food source for male mosquitoes, but it increases longevity of females (Gary & Foster 2001, 2004; Impoinvil et al. 2004; Stone et al. 2012a). Previous studies have reported higher biting rates of females (Straif & Beier 1996; Gary & Foster

2001, 2004) and lower survival and mating capability of males (Gary et al. 2009, Stone et al. 2009) in absence of sugar. Consequently, sugar deprivation can decrease insemination rate and fecundity of females. Mosquito density relative to human, biting frequency, survival rate, and duration of the pathogen extrinsic cycle contribute to a function called vectorial capacity, which describes the number of secondary infections caused by a population of mosquitoes per daily exposure to an infected host (Garrett-

Jones 1964; Dye 1992).

Stone et al. (2012a) found a higher survival and a lower biting rate of the malaria vector Anopheles gambiae s.s. in the environment with a higher sugar production.

However, they found substantially different vectorial capacities among four replicates, suggesting that either the sugar production of plants varied throughout all replicates, the females’ access to blood was too short (i.e., half an hour) and only in the early mornings, or other physiological or behavioral variables of the mosquitoes can affect vectorial capacity. To assess vectorial capacity, an experiment following Stone et al. (2012a) was

121 carried out, but the blood-feeding was allowed in the middle of the night for 1 hr per mesocosm, between 23.00 and 1.00 hr. Also the amount of vegetation in the mesocosms was increased by adding more plants. The first two replicates had a high number of

Parthenium hysterophorus with flowers in the sugar-poor treatment, whereas replicates 3 and 4 had a lower number of P. hysterophorus with no flowers, in order to determine whether a high quantity of “sugar-poor” plants could make up for their poor nectar status, i.e., whether mosquitoes can locate sufficient sugar if they have access to a greater number of plants.

All experiments were conducted in two similar mesocosms. In recent years the use of semi-field systems has gained recognition and popularity among vector ecologists, in part driven by a need to understand aspects of gene flow and relative fitness of genetically- or biologically-modified mosquitoes (Knols et al., 2002; Ferguson et al.,

2008; Fachinelli et al., 2011; Ritchie et al., 2011). Such evaluations depend on accurate and meaningful measures of life-history parameters, such as age-specific survivorship, fecundity, and mating behavior. Several studies have demonstrated that the decreased flight activity associated with small laboratory cages can misrepresent natural energetic intake and expenditure (Stone et al., 2011), resulting in artificially high insemination and survivorship rates (Gary et al., 2009; Ponlawat and Harrington, 2009; Stone et al.,

2009b). It is also paramount that the design allows for an accurate assessment of the standing population of both males and females. Before conducting vectorial capacity

122 experiments in the mesocosms, we assessed their suitability for sugar feeding and mating behavior of An. gambiae in one of them.

Materials and Methods

Mosquito Rearing and Maintenance

The Mbita strain of An. gambiae was reared following the method described in chapter 2. To obtain approximately 2,000 1-d-old adults for each replicate, 4,500 larvae were reared. Our earlier trials showed that 65-70% of the larvae pupated during the same

24-hr period, out of which same-day emergence was approximately 75-80%.

Mesocosm Construction

From plans we developed, MegaView Science Co., Ltd. (Taichung, Taiwan) manufactured a pair of one-piece customized enclosures that required only our construction of supporting frames before installation. These two mesocosms were erected within adjacent 44.5 m2 rooms of The Ohio State University Biological Sciences

Greenhouse, each with a concrete floor, a 1-m-high concrete block wall on all four sides, with glass panels above them and on the ceiling. The pair of mesocosms later would allow simultaneous comparison of different experimental treatments (see Stone et al.,

2012a). Dimensions of a single mesocosm were 5.66 by 4.87 by 3.00 m (L x Wx H) for a total of 82.69 m3 (Fig. 5.1). The mesocosm sides, ceiling, and sleeves were made of white

123 polyester netting (42 by 12 mesh per sq. cm, 470 µm mesh aperture), whereas the floor material was white vinyl. To protect the vinyl floor, a 4.57 by 6.10 m white tarp

(Tarpaflex US, Naples, FL) was placed inside the mesocosm, and two more tarps were placed side-by-side beneath it on the concrete floors. The mesocosm was suspended on a framework of PVC connectors and pipes cut from 40 pieces of 3.81 by 304.8 cm (1.5 in by 10 ft) PVC pipe, purchased from a local hardware store, and also specialty 3-way and

4-way connectors (AFC Greenhouses, Buffalo Junction, VA). Pipes and connectors were secured with small screws, rather than glue, allowing adjustments and disassembly. Pairs of 0.46-m-long ties, located every 0.4 - 0.5 m along all horizontal and vertical corners and across the ceiling, allowed the mesocosm to be tied to the PVC framework and the greenhouse ceiling joists. Nine cylindrical sleeves (0.45 m diameter and 0.5 m length) made from the polyester netting were located in the ceiling to allow 400 W metal-halide wide-spectrum grow lights to be suspended inside the mesocosm. These lights were suspended by chains from the metal superstructure of the greenhouse. The cylindrical sleeves were then wrapped around the chains and held in place with plastic zip ties. Also located in the ceiling was a 1-m-long D-shaped zippered door made from the same material as the ceiling. This allowed access to mechanical parts above the mesocosm, in case repairs were required. Located on the floor were two zippered drains, covered by vinyl flaps. Each drain was 0.18 m2, with netting beneath the flap to allow water to drain through. The position of the drains on the vinyl floor corresponded to drains in the concrete greenhouse floor. Holes were cut in the previously mentioned tarps, above the

124 drains, to allow for proper drainage. Access to the mesocosm was through a large D- shaped zippered door (1.8 m high and 1.2 m wide) made from clear vinyl, positioned in the middle of one of the short walls. Outside of the door was a small antechamber to prevent mosquito escape. The 1 by 2 by 2 m antechamber consisted of netting on all four walls and ceiling and had a vinyl floor. Two of the short walls had clear vinyl zippered access doors. The doors could be partially unzipped to allow a person to enter or exit while minimizing the chance of escaping mosquitoes. When fully unzipped, they allowed import of large plants. On the vertical wall, left of the antechamber, were two sleeves, one for the humidifier exhaust pipe, the other for the humidistat controls (see below).

Inside of the mesocosm were four resting sites and two oviposition sites. Clay pots, placed on their sides, served as resting sites. Each measured 32 by 37 by 22 cm

(height by opening diameter by base diameter). The pot opening was covered with a circular plywood insert with a 13-cm hole in the middle, which allowed the mosquitoes to enter and exit the pots. Oviposition sites were 31-liter clear plastic storage bins (58 by 38 by 17 cm), filled with 5 L aged tap water and seated on dark fabric to contrast with the white floor.

Temperature was controlled within broad limits by the greenhouse heating and cooling systems. Wall-mounted steam radiators heated the room while cooler temperatures were maintained by louvered roof vents and a 1/3 horsepower exhaust fan that pulled air through an evaporative cooling system spanning the opposite wall of the greenhouse. Both systems were controlled by a thermostat within a sensor module

125 suspended in the middle of the mesocosm ca. 1 m from the floor. Humidity was maintained with an Ocean Mist ® MH3 industrial ultrasonic humidifier (Mico Inc., El

Monte, CA), which was controlled by a humidistat located adjacent to the sensor module.

A solenoid actuated gate was fitted to the end of the humidifier exit pipe to prevent mosquitoes from flying into the humidifier when it was not running. The grow lights were programmed to turn on 30 min after sunrise and turn off 30 min prior to sunset, but within that photophase they were on only when outdoor lighting dropped below 15,000 lux. The delayed onset and early termination of lighting allowed for more natural morning and evening crepuscular periods.

As needed, plant debris was removed by vacuum cleaner from the mesocosm floor, and the covering tarp was taken outside for washing. Accidental tears were easily repaired with fabric glue. Even with a scale inhibition system (CUNO, Meriden, CT) attached to the water intake of the humidifier, the minerals in the greenhouse water lines created problems, requiring monthly maintenance to remove excess mineral deposits from the tank. The humidifier’s ceramic disks were replaced every 6 mo and the float valve every 8 -12 mo.

The two side walls of the greenhouse room were covered with reflective insulation (FarmTek, Dyersville, IA) and the hallway-facing wall with 6 mil black plastic sheeting to prevent excess light after sunset. The outer sunset-facing wall was covered with black polypropylene commercial shade cloth (60% shade) (Hummert International,

Earth City, MO). It also was suspended above the mesocosm. The shade cloth shielded

126 the cage from direct sunlight, helping to maintain temperatures within the specified range and minimizing damaging effects of ultraviolet light on the netting. At mid-day on a sunny day, with the grow lights turned off, the light intensity at floor level of an empty greenhouse room, lacking shade cloth and mesocosm netting, was 79,600 lux, as measured by light meter (LX-1010BS). Inside the fully constituted mesocosm the light intensity was 14,830 lux, illustrating the need for the grow lights to provide more light.

The total cost of a single mesocosm setup was approximately US $3,123.00. This included the prefabricated mesocosm, PVC framework, humidifier and its accompanying parts, tarps, resting sites, and oviposition sites. Having available space in a greenhouse allowed us to keep the cost down, because we did not have to provide a suitable structure to house the mesocosm and control temperature.

Preliminary observations consisted of the release of cohorts of mosquitoes into the mesocosm containing resting sites and potted plants. In the earliest trials, to determine whether males were forming mating swarms just before and after sunset, no mosquito flight activity was observed. A large proportion of dead mosquitoes was found in both corners of the outer sunset-facing wall – which also received minor light pollution of nearby buildings in the evening sky. To minimize extraneous artificial light, a long and wide strip of reflective insulation was installed on the window at each corner of this wall.

An artificial horizon (inspired by Marchand, 1985) was created along the exposed glass windows by covering their lower portions, up to ca. 1.5 m, with an opaque black cloth.

Following these modifications, male swarms were observed at dusk along this wall,

127 particularly at the dark-light horizon and near the corners where the window met the insulation.

Testing Mesocosms

Four replicates of overnight assays were performed to assess whether sugar feeding occurred in a mesocosm when mosquitoes had access only to various potted plants (5 Tithonia diversifolia and 1 Senna didymobotrya). Mosquitoes (200 pupae were used for each replicate) were introduced into the mesocosm the afternoon after they emerged and were collected by backpack and mouth aspirators 1-2 h after sunrise the following morning, following the same protocol as used in Stone et al. (2012b). Collected mosquitoes (n = 239) were kept in a -40˚C freezer until assayed by cold-anthrone test for the presence and estimated amount of ingested fructose (Haramis & Foster, 1983), indicating undigested plant sugar. Surviving females were dissected to detect the presence of sperm in the spermatheca. A survival experiment was conducted, in which 854 An. gambiae adults emerged from pupae in the mesocosm, where they had access to eight nectariferous plants identified as possible hosts for this mosquito (Manda et al., 2007), four resting sites, a human blood source (B.E.) (1 h nightly) (IRB permit 2004H0193,

IBC permit 2005R0020), and two oviposition sites.

Measuring Vectorial Capacity

128 Following Stone et al. (2012a), four replicates of a vectorial-capacity assay were conducted, comparing one sugar-rich and one sugar-poor mesocosm. Table 5.1. shows the plant composition for each replicate. Because the size of plants varied, their number was changed so that the two mesocosms in all replicates had similar amounts of vegetation, so that we could assume that the amount of sugar production in the sugar-rich mesocosm was constant among four replicates. To increase the chance of finding dead mosquitoes , the tops of all resting pots were covered with long pieces of white stockinette. Every 3 days, the sugar-poor plants in replicates 3 and 4, in which the access to sugar was more limited than in replicates 1 and 2, were checked for pests and honeydew on leaves, and they were removed if found. Plants were watered daily throughout the 21 d of each experiment.

After providing the mesocosms with plants, resting pots, and oviposition containers, equal numbers of pupae (1100 -1300) were placed in the two mesocosms in the afternoon. The following day (designated d 0), all unemerged pupae were removed from the mesocosms and counted to find out how many adults had emerged. To blood feed females , a human blood host (B.E.) was available for 1 hour per mesocosm between the hours of 2300 and 0100 every night. The first host exposure started before midnight of d 0, when adults were < 28 hr old. The host wore a Tyvek ® coverall suit with a 8 ×

30 cm opening at the front part of each leg, so that mosquito would have access to blood only from the shins of the host's legs. A Sony ® Camcorder (HDR-SR11) with night vision feature and a built-in infra-red lamp, was fixed on a tripod and recorded the front

129 part of the host's legs during blood-feeding. Biting rate was measured by counting number of engorged mosquitoes on the legs, divided by estimated number of living females. Laid eggs were transferred to a round filter paper (Whatman 1001-320), scanned

(HP ® Deskjet 1051), and their numbers estimated with ImageJ software (Mains et al.

2008). Every morning, the mesocosms were inspected for dead mosquitoes. Survival rates of males and female were calculated by subtracting the dead mosquitoes from the initial release, as a life table was constructed. The dead were noticeable in contrast to the white floor and thus could be collected easily. Resting sites as well as plant leaves and pots also were checked for dead individuals. Wing length of both sexes and insemination status of females were determined by compound microscope. If the number of collected bodies was < 10, all of them were used for measuring wing length and dissected for presence of sperm in the spermatheca. Only 10 of 10-35 collected bodies, and 20-25% of

>35 bodies, were randomly selected for dissection and wing-length measurements.

Mosquitoes recovered alive at the end of the experiment (d 21), and those whose time of death was not known (the discrepancy between number released and number alive plus dead mosquitoes recovered, and also mosquitoes accidentally killed during the experiment), were entered as censored data points.

In a separate trial, the wings of dead males and females that were collected daily from a sugar-rich mesocosm were measured to establish a correlation between the mortality and the mosquito size. The plants composition in the sugar-rich mesocosm was similar to the replicated 1.

130 The calculated vectorial capacity used this basic formula:

C = (m ∙ a2 ∙ p n) / (-log(p)) with m (density of mosquitoes per host) set to an arbitrary 5 and n (extrinsic incubation period) at 12. The values for a (host biting rate) came directly from the experimental results, whereas a constant survivorship p was calculated from the predicted mean age at death (assuming that this is equivalent to 1/-ln(p)).

Statistical Analyses

Kaplan-Meier survivorship curves were constructed for males and females of each mesocosm test. Differences in survivorship were analyzed using Cox proportional-hazard analysis in R. Differences in biting rates between environments were analyzed for each replicate separately with GLS (generalized least squares) in R, allowing different variances for the treatments (sugar-rich and sugar-poor). A temporal auto-correlation structure was included in the regression model and allowed to differ for the rich and poor mesocosms. A paired t-test was used to compare daily egg output per female between the rich and poor environments. One-way ANOVA with Tukey b post-hoc test was used to compare mean wing size of survived females in the sugar-rich mesocosms that were collected on the last day of each replicate. If data were not normal, Mann-Whitney U and

Kruskal-Wallis tests were used instead of the t-test and one-way ANOVA.

131 Results

Testing Mesocosms

The results of the mosquito-performance test in a mesocosm are shown in Figs.

5.2A, B. Fructose positivity was affected by both sex (Binomial logit: χ2 = 40.3; df = 1; P

<0.0001) and replicate night (Binomial logit: χ2 = 24.6; df = 1; P < 0.0001). The rate of sucrose positivity of all replicates, combined, was 74.7% for males and 36.6% for females. A comparison of the amounts of fructose detected a significant interaction between sex and replicates (Kruskal-Wallis = 15.64, df = 5, P = 0.008). Differences in fructose positivity and fructose amount among replicates may have been caused by changes in plant health and nectar production.

At day 21, 54% of the males and 80% of the females were still alive (Fig. 5.3). At the termination of the experiment, all surviving females were found to be inseminated.

The average recorded temperature and relative humidity inside the mesocosm were

25.2°C (range: 20.7 - 33.7°C) and 71% (range: 50 - 81%), respectively.

Measuring Vectorial Capacity

For both males and females, a model with an interaction term for replicate and treatment was the minimal adequate model (Table 5.2, Fig. 5.4). In general, mosquitoes of both sexes in the sugar-rich environment survived longer. Sugar deprivation had a greater effect on males than on females.

132 The mean biting rates of mosquitoes in the rich and poor mesocosms were not consistent. The females in replicates 1 and 3 showed a tendency toward a higher biting rate in the sugar-poor mesocosm. And in replicates 2 and 4 the biting rate tended to be higher in sugar-rich one . Only in replicate 2 was a significant difference detected (Table

5.3, Fig. 5.5). When combined, the mean biting rates in the rich and poor mesocosms were, respectively, 0.39 (SE = 0.04) and 0.40 (SE = 0.04) per night (paired t-test = 0.22 , df = 3, P = 0.84).

The survival values and biting rates were used to calculate vectorial capacity

(Table 5.4) for the sugar-rich and sugar-poor environments of each replicate. The lowest and highest vectorial capacity were observed in the sugar-poor and sugar-rich mesocosms, respectively. Mean vectorial capacity of the sugar-poor and sugar-rich mesocosms were 35.7 (SE = 9.4) and 82.5 (SE = 41.2) respectively. However, the difference was not significant (paired t-test = 1.33, df = 3, P = 0.28).

In all replicates except for replicate 3, insemination rate was higher in the sugar- rich mesocosm than in the sugar-poor one (Fig. 5.7). Nonetheless, daily egg output per female of the two environments of each replicate did not differ (Table 5.5). The mean wing lengths of surviving females that were collected on day 21 in the sugar-rich mesocosms of each replicate, varied among the four replicates. Although all were reared under the same conditions, females in replicates 1 and 3 had, respectively, the smallest and largest wing lengths (Table 5.6). The wings of dead males and females of the separate trial, are presented in Fig. 5.8. The mean wing size of dead females and males increased

133 between day 1 and 6 (P < 0.05), but there was no difference in the mean size of dead mosquitoes in the following days (P > 0.05).

Discussion

Testing Mesocosms

In our small mesocosm design (Stone et al., 2009a), we relied on random subsamples of the resting population to estimate total population size. Drawbacks of this method are inherent sampling error and either a consequently diminished population or the mosquito stress created by removal-replacement sampling. Even total counts of resting mosquitoes are imprecise unless every individual is located, and distinguishing between males and females within their resting sites can be difficult. Here we describe a cost-effective, easily erected enclosure that addresses the issues of flight space, population estimation, and exclusion of unwanted predators and nectar thieves . We present data on both sexes of Anopheles gambiae from an experiment that tested their ability to locate and feed on plants and from an experiment on the survival, biting, and reproductive behavior over an extended period of time when they had access to plants and a human host.

While we did not attempt to maintain stable overlapping generations within this enclosure, we showed that the equatorial malaria vector An. gambiae survived and engaged in its normal behaviors related to mating, foraging for blood and sugar, and resting (Takken & Knols, 1999). The enclosure described here is easy to set up and

134 maintain, inexpensive, and sufficiently spacious to allow for a variety of realistic behavioral and population level investigations. Our approach to recreating semi-natural situations in temperate zones is likely to be useful for studies of other mosquito species, and it will be particularly relevant when natural energetic demands and the implications thereof on survival, reproductive potential, and behavior could significantly influence the outcome of the study.

Measuring Vectorial Capacity

The sugar production of plants affected the survival of mosquitoes, particularly males.

Increased longevity of mosquitoes in the presence of sugar has been seen in all previous studies (Straif & Beier 1996; Gary & Foster 2001, 2004; Stone et al. 2012a). At the end of day 21 of this study, the survival rate of females in the sugar-rich mesocosm was approximately 15-20% higher than that of females in the sugar-poor mesocosm, despite nightly access to blood. We observed a higher female survival in both environments compared to what Stone et al. (2012a) reported, which might have been because of the difference in the biting rates.

It was surprising, therefore, to see that the daily biting rate did not change between the two mesocosms, except for one case (replicate 2), in which the biting rate in the sugar-rich mesocosm was marginally higher. The highest and the lowest daily biting rates were respectively observed in the sugar-rich environment of replicate 4 and in the sugar-poor mesocosm of replicate 2. The mean biting rates between the sugar-rich and the

135 sugar-poor mesocosms were very similar (0.4 bite per female per night), and about twice as much as what Stone et al. (2012) have found in the sugar-poor mesocosm. One of the main differences between these two studies is that we let the mosquitoes have access to human blood at a time of night more typical of biting in the field and for a longer period of time each night night. Stone et al. (2012b) reported that when the access to a blood host was restricted using a bed net, mosquitoes more frequently took sugar meals. It is likely that when access to blood is not so restricted, sugar deprivation does not affect the biting rates of the mosquitoes but rather their survival.

The results indicated that vectorial capacity was mainly affected by the female survival, because there was no difference between the daily biting rates of sugar-rich and sugar-poor environments (Tables 5.2 and 5.4). Gu et al. (2011) and Beier et al. (2012) have also reported a higher vectorial capacity for Anopheles sergentii in a sugar-rich environment.

The sugar production of plants did not influence the daily fecundity per female.

Also, the insemination rates of the two mesocosms were similar. It should be noted that the total eggs production was about 20% higher in the sugar-rich environment which was most likely due to a higher female survival. The results of our study, along with Stone et al. (2012a) indicate that if males can survive at a high rate during the first 3 days post- emergence, females may have a higher chance of mating with males to store enough sperm to fertilize their eggs for at least 21 days.

136 Because the sugar-rich mesocosms of all four replicates were quite similar, the mean wing lengths of 20-35% of the survived females, in the last day of experiment were measured as a proxy of the the mosquito sizes. The difference in the mean wing lengths of replicates might have affected the mosquitoes survival and vectorial capacity.

But measuring the wing length of surviving females had a drawback – the wing lengths of dead mosquitoes were not included. The correlation between the mortality and the mosquito size showed that smaller mosquitoes died within the first week, suggesting that to determine if vectorial capacity is indeed influenced by the mosquito size, the size of each dead female should be included when estimating survival and biting rate.

It is worthwhile to investigate how an integration of deploying bed nets and attractive toxic sugar baits (ATSB) (Müller et al. 2010; Gu et al. 2011), and eliminating natural sugar sources may influence vectorial capacity. Bed nets will force female mosquitoes to depend on sugar for survival. If they are left with ATSB, and other sources of sugar are removed from an area, mosquito survival will drop dramatically, resulting in a very small vectorial capacity. Stone et al. (2012b) suggested that using ATSB might be more effective where larval development sites are close to human habitations. We think that using ATSB around blood hosts might be more effective if larval development sites are located farther. Because energy-depleted mosquitoes that are unsuccessful in taking blood meals may quickly switch to searching for sugar (Zappia & Roitberg 2012), their probability of feeding on the toxic sugar will increase.

137 Acknowledgements

We thank Eddy Lin of MegaView Science Co., Ltd., for his help on the design and manufacture of the mesocosms, as well as Joan Leonard and Emily Yoders-Horn for their support in the OSU Biological Sciences Greenhouse. This research was supported by

National Institutes of Health (NIH) grant R01-AI077722 from the National Institute of

Allergy & Infectious Diseases (NIAID) to W.A.F. Its content is solely the responsibility of the authors and does not represent the official views of NIAID or NIH.

138 Table 5.1. Plant compositions in high sugar or low sugar mesocosms. Number of used plants in each mesocosm is before each species name.

Replicate High sugar (sugar-rich) Low sugar (sugar-poor) Number 4 Senna didymobotrya (-)

4 Ricinus communis (-) 17 Parthenium hysterophorus (+) 1 2 Senna occidentalis (+) 4 Lantana camara (-)

2 Tecoma stans (+) 3 Senna didymobotrya (-)

3 Ricinus communis (-) 18 Parthenium hysterophorus (+) 2 1 Senna occidentalis (+) 4 Lantana camara (-)

1 Tecoma stans (+) 3 Senna didymobotrya (-) 3 Parthenium hysterophorus (-)

3 Ricinus communis (-) 3 Lantana camara (-) 3 1 Senna occidentalis (+) 4 Schefflera soleil (-)

1 Tecoma stans (+) 2 Schefflera actinophylla (-) 4 Senna didymobotrya (-) 3 Parthenium hysterophorus (-)

3 Ricinus communis (-) 2 Lantana camara (-) 4 1 Senna occidentalis (+) 3 Schefflera soleil (-)

1 Tecoma stans (+) 2 Schefflera actinophylla (-) (-) Without flowers (+) With flowers

139 Table 5.2. Predicted mean age at death of females in each replicate

Cox Proportional Replicate Poor Rich P -value Hazard 1 28.9 40.9 2.86 0.004 2 53.3 53.0 -0.51 0.61 3 80.5 143.1 -4.55 < 0.0001 4 53.9 169.3 -2.53 0.011

Table 5.3. Mean daily biting rates per female in sugar rich and sugar-poor mesocosms.

Replicate Poor Rich Coeff. SE Z P -value 1 0.426 0.381 -0.044 0.03 -1.5 0.14 2 0.305 0.371 0.066 0.03 1.99 0.05 3 0.362 0.344 -0.019 0.03 -0.59 0.56 4 0.486 0.504 0.013 0.033 0.39 0.70

Table 5.4. Vectorial capacity, calculated based on C = (m ∙ a 2 ∙ p n) / (-log(p)) when m =

5, n = 12, p = predicted mean age.

Replicate Poor Rich 1 8.87 20.99 2 37.53 30.67 3 45.44 77.85 4 50.94 200.3

140 Table 5.5. Mean daily egg output per female for each replicate after 20 days of blood- feeding.

Replicate Poor Rich Paired t-test P -value 1 38.66 34.89 2 29.05 39.08 1.006 0.389 3 42.62 49.47 (df = 3) 4 35.92 35.59

Table 5.6. Mean wing size of surviving females in sugar rich mesocosms, collected on day 21, for each replicate. Mean wing sizes of 4 replicates were tested using one-way

ANOVA with Tukey post hoc.

Mean Wing Size Post Hoc test Replicate SE ANOVA (mm) (Tukey B) 1 2.80 0.02 a F = 364.6 2 3.33 0.02 c df = 3, 212 3 3.40 0.01 d 4 2.90 0.02 b P < 0.0001

141 Fig. 5.1. (A) Schematic drawing of a mesocosm as it appeared during experimentation identifying the following: (a) metal-halide grow light, (b) humidifier exhaust pipe, (c) resting site, (d) oviposition site, and (e) temperature and humidity sensor. (B) A panoramic view from the antechamber looking into the mesocosm during an experiment.

142 Fig. 5.2. Anopheles gambiae males and females were held in a mesocosm overnight where they had access to six plants. This was replicated four times. (A) Proportion of males and females positive for fructose. (B) Mean amount of fructose (µg) ±SD for both males and females.

143

Fig. 5.4. Kaplan-Meier survivorship curves for male and female An. gambiae when exposed to sugar-rich or sugar-poor plants. Each line represents one replicate.

145

Fig. 5.5. Daily biting rates of mosquitoes in sugar-rich and sugar-poor mesocosms

(circles) of four replicates (colors), and the associated 3-d moving averages (lines).

Biting rates were calculated by dividing the number of observed bites per night by the number of females predicted to be present (initial number of females times the predicted survivorship).

146 1 2

3 4

Fig. 5.6. Distribution of biting rates among replicates. In replicate 2, the biting in sugar rich mesocosm was “marginally” higher than in sugar-poor.

147

Fig. 5.7. Insemination rates (3 days cumulative) of dead females collected daily. Blue line: sugar-poor mesocosm; Red line: sugar-rich mesocosm.

148

Fig. 5.8. Mean wing size of dead females and male collected daily from a sugar rich environment.

149 Chapter 6 : Delayed Egg Hatching of Anopheles gambiae (Diptera: Culicidae), Pending

Water Agitation*

Abstract

Mosquito eggs laid on water surfaces typically hatch spontaneously soon after the embryos within them become fully formed 1st-instar larvae. However, we have found that

Anopheles gambiae Giles, an important vector of malaria in Africa, exhibits delayed hatching until the water surface is agitated, a feature overlooked in most laboratory colonies. Agitation within 24 h post-oviposition, before embryonation was complete, failed to stimulate delayed post-embryonic hatching of isolated eggs on the following day

(day 2), when <1% had hatched spontaneously. However, 5 min of water agitation of these dormant pharate 1st-instar larvae on day 2 resulted in an almost immediate hatch of

63.3%, vs. 0% of non-agitated controls, plus another 3.9% vs. 0.3%, respectively, during the following 24 h. With daily agitation, installment hatching occurred mainly during 2-6 d post-oviposition. The mean cumulative hatch after 7 d of daily agitation was 83.1%, vs.

1.1% of non-agitated eggs. Experiments with eggs in groups demonstrated that egg density and activity of already-hatched larvae had no stimulatory effect. Eggs stored 1-4

* B. Ebrahimi, S. Shakibi, W.A. Foster. Submitted to the Journal of Medical Entomology 150 wk at 25.5°C or at 15.5°C, and then agitated daily for 6 d at 25.5°C, showed a gradual decline in viability. Viability was sustained longer at the lower temperature. Implications of agitation-induced egg hatching for rainy-season and dry-season ecology of An. gambiae are discussed. Suspended hatching and cool storage already are proving convenient for colony maintenance and experiments on this species. It also may be useful for efficient mass rearing and accurate modeling of weather-based population dynamics.

Keywords: Anopheles gambiae, egg, hatch, agitation, temperature, mass rearing

Introduction

Anopheles gambiae Giles sensu stricto is a major vector of malaria in equatorial Africa

(Coetzee 2004), where about 175.5 million cases and 91% of worldwide deaths by the disease were reported in 2009 (World Health Organization. 2010). Epidemics are correlated with wet years (Craig et al. 1999). Rainfall provides this species both with developmental sites (e.g., Gimnig et al. 2001) and with high humidity that probably favors adult survival and flight activity. About 2-3 d after each blood meal, a female lays about 50-200 boat-shaped eggs bearing the distinctive anopheline floats and protruding tubercles. The eggs are laid singly on the surfaces of open bodies of fresh water or on hypoosmotic damp soil (Valencia et al. 1996, Minakawa et al. 2001, Munga et al. 2005).

If eggs are kept in damp soil, they remain viable for about 14-18 d, depending on the soil

151 type (Deane and Causey 1943, Beier et al. 1990, Shililu et al. 2004). Eggs of An. gambiae are incapable of hatching until about 2 d post-oviposition; most hatch (ca. 90%)

2-3 d post oviposition (Beier et al. 1990, Yaro et al. 2006). Environmental factors such as atmospheric humidity, type of water (Yaro et al. 2006), and temperature (Valencia et al.

1996, de Carvalho et al. 2002, Bayoh and Lindsay 2003, Huang et al. 2006, Impoinvil et al. 2007) all affect the proportion that hatch. The eggs’ wide tolerable temperature range,

4-40°C (Beier et al. 1990, Huang et al. 2006, Impoinvil et al. 2007), makes them the most cold- and heat-tolerant of An. gambiae’s developmental stages. Successful completion of development, from egg to adult, occurs only between ca. 18 and 34°C (Bayoh and

Lindsay 2003).

Most research on egg hatching of mosquitoes has been conducted on aedine eggs.

Because aedine eggs typically are laid out of water, on container walls, plant debris, or soil, hatching generally requires submergence in water containing organic matter and microorganisms, which reduces oxygen tension (Gjullin et al. 1941, Borg and Horsfall

1953, Judson 1960, Judson et al. 1965, Ponnusamy et al. 2011). In addition, bacteria themselves may promote egg hatching (e.g., Ponnusamy et al. 2011), and previous exposure to high humidity, warm temperature, and long photoperiod also may be required

(Shroyer and Craig 1980, Campos and Sy 2006).

The first record of egg hatching stimulated by water agitation was reported for the aedine Psorophora ferox (Dupree and Morgan 1902). Two decades later, Young (1922) found that more hatching occurred when Ae. aegypti eggs were agitated in water or left in

152 rain for 5 min. Borg and Horsfall (1953) referred to studies on agitation-stimulated hatching in other aedines but concluded that chemical stimuli outweighed the physical ones. In later studies, water agitation was suggested as a stimulus for egg hatching of

Aedes caspius, Aedes vittatus (Roberts 2001), and Aedes atropalpus (G.F. O’Meara, personal communication).

Among anophelines, James and Liston (1904) found that An. gambiae eggs laid on water hatched in 2 d, and those lying dormant on mud would hatch if they were stimulated by adding water. However, there was no mention of any effect of water- induced agitation or disturbance until Muirhead Thomson (1946) suggested that a film on the water surface, produced by iron bacteria, was responsible for keeping dormant eggs unhatched until the thin film was broken-up by rain. Other studies have demonstrated that an abrupt increase in water temperature or a drastic change in water salinity stimulates hatching in Anopheles melas (Giglioli 1965). Water agitation was first suggested as a natural anopheline hatching stimulus for An. squamifemur (Boreham and Baerg 1974). In

An. gambiae sensu lato, most studies have concluded that flooding is the hatching stimulus for dry or semi-dry eggs (Deane and Causey 1943, Beier et al. 1990, Shililu et al. 2004, Impoinvil et al. 2007), and it is generally assumed that eggs laid on water will hatch as soon as embryonation is complete (Minakawa et al. 2001). However, a detailed study of the impact of water agitation on floating An. gambiae eggs has not been conducted, except for a recent examination of the genetic connection between staggered time to hatch and insecticide resistance (Kaiser et al. 2010).

153 The following study was prompted by a 2006 observation in our laboratory that a large proportion of An. gambiae eggs, left undisturbed in oviposition cups, did not hatch until they were rinsed in a jet of tap water more than 1 wk after being laid. We surmised that the hatching stimulus was mechanical agitation of water and that heavy rainfall might provide a similar stimulus under natural conditions. To confirm this and explore its ecological and practical implications, we measured 1) the effect of water agitation on both daily hatch proportion and total proportion of eggs hatching after 1 wk of daily agitations, 2) the effect of other eggs and of previously hatched first-instar larvae on hatching, and 3) the effect of cool storage for up to 4 wk on the proportion hatching.

Materials and Methods

Rearing and Maintenance. The Mbita strain of An. gambiae s.s. (S form) originally was colonized and identified from locally collected material in 2001 by the staff of the International Centre of Insect Physiology and Ecology (ICIPE) at Mbita Point,

Suba District, Nyanza, Kenya. A colony of this strain, maintained at The Ohio State

University for several years (Biosafety protocol No. 2005R0020), provided the eggs used in the present study, which was conducted during 2008-2011. All stages were held in a rearing and colony maintenance room at 25.5 ± 1°C, 70 ± 5% RH, and L:D about 12:12 h. Total darkness occurred between 2000 and 0730 hours, preceded and followed, respectively, by 45 min crepuscular periods of gradual dimming and brightening. Larvae

154 were reared in shallow pans in aged tap water, as described by Gary and Foster (2001).

Adults were held in 30 by 30 by 50 cm clear acrylic cages, with cotton wicks providing water and 10% sucrose.

Egg Collection and Preparation. To obtain eggs, one of the authors (B.E.) allowed 50-70 females to blood-feed on his arm between 1600 and 1800 hours

(Biomedical IRB protocol No. 200440193, FWA No. 00006378). Exposure was terminated after ca. 15 min, when about 70-80% of mosquitoes had engorged and ceased feeding. A shallow plastic oviposition cup (9 cm diam.), half-filled with 50 mL aged tap water, was placed in the cage 3 d post blood-meal, when the mosquitoes would be gravid.

The cup containing the eggs laid that night was removed slowly in the morning, and the eggs were transferred gently to experimental containers (see below) at 1400-1600 hours, about 18-20 h post oviposition (referred to as “1 d post-oviposition” in this study), assuming that most oviposition occurred in early night under these conditions (Haddow and Ssenkubuge 1962, Sumba et al. 2004). At this age and temperature, embryonic development would not yet be complete (Valle et al. 1999).

Floating eggs were transferred from the oviposition cup to 3.5 mL wells in 24- well Linbro® tissue-culture plates, McLean, Virginia, that each held 1.5 mL water. Each well was 1.7 cm diameter by 1.6 cm deep. Each egg was transferred individually by clean glass Pasteur pipette tip, which was inserted into the water near the egg and moved toward it until the pipette touched it. The egg adhered to the side of the pipette tip, which

155 was then slowly lifted out of the water and reinserted into one of the plate wells, where it detached and floated free when the pipette tip was drawn slowly down and away. When all wells contained eggs, the plate was covered with a fitted lid to prevent evaporation.

Total time to prepare each plate was about 10 min.

Exp. 1: Effect of Daily Agitation on Isolated Eggs. Three batches of eggs were used, each from a different mosquito generation. From each batch, ten 24-well plates were prepared, each with one egg per well. The plates were randomly assigned to treatment and control groups, five plates each, and stored together on an undisturbed shelf. The treatment plates were agitated daily for 7 d, starting 1 h after plate preparation.

Both control and treatment plates were inspected by naked eye, both immediately before and about 10 min after each agitation, and hatched first-instar larvae were counted. The final inspection was 24 h after the seventh agitation. For analysis, the Daily Hatch (DHi), where 0 ≤ i ≤ 6, was defined as the number of eggs that hatched within 24 h after each agitation, divided by the total number of eggs in each plate. For example, DH0 was the proportion hatching before the first agitation of pharate first-instar (fully developed) larvae, examined 2 d post-oviposition. The Daily Hatch on day 1 (DH1) was the sum of hatched eggs after the second agitation but before the third agitation, divided by the number of eggs. Numbers hatched in non-agitated controls were counted at the same times. Total Hatch (TH) for each plate was calculated by dividing the total number of hatched eggs after 7 d by the number of eggs.

156 At each agitation, each plate in the treatment group was uncovered, placed on a bench surface, and swirled by hand in a 10-cm-diameter circle at about 100 revolutions per min for 5 min. The swirling was slow enough to prevent the water and eggs from spilling out of the wells. If an egg became lodged on the side of a well, it was washed by pipette back onto the water surface by one or two drops of water from its respective well.

Five min of agitation was chosen, because preliminary experiments demonstrated that egg hatch did not increase significantly between 4 and 10 min of agitation. Food was not provided in the wells, so hatched larvae died within 3 d.

Exp. 2: Effect of the Presence of Other Eggs and of First-instar Larvae. To determine whether the presence of other eggs increases the hatching rate in the absence of water agitation, multiple eggs were added to wells in plates, as described above for single eggs. On the day after oviposition, from each of the three replicate batches of eggs, five wells of a plate (n = 5) were supplied with 10 eggs per well and four wells (n = 4) were supplied with 30 - 40 eggs per well. Results were compared to isolated non-agitated eggs in the previous experiment, which served as controls. The plate was held undisturbed on a shelf for 7 d post-oviposition without agitation, after which the wells were inspected.

Total Hatch was calculated for each well, and the two treatments were compared with each other and with the controls.

In a separate experiment involving a single batch of eggs, to determine whether the activity of newly hatched larvae increases the hatching rate, 10 wells of a plate were

157 supplied with 10 eggs each. Five of these wells (n = 5) received one newly hatched larva from another egg batch, and the other five received (n = 5) two larvae. The plate was held for 7 d post-oviposition without agitation, after which the wells were inspected and Total

Hatch of the two treatments were compared to each other and to the non-agitated isolated-egg controls of Exp. 1.

Exp. 3: Effect of Long-term Storage at Two Temperatures. To determine whether prolonged egg storage (i.e., egg age) affects agitation-induced hatching and whether low temperature extends egg viability during dormancy, eggs were stored at either 15.5°C or 25.5°C for up to 4 wk post-oviposition, then agitated after acclimation at the higher temperature. In this study15.5°C was chosen, because it is the approximate mean minimum temperature in malarious equatorial Africa (Bayoh and Lindsay 2003).

They were prepared and tested as follows: Ten eggs were transferred 1 d post-oviposition

(i.e., embryos <24 hr old) by pipette (see Exp. 1) to each of fifty 10-mL glass test tubes

(10 cm x 12 mm diam.) containing 5 mL of aged tap water. Each tube was sealed with

Parafilm M®, placed in one of two tube racks, then the racks were placed on their sides so that the tubes were horizontal. If an egg adhered to the wall of the tube, the tube was rotated gently within the rack until the egg floated free. One rack of 25 tubes was stored in the rearing room at 25.5 ± 1°C. The other rack immediately was placed in a 15.5 ±

0.5°C refrigerator, with 12:12h L:D fluorescent lighting but no crepuscular period.

Storage commenced on day 1 post-oviposition, when the embryos were incompletely

158 developed, so that mechanical disturbance during transfer of the eggs to tubes would not induce hatching.

The eggs were not agitated until the assigned day. On days 2, 8, 15, 22, and 29 d post-oviposition, five randomly selected tubes from each rack were transferred to a tray, while remaining horizontal, and were agitated by swirling the tray for 5 min at 100 revolutions per min. The samples stored in the warm room were agitated immediately, whereas those stored in the refrigerator were allowed to acclimate for about 30 min in the warm room before receiving the first agitation, because a preliminary experiment showed that egg agitation at the low temperature did not stimulate hatching (unpublished data).

Tubes from each storage temperature then were kept in the warm room for the remainder of the experiment, where they were agitated daily for another 5 d. The number of hatched larvae were counted by naked eye before, and 10 min after, each agitation. (One of the 25 tubes stored at 15.5°, tested after 1 day of storage, broke at the second agitation and was not part of the analysis.) The last count occurred on day 6 after removal from storage, 10 min after the sixth agitation. For each tube, the Daily Hatch and Total Hatch were calculated to assess the effect of egg storage at each temperature on hatching. Daily

Hatch (DHi where 0 ≤ i ≤ 6) was defined as the number of eggs that hatched within 24 h after each agitation, divided by the total number of eggs in each test tube. Similar to Exp.

1, DH0 was the proportion hatching before the first agitation of pharate first-instar larvae

(i.e., 2 d post-oviposition). Because Total Hatch (TH) included a few hatches before the

159 first agitation, Total Hatch by Agitation (THA) also was calculated, by deducting the small number of eggs hatched prior to the first agitation from Total Hatch for each tube.

A separate study evaluated the long-term viability of eggs stored at 25.5° for 1, 3,

6 and 12 mo post oviposition. Eggs collected 1 d after oviposition were placed, as previously described, on aged tap water in four small half-filled lidded cups (5.3 cm diam., 3.5 cm high), 100-150 eggs per container. The containers were inspected daily for hatched larvae during the first week, then monthly. The water level was checked weekly, and more was added by pipette as needed to compensate for evaporation from the loose- fitting lids. To avoid agitation, the tip of the pipette was first submerged, then water was expelled into the container. No eggs hatched immediately as a result of these replenishments. At the end of each storage duration, the lids were tightened and the cups vigorously agitated for 10 min on each of three consecutive days.

Exp. 4: Effect of Vigorous Agitation During Embryonic Development. To confirm the effect of egg agitation during embryonic development on hatching likelihood after development is complete, eggs were treated as follows: Vigorous and prolonged shaking was applied, rather than swirling. Ten day-1 eggs (i.e., <24 h post oviposition) were placed in each of five tubes as above (Exp. 3) at 25.5°C. After 10 min, they were shaken by hand for 15 min. Tubes were examined for the presence of larvae 10 min after agitation, and again 24 hr later, when they would be pharate first-instar larvae, capable of hatching spontaneously.

160 Statistical Analysis. For analysis, proportions hatching at various times were analyzed as mathematical ratios between 0 and 1 but were converted to percentages for reporting results.

In Exp. 1, hatching data did not differ significantly among generations and were combined for analysis. Total Hatch for each plate was the cumulative hatch over 7 d, divided by the 24 eggs in the plate. Each plate provided a single unit of measure, so in this case n = 15 (5 plates x 3 generations) for 360 agitated eggs and 360 control eggs.

Data from the other experiments were treated similarly.

Following the methods of Bayoh and Lindsay (2003), Koenraadt et al (2003),

Huang et al. (2006), and Impoinvil et al. (2007), the pattern of Daily Hatch (DHi) after each agitation (i.e., installment hatching) was described by a model (model 1) employing the following independent variables: agitation day (0 ≤ i ≤ 6) and its log- and square- transformed data. A stepwise method eliminated insignificant independent variables from the equation. Independent variables of the best equation were described by their coefficients and their Standardized Partial Regression Coefficients (β) and are reported in the results. The greater the absolute value of β of a variable, the larger was its impact

(Sokal and Rohlf 1995).

When data were not normally distributed, as determined by Kolmogorov-Smirnov test, they were transformed (e.g., log [100 DHi + 0.1]), and their normality was re-tested.

Student’s t-test and one-way ANOVA were applied to normalized data. If normality was

161 not achieved, differences between two groups or among multiple groups were evaluated by Mann-Whitney U or Kruskal-Wallis tests, respectively (Sokal and Rohlf 1995). For all tests, P = 0.05 was the criterion for statistical significance.

Two types of tests were applied to compare long-term hatch patterns at two temperatures (Exp. 3). One-Way Analysis of Variance (ANOVA) was used to compare proportions hatching after storage at each temperature, before the first agitation and again after it (Table 6.3). Two stepwise linear regression models (models 2 and 3) were run to predict log-transformed Daily Hatch on day i (log [100 DHi + 0.1]) at the two temperatures. In model 4, temperature (temp) was included. The following independent variables and their log- and square-transformed data were employed in the models: agitation day (0 ≤ i ≤ 6), egg storage duration ( 2 ≤ age ≤ 29), storage temperature (temp

= 15.5 or 25.5), and agitation day by egg storage duration (i · age). A stepwise linear regression line was estimated for Total Hatch by Agitation (THA). SPSS (ver. 19.0; IBM,

Armonk, NY) software package was used for all tests and to create 3-D graphics.

Results

Exp. 1: Effect of Daily Agitation on Isolated Eggs. No eggs hatched among agitated or control groups after agitation of the treated group on day 1 (Fig. 6.1).

Furthermore, before agitation on day 2, the proportion hatching spontaneously was <1%.

But immediately after agitation on day 2, 63% of the treated eggs hatched. Declining

162 proportions of those remaining unhatched after the first agitation had hatched immediately after each subsequent daily agitation, and negligible proportions hatched between agitations (Table 6.1). This pattern of delayed hatching after two or more applications of a hatching stimulus will be referred to as “installment hatching.” A linear regression model (model 1) of the DHi decline in hatching that fits the data is as follows: r = 0.955 and Adjusted R2 = 90.9% (F = 298.45; df = 3, 86; P < 0.0001), where

2 DHi = 1.12 (i) – 4.92 log i – 0.07 (i) – 0.38

(1)

The β for (i), log(i), and (i)2 were as follows: 7.62 , - 5.15, and -3.39, respectively. It means that Daily Hatch was mainly affected by agitation day (i) and its log-transform.

The mean Total Hatch on day 8, after 7 periods of agitation, was 83.1% (SE = 1.7). By contrast, the mean Total Hatch of non-agitated controls was 1.1% (SE = 0.5) (U = 0; n1, n2 = 15; P < 0.0001). Daily differences in hatching proportions between treated and control groups were statistically significant after agitations on days 2 and 3 and before agitations on days 3 and 4 (Table 6.1).

Exp. 2: Effect of the Presence of Other Eggs and of First-instar Larvae. After

1 wk, groups of 10 eggs per well had a Total Hatch of 6.7% (SE = 2.4). In groups of 30-

40 per well, the Total Hatch was 3.5% (SE = 0.5). These values were close to the spontaneous Total Hatch of isolated eggs in Exp. 1 (1.1%, SE = 0.5). The differences among them are not significant (Kruskal-Wallis = 1.06, df = 3, P = 0.79). In an ancillary

163 analysis, hatch rates of groups of 10 eggs per tube (see Exp. 3, below), both prior to the first agitation of pharate first-instar larvae (DH0) and also after 1 wk of daily agitations

(THA) (see below, Exp. 3), were compared to isolated eggs of Exp. 1 (Table 6.2). There was no difference between the results, confirming that egg density did not affect hatching. In the experiment to detect the effect of larvae, among wells containing one larva in addition to the 10 eggs, none had hatched by the end of the week. Among those with two larvae, 4.7% (SE = 1.9) had hatched. Again, the differences among these treatments, including the single-egg spontaneous hatch rate (Exp. 1), were not significant

(Kruskal-Wallis = 2.79, df = 2, P = 0.25).

Exp. 3: Effect of Long-term Storage at Two Temperatures. The proportion hatching prior to the first agitation (i.e., spontaneously) was very low and did not differ among the five age groups stored at the two temperatures (P = 0.20 at 15.5° and P = 0.27 at 25.5°) (Table 6.3). Among the 2-d-old eggs stored at 25.5°, 82% hatched after the first agitation, whereas those stored at 15.5° first hatched between the first and second agitations (2.5%), followed by substantial numbers after the second agitation (57.5%)

(Fig. 6.2). Among 8-d-old samples held at 25.5°, a few eggs hatched prior to the first agitation (2.0%) but many more afterward (59.5%), whereas among those stored at 15.5°, only 20% hatched after the first agitation and many more after subsequent agitations.

Installment hatching was particularly common among eggs stored for 8 d and was more common among those stored at 15.5°. After progressively longer storage durations, fewer

164 hatched, and installment hatching became inapparent. No delayed hatching occurred in

29-d-old eggs at 15.5°, i.e., they hatched in response to day-1 agitation only. No eggs hatched on day 6 of egg agitation, regardless of storage temperature or storage duration.

To predict Daily Hatch (DHi) after egg storage, two stepwise linear regression models (2 and 3) were estimated separately for 15.5° and 25.5°. In another model (4), temperature (temp) was included as an independent variable (see Table 6.4). In the 15.5° model, the greatest absolute value of β belonged to log-transformed agitation day (log

[i]), and then secondarily to the square of the storage duration (age2 ). This means that the

Daily Hatch at the lower temperature was affected only by these two variables. By contrast, in the 25.5° model, age was more important than the log-transformed agitation day. When temp was included in the Daily Hatch model (model 4), log-transformed agitation day and storage duration were found to be more important than the other variables.

The following model (model 5) estimates Total Hatch by Agitation (THA), with storage duration (age) and temperature (temp) as independent variables:

THA = 1.165 – 0.032 age – 0.011 temp

(5)

The regression line observed was r = 0.877, with Adjusted R2 = 75.9% (F = 76.57; df =

2, 46; P < 0.0001). In the above model, the β value of age (-0.867) was larger than temp

(-0.161), indicating that age (storage duration) had the greater (and negative) effect on

Total Hatch by Agitation.

165 After storage of batches of eggs for 1, 3, 6, and 12 months post-oviposition at

25.5°, then agitation for 10 min each day for 1 wk, no eggs hatched. Re-agitating these eggs 18 mo post-oviposition did not stimulate hatch. Microscopic examination of these eggs, prepared according to Shililu (2004), demonstrated that the embryos within them were dead but had developed fully, i.e., they had become pharate first-instar larvae, capable of hatching had they been agitated at an earlier age.

Exp. 4: Effect of Vigorous Agitation During Embryonic Development. After vigorous and prolonged shaking of eggs laid < 24 h earlier, no eggs hatched right after the agitation, and only one out of 50 eggs (2.0%, SE = 2.0) hatched without further agitation the following day. This is not statistically different from the 1.1% rate of its comparable group in Exp. 1. This result confirmed the absence of a delayed effect of agitation, if applied during embryonic development.

Discussion

Hatching Stimulus and Installment Hatching. The results demonstrated that the eggs of An. gambiae, which usually are laid on the surface of open bodies of water and undergo embryonic development there, rarely hatch without the stimulus of mechanical agitation. Within minutes of agitation, the pharate first-instar larvae dehisce from their chorions. Embryonated but unstimulated eggs almost always remain unhatched, and the

166 larvae within them can survive in this dormant condition for up to several weeks. Most eggs hatch right after the initial stimulus, but smaller numbers hatch after a second agitation 1 day later, and still smaller numbers upon successive agitations. This is the installment-hatching pattern commonly observed after each of a series of successive inundations in eggs of anophelines (Deane And Causy 1943, Holstein 1954, Beier 1990,

Shililu 2004, Yaro et al. 2006) and aedines (e.g., Wilson and Horsfall 1970, and many others). How the pattern might change if anopheline agitations were more frequent than once a day, or if the water temperature had been higher or lower, remains to be investigated. Total Hatch averaged 1.1% without immediately preceding agitation and

83.1% including agitation. The observed hatching pattern with agitation is quite similar to the results of Shililu et al. (2004), in eggs which might have been agitated, when incubated up to 20 d in a high-moisture environment. Nevertheless, our hatch rates at each egg age were ca. 30% higher.

Tests of the effect of other eggs and of hatched larvae on spontaneous (non- agitated) hatching showed that these are not factors that affect hatching. Furthermore, agitation prior to the completion of embryonic development (Exp. 1 and 4) demonstrated that agitation before the larvae are fully developed and capable of dehiscence has no effect on the spontaneous hatching rate after that period has been completed. We tentatively conclude that the mechanism for triggering hatching has no delayed-action component.

167 The mechanism by which agitation causes hatching remains obscure. We have considered the possibility that the stimulus is actually chemical or physico-chemical, just as reduced oxygen tension has been demonstrated to be a hatching trigger in aedines

(Gjullin et al. 1941, Judson 1960, Judson et al. 1965). However, our observations on An. gambiae suggest otherwise: the water we used was clean aged tap water. The eggs floated on the surface of that water, where the upper chorion was in contact with the atmosphere throughout development and during undisturbed storage. Falling droplets of water, which contain more, not less, dissolved oxygen (Graedel and Weschler 1981), provided a reliable egg-hatch stimulus (B.E., unpublished data). And the floating eggs hatched immediately after they have been prodded by the tip of a pipette.

Therefore, we conclude that the stimulus for egg-hatching in this study was entirely mechanical. The mechanism cannot involve removal of the exochorion, other structural changes, or enhanced water permeability. Otherwise, eggs shaken on day 1 would have hatched in high numbers in the absence of further agitation after embryonation was complete on day 2. Furthermore, a fully tanned chorion would seem unlikely to be susceptible to changes in permeability so long after oviposition

(Schreuders et al. 1996, Valencia et al. 1996), when our eggs nonetheless hatched in large numbers after a single agitation.

The hatching behavior of the Mbita strain of An. gambiae, from western Kenya, is described here, but we found similar characteristics of egg dormancy and agitation- induced hatching in the Suakoko strain from Liberia (unpublished data). Furthermore,

168 Kaiser et al. (2010) recently described a similar phenomenon in the GAH strain from

Ghana, though with a higher rate of spontaneous hatching. All three examples are S-form

An. gambiae. The presence of this characteristic in M-form An. gambiae (i.e., An. coluzzii Coetzee and Wilkerson)(Coetzee et al. 2013) and in other species of the An. gambiae complex, particularly An. arabiensis, is worth investigating.

Dormancy and Egg Storage. When stored at a cool temperature (15.5°), dormant eggs remained viable for a longer period of time than when stored undisturbed at room temperature (25.5°). Spontaneous hatching remained uncommon, regardless of storage temperature or duration. These results suggest the possibility of a practical method for retaining unhatched eggs for extended periods, for its convenience in laboratory research or for the accumulation of unhatched eggs prior to a mass rearing. Yet, under the conditions we used, viability declined at both temperatures by the second and third weeks of storage, even at the lower temperature. None was capable of stimulated hatching beyond 1 month of dormancy, at least at the warmer temperature. Similar viability results, without intentional agitation, were reported by Deane and Causey (1943), Beier et al.

(1990), and Shililu et al. (2004). Storage under other conditions, including lower temperatures, followed by agitations, needs to be explored.

The statistical shape of installment hatching was affected by storage time (egg age), storage temperature, and agitation day (Fig. 6.2). Among eggs stored warm, maximum hatch occurred at the youngest age, having been stored the least amount of

169 time (i.e., 2 d post-oviposition), and after the first agitation. Among those stored cool, maximum hatch occurred in the youngest group, but only after the second agitation, i.e., fewer cool-stored eggs were immediately capable of agitation-induced hatching when returned to the warmer temperature, almost certainly a result of chilling while they were still embryonic. The fact that eggs did become competent to respond to agitation after a further period of cold storage demonstrates that embryonic development continued at

15.5°, but more slowly. The minimum temperature at which the eggs of An. gambiae can hatch is reported to be 4°C at 5 days post-oviposition (Beier et al. 1990). Bayoh and

Lindsay (2003) reported 18°C as the minimum temperature threshold for An. gambiae development to adulthood. Another notable result of our storage test was that among the eggs stored for longer periods of time (1-3 wk), the installment hatchings of cool-stored eggs tended to be more spread out over 2-3 daily agitations. This difference from eggs stored at room temperature suggests lingering effects of the cold treatment, retarding readiness to hatch.

Eggs at both temperatures reached the same Total Hatch by Agitation (ca. 66%) at around day 10 (Fig. 6.3). After that day, the slope downward at 25.5 °C was almost twice as steep as at 15.5 °C. The model showed that a 50% hatch was obtained around day 12 at

25.5 °C, similar to that obtained by Deane and Causey (1943). When we applied 15.5 and

25.5 °C to the Impoinvil et al. (2007) model, with target hatches of 10 and 50%, shorter periods of incubation resulted, suggesting a shorter survival than we obtained, probably due to a difference in temperature and our use of agitation. Our Total Hatch by Agitation

170 of embryos <1 wk old at 25.5 °C supports the results of Shililu et al. (2004), indicating that there is a time window of 1 wk for embryos to hatch in maximum numbers in response to the stimulus, and after 1 wk the embryos start to die. At 15.5 °C, however, the slower but steady decline in viability reflected the low metabolism and slower embryonic development at this temperature (Valencia et al. 1996).

Ecological Significance. Many studies (e.g., Depinay et al. 2004, Parham et al.

2012) have focused on predicting mosquito populations based on environmental variables. The models of this study take into account agitation and dormancy at two temperatures to estimate future populations of the first-instar larvae. The adaptive value of agitation-induced hatching appears similar to that of inundation for aedine mosquitoes.

In anophelines, the impact of rain drops on the water surface signals an influx of water and nutrients to rain pools, ditches, and borrow pits, which are commonly used by An. gambiae. When these pools provide nutrients to a finite number of healthy larvae and also are at risk of drying out, a pool that is increasing in size offers a larva better prospects for survival. The installment-hatching feature, as in aedine eggs, appears to be a form of bet-hedging strategy against the possibility that the rain is insufficient to sustain the pool or enhance its nutrient base. Other inferences, including the effect of delayed hatching on avoidance of predators or eventual adult body size, also have been invoked

(e.g., Lyimo et al. 1992, Minakawa et al. 2001, Gimnig et al. 2002, Koenraadt et al.

2004).

171 If a pool should become entirely dry, unhatched eggs of An. gambiae appear to be able to survive in drying mud longer than small larvae (Deane and Causey 1943, Beier et al. 1990, Koenraadt et al. 2003, Shililu et al. 2004), though they cannot resist the long periods of drought that aedine eggs can (e.g., Gjullin et al. 1950). Because evidence exists that mechanical agitation also promotes egg hatching in aedines (Dupree and

Morgan 1902, Young 1922, Borg and Horsfall 1953, Roberts 2001, G.F. O’Meara pers. comm.), for those species rain probably likewise contributes a mechanical signal indicating an influx of water and nutrients in container and floodwater habitats. Our observation that the eggs of An. gambiae failed to hatch after several months and up to 1 yr of storage supports the speculation of Koenraadt et al. (2003) that they cannot survive several months of a dry season.

Other An. gambiae Studies. Previous workers have reported total hatching rates of An. gambiae that ranged from 73% to 93% under various conditions of temperature and water quality (Yaro et al. 2006, Impoinvil et al. 2007), in the same range as our average of ca. 83% after agitation. This suggests the possibility that eggs typically are agitated inadvertently during their handling after oviposition, as was the case in our laboratory. Alternatively, Munga et al. (2005) and Yaro et al. (2006) suggested that chemicals in the water may affect hatching. Despite the importance of agitation, which has gone largely unnoticed, delayed hatching, i.e., installment hatching, is well known in

An. gambiae (see above). Its genetic basis has been explored by Kaiser et al. (2010), who

172 did notice that its frequency is changed by egg agitation. The other studies appear to describe spontaneous anopheline installment hatching, but the methods presented do not preclude the possibility that they were agitated after reaching full embryonic development, either by being poured from oviposition cups into large larval pans or by being sprayed with a jet of water to knock off eggs lodged on the walls of cups and pans.

Laboratory Use of Agitation-Induced Hatching. Planning and preparations for both small-scale experiments and mass rearings of An. gambiae are more efficient and convenient when eggs of several ages can be accumulated and hatched at a designated time. Furthermore, cage densities of adults can be kept constant, and fewer blood-meals are required. In our insectary, we collect eggs 10-20 h post-oviposition (next morning) from colony cages and divide them into two to five small containers, based on the number of first-instar larvae needed on subsequent days. The water level is raised so that no eggs are stranded on the side. Food is not added to water until the first day of agitation. The containers are covered and sealed to minimize evaporation and then stored in an undisturbed location. The egg containers are shaken 2-15 d post-oviposition, to obtain a predictable number of first instars on demand. Hatching success drops substantially between the first and second weeks of storage, particularly at higher temperatures, so age and temperature must be taken into account when estimating numbers of larvae derived from eggs older than 8 d. The very few larvae that hatch spontaneously are negligible and die within 2 d in the absence of food.

173 Acknowledgments

The authors extend thanks to Christopher M. Stone, Adrea C. Rodriguez-Lovejoy, and

David L. Denlinger, Department of Entomology, The Ohio State University (OSU), for their suggestions on experimental design; to Leila Farivar, Department of Economics,

OSU, for statistical consulting; and to Robert L. Aldridge for maintaining mosquito colonies and providing experimental eggs in our vector behavior laboratory. This work was supported, in part, by NIH grant R01-AI077722 from the National Institute of

Allergy & Infectious Diseases (NIAID) to W.A.F. Its content is solely the responsibility of the authors and does not represent the official views of the NIAID or the National

Institutes of Health.

174 Table 6.1. Details of the effect of seven consecutive daily agitations on mean proportion (%) hatching of Anopheles gambiae eggs

(1egg / 1.5 mL in well). Three data sets are compared: 1) daily spontaneous hatch rates of control (non-agitated) groups on

successive days at the same times as the treatment (5 min/d agitation) group counts, and 2) daily stimulated hatch rates of

treatment (agitated) groups a) immediately after (<10 min) each agitation, and b) during the 24-h period before the next one.

Graphic summary of data with daily times combined (i) is presented in Fig. 6.1. (n = 15)

i a Agitation Hatch Count Group Mean ± SE Mann-Whitney U P-value c

1

7

5 Day b Before/After

Agitation

0 1 After Control 0 112.5 1.00

Treatment 0

(Continued) Table 6.1: Continued

2 Before Control 0.6 ± 0.4 105 0.775

Treatment 0.3 ± 0.3

1 2 After Control 0 0 <0.0001

Treatment 63.3 ± 4.4

1

7 3 Before Control 0.3 ± 0.3 50.5 0.009

6 Treatment 3.9 ± 1.0

2 3 After Control 0 45 0.004

Treatment 5.8 ± 1.8

4 Before Control 0 60 0.029

(Continued) Table 6.1: Continued

Treatment 2.5 ± 0.8

3 4 After Control 0 82.5 0.217

Treatment 1.4 ± 0.7

5 Before Control 0.3 ± 0.3 97.5 0.539

1

7

7 Treatment 0.8 ± 0.5

4 5 After Control 0 90 0.367

Treatment 0.8 ± 0.5

6 Before Control 0 75 0.126

Treatment 1.7 ± 0.7

(Continued) Table 6.1: Continued

5 6 After Control 0 97 0.539

Treatment 1.4 ± 1.1

7 Before Control 0 105 0.775

Treatment 0.3 ± 0.3

1

7

8 6 8 After Control 0 90 0.367

Treatment 0.8 ± 0.5

a i = time period after each agitation, and before next one, that was used to define DHi .

b In this experiment (Exp. 1), agitation day = day post-oviposition.

c P : probability of no difference between control (non-agitated) and treatment (agitated) groups, by Mann-Whitney U test, before

and after each agitation time. Table 6.2. Effect of grouped eggs on proportion of Anopheles gambiae eggs that hatched (% mean ± SE). Comparison of hatching

rates of solitary eggs (1egg / 1.5 mL in well)a and grouped eggs (10 eggs / 5mL in test tube)a before the first agitation of pharate

first-instar larvae (DH0 , 2 d post-oviposition) and Total Hatch by Agitation after 1 wk consecutive daily agitations (THA).

Temperature 25.5°C.

Hatch Count Isolated Eggs Groups of 10 Eggs t-test df P-value b

1

7 (n = 15) (n = 5)

9

Before the 1st agitation of 0.3 ± 0.3 0 0.567 18 0.578

pharate first-instar larvae

After 1 wk of daily agitation 82.8 ± 1.7 84.0 ± 5.1 0.301 18 0.767

(Continued) Table 6.2: Continued

a Solitary data drawn from Exp. 1, grouped data from Exp. 3, so container sizes differ.

b P : probability of no difference between isolated and grouped eggs, by Student’s t test, before and after each agitation time.

1

8

0 Table 6.3. Details of effect of storage duration and temperature on proportion of Anopheles gambiae eggs hatching (10 eggs / 5mL

in test tube). Comparison of mean hatch rates (± SE, n = 5 a) after storage for five durations (2-29 d) at 15.5 and 25.5°C, before the

first agitation (DH0) and again <10 min after the first agitation of pharate first-instar larvae, and of Total Hatch (TH) after 6

consecutive days of agitation. Post-storage (6-d agitation period) temperature 25.5°C.

Storage Duration (day)

1

8 a b

1 Temperature Hatch Count 2 8 15 22 29 F df P-value

15.5 °C DH0 0 0 5.6 ± 3.7 8.0 ± 3.7 4.0 ± 2.5 1.66 4, 19 0.201

<10 min After 0 20 ± 6.3 41.5 ± 8.8 7.6 ± 4.7 0 10.06 4, 19 <0.0001

TH 77.5±9.5 72.6±1.9 60.9 ± 12.1 43.3 ± 6.9 16.0 ± 6.8

(Continued) Table 6.3: Continued

25.5 °C DH0 0 2.0 ± 2.0 5.8 ± 2.4 8.0 ± 4.9 2.0 ± 2.0 1.405 4, 20 0.268

<10 min After 82 ± 5.8 59.5 ± 2.5 23.6 ± 5.2 0 0 99.96 4, 20 <0.0001

TH 84.0 ± 5.1 88.4 ± 5.7 35.3 ± 2.3 8.0 ± 4.9 2.0 ± 2.0

1

8

2

a n = 4, for 1 d-old eggs at 15.5 °C.

b P : probability of no difference among storage durations, by one-way ANOVA, before and immediately after the first agitation at

each temperature. Table 6.4. Regression models describing relationship between Daily Hatch (DHi) of Anopheles gambiae eggs (10 eggs / 5mL in

test tube) after 5 min agitation per day (0 ≤ i ≤ 6), storage duration (2 ≤ age ≤ 29), and temperature (temp = 15.5 or 25.5). Daily

Hatch (DHi) ranged between 0 and 1.

2 log(100 DHi + 0.1) = a log(i) + b (i)∙(age) + c (age) + d (age) + e log(age) + f (temp) + g

log (i) (i)∙(age) (age) (age) 2 log (age) (temp)

1

8 a 2

3 Model r Adjusted R F*** df a b c d e f g

2 0.702 48.5 68.33 2, 141 -2.742 0 0 -0.001 0 - 1.198

(β) (-0.674) (0) (0) (-0.194) (0) -

3 0.806 63.7 49.03 4, 132 -4.074 0.022 -0.38 0.005 3.19 -

1.152

(β) (-1.154) (1.085) (-3.849) (1.517) (1.37) - (Continued) Table 6.4: Continued

4 0.724 51.5 60.64 5, 276 -3.679 0.013 -0.101 0 0.710 -0.028 1.93

( β ) ( -0.959 ) ( 0.611 ) ( -0.943 ) ( 0 ) ( 0.279 ) ( -0.136 )

a Model 2: 15.5 °C; Model 3: 25.5 °C; Model 4: both temperatures were included.

1

8

4 β: Standardized Partial Regression Coefficient.

*** : P-value <0.0001 for all three models. 80

70

60 ) % (

d 50 e h c t a

H 40 Non-agitated y l i

a Agitated D

30 n a e

M 20

10

0 0 1 2 3 4 5 6

Day (i)

Fig. 6.1. Effect of seven consecutive daily agitations on mean daily hatch (% DHi).

Comparison of DHi of non-agitated and agitated (5min/d) Anopheles gambiae eggs

(1egg / 1.5 mL in well) from i = 0 (1 d post-oviposition) until i = 6 (7 d post- oviposition). Temperature 25.5°C. Details presented in Table 6.1.

185 Fig 6.2. Effect of storage duration and temperature on mean daily hatch (% DHi) of

Anopheles gambiae eggs (10 eggs / 5mL in test tube) agitated 5 min/d on six consecutive days, from i = 0 (before the first agitation) until i = 6 (after the sixth agitation).

Comparisons after each of five different storage durations (2-29 d) at either 15.5° or

25.5°C. Post-storage (6-d agitation period) temperature 25.5°C. 186 100 ) %

( 90

n o

i 80 t a t i 70 g A

y 60 b

d

e 50 15.5 °C h c t 40 25.5 °C a H

l 30 a t o

T 20

n

a 10 e

M 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Storage Duration (day)

Fig 6.3. Effect of storage duration and temperature on mean (± SE) Total Hatch by

Agitation (% THA) of Anopheles gambiae eggs (10 eggs / 5mL in test tube) after six consecutive days of agitation (5 min/day) following removal from storage. Comparisons of THA after each of five different storage durations (2-29 d) at either 15.5° or 25.5°C.

Post-storage (6-d agitation period) temperature 25.5°C.

187 Chapter 7: A non-flickering and programmable dimmer for LED lights *

Abstract

A simple electrical system is described that can produce a flicker-free lighting system for rearing rooms, growth chambers, and light-related insect studies. It utilizes energy-efficient light-emitting diodes (LEDs) and a relatively inexpensive AC power source. Because the core of the system is based on a microcontroller, light intensities can be programmed to simulate light transitions in dawn and dusk and also to change photoperiod in long-term experiments to simulate seasonal changes in daylength. This system is inexpensive and can be installed easily.

Keywords: dimmer, LED lights, rearing room, growth chamber

* B. Ebrahimi, R. Farivar, and W. A. Foster. In preparation for Medical and Veterinary Entomology 188 Introduction

Artificial lights are used in most entomological laboratories, both for rearing and for studying insects. It is essential that these light sources resemble natural light as closely as possible. There are two types of artificial lights traditionally used in labs: incandescent lights and fluorescent lights.

To control the light intensity of the fluorescent and incandescent lights, the prominent technology is based on dimmers that operate under the principle of modifying the AC waveform that powers the light (Byers & Unkrich 1983). Because AC power has a frequency of 50 Hz, the controlled light exhibits flickers at the same frequency. AC- powered incandescent and light-emitting diode (LED) lights both flicker at 100-120 cycles per sec (cps), and florescent lights flicker at 50-60 cps.

To human eyes, the light with flicker frequencies higher than 45-53 per sec appears steady. This threshold frequency is called Flicker Fusion Frequency (FFF) (Miall

1978). FFF has been reported to be relatively high for insects, between 20 and 300 cps, depending on species (Shields 1989; Lehane 2005). As a result, the visible flickering of incandescent and florescent lights may influence behavior of insects in lab experiments, and in some cases it influences trap catches (Syms & Goodman 1987), e.g., mosquitoes

(Silver 2008). The minute pirate bug, Orius tristicolor, showed different activity levels and turning ratios under LED lights compared to the traditional lighting, even when the light intensity was controlled (Shields 1989). Among other studies on other animals,

189 Evans et al. (2006) found that the low-frequency flickers of fluorescent lights, commonly used in the housing of captive birds, affected their welfare and performance in experiments. Also, human subjects showed a decrease in accuracy of performance when exposed to flicker (Kuller & Laike 1998).

Traditionally, a dimmer is used to control the amount of light in lab experiments.

In these systems, the whole electric current that powers the incandescent or fluorescent lights is controlled by a triac-based dimmer (Rand et al. 2007; Fitt & Thornley 2013).

Figure 7.1 shows how a dimmer-based system cuts the standard sinusoidal AC current at appropriate time intervals to control the light. Triac-based dimmers make the design of a low-pass filter a hard task, because low-pass filter components must dissipate a lot of energy to eliminate the alternating values and use high-power components for the filter.

Also, the frequency of flickering in the traditional dimming systems remains at 50-60 Hz.

Instead of incandescent and fluorescent lights, the lighting system proposed is

LED-based. Morrow (2008) has described many benefits of LED lights for horticulture.

With the development of LED and transistor technology in recent decades, LED light sources are gaining popularity in laboratory applications. LED lighting has many advantages over the incandescent or fluorescent lighting. First, LED lights operate at lower energy consumption levels, which makes them cost-efficient, especially in the applications that require prolonged use (Bourget 2008). Second, they are environmentally friendly, considering their long working life-span and lower carbon footprint in their production (Jung et al. 2010). Third, because of the absence of hazardous materials in

190 their construction and their cool light emission, LED lights are safer to operate and offer minimal interference with the controlled lab temperature (Morrow 2008). However, despite their many favorable features, AC-powered LED lights do flicker.

Our system eliminates the potentially intrusive flickers in lighting by first increasing the flicker frequency to 500cps, which is above the FFF of most insects, and by subsequently eliminating flicker altogether. The system is designed for use in entomology laboratory experiments, temperature-sensitive incubators, and rearing rooms.

Thus it provides economically efficient artificial lighting that provides smooth crepuscular transitions between night and day and is programmable for different daylengths.

Materials and methods

In this section, the hardware platform and the design of the required additional circuitry are presented, followed by a software algorithm that controls the system. All components combined cost < $60.

Hardware design principles

Because of their much lower power consumption, powering and controlling LED lights through programmable digital circuitry is feasible. A microcontroller-based system was used to create a controlling signal, with a higher frequency than the standard 50-Hz

191 alternating current (AC) used in most available lighting devices. This microcontroller regulates the amount of light generated by LEDs. Its pulse-width-modulated (PWM) signal controls programmable and varying light intensities. The output signal of the microcontroller alternates at approximately 490Hz, which is then passed through a first- degree low-pass filter. The low-pass filter eliminates all high- frequency flickering of the signal and ensures that the final light created is flicker-free. Finally, the signal is amplified by a transistor-based amplifier and sent to one or more LED lights. As shown in the next section, the final light level is completely controllable.

Hardware platform

This system is based on a microcontroller reference system, the Arduino system

(Duemilanove Board, ATmega328P). Arduino is an inexpensive open-source single-board microcontroller design that includes an Atmel AVR microcontroller as its main computing platform. It has a few supporting electrical components, including power- supply voltage regulation, crystal-oscillation-clock circuitry, and a boot loader that simplifies uploading of programs to the on-chip flash memory through a USB interface.

The PWM signal is created by the microcontroller (software described below). The ratio of the time when the signal has a logic-high-value (5-volt signal) to the time when it has a logic-low-value (zero-volt signal) determines the intensity of the light emitted. For example, if during each period of the signal (which takes approximately 2 milliseconds) the signal is high for 10% (0.2 milliseconds) and low for 90% (1.8 milliseconds), the

192 LEDs are lighted with 10% of their maximum luminosity. Figure 7.3 depicts a simplified view of the additional circuitry required.

The values of R (Resistor) and C (Capacitor) were selected to eliminate the 490

Hz high-frequency component of the signal. The transition diagram of the RC filter is shown in Figure 7.4. In designing the filter, we reduced the high-frequency flicker value to less than 1% of the final signal. As such, and based on the design parameters of a first- order RC circuit, we set the breakpoint frequency of the filter to a value of 3Hz, so that the first-order filter, whose attenuation rate is 6 decibel per decade in frequency, can attenuate the 490HZ signal to about 0.95%, which is practically a flicker-free signal. The value of the resistor, about 220 ohms, is controlled by the characteristics of the amplifier transistor and the LED load. The value of the capacitor is the control parameter for the filter. In this system, a 220 micro-farad electrolytic capacitor achieved the required filter characteristics. This filter has a peak-to-peak ripple voltage of less than 1% and has a 0-

90% settling time of approximately 0.125 seconds [http://sim.okawa- denshi.jp/en/PWMtool.php]. The system is designed to handle time-based light-value controls of 1-sec granularity, which is an order of magnitude larger than the settling time.

As such, the filter settling time does not interfere with the time-based control logic of the system.

The amplifier stage uses a generic medium-power transistor in TO-220 packaging.

We used a common TIP-31 NPN transistor with a typical current gain of 25 to 50, but other medium-powered parts could be substituted, depending on the watt usage of LEDs.

193 Finally, we used a commercially available flexible LED strip-light (300 SMD White LED

Ribbon, 2026wh), provided by Ledwholesalers. Each three LED lights were grouped together in a parallel configuration, with one resistor in serial configuration (Figure 7.5).

The forward voltage of each individual LED in these strips is 3.3 volts. Given that the saturation collector-emitter voltage of the amplifier transistor is 1.2 volts, we needed at least a 11.2-volt power source for the circuit. We used a typical 12-volt laptop power supply, which is very common and relatively inexpensive. It is strong enough to provide power to multiple strips of LEDs.

Each strip takes approximately 1 ampere of current, at 12 volts, to give full light.

Given our choice of a single-stage transistor-amplifier and its current gain (between 25 and 50), the microcontroller must supply a maximum current of approximately 40 milli- amperes. This is equal to the rated value of the microcontroller used in the Arduino board.

Therefore this circuit can at most power one strip of 300 LEDs. To power more LED strips, we can employ either a better medium-power transistor with a higher current gain or a darlington transistor configuration that uses two transistors. However, the current set- up was powerful enough for our laboratory tests, and therefore we did not implement these possible extensions to the circuit.

Software

The main algorithm was written as a continuous loop that runs in the microcontroller and a set of time-based conditions that are checked against the real-time

194 clock in this loop. Each time-based condition includes a starting and finishing time and a luminosity value (between 0 and 100) for the start time and the end time. The condition states whether during this time period the light value should ramp up or down, or whether the light should remain constant (see lines 38-44 of Table 7.1). As such, to control the system the end user simply writes the proper set of conditions for each experiment. For example, the following conditions represent an experiment where we want to simulate darkness (i.e. night) from 7 PM to 7AM, simulate sunrise during a 45 min period from 7

AM to 7:45 AM, simulate daytime until 6 PM, and finally simulate a 1 hour sun-set from

6 PM to 7 PM.

When the system is connected to a computer, it can be programmed with a 1- second granularity. Much more complex LED lighting can be programmed easily with this system.

Running and testing the system

Arduino software can be downloaded free from the internet (www.arduino.cc) and installed on main operating systems. First, the DateTime library should be added to the software to make the remaining code-writing simple and short. The DateTime library can be downloaded from http://www.arduino.cc/en/Reference/Libraries. The codes for running the system can be copied and pasted on the blank page of the software (see Table

7.1 for code). First, the microcontroller is connected to the computer via a USB A-Male to B-Male cable. When the Upload button of the software is clicked, all codes are

195 transferred to the microcontroller, after which the cable between computer and the microcontroller can be disconnected. It should be noted that the microcontroller clock must be reset to the time set in the code every time the system is connected to a computer or a power source.

We measured light intensity of 6 LEDs during ramp up (i.e., increasing current and light intensity) by putting the lights on a 2-inch (inner diameter: 4-cm) by 15-cm long

PVC tube. The sensor of a lux meter (Mastech Light Meter LX1010BS) was put on the others side of the PVC tube to measure the light intensities as the luminosity values

(PWM active time: 0 -100) increased. When in operation and connected to a computer, the 'Serial Monitor' of the software showed the luminosity values (i.e., the PWM value of the microcontroller). The lux value for each PWM was read, and a curve between these two variables was plotted.

Results and Discussion

The system has been working flawlessly in our mosquito rearing room at The

Ohio State University for more than 1 year. The flexibility of the LED light strips allowed us to install them on the underside of each shelf to illuminate the rearing pans and cages below. Whereas a fluorescent light tube or a 40-watt incandescent light bulb provide, respectively, 2000 or 1500 lux at a distance of 15 cm, 6 LED lights produced approximately 1500 lux.

196 Figure 7.6 shows the lux of the 6 LED lights that were measured for each luminosity value (i.e., PWM value). Although the correlation between current and PWM is linear, our results showed that the relation between PWM and lux is not.

To linearize the light emission value based on the input, we approximated the circuit behavior with a logarithmic function. As seen in Figure 7.6, this approximation closely matches the real circuit behavior from roughly 55/255 onwards.

The next step was to counter the logarithmic behavior of the circuit with an exponential function in the software. After the software calculated a light ratio based on the time-based conditions provided by the user, this value was transformed by an exponential function and then was used to activate the circuits. The correlation between lux and the new PWM, produced by the system, was measured again. The final results are shown in Figure 7.7, which depicts a better approximation of a linear relationship between the input value that the user provides and the real light value emitted by the system.

This final adjustment shows that the system is so flexible that almost any pattern can be applied for ramp up (increasing light intensity) or ramp down (decreasing light intensity), as well as intensity and duration during days and nights. As an example, one can fit two or three “days” into a 24-hr period and test the impact of these shorter cycles on test subjects. Because the microcontroller that was used in this study has 6 PWM ports, up to 6 sets of LED lights, with a low filter pass and amplifier for each, can be added to the system. Then the software can be modified so that the lights can be

197 controlled independently of each other. Also, a simple LCD module and pertinent codes can be added to display time, date, and the output (PWM). If the goal of a study is to produce flickers at specific frequencies, the capacitor of the low pass filter can be substituted with a variable capacitor (Figure 7.2). In this case, the user may need an oscilloscope to measure the flicker frequency.

The narrow wavelengths and narrow range of brightness (i.e., light intensity) of

LED lights give researchers more flexibility when designing their experiments. There are various commercially available LED light strips, with different features. Some are covered with silicon so that they can be submerged in water with minimal effect on temperature. As more companies invest in LED lights, we expect cheaper, more efficient, and more advanced types to appear on the market.

Acknowledgements

The authors thank Leila Farivar, Department of Economics, The Ohio State University

(OSU) for drawing the figures. We also thank Mohammad Shakiba, Department of

Electrical and Computer Engineering, OSU, for his assistance in testing the system with an oscilloscope.

198 Table 7.1: Codes to control LED lights. Parts of the codes that can be changed are boldfaced.

1 #include 2 3 int transition; 4 int brightness = 0; // determines how bright the LED should be; range: 0-255 5 int startHours, startMinutes, startSeconds; 6 int endHours, endMinutes, endSeconds; 7 int startValue, endValue; 8 9 void setup() { 10 setTime(13,10,5,8,7,2013); // (hr,min,sec,day,month,yr); 11 12 Serial.begin(9600); 13 pinMode(13, OUTPUT); // Pin 13 has an LED connected on most Arduino 14 boards 15 pinMode(11, OUTPUT); // Pin 11 is our analog output; The resistor of the low 16 pass filter is connected to this pin. 17 } 18 19 int condition (int startH, int startM, int startS, int endH, int endM, int endS, int 20 state, int startV, int endV) { 21 int chm = hour() * 3600 + minute() * 60 + second(); //chm = current hour and 22 minute 23 int startT = startH * 3600 + startM * 60 + startS; 24 int endT = endH * 3600 + endM * 60 + endS;

(Continued)

199 Table 7.1: Continued

24 if( chm >= startT && chm < endT ) { 25 transition = state; 26 if (state ==true) { 27 startHours = startH; startMinutes = startM; startSeconds = startS; 28 endHours = endH; endMinutes = endM; endSeconds = endS; 29 startValue = startV; 30 endValue = endV; 31 } else 32 startValue = startV; 33 } 34 } 35 36 void timeLogic() { 37 38 condition (0,0,0,7,0,0, false,0,0); // startHours, startMinutes, startSeconds, 39 endHours, endMinutes, endM; endSeconds, transitionState, startValue, 40 endValue (ignored if false) 41 condition (7,0,0,7,30,0, true,0,255); 42 condition (7,30,0,19,0,0, false,255,255); 43 condition (19,0,0,19,30,0, true, 255, 0); 44 condition (19,30,0,23,59,59, false, 0,0); 45 } 46 47 void setLight() { 48 if (transition == true) {

(Continued)

200 Table 7:1: Continued

49 int ctime = (hour() * 3600) + (minute() * 60) + second(); 50 int startT = (startHours * 3600) + (startMinutes * 60) + startSeconds ; 51 int endT = (endHours * 3600) + (endMinutes * 60) + endSeconds; 52 brightness = (float)(endValue - startValue) * ( (float)(ctime - startT) / (float) 53 (endT - startT) ) + startValue; 54 } else 55 brightness = startValue; 56 57 //up to this point, the brightness is a linear value between 0 and 58 100 now we de-linearize it to map to our log-based transistor circuit . The 59 formula to set the value is 0.01*ln(255/25) 60 float flBrightness = 25*pow(2.718281828,0.023223877*(float)brightness); 61 analogWrite(11, (int)flBrightness); // analogWrite(11, brightness); 62 It is the alternative code if above line is not needed 63 } 64 65 void loop() { 66 timeLogic(); 67 setLight(); 68 digitalWrite(13, transition); // set the LED on 69 Serial.println(brightness, DEC); 70 Serial.println(minute() * 60 + second(), DEC); 71 72 delay(500); // wait for a second 73 digitalWrite(13, LOW); // sets the on-board LED off 74 delay(500); // wait for a second 75 } 201

Figure 7.6. Electrical relationship between lux and the PWM duty cycle, showing non- linearity.

Figure 7.7. Correlation between lux and requested PWM value after adding exponential function to the codes.

204 Bibliography

Chapter 1:

Abdel-Malek, A. A. 1964. Study of the feeding habits of male Anopheles sergentii Theo.

at Siwa Oasis using radio-phosphorus. Bull. World Health Organ. 30: 137.

Abdel-Malek, A. A., and W. F. Baldwin. 1961. Specificity of plant feeding in

mosquitoes as determined by radioactive phosphorus. Nature. 192: 178–179.

Beier, J. C. 1996. Frequent Blood-Feeding and Restrictive Sugar-Feeding Behavior

Enhance the Malaria Vector Potential of Anopheles gambiae s.l. and An. funestus

(Diptera: Culicidae) in Western Kenya. J. Med. Entomol. 33: 613–618.

Beier, J. C., G. C. Müller, W. Gu, K. L. Arheart, and Y. Schlein. 2012. Attractive toxic

sugar bait (ATSB) methods decimate populations of Anopheles malaria vectors in

arid environments regardless of the local availability of favoured sugar-source

blossoms. Malar. J. 11: 1–7.

Bowen, M. F., and J. Romo. 1995. Host-seeking and sugar-feeding in the autogenous

mosquito Aedes bahamensis (Diptera: Culicidae). J. Vector Ecol. 20: 195–202.

Braks, M. A. H., S. A. Juliano, and L. P. Lounibos. 2006. Superior reproductive

success on human blood without sugar is not limited to highly anthropophilic

mosquito species. Med. Vet. Entomol. 20: 53–59.

205 Burkett, D. A., D. L. Kline, and D. A. Carlson. 1999. Sugar meal composition of five

north central Florida mosquito species (Diptera: Culicidae) as determined by gas

chromatography. J. Med. Entomol. 36: 462–467.

Cardé, R. T. 1996. Odour plumes and odour-mediated flight in insects, pp. 54-70. In: G.

R. Bock and G. Cardew (eds.), Olfaction in Mosquito-Host Interactions. John

Wiley & Sons Ltd., Chichester, UK.

Chilaka, N., E. Perkins, and F. Tripet. 2012. Visual and olfactory associative learning

in the malaria vector Anopheles gambiae sensu stricto. Malar. J. 11: 27.

Dao, A., A. Adamou, A. S. Yaro, H. M. Maïga, Y. Kassogue, S. F. Traoré, and T.

Lehmann. 2008. Assessment of alternative mating strategies in Anopheles

gambiae: Does mating occur indoors? J. Med. Entomol. 45: 643–652.

DeGennaro, M., C. S. McBride, L. Seeholzer, T. Nakagawa, E. J. Dennis, C.

Goldman, N. Jasinskiene, A. A. James, and L. B. Vosshall. 2013. orco mutant

mosquitoes lose strong preference for humans and are not repelled by volatile

DEET. Nature. 498: 487–491.

Dötterl, S., K. Jahreiß, U. S. Jhumur, and A. Jürgens. 2012. Temporal variation of

flower scent in Silene otites (Caryophyllaceae): a species with a mixed pollination

system. Bot. J. Linn. Soc. 169: 447–460.

Dudareva, N. A., and E. Pichersky. 2006. Biology of floral scent. CRC Press, Boca

Raton, FL.

206 Dye, C. 1992. The analysis of parasite transmission by bloodsucking insects. Annu. Rev.

Entomol. 37: 1–19.

Eischen, F. A., and W. A. Foster. 1983. Life span and fecundity of adult female Aedes

aegypti (Diptera: Culicidae) fed aqueous extracts of pollen. Ann. Entomol. Soc.

Am. 76: 661–663.

Feinsod, F. M., and A. Spielman. 1980. Nutrient-mediated juvenile hormone secretion

in mosquitoes. J. Insect Physiol. 26: 113–117.

Fernandes, L., and H. Briegel. 2005. Reproductive physiology of Anopheles gambiae

and Anopheles atroparvus. J. Vector Ecol. 30: 11.

Foster, W. A. 1995. Mosquito Sugar Feeding and Reproductive Energetics. Annu. Rev.

Entomol. 40: 443–474.

Foster, W. A. 2008. Phytochemicals as population sampling lures. J. Am. Mosq. Control

Assoc. 24: 138–146.

Foster, W. A., and R. G. Hancock. 1994. Nectar-related olfactory and visual attractants

for mosquitoes. J. Am. Mosq. Control Assoc. 10: 288.

Foster, W. A., and W. Takken. 2004. Nectar-Related vs. human-helated volatiles:

behavioural response and choice by female and male Anopheles gambiae

(Diptera: Culicidae) between emergence and first feeding. Bull. Entomol. Res. 94:

145–157.

Garrett-Jones, C. 1964. Prognosis for interruption of malaria transmission through

assessment of the mosquito’s vectorial capacity. Nature. 204: 1173–1175.

207 Gary, R. E., J. W. Cannon III, and W. A. Foster. 2009. Effect of sugar on male

Anopheles gambiae mating performance, as modified by temperature, space, and

body size. Parasit. Vectors. 2: 19.

Gary, R. E., and W. A. Foster. 2001. Effects of available sugar on the reproductive

fitness and vectorial capacity of the malaria vector Anopheles gambiae (Diptera:

Culicidae). J. Med. Entomol. 38: 22–28.

Gary, R. E., and W. A. Foster. 2004. Anopheles gambiae feeding and survival on

honeydew and extra‐floral nectar of peridomestic plants. Med. Vet. Entomol. 18:

102–107.

Gouagna, L. C., R. S. Poueme, K. R. Dabiré, J. B. Ouédraogo, D. Fontenille, and F.

Simard. 2010. Patterns of sugar feeding and host plant preferences in adult males

of An. gambiae (Diptera: Culicidae). J. Vector Ecol. 35: 267–276.

Gu, W., G. Müller, Y. Schlein, R. J. Novak, and J. C. Beier. 2011. Natural plant sugar

sources of Anopheles Mosquitoes strongly impact malaria transmission potential.

PLoS ONE. 6: e15996.

Guerenstein, P. G., E. A.Yepez, J. van Haren, D. G. Williams, and J. G. Hildebrand.

2004. Floral CO2 emission may indicate food abundance to nectar-feeding moths.

Naturwissenschaften. 91: 329–333.

Hancock, R. G., and W. A. Foster. 1997. Larval and adult nutrition effects on

blood/nectar choice of Culex nigripalpus mosquitoes. Med. Vet. Entomol. 11:

112–122.

208 Healy, T. P., and P. C. Jepson. 1988. The Location of floral nectar sources by

mosquitoes: The long-range responses of Anopheles arabiensis Patton (Diptera:

Culicidae) to Achillea Millefolium flowers and isolated floral odour. Bull.

Entomol. Res. 78: 651–657.

Impoinvil, D. E., J. O. Kongere, W. A. Foster, B. N. Njiru, G. F. Killeen, J. I.

Githure, J. C. Beier, A. Hassanali, and B. G. J. Knols. 2004. Feeding and

survival of the malaria vector Anopheles gambiae on plants growing in Kenya.

Med. Vet. Entomol. 18: 108–115.

Jhumur, U. S., S. Dötterl, and A. Jürgens. 2006. Naïve and Conditioned Responses of

Culex pipiens pipiens Biotype molestus (Diptera: Culicidae) to Flower Odors. J.

Med. Entomol. 43: 1164–1170.

Jhumur, U. S., S. Dötterl, and A. Jürgens. 2008. Floral Odors of Silene otites: Their

Variability and Attractiveness to Mosquitoes. J. Chem. Ecol. 34: 14–25.

Joseph, S. R. 1970. Fruit feeding of mosquitoes in nature. Proc. New Jersey Mosq.

Exterm. Assoc. 57: 125–131.

Jürgens, A., T. Witt, and G. Gottsberger. 2002. Flower scent composition in night-

flowering Silene species (Caryophyllaceae). Biochem. Syst. Ecol. 30: 383–397.

Kaufmann, C., and H. Briegel. 2004. Flight performance of the malaria vectors

Anopheles gambiae and Anopheles atroparvus. J. Vector Ecol. 29: 140–153.

209 Knudsen, J. T., and L. Tollsten. 1993. Trends in floral scent chemistry in pollination

syndromes: floral scent composition in moth-pollinated taxa. Bot. J. Linn. Soc.

113: 263–284.

Lindh, J. M., O. Terenius, K. Eriksson-Gonzales, B. G. Knols, and I. Faye. 2006. Re-

introducing bacteria in mosquitoes—a method for determination of mosquito

feeding preferences based on coloured sugar solutions. Acta Trop. 99: 173–183.

Lyimo, E. O., and W. Takken. 1993. Effects of adult body size on fecundity and the

pre‐gravid rate of Anopheles gambiae females in Tanzania. Med. Vet. Entomol.

7: 328–332.

Ma, B. O., and B. D. Roitberg. 2008. The role of resource availability and state-

dependence in the foraging strategy of blood-feeding mosquitoes. Evol. Ecol. Res.

10: 1111–1130.

Manda, H., L. C. Gouagna, W. A. Foster, R. R. Jackson, J. C. Beier, J. I. Githure,

and A. Hassanali. 2007a. Effect of discriminative plant-sugar feeding on the

survival and fecundity of Anopheles gambiae. Malar. J. 6: 113.

Manda, H., L. C. Gouagna, E. Nyandat, E. W. Kabiru, R. R. Jackson, W. A. Foster,

J. I. Githure, J. C. Beier, and A. Hassanali. 2007b. Discriminative feeding

behaviour of Anopheles gambiae s.s. on endemic plants in western Kenya. Med.

Vet. Entomol. 21: 103–111.

McCall, P. J., and G. Eaton. 2001. Olfactory memory in the mosquito Culex

quinquefasciatus. Med. Vet. Entomol. 15: 197–203.

210 McCall, P. J., and D. W. Kelly. 2002. Learning and memory in disease vectors. Trends

Parasitol. 18: 429–33.

De Meillon, B., A. Sebastian, and Z. H. Khan. 1967. Cane-sugar feeding in Culex

pipiens fatigans. Bull. World Health Organ. 36: 53.

Menda, G., J. H. Uhr, R. A. Wyttenbach, F. M. Vermeylen, D. M. Smith, L. C.

Harrington, and R. R. Hoy. 2013. Associative learning in the dengue vector

mosquito, Aedes aegypti: avoidance of a previously attractive odor or surface

color that is paired with an aversive stimulus. J. Exp. Biol. 216: 218–223.

Metcalf, R. L., and M. Kogan. 1987. Plant volatiles as insect attractants. Crit. Rev. Plant

Sci. 5: 251–301.

Müller, G., J. Beier, S. Traore, M. Toure, M. Traore, S. Bah, S. Doumbia, and Y.

Schlein. 2010a. Successful field trial of attractive toxic sugar bait (ATSB) plant-

spraying methods against malaria vectors in the Anopheles gambiae complex in

Mali, West Africa. Malar. J. 9: 210.

Müller, G. C., A. Junnila, W. Qualls, E. E. Revay, D. L. Kline, S. Allan, Y. Schlein,

and R. D. Xue. 2010b. Control of Culex quinquefasciatus in a storm drain system

in Florida using attractive toxic sugar baits. Med. Vet. Entomol. 9999.

Müller, G. C., A. Junnila, and Y. Schlein. 2010c. Effective control of adult Culex

pipiens by spraying an attractive toxic sugar bait solution in the vegetation near

larval habitats. J. Med. Entomol. 47: 63–66.

211 Müller, G. C., and Y. Schlein. 2008. Efficacy of toxic sugar baits against adult cistern-

dwelling Anopheles claviger. Trans. R. Soc. Trop. Med. Hyg. 102: 480–484.

Müller, G., and Y. Schlein. 2006. Sugar questing mosquitoes in arid areas gather on

scarce blossoms that can be used for control. Int. J. Parasitol. 36: 1077–1080.

Nayar, J. K., and D. M. Sauerman. 1975. The effects of nutrition on survival and

fecundity in Florida mosquitoes Part 2. Utilization of a blood meal for survival. J.

Med. Entomol. 12: 99–103.

Nyasembe, V. O., P. E. A. Teal, W. R. Mukabana, J. H. Tumlinson, B. Torto, and K.

Nairobi. 2012. Behavioural response of the malaria vector Anopheles gambiae to

host plant volatiles and synthetic blends. Parasit. Vectors. 5: 234.

Otienoburu, P., B. Ebrahimi, P. Phelan, and W. Foster. 2012. Analysis and

optimization of a synthetic milkweed floral attractant for mosquitoes. J. Chem.

Ecol. 38: 873–881.

Patterson, R. S., B. J. Smittle, and R. T. Deneve. 1969. Feeding habits of male southern

house mosquitoes on 32P-labeled and unlabeled plants. J. Econ. Entomol. 62:

1455–1458.

Raguso, R. A. 2001. Floral scent, olfaction, and scent-driven foraging behavior. Cogn.

Ecol. Pollinat. Camb. Univ. Press Camb. 83–105.

Sachs, J., and P. Malaney. 2002. The economic and social burden of malaria. Nature.

415: 680–685.

212 Sanford, M. R., and J. K. Tomberlin. 2011. Conditioning individual mosquitoes to an

odor: sex, source, and time. PLoS ONE. 6: e24218.

Schlein, Y., and G. C. Müller. 2008. An approach to mosquito control: using the

dominant attraction of flowering Tamarix jordanis trees against Culex pipiens. J.

Med. Entomol. 45: 384–390.

Shaukat, A. M., J. G. Breman, and F. E. McKenzie. 2010. Research using the

entomological inoculation rate to assess the impact of vector control on malaria

parasite transmission and elimination.

Snow, R. W., C. A. Guerra, A. M. Noor, H. Y. Myint, and S. I. Hay. 2005. The global

distribution of clinical episodes of Plasmodium falciparum malaria. Nature. 434:

214–217.

Stanczyk, N. M., J. F. Y. Brookfield, L. M. Field, and J. G. Logan. 2013. Aedes

aegypti mosquitoes exhibit decreased repellency by DEET following previous

exposure. PLoS ONE. 8: e54438.

Stone, C. M., B. T. Jackson, and W. A. Foster. 2012. Effects of plant-community

composition on the vectorial capacity and fitness of the malaria mosquito

Anopheles gambiae. Am. J. Trop. Med. Hyg. 87: 727–736.

Stone, C. M., R. M. Taylor, B. D. Roitberg, and W. A. Foster. 2009. Sugar deprivation

reduces insemination of Anopheles gambiae (Diptera: Culicidae), despite daily

recruitment of adults, and predicts decline in model populations. J. Med. Entomol.

46: 1327–1337.

213 Straif, S. C., and J. C. Beier. 1996. Effects of sugar availability on the blood-feeding

behavior of Anopheles gambiae (Diptera: Culicidae). J. Med. Entomol. 33: 608–

612.

Thorsteinson, A. J., and R. A. Brust. 1962. The influence of flower scents on

aggregations of caged adult Aedes aegypti. DTIC Document.

Tomberlin, J. K., G. C. Rains, S. A. Allan, M. R. Sanford, and W. J. Lewis. 2006.

Associative learning of odor with food- or blood-meal by Culex quinquefasciatus

Say (Diptera: Culicidae). Naturwissenschaften. 93: 551–6.

Van Handel, E. 1965. The obese mosquito. J. Physiol. 181: 478–486.

Van Handel, E. 1972. The detection of nectar in mosquitoes. Mosq. News. 32: 458–458.

Vargo, A. M., and W. A. Foster. 1982. Responsiveness of female Aedes aegypti

(Diptera: Culicidae) to flower extracts. J. Med. Entomol. 19: 710–718.

Wada-Katsumata, A., J. Silverman, and C. Schal. 2013. Changes in Taste Neurons

Support the Emergence of an Adaptive Behavior in Cockroaches. Science. 340:

972–975.

Wittie, J. L. 2003. The effect of sugar and blood-host type on the survival, fecundity,

biting frequency, and persistence of the eastern tree-hole mosquito, Ochlerotatus

triseriatus (Diptera: Culicidae).

Yee, W. L., W. A. Foster, M. J. Howe, and P. . Hancock. 1992. Simultaneous field

comparison of evening temporal distributions of nectar and blood feeding by

214 Aedes vexans and Ae. trivittatus (Diptera: Culicidae) in Ohio. J. Med. Entomol.

29: 356–360.

Yuval, B., M. L. Holliday-Hanson, and R. K. Washing. 1994. Energy budget of

swarming male mosquitoes. Ecol. Entomol. 19: 74–78.

Chapter 2:

Ahmadi, A., and G. A. H. McClelland. 1985. Mosquito-mediated attraction of female

mosquitoes to a host. Physiol. Entomol. 10: 251–255.

Allan, S. A., U. R. Bernier, and D. L. Kline. 2006a. Laboratory Evaluation of Avian

Odors for Mosquito (Diptera: Culicidae) Attraction. J. Med. Entomol. 43: 225–

231.

Allan, S. A., U. R. Bernier, and D. L. Kline. 2006b. Attraction of mosquitoes to

volatiles associated with blood. J. Vector Ecol. 31: 71–78.

Bateson, M., S. Desire, S. E. Gartside, and G. A. Wright. 2011. Agitated Honeybees

Exhibit Pessimistic Cognitive Biases. Curr. Biol. 21: 1070–1073.

Bosch, O. J., M. Geier, and J. Boeckh. 2000. Contribution of Fatty Acids to Olfactory

Host Finding of Female Aedes aegypti. Chem. Senses. 25: 323–330.

Braks, M. A. H., and W. Takken. 1999. Incubated Human Sweat but not Fresh Sweat

Attracts the Malaria Mosquito Anopheles gambiae sensu stricto. J. Chem. Ecol.

25: 663–672.

215 Cabrera, M., and K. Jeffe. 2007. An aggregation pheromone modulates lekking

behavior in the vector mosquito Aedes aegypti (Diptera: Culicidae). J. Am. Mosq.

Control Assoc. 23: 1–10.

Carlson, D. A., N. Smith, H. K. Gouck, and D. R. Godwin. 1973. Yellowfever

Mosquitoes: Compounds Related to Lactic Acid that Attract Females. J. Econ.

Entomol. 66: 329–331.

DeGennaro, M., C. S. McBride, L. Seeholzer, T. Nakagawa, E. J. Dennis, C.

Goldman, N. Jasinskiene, A. A. James, and L. B. Vosshall. 2013. orco mutant

mosquitoes lose strong preference for humans and are not repelled by volatile

DEET. Nature. advance online publication.

Dekker, T., W. Takken, and R. T. Cardé. 2001. Structure of host-odour plumes

influences catch of Anopheles gambiae s.s. and Aedes aegypti in a dual-choice

olfactometer. Physiological Entomology. 26: 124–134.

Dogan, E. B., and P. A. Rossignol. 1999. An Olfactometer for Discriminating Between

Attraction, Inhibition, and Repellency in Mosquitoes (Diptera: Culicidae). J. Med.

Entomol. 36: 788–793.

Ebrahimi, B., S. Shakibi, and W. A. Foster. 2013. Delayed egg hatching of Anopheles

gambiae (Diptera: Culicidae), pending water agitation. J. Med. Entomol.

Submitted.

Foster, W. A., and W. Takken. 2004. Nectar-Related vs. human-helated volatiles:

behavioural response and choice by female and male Anopheles gambiae

216 (Diptera: Culicidae) between emergence and first feeding. Bull. Entomol. Res. 94:

145–157.

Geier, M., and J. Boeckh. 1999. A new Y-tube olfactometer for mosquitoes to measure

the attractiveness of host odours. Entomol. Exp. Appl. 92: 9–19.

Geier, M., O. J. Bosch, and J. Boeckh. 1999. Ammonia as an Attractive Component of

Host Odour for the Yellow Fever Mosquito, Aedes aegypti. Chem. Senses. 24:

647–653.

Hancock, R. G., and W. A. Foster. 1993. Effects of Preblood-Meal Sugar on Sugar

Seeking and Upwind Flight by Gravid and Parous Aedes aegypti (Diptera:

Culicidae). J. Med. Entomol. 30: 353–359.

Hancock, R. G., and W. A. Foster. 1997. Larval and adult nutrition effects on

blood/nectar choice of Culex nigripalpus mosquitoes. Med. Vet. Entomol. 11:

112–122.

Healy, T. P., and P. C. Jepson. 1988. The Location of floral nectar sources by

mosquitoes: The long-range responses of Anopheles arabiensis Patton (Diptera:

Culicidae) to Achillea Millefolium flowers and isolated floral odour. Bull.

Entomol. Res. 78: 651–657.

Hern, A., and S. Dorn. 2001. Statistical modelling of insect behavioural responses in

relation to the chemical composition of test extracts. Physiol. Entomol. 26: 381–

390.

217 Jennions, M. D., and A. P. Møller. 2003. A survey of the statistical power of research in

behavioral ecology and animal behavior. Behav. Ecol. 14: 438–445.

Jepson, P. C., and T. P. Healy. 1988. The Location of Floral Nectar Sources by

Mosquitoes: An Advanced Bioassay for Volatile Plant Odours and Initial Studies

with Aedes Aegypti (L.) (Diptera: Culicidae). Bull. Entomol. Res. 78: 641–650.

Jhumur, U. S., S. Dötterl, and A. Jürgens. 2006. Naïve and Conditioned Responses of

Culex pipiens pipiens Biotype molestus (Diptera: Culicidae) to Flower Odors. J.

Med. Entomol. 43: 1164–1170.

Jhumur, U. S., S. Dötterl, and A. Jürgens. 2007. Electrophysiological and behavioural

responses of mosquitoes to volatiles of Silene otites (Caryophyllaceae).

Arthropod-Plant Interactions. 1: 245–254.

Knols, B. G. J., R. D. Jong, and W. Takken. 1994. Trapping system for testing olfactory

responses of the malaria mosquito Anopheles gambiae in a wind tunnel. Med. Vet.

Entomol. 8: 386–388.

Mauer, D. J., and W. A. Rowley. 1999. Attraction of Culex pipiens pipiens (Diptera:

Culicidae) to flower volatiles. J. Med. Entomol. 36: 503–507.

Mayer, M. S., and J. D. James. 1970. Attraction of Aedes aegypti II. Velocity of reaction

to host with and without additional carbon dioxide. Entomol. Exp. Appl. 13: 47–

53.

McIndoo, N. E. 1926. An insect olfactometer. J. Econ. Entomol. 19: 545–571.

218 Nyasembe, V. O., P. E. A. Teal, W. R. Mukabana, J. H. Tumlinson, B. Torto, and K.

Nairobi. 2012. Behavioural response of the malaria vector Anopheles gambiae to

host plant volatiles and synthetic blends. Parasites & Vectors. 5: 234.

Otienoburu, P., B. Ebrahimi, P. Phelan, and W. Foster. 2012. Analysis and

Optimization of a Synthetic Milkweed Floral Attractant for Mosquitoes. J. Chem.

Ecol. 38: 873–881.

Schreck, C. E., H. K. Gouck, and N. SMITH. 1967. An improved olfactometer for use

in studying mosquito attractants and repellents. J. Econ. Entomol. 60: 1188–1190.

Sokal, R. R., and F. J. Rohlf. 1995. Biometry. Macmillan.

Takken, W. 1991. The Role of Olfaction in Host-Seeking of Mosquitoes: A Review. Int.

J. Trop. Insect Sci. 12: 287–295.

Torto, B., A. T. Fombong, D. M. Mutyambai, E. Muli, R. T. Arbogast, and P. E. A.

Teal. 2010. Aethina tumida (Coleoptera: Nitidulidae) and Oplostomus haroldi

(Coleoptera: Scarabaeidae): Occurrence in Kenya, Distribution Within Honey Bee

Colonies, and Responses to Host Odors. Ann. Entomol. Soc. Am. 103: 389–396.

Verhulst, N. O., W. Takken, and R. C. Smallegange. 2008. Structural design affects

entry response of mosquitoes in olfactometers. Proc Neth Entomol Soc. 19: 129–

136.

Zar, J. H. 2010. Biostatistical analysis. Prentice-Hall/Pearson.

219 Chapter 3:

Becker, N., M. Zgomba, D. Petric, and M. Ludwig. 1995. Comparison of carbon

dioxide, octenol and a host-odour as mosquito attractants in the Upper Rhine

Valley, Germany. Med. Vet. Entomol. 9: 377–380.

Brown, H. E., M. Paladini, R. A. Cook, D. Kline, D. Barnard, and D. Fish. 2008.

Effectiveness of mosquito traps in measuring species abundance and composition.

J. Med. Entomol. 45: 517–521.

Cooperband, M. F., and R. T. Cardé. 2006. Comparison of plume structures of carbon

dioxide emitted from different mosquito traps. Med. Vet. Entomol. 20: 1–10.

Dudareva, N., and E. Pichersky. 2000. Biochemical and molecular genetic aspects of

floral scents. Plant Physiol. 122: 627–634.

Foster, W. A. 2008. Phytochemicals as population sampling lures. J. Am. Mosq. Control

Assoc. 24: 138–146.

Foster, W. A., and W. Takken. 2004. Nectar-Related vs. human-related volatiles:

behavioural response and choice by female and male Anopheles gambiae

(Diptera: Culicidae) between emergence and first feeding. Bull. Entomol. Res. 94:

145–157.

Gillies, M. T., and T. J. Wilkes. 1969. A comparison of the range of attraction of animal

baits and of carbon dioxide for some West African mosquitoes. Bull. Entomol.

Res. 59: 441–456.

220 Gillies, M. T., and T. J. Wilkes. 1970. The range of attraction of single baits for some

West African mosquitoes. Bull. Entomol. Res. 60: 225–235.

Gillies, M. T., and T. J. Wilkes. 1972. The range of attraction of animal baits and carbon

dioxide for mosquitoes. Studies in a freshwater area of West Africa. Bull.

Entomol. Res. 61: 389–404.

Grimstad, P. R., and G. R. DeFoliart. 1975. Mosquito nectar feeding in Wisconsin in

relation to twilight and microclimate. J. Med. Entomol. 11: 691–698.

Hancock, R. G., and W. A. Foster. 1993. Effects of preblood-meal sugar on sugar

seeking and upwind flight by gravid and parous Aedes aegypti (Diptera:

Culicidae). J. Med. Entomol. 30: 353–359.

Jhumur, U. S., S. Dötterl, and A. Jürgens. 2008. Floral Odors of Silene otites: their

variability and attractiveness to mosquitoes. J. Chem. Ecol. 34: 14–25.

Jürgens, A., T. Witt, and G. Gottsberger. 2002. Flower scent composition in night-

flowering Silene species (Caryophyllaceae). Biochem. Syst. Ecol. 30: 383–397.

Kline, D. L. 2002. Evaluation of various models of propane-powered mosquito traps. J.

Vector Ecol. 27: 1–7.

Kline, D. L. 1999. Comparison of two American biophysics mosquito traps: the

professional and a new counterflow geometry trap. J. Am. Mosq. Control Assoc.

15: 276–282.

221 Kline, D. L., W. Takken, J. R. Wood, and D. A. Carlson. 1990. Field studies on the

potential of butanone, carbon dioxide, honey extract, l-octen-3-ol, L-lactic acid

and phenols as attractants for mosquitoes. Med. Vet. Entomol. 4: 383–391.

Knudsen, J. T., and L. Tollsten. 1993. Trends in floral scent chemistry in pollination

syndromes: floral scent composition in moth-pollinated taxa. Bot. J. Linn. Soc.

113: 263–284.

Mauer, D. J., and W. A. Rowley. 1999. Attraction of Culex pipiens pipiens (Diptera:

Culicidae) to flower volatiles. J. Med. Entomol. 36: 503–507.

Mboera, L. E. G., B. G. J. Knols, M. A. H. Braks, and W. Takken. 2000. Comparison

of carbon dioxide-baited trapping systems for sampling outdoor mosquito

populations in Tanzania. Med. Vet. Entomol. 14: 257–263.

De Meillon, B., A. Sebastian, and Z. H. Khan. 1967. Cane-sugar feeding in Culex

pipiens fatigans. Bull. World Health Organ. 36: 53.

Müller, G., J. Beier, S. Traore, M. Toure, M. Traore, S. Bah, S. Doumbia, and Y.

Schlein. 2010. Successful field trial of attractive toxic sugar bait (ATSB) plant-

spraying methods against malaria vectors in the Anopheles gambiae complex in

Mali, West Africa. Malar. J. 9: 210.

Müller, G. C., and Y. Schlein. 2008. Efficacy of toxic sugar baits against adult cistern-

dwelling Anopheles claviger. Trans. R. Soc. Trop. Med. Hyg. 102: 480–484.

Müller, G., and Y. Schlein. 2006. Sugar questing mosquitoes in arid areas gather on

scarce blossoms that can be used for control. Int. J. Parasitol. 36: 1077–1080.

222 Okumu, F. O., G. F. Killeen, S. Ogoma, L. Biswaro, R. C. Smallegange, E. Mbeyela,

E. Titus, C. Munk, H. Ngonyani, W. Takken, H. Mshinda, W. R. Mukabana,

and S. J. Moore. 2010. Development and field evaluation of a synthetic mosquito

lure that is more attractive than humans. PloS ONE. 5: e8951.

Otienoburu, P. E. 2011. Plant Semiochemicals as Mosquito Attractants (thesis).

Otienoburu, P., B. Ebrahimi, P. Phelan, and W. Foster. 2012. Analysis and

optimization of a synthetic milkweed floral attractant for mosquitoes. J. Chem.

Ecol. 38: 873–881.

Raguso, R. A. 2001. Floral scent, olfaction, and scent-driven foraging behavior. Cogn.

Ecol. Pollinat. Camb. Univ. Press Camb. 83–105.

Schlein, Y., and G. C. Müller. 2008. An approach to mosquito control: Using the

dominant attraction of flowering Tamarix jordanis trees against Culex pipiens. J.

Med. Entomol. 45: 384–390.

Schmied, W., W. Takken, G. Killeen, B. Knols, and R. Smallegange. 2008. Evaluation

of two counterflow traps for testing behaviour-mediating compounds for the

malaria vector Anopheles gambiae s.s. under semi-field conditions in Tanzania.

Malar. J. 7: 230.

Thorsteinson, A. J., and R. A. Brust. 1962. The influence of flower scents on

aggregations of caged adult Aedes aegypti. DTIC Document.

Vargo, A. M., and W. A. Foster. 1982. Responsiveness of female Aedes aegypti

(Diptera: Culicidae) to flower extracts. J. Med. Entomol. 19: 710–718.

223 Wensler, R. J. 1972. The effect of odors on the behavior of adult Aedes aegypti and some

factors limiting responsiveness. Can. J. Zool. 50: 415–420.

Yee, W. L., W. A. Foster, M. J. Howe, and P. . Hancock. 1992. Simultaneous field

comparison of evening temporal distributions of nectar and blood feeding by

Aedes vexans and Ae. trivittatus (Diptera: Culicidae) in Ohio. J. Med. Entomol.

29: 356–360.

Chapter 4:

Abramson, C. I. 1990. Invertebrate Learning: A Laboratory Manual and Source Book.

American Psychological Association.

Abramson, C. I. 1994. A Primer of Invertebrate Learning: The Behavioral Perspective.

American Psychological Association.

Alonso, W. J., T. D. Wyatt, and D. W. Kelly. 2003. Are vectors able to learn about their

hosts? A case study with Aedes aegypti mosquitoes. Memórias Inst. Oswaldo

Cruz. 98: 665–72.

Andersson, S., L. A. A. Nilsson, I. Groth, and G. Bergstrom. 2002. Floral scents in

butterfly-pollinated plants: possible convergence in chemical composition. Bot. J.

Linn. Soc. 140: 129–153.

Bateson, M., S. Desire, S. E. Gartside, and G. A. Wright. 2011. Agitated honeybees

exhibit pessimistic cognitive biases. Curr. Biol. 21: 1070–1073.

224 Benedict, M. Q., R. C. Hood-Nowotny, P. I. Howell, and E. E. Wilkins. 2009.

Methylparaben in Anopheles gambiae sl sugar meals increases longevity and

malaria oocyst abundance but is not a preferred diet. J. Insect Physiol. 55: 197–

204.

Bitterman, M. E., R. Menzel, A. Fietz, and S. Schäfer. 1983. Classical conditioning of

proboscis extension in honeybees (Apis mellifera). J Comp Psychol. 97: 107–119.

Carey, A. F., G. Wang, C.-Y. Su, L. J. Zwiebel, and J. R. Carlson. 2010. Odorant

reception in the malaria mosquito Anopheles gambiae. Nature. 464: 66–71.

Chilaka, N., E. Perkins, and F. Tripet. 2012. Visual and olfactory associative learning

in the malaria vector Anopheles gambiae sensu stricto. Malar. J. 11: 27.

Dudareva, N. A., and E. Pichersky. 2006. Biology of floral scent. CRC Press.

Erdmann, G. R., and S. K. W. Khalil. 1986. Isolation and identification of two

antibacterial agents produced by a strain of Proteus mirabilis isolated from larvae

of the screwworm (Cochliomyia hominivorax) Diptera: Calliphoridae). J. Med.

Entomol. 23: 208–211.

Gary Jr, R. E. G., and W. A. Foster. 2006. Diel timing and frequency of sugar feeding

in the mosquito Anopheles gambiae, depending on sex, gonotrophic state and

resource availability. Med. Vet. Entomol. 20: 308–316.

Howse, E. P. 2003. INSECT ATTRACTANT.

Ignell, R., S. Okawa, J. Englund, and S. R. Hill. 2010. Assessment of diet choice by

the yellow fever mosquito Aedes aegypti. Physiol. Entomol. 35: 274–286.

225 Jhumur, U. S., S. Dötterl, and A. Jürgens. 2006. Naïve and Conditioned Responses of

Culex pipiens pipiens Biotype molestus (Diptera: Culicidae) to Flower Odors. J.

Med. Entomol. 43: 1164–1170.

Jhumur, U. S., S. Dötterl, and A. Jürgens. 2007. Electrophysiological and behavioural

responses of mosquitoes to volatiles of Silene otites (Caryophyllaceae).

Arthropod-Plant Interactions. 1: 245–254.

Jhumur, U. S., S. Dötterl, and A. Jürgens. 2008. Floral odors of Silene otites: their

variability and attractiveness to mosquitoes. J. Chem. Ecol. 34: 14–25.

Kaur, J. S., Y. L. Lai, and A. D. Giger. 2003. Learning and memory in the mosquito

Aedes aegypti shown by conditioning against oviposition deterrence. Med. Vet.

Entomol. 17: 457–460.

Knudsen, J. T., and L. Tollsten. 1993. Trends in floral scent chemistry in pollination

syndromes: floral scent composition in moth-pollinated taxa. Bot. J. Linn. Soc.

113: 263–284.

Lothrop, H. D., S. S. Wheeler, Y. Fang, and W. K. Reisen. 2012. Use of scented sugar

bait stations to track mosquito-borne arbovirus transmission in California. J. Med.

Entomol. 49: 1466–1472.

McCall, P. J., and G. Eaton. 2001. Olfactory memory in the mosquito Culex

quinquefasciatus. Med. Vet. Entomol. 15: 197–203.

McCall, P. J., and D. W. Kelly. 2002. Learning and memory in disease vectors. Trends

Parasitol. 18: 429–33.

226 McCall, P. J., F. W. Mosha, K. J. Njunwa, and K. Sherlock. 2001. Evidence for

memorized site-fidelity in Anopheles arabiensis. Trans. R. Soc. Trop. Med. Hyg.

95: 587–590.

Meijerink, J., M. A. H. Braks, A. A. Brack, W. Adam, T. Dekker, M. A. Posthumus,

T. A. Van Beek, and J. J. A. Van Loon. 2000. Identification of olfactory

stimulants for Anopheles gambiae from human sweat samples. J. Chem. Ecol. 26:

1367–1382.

Menda, G., J. H. Uhr, R. A. Wyttenbach, F. M. Vermeylen, D. M. Smith, L. C.

Harrington, and R. R. Hoy. 2013. Associative learning in the dengue vector

mosquito, Aedes aegypti: avoidance of a previously attractive odor or surface

color that is paired with an aversive stimulus. J. Exp. Biol. 216: 218–223.

Mukabana, W. R., C. K. Mweresa, P. Omusula, B. O. Orindi, R. C. Smallegange, J.

J. van Loon, and W. Takken. 2012. Evaluation of low density polyethylene and

nylon for delivery of synthetic mosquito attractants. Parasit. Vectors. 5: 1–8.

Mwandawiro, C., M. Boots, N. Tuno, W. Suwonkerd, Y. Tsuda, and M. Takagi. 2000.

Heterogeneity in the host preference of Japanese encephalitis vectors in Chiang

Mai, northern Thailand. Trans. R. Soc. Trop. Med. Hyg. 94: 238–242.

Nyasembe, V. O., P. E. A. Teal, W. R. Mukabana, J. H. Tumlinson, B. Torto, and K.

Nairobi. 2012. Behavioural response of the malaria vector Anopheles gambiae to

host plant volatiles and synthetic blends. Parasit. Vectors. 5: 234.

227 Papaj, D. R., and A. C. Lewis. 1993. Insect Learning: Ecological and Evolutionary

Perspectives. Springer.

Riffell, J. A., R. Alarcón, L. Abrell, G. Davidowitz, J. L. Bronstein, and J. G.

Hildebrand. 2008. Behavioral consequences of innate preferences and olfactory

learning in hawkmoth–flower interactions. Proc. Natl. Acad. Sci. 105: 3404–3409.

Sanford, M. R., J. K. Olson, W. J. Lewis, and J. K. Tomberlin. 2012. The effect of

sucrose concentration on olfactory-based associative learning in Culex

quinquefasciatus Say (Diptera: Culicidae). J. Insect Behav. 1–20.

Sanford, M. R., and J. K. Tomberlin. 2011. Conditioning individual mosquitoes to an

odor: sex, source, and time. PloS One. 6: e24218.

Stanczyk, N. M., J. F. Y. Brookfield, L. M. Field, and J. G. Logan. 2013. Aedes

aegypti mosquitoes exhibit decreased repellency by DEET following previous

exposure. PloS One. 8: e54438.

Stone, C. M., R. M. Taylor, and W. A. Foster. 2009. An effective indoor mesocosm for

studying populations of Anopheles gambiae in temperate climates. J. Am. Mosq.

Control Assoc. 25: 514–516.

Syed, Z., and W. S. Leal. 2009. Acute olfactory response of Culex mosquitoes to a

human- and bird-derived attractant. Proc. Natl. Acad. Sci. 106: 18803–18808.

Tomberlin, J. K., G. C. Rains, S. A. Allan, M. R. Sanford, and W. J. Lewis. 2006.

Associative learning of odor with food- or blood-meal by Culex quinquefasciatus

Say (Diptera: Culicidae). Naturwissenschaften. 93: 551–6.

228 Torr, S. J., D. R. Hall, R. J. Phelps, and G. A. Vale. 1997. Methods for dispensing

odour attractants for tsetse (Diptera: Glossinidae). Bull. Entomol. Res. 87:

299–311.

Vet, L., W. J. Lewis, and R. Cardé. 1995. Parasitoid Foraging and Learning, pp. 65–

101. In Chem. Ecol. Insects 2. Chapman & Hall, New York.

Chapter 5:

Beier, J. C., G. C. Müller, W. Gu, K. L. Arheart, and Y. Schlein. 2012. Attractive toxic

sugar bait (ATSB) methods decimate populations of Anopheles malaria vectors in

arid environments regardless of the local availability of favoured sugar-source

blossoms. Malar. J. 11: 1–7.

Dye, C. 1992. The analysis of parasite transmission by bloodsucking insects. Annu. Rev.

Entomol. 37: 1–19.

Facchinelli, L., L. Valerio, J. G. Bond, M. R. W. de Valdez, L. C. Harrington, J. M.

Ramsey, M. Casas-Martinez, and T. W. Scott. 2011. Development of a semi-

field system for contained field trials with Aedes aegypti in southern Mexico. Am.

J. Trop. Med. Hyg. 85: 248.

Ferguson, H. M., K. R. Ng’habi, T. Walder, D. Kadungula, S. J. Moore, I. Lyimo, T.

L. Russell, H. Urassa, H. Mshinda, and G. F. Killeen. 2008. Establishment of a

large semi-field system for experimental study of African malaria vector ecology

and control in Tanzania. Malar. J. 7: 158.

229 Garrett-Jones, C. 1964. Prognosis for interruption of malaria transmission through

assessment of the mosquito’s vectorial capacity. Nature. 204: 1173–1175.

Gary, R. E., J. W. Cannon III, and W. A. Foster. 2009. Effect of sugar on male

Anopheles gambiae mating performance, as modified by temperature, space, and

body size. Parasit. Vectors. 2: 19.

Gary, R. E., and W. A. Foster. 2001. Effects of available sugar on the reproductive

fitness and vectorial capacity of the malaria vector Anopheles gambiae (Diptera:

Culicidae). J. Med. Entomol. 38: 22–28.

Gary, R. E., and W. A. Foster. 2004. Anopheles gambiae feeding and survival on

honeydew and extra‐floral nectar of peridomestic plants. Med. Vet. Entomol. 18:

102–107.

Gu, W., G. Müller, Y. Schlein, R. J. Novak, and J. C. Beier. 2011. Natural Plant Sugar

Sources of Anopheles Mosquitoes Strongly Impact Malaria Transmission

Potential. PLoS ONE. 6: e15996.

Haramis, L., and W. Foster. 1983. Survival and population density of Aedes triseriatus

(Diptera: Culicidae) in a woodlot in central Ohio, USA. J Med Entomol. 20: 391–

398.

Impoinvil, D. E., J. O. Kongere, W. A. Foster, B. N. Njiru, G. F. Killeen, J. I.

Githure, J. C. Beier, A. Hassanali, and B. G. J. Knols. 2004. Feeding and

survival of the malaria vector Anopheles gambiae on plants growing in Kenya.

Med. Vet. Entomol. 18: 108–115.

230 Knols, B. G., B. N. Njiru, E. M. Mathenge, W. R. Mukabana, J. C. Beier, and G. F.

Killeen. 2002. MalariaSphere: A greenhouse-enclosed simulation of a natural

Anopheles gambiae (Diptera: Culicidae) ecosystem in western Kenya. Malar. J. 1:

19.

Mains, J. W., D. R. Mercer, and S. L. Dobson. 2008. Digital image analysis to estimate

numbers of Aedes eggs oviposited in containers. J. Am. Mosq. Control Assoc. 24:

496.

Manda, H., L. C. Gouagna, E. Nyandat, E. W. Kabiru, R. R. Jackson, W. A. Foster,

J. I. Githure, J. C. Beier, and A. Hassanali. 2007. Discriminative feeding

behaviour of Anopheles gambiae s.s. on endemic plants in western Kenya. Med.

Vet. Entomol. 21: 103–111.

Marchand, R. P. 1985. A new cage for observing mating behavior of wild Anopheles

gambiae in the laboratory. J. Am. Mosq. Control Assoc. 1: 234.

Müller, G., J. Beier, S. Traore, M. Toure, M. Traore, S. Bah, S. Doumbia, and Y.

Schlein. 2010. Successful field trial of attractive toxic sugar bait (ATSB) plant-

spraying methods against malaria vectors in the Anopheles gambiae complex in

Mali, West Africa. Malar. J. 9: 210.

Ponlawat, A., and L. C. Harrington. 2009. Factors associated with male mating success

of the dengue vector mosquito, Aedes aegypti. Am. J. Trop. Med. Hyg. 80: 395–

400.

231 Reisen, W. K. 2010. Landscape Epidemiology of Vector-Borne Diseases. Annu. Rev.

Entomol. 55: 461–483.

Ritchie, S. A., P. H. Johnson, A. J. Freeman, R. G. Odell, N. Graham, P. A. DeJong,

G. W. Standfield, R. W. Sale, and S. L. O’Neill. 2011. A secure semi-field

system for the study of Aedes aegypti. PLoS Negl. Trop. Dis. 5: e988.

Stone, C. M., I. M. Hamilton, and W. A. Foster. 2011. A survival and reproduction

trade-off is resolved in accordance with resource availability by virgin female

mosquitoes. Anim. Behav. 81: 765–774.

Stone, C. M., B. T. Jackson, and W. A. Foster. 2012a. Effects of plant-community

composition on the vectorial capacity and fitness of the malaria mosquito

Anopheles gambiae. Am. J. Trop. Med. Hyg. 87: 727–736.

Stone, C. M., B. T. Jackson, and W. A. Foster. 2012b. Effects of bed net use, female

size, and plant abundance on the first meal choice (blood vs sugar) of the malaria

mosquito Anopheles gambiae. Malaria Journal. 11: 3.

Stone, C. M., R. M. Taylor, and W. A. Foster. 2009a. An effective indoor mesocosm for

studying populations of Anopheles gambiae in temperate climates. J. Am. Mosq.

Control Assoc. 25: 514–516.

Stone, C. M., R. M. Taylor, B. D. Roitberg, and W. A. Foster. 2009b. Sugar

deprivation reduces insemination of Anopheles gambiae (Diptera: Culicidae),

despite daily recruitment of adults, and predicts decline in model populations. J.

Med. Entomol. 46: 1327–1337.

232 Straif, S. C., and J. C. Beier. 1996. Effects of sugar availability on the blood-feeding

behavior of Anopheles gambiae (Diptera: Culicidae). J. Med. Entomol. 33: 608–

612.

Takken, W., and B. G. Knols. 1999. Odor-mediated behavior of Afrotropical malaria

mosquitoes. Annu. Rev. Entomol. 44: 131–157.

Zappia, S. P. W., and B. D. Roitberg. 2012. Energy-state dependent responses of

Anopheles gambiae (Diptera: Culicidae) to simulated bednet-protected hosts. J.

Vector Ecol. 37: 172–178.

Chapter 6:

Bayoh, M. N., and S. W. Lindsay. 2003. Effect of temperature on the development of

the aquatic stages of Anopheles gambiae sensu stricto (Diptera: Culicidae). Bull.

Entomol. Res. 93: 375-381.

Beier, J. C., R. Copeland, C. Oyaro, A. Masinya, W. O. Odago, S. Oduor, D. K.

Koech, and C. R. Roberts. 1990. Anopheles gambiae complex egg-stage

survival in dry soil from larval development sites in western Kenya. J. Am. Mosq.

Control Assoc. 6: 105-109.

Boreham, M. M., and D. C. Baerg. 1974. Description of the larva, pupa and egg of

Anopheles (Lophopodomyia) squamifemur Antunes with notes on development

(Diptera: Culicidae). J. Med. Entomol. 11: 564-569.

233 Borg, A. F., and W. R. Horsfall. 1953. Eggs of floodwater mosquitoes II. Hatching

stimulus. Ann. Entomol. Soc. Am. 46: 472-478.

Campos, R. E., and V. E. Sy. 2006. Variation in the hatching response of Ochlerotatus

albifasciatus egg batches (Diptera: Culicidae) in temperate Argentina. Mem. Inst.

Oswaldo Cruz. 101.

Coetzee, M. 2004. Distribution of the African malaria vectors of the Anopheles gambiae

complex. Am. J. Trop. Med. Hyg. 70: 103-104.

Coetzee, M., R. H. Hunt, R. Wilkerson, A. Della Torre, M. B. Coulibaly, and N. J.

Besansky. 2013. Anopheles coluzzii and Anopheles amharicus, new members of

the Anopheles gambiae complex. Zootaxa 3619: 246-274.

Craig, M. H., R. W. Snow, and D. le Sueur. 1999. A Climate-based distribution model

of malaria transmission in sub-Saharan Africa. Parasitol. Today. 15: 105-111.

Deane, M. P., and O. R. Causey. 1943. Viability of Anopheles gambiae eggs and

morphology of unusual types found in Brazil. Am. J. Trop. Med. S1-23: 95-103. de Carvalho, S. C. G., A. J. Martins Junior, J. B. P. Lima, and D. Valle. 2002.

Temperature influence on embryonic development of Anopheles albitarsis and

Anopheles aquasalis. Mem. Inst. Oswaldo Cruz. 97: 1117-1120.

Depinay, J. M. O., C. M. Mbogo, G. Killeen, B. Knols, J. Beier, J. Carlson, J.

Dushoff, P. Billingsley, H. Mwambi, J. Githure, A. M. Toure, and F .E.

McKenzie. 2004. A simulation model of African Anopheles ecology and

population dynamics for the analysis of malaria transmission. Malaria J. 3: 29.

234 Dupree, J. W., and H. A. Morgan. 1902. Mosquito development and hibernation.

Science. 16: 1036.

Gary, R. E., and W. A. Foster. 2001. Effects of available sugar on the reproductive

fitness and vectorial capacity of the malaria vector Anopheles gambiae (Diptera:

Culicidae). J. Med. Entomol. 38: 22-28.

Giglioli, M. E. C. 1965. Oviposition by Anopheles melas and its effect on egg survival

during the dry season in the Gambia, West Africa. Ann. Entomol. Soc. Am. 58:

885-891.

Gillies, M. T., and B. de Meillon. 1968. The Anophelinae of Africa south of the Sahara

(Ethiopian Zoogeographical Region). Pub. S. Afr. Inst. Med. Res. 54: 343.

Gimnig, J. E., M. Ombok, L. Kamau, and W. A. Hawley. 2001. Characteristics of

larval anopheline (Diptera: Culicidae) habitats in western Kenya. J. Med.

Entomol. 38: 282-288.

Gimnig, J. E., M. Ombok, S. Otieno, M. G. Kaufman, J. M. Vulule, and E. D.

Walker. 2002. Density-dependent development of Anopheles gambiae (Diptera:

Culicidae) larvae in artificial habitats. J. Med. Entomol. 39: 162-172.

Gjullin, C. M., C. P. Hegarty, and W. B. Bollen. 1941. The necessity of a low oxygen

concentration for the hatching of Aedes mosquito eggs. J. Cell Com. Physiol. 17:

193-202.

235 Gjullin, C. M., W. W. Yates, and H. H. Stage. 1950. Studies on Aedes vexans (Meig.)

and Aedes sticticus (Meig.), flood-water mosquitoes, in the lower Columbia River

Valley. Ann. Entomol. Soc. Am. 43: 262-275.

Graedel, T. E., and C. J. Weschler. 1981. Chemistry within aqueous atmospheric

aerosols and raindrops. Rev. Geophys. 19: 505–539.

Haddow, A. J., and Y. Ssenkubuge. 1962. Laboratory observations on the oviposition-

cycle in the mosquito Anopheles (Cellia) gambiae Giles. Ann. Trop. Med.

Parasitol. 56: 352-355.

Holstein, M. H. 1954. Biology of Anopheles gambiae: Research in French West Africa.

World Health Organization, Geneva, Switzerland.

Huang, J., E. Walker, J. Vulule, and J. Miller. 2006. Daily temperature profiles in and

around Western Kenyan larval habitats of Anopheles gambiae as related to egg

mortality. Malaria J. 5: 87.

Impoinvil, D. E., G. A. Cardenas, J. I. Githure, C. M. Mbogo, and J. C. Beier. 2007.

Constant temperature and time period effects on Anopheles gambiae egg hatching.

J. Am. Mosq. Control Assoc. 23: 124-130.

James, S. P., and W. G. Liston. 1904. A monograph of the Anopheles mosquitoes of

India,1st ed. Thacker Spink. Calcutta.

Judson, C. L. 1960. The physiology of hatching of aedine mosquito eggs: hatching

stimulus. Ann. Entomol. Soc. Am. 53: 688–691.

236 Judson, C. L., Y. Hokama, and I. Haydock. 1965. The physiology of hatching of

aedine mosquito eggs: Some larval responses to the hatching stimulus. J. Insect

Physiol. 11: 1169-1177.

Kaiser, M. L., L. L. Koekemoer, M. Coetzee, R. H. Hunt, and B. D. Brooke. 2010.

Staggered larval time-to-hatch and insecticide resistance in the major malaria

vector Anopheles gambiae S form. Malaria J. 9: 360.

Koenraadt, C. J. M., A. K. Githeko, and W. Takken. 2004. The effects of rainfall and

evapotranspiration on the temporal dynamics of Anopheles gambiae s.s. and

Anopheles arabiensis in a Kenyan village. Acta Trop. 90: 141-153.

Koenraadt, C., K. P. Paaijmans, A. K. Githeko, B. G. Knols, and W. Takken. 2003.

Egg hatching, larval movement and larval survival of the malaria vector

Anopheles gambiae in desiccating habitats. Malaria J. 2: 20.

Lyimo, E. O., W. Takken, and J. C. Koella. 1992. Effect of rearing temperature and

larval density on larval survival, age at pupation and adult size of Anopheles

gambiae. Entomol. Exp. Appl. 63: 265-271.

Minakawa, N., J. I. Githure, J. C. Beier, and G. Yan. 2001. Anopheline mosquito

survival strategies during the dry period in western Kenya. J. Med. Entomol. 38:

388-392.

Muirhead Thomson, R. C. 1946. Studies on the breeding places and control of

Anopheles gambiae and A. gambiae var. melas in coastal districts of Sierra Leone.

Bull. Entomol. Res. 36: 185-252.

237 Munga, S., N. Minakawa, G. Zhou, O. J. Barrack, A. K. Githeko, and G. Yan. 2005.

Oviposition site preference and egg hatchability of Anopheles gambiae: effects of

land cover types. J. Med. Entomol. 42: 993-997.

Parham, P. E., D. Pople, C. Christiansen-Jucht, S. Lindsay, W. Hinsley, and E.

Michael. 2012. Modeling the role of environmental variables on the population

dynamics of the malaria vector Anopheles gambiae sensu stricto. Malaria J. 11:

271.

Ponnusamy, L., K. Böröczky, D. M. Wesson, C. Schal, and C. S. Apperson. 2011.

Bacteria stimulate hatching of yellow fever mosquito eggs. PLoS One. 6: e24409.

Roberts, D. M. 2001. Egg hatching of mosquitoes Aedes caspius and Ae. vittatus

stimulated by water vibrations. Med. Vet. Entomol. 15: 215-218.

Schreuders, P.D., E.D. Smith, K.W. Cole, and P. Mazur. 1996. Characterization of

intraembryonic freezing in Anopheles gambiae embryos. Cryobiology. 33: 487-

501.

Shililu, J. I., W. B. Grueber, C. M. Mbogo, J. I. Githure, L. M. Riddiford, and J. C.

Beier. 2004. Development and survival of Anopheles gambiae eggs in drying soil:

influence of the rate of drying, egg age, and soil type. J. Am. Mosq. Control

Assoc. 20: 243-247.

Shroyer, D. A., and G. B. Craig. 1980. Egg hatchability and diapause in Aedes

triseriatus (Diptera: Culicidae): temperature and photoperiod-induced latencies.

Ann. Entomol. Soc. Am. 73: 39-43.

238 Sokal, R. R., and F. J. Rohlf. 1995. Biometry: The Principles and Practice of Statistics

in Biological Research, W.H. Freeman & Company, New York.

Sumba, L. A., K. Okoth, A. L. Deng, J. Githure, B. G. Knols, J. C. Beier, and A.

Hassanali. 2004. Daily oviposition patterns of the African malaria mosquito

Anopheles gambiae Giles (Diptera: Culicidae) on different types of aqueous

substrates. J. Circadian Rhythms. 2: 6.

Valencia, M., L. H. Miller, and P. Mazur. 1996. Permeability of intact and

dechorionated eggs of the Anopheles mosquito to water vapor and liquid water: a

comparison with Drosophila. Cryobiology. 33: 142-148.

Valle, D., A. T. Monnerat, M. J. Soares, M. G. Rosa-Freitas, M. Pelajo-Machado, B.

S. Vale, H. L. Lenzi, R. Galler, and J. B. P. Lima. 1999. Mosquito embryos and

eggs: polarity and terminology of chorionic layers. J. Insect Physiol. 45: 701-708.

W.H.O. 2010. World malaria report 2010. World Health Organization, Geneva,

Switzerland.

Wilson, G. R., and W. R. Horsfall. 1970. Eggs of floodwater mosquitoes XII.

Installment hatching of Aedes vexans (Diptera: Culicidae). Ann. Entomol. Soc.

Am. 63: 1644-1647.

Yaro, A. S., A. Dao, A. Adamou, J. E. Crawford, J. M. Ribeiro, R. Gwadz, S. F.

Traoré, and T. Lehmann. 2006. The distribution of hatching time in Anopheles

gambiae. Malaria J. 5: 19.

239 Young, C. J. 1922. Notes on the bionomics of Stegomyia calopus Meigen, in Brazil, Part

I. Ann. Trop. Med. Parasitol. 16: 389-406.

Chapter 7:

(Sample)RC Low-pass Filter Design for PWM - Result - . 2013. Result -.

http://sim.okawa-denshi.jp/en/PWMtool.php.

Bourget, C. M. 2008. An Introduction to Light-emitting Diodes. HortScience. 43: 1944–

1946.

Byers, J. A., and M. A. Unkrich. 1983. Electronic light intensity control to simulate

dusk and dawn conditions. Ann. Entomol. Soc. Am. 76: 556–558.

Evans, J. E., I. C. Cuthill, and A. T. D. Bennett. 2006. The effect of flicker from

fluorescent lights on mate choice in captive birds. Anim. Behav. 72: 393–400.

Fitt, B., and J. Thornley. 2013. Lighting Technology. Taylor & Francis.

Jung, H.-M., J.-H. Kim, B.-K. Lee, and D.-W. Yoo. 2010. A new PWM dimmer using

two active switches for AC LED lamp, pp. 1547 –1551. In Power Electron. Conf.

IPEC 2010 Int. Presented at the Power Electronics Conference (IPEC), 2010

International.

Kuller, R., and T. Laike. 1998. The impact of flicker from fluorescent lighting on well-

being, performance and physiological arousal. Ergonomics. 41: 433–447.

Lehane, M. J. 2005. The Biology of Blood-Sucking in Insects. Cambridge University

Press.

240 Miall, R. C. 1978. The flicker fusion frequencies of six laboratory insects, and the

response of the compound eye to mains fluorescent “ripple”. Physiol. Entomol. 3:

99–106.

Morrow, R. C. 2008. LED Lighting in Horticulture. HortScience. 43: 1947–1950.

Rand, D., B. Lehman, and A. Shteynberg. 2007. Issues, models and solutions for triac

modulated phase dimming of LED lamps, pp. 1398–1404. In Power Electron.

Spec. Conf. 2007 PESC 2007 IEEE.

Shields, E. J. 1989. Artificial Light: Experimental Problems with Insects. Bull. ESA. 35:

40–45.

Silver, J. B. 2008. Mosquito Ecology: Field Sampling Methods. Springer.

Syms, P. R., and L. J. Goodman. 1987. The effect of flickering UV light output on the

attractiveness of an insect electrocutor trap to the house-fly, Musca domestica.

Entomol. Exp. Appl. 43: 81–85.

241