TESTING EFFECTS OF AERIAL SPRAY TECHNOLOGIES ON BITING

AND NONTARGET AT THE PARRIS ISLAND MARINE CORPS

RECRUIT DEPOT, SOUTH CAROLINA, USA.

A dissertation submitted to Kent State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy

by

Mark S. Breidenbaugh

December 2008

Dissertation written by

Mark S. Breidenbaugh

B.S., California State Polytechnic University, Pomona 1994

M.S., University of California, Riverside, 1997

Ph.D., Kent State University, 2008

Approved by

______, Chair, Doctoral Dissertation Committee Ferenc A. de Szalay

______, Members, Doctoral Dissertation Committee Benjamin A. Foote

______Mark W. Kershner

______Scott C. Sheridan

Accepted by

______, Chair, Department of Biological Sciences James L. Blank

______, Dean, College of Arts and Sciences John R.D. Stalvey

ii

TABLE OF CONTENTS

Page

LIST OF FIGURES……………………………………………………………………viii

LIST OF TABLES………………………………………………………………………xii

ACKNOWLEDGEMENTS………………….…………………………………………xiv

CHAPTER I. An introduction to the biting flies of Parris Island and the use of aerial spray technologies in their control……………………………………………..1

Biology of biting midges .....……..……………………………………………..1

Culicoides as nuisance pests and vectors……………………………3

Biology of mosquitoes…………………………………………………………..5

Mosquitoes as nuisance pests and vectors…………………………..6

Integrated pest management…………………………………………………..7

Physical barriers…………………………………………………………8

Cultural control of midges and mosquitoes…………………………..8

Biological control………………………………………………………...9

Chemical control……………………………………………………….10

Biting adulticides applied by aircraft……………………...11

Content and scope of dissertation…………………………………………...13

References cited……………………………………………………………….14

iii Page

CHAPTER II. Seasonal and diel patterns of biting midges () and mosquitoes (Culicidae) on the Marine Corps Recruit Depot,

Parris Island, SC……………………………………………………………………….22

Abstract………………………………………………………………………....22

Introduction……………………………………………………………………..23

Methods…………………………………………………………………………26

Research site description……………………………………………..26

Seasonal biting fly abundance……………………………………….26

Diel activity of biting flies……………………………………………...28

Results………………………………………………………………………….29

Seasonal patterns of biting fly abundance………………………….29

Diel activity of biting flies……………………………………………...33

Discussion……………………………………………………………………...34

Management implications…………………………………………….37

Acknowledgements……………………………………………………………39

References cited…………………………………………………………….…39

CHAPTER III. Characterization of a new ultra-low volume fuselage spray configuration on Air Force C-130H aircraft used for adult control…….63

Abstract………………………………………………………………...... 63

iv Page

Introduction……………………………………………………………………..64

Methods…………………………………………………………………………67

Study site description………………………………………………….67

Field characterization of droplet size………………………….…..…67

AGDISP computer model…………………………………………..…70

Results……………………………………………………………………….…71

AGDISP computer model predictions and field-trial results…….…72

Discussion………………………………………………………………………74

References cited……………………………………………………………….79

CHAPTER IV. Efficacy of aerial spray applications with fuselage booms on

Air Force C-130 aircraft against mosquitoes ( Culicidae) and biting midges

(Ceratopogonidae)…………………………………………………………………….96

Abstract………………………………………………………………...... 96

Introduction……………………………………………………………………..97

Methods…………………………………………………………………………99

Study site descriptions………………………………………………...99

Bioassays……………………………………………………...... 100

Aircraft applications…………………………………………..………101

Craney Island trials…………………………………………….…….101

Parris Island trials………………………………………………….…103

Dibrom applications………………………………………….103

v Page

Anvil applications……………………………………………………..103

Statistical analyses…………………………………………………..104

Results………………………………………………………………………...105

Craney Island trials………………………………………………..…105

Parris Island trials…………………………………………………….106

Dibrom applications………………………………………….106

Anvil applications……………………………………………..107

Discussion…..………………………………………………………………...108

Craney Island trials…………………………………………………..108

Parris Island trials…………………………………………………….110

References……………………………………………………………………113

CHAPTER V. Effects of Aerial Applications of Naled on Nontarget Insects at

Parris Island, South Carolina………………………………………………………..131

Abstract………………………………………………………………...... 131

Introduction……………………………………………………………………132

Methods…………………………………………………………………….…135

Insect trapping methods……………………………………………..136

Experimental design…………………………………………………137

Statistical analyses…………………………………………………..138

Results………………………………………………………………………..140

vi Page

Nontarget species……………………………………………………140

Target species………………………………………………………..141

Discussion…………………………………………………………………….142

Acknowledgements…………………………………………………………..149

References cited……………………………………………………………..149

CHAPTER VI. General Discussion and Conclusions…………………………….169

General discussion and conclusions……………………………………….169

General conclusions…………………………………………………………176

References cited……………………………………………………………..177

vii LIST OF FIGURES

Page

CHAPTER II. Seasonal and diel patterns of biting midges (Ceratopogonidae) and mosquitoes (Culicidae) on the Marine Corps Recruit Depot,

Parris Island, SC

Figure 1. The Marine Corps Recruit Depot on Parris Island, South Carolina…..49

Figure 2. Seasonal abundance of furens at Parris Island …..………50

Figure 3. Seasonal abundance of Culicoides hollensis at Parris Island………...51

Figure 4. Seasonal abundance of Culicoides melleus at Parris Island…………..52

Figure 5. Seasonal abundance of taeniorhynchus at Parris Island……..53

Figure 6. Seasonal abundance of Aedes sollicitans at Parris Island……………54

Figure 7. Seasonal abundance of Culex salinarius at Parris Island……………..55

Figure 8. Seasonal abundance of Aedes vexans at Parris Island……………….56

Figure 9. Seasonal abundance of Culex quinquefasciatus at Parris Island…….57

Figure 10. Correlations between Culicoides numbers and meteorological data.

A. Culicoides furens vs. air temperature B. C. hollensis vs. air temperature C.

C. melleus vs. air temperature D. Culicoides furens vs. rainfall E. C. hollensis vs. rainfall……...……………………………………………………………………58-59

Figure 11. Correlations between Culicidae numbers and meteorological data. A.

Aedes taeniorhynchus vs. temperature B. Ae. taeniorhynchus vs. rainfall

C. Culex salinarius vs. Palmer's Drought Severity Index (PDSI)………………..60

viii Page

Figure 12. Diel activity patterns of Culicoides furens, C. hollensis, and C. melleus…………………………………………………………………………………61

Figure 13. Diel activity patterns of Aedes spp. and Culex spp………………….. 62

CHAPTER III. Characterization of a new ultra-low volume fuselage spray configuration on Air Force C-130H aircraft used for adult mosquito control

Fig. 1. View of field trial location at Avon Park Air Force Range, Florida……….84

Fig. 2. Drop size distribution for droplet spectra produced by 8001 and 8005 flat- fan nozzles……………………………………………………………………………..85

Fig. 3. Distribution of drop sizes in spray cloud from A) 8001 nozzles; B) 8005 nozzles……………………………………………………………………...... 86

Fig. 4. Average droplet size recovered from USAF C-130 fuselage sprays using

8005 flat-fan nozzles…………………………………………………………………..87

Fig. 5. Predicted deposition from AGDISP model of droplets sprayed with C-130 fuselage booms released at (A) 46 m and (B) 91 m above ground………………88

Fig. 6. AGDISP model predicted trajectory of an A) 22.3 µm droplet (DV10 );

B) 54.3 µm droplet (DV50); and C) 104.7 µm droplet (DV90) from 46 m release height………………………………………………………………………………..89-90

Fig. 7. AGDISP model predicted trajectory of an A) 22.3 µm droplet (DV10 );

B) 54.3 µm droplet (DV50); and C) 104.7 µm droplet (DV90) from 91 m release height………………………………………………………………………………..91-92

ix Page

CHAPTER IV. Efficacy of aerial spray applications with fuselage booms on

Air Force C-130 aircraft against mosquitoes ( Culicidae) and biting midges

(Ceratopogonidae)

Fig. 1. Overview of experimental design for USAF C-130 single-pass Dibrom sprays………………………………………………………………………………….119

Fig. 2. Diagrammatic representation of flight path, sampling line, and wind direction on a magnetic grid for A. June 3-trial 1. B. June 3-trial 2. C. October 6

D. October 7, 2004……………………………………………………………..……120

Fig. 3. Relationship between (A) droplet density and mortality (B) droplet size and mortality (B) determined from bioassay cages…………………………124-125

Fig. 4. Relationship between mosquito mortality and distance downwind from pesticide release point following Dibrom applications…………………………….126

Fig 5. Relative numbers of Culicoides spp. before and after Anvil sprays on Parris

Island and a no-spray area……………………………………………………..……127

Fig. 6. Flight path of aircraft (black lines) while making an Anvil insecticide application……...…………………………………………………………………128-129

Fig. 7. Daily biting midge density monitored on Parris Island and a no-spray area……………………………………………………………………………………..130

CHAPTER V. Effects of Aerial Applications of Naled on Nontarget Insects at

Parris Island, South Carolina

x Page

Fig. 1. Map of study area showing Malaise, pan, and CDC trap locations on the

Parris Island Marine Corps Recruit Depot, South Carolina………………...156-157

Fig. 2. Mean total nontarget abundance collected in pan traps before and after a single naled application in 2005……………………………………………166

Fig. 3. Number of target pests 5 days before and 1 day after naled sprays in

2003 at Parris Island MCRD, SC…………………………………………………...167

Fig. 4. Number of target pests (biting midges and mosquitoes) 1 day before and

2 days after a naled spray in 2005 at Parris Island MCRD, SC…………………168

xi

LIST OF TABLES Page

CHAPTER II. Seasonal and diel patterns of biting midges (Ceratopogonidae) and mosquitoes (Culicidae) on the Marine Corps Recruit Depot, Parris Island, SC

Table 1. Abundance of adult biting midges (Ceratopogonidae) and mosquitoes

(Culicidae) collected from November 2001 to November 2004 at four trapping locations on the Marine Corps Recruit Depot, Parris Island…………………….48

CHAPTER III. Characterization of a new ultra-low volume fuselage spray configuration on Air Force C-130H aircraft used for adult mosquito control

Table 1. Parameters input into the AGDISP model to predict droplet fate from fuselage boom spray applications………………………………………………….93

Table 2. Parameters during BVA-13 spray-oil field characterization trials of

USAF ULV fuselage boom configurations…………………………………………94

Table 3. Average droplet size (µm) collected at stations downwind of release point……………………………………………………………………………………..95

CHAPTER IV. Efficacy of aerial spray applications with fuselage booms on

Air Force C-130 aircraft against mosquitoes ( Culicidae) and biting midges

(Ceratopogonidae)

xii Page

Table 1. Percent species composition of field-collected mosquitoes used in caged bioassays at Craney Island, VA in 2004……………………………...……121

Table 2. Values for 24 hr mortality of mosquitoes, volume median diameter, and droplet density at sampling stations……………………………………..……122-123

CHAPTER V. Effects of Aerial Applications of Naled on Nontarget Insects at

Parris Island, South Carolina

Table 1. Parameters measured during naled application at Parris Island October

2003 and April 2005………………………………………………………………….158

Table 2. Means, Shannon's diversity and P-values of nontarget insect abundance collected in malaise traps before and after naled applications in

2003……………………………………………………………………...... ………….159

Table 3. Means (±1 SE) and P-values of nontarget insect abundance collected in pan traps before and after an aerial naled application in 2005……………160-163

Table 4. Means (±1 SE) and P-values of Shannon Diversity Indices (H’) for all insects collected in pan traps before and after an aerial naled application in

2005……………………………………………………………………………...... 164

Table 5. Means and P-values of primary target insect abundance collected in pan traps before and after an aerial naled application in 2005……………..….165

xiii

ACKNOWLEDGEMENTS

It is with great pleasure that I am able to thank those individuals that contributed their knowledge, time, and encouragement toward the completion of these research projects.

I would like to thank my dissertation committee for their insight and direction, especially during the planning stages of the research projects. My advisor, Ferenc de Szalay displayed patience and understanding while the dissertation was interrupted by military deployments and an extensive travel schedule. I appreciated his attention to detail and the many invitations to work and play at Meadowhawk farm. Mark Kershner taught me population ecology and took the time for many engaging conversations. Ben Foote’s incurable interest in wetland ecology was pleasantly infectious. Scott Sheridan offered encouragement and valuable insight into analyzing soil moisture conditions.

Members from the 910 Airlift Wing provided me the flexibility to complete a degree at Kent State: Brian Spears, Mike Deckman, Eric Crabtree, Marty Davis,

Bill Whittenberger, and Daryl Hartman. I would also like to thank the men and women of the Air Force Spray Flight who carefully and patiently flew the spray sorties to make these projects a scientific success. Others provided specific technical support: Karl Haagsma, Steve Olson, Don Teig, Susan Kintz, Lyle

Hefner, and Tom Kocis.

xiv In South Carolina, Jim Clark, Robert Brodeur, Charles Pinckney, Cindy

Zapotoczny, and Joanna Lake of Parris Island MCRD were involved in data acquisition by managing biting fly traps and meteorological information. I am grateful to Bruce Lampwright of the Spring Island Trust for cooperation in the placement of insect trapping on Spring Island. Thanks also go out to Elizabeth

Hager and Gregg Hunt of Beaufort County Mosquito Control.

In Virginia, George Wojcik and his staff at Portsmouth Mosquito Control supported the single pass tests there. Fran Krenick of Clark Mosquito Control

Products, Inc. provided the supplies for making bioassay cages. Mark Latham of

Manatee County Mosquito Control District participated in the droplet characterization tests at Avon Park and instructed me in the use of the AGDISP model.

Finally, I would like to thank my immediate family, Caralisa and Toft, who endured my time away from home at the project field sites with patience and encouragement. I sincerely appreciate their longsuffering during my absence which was in addition to an already extended business related travel schedule.

xv CHAPTER I

AN INTRODUCTION TO THE BITING FLIES OF PARRIS ISLAND AND THE USE OF AERIAL SPRAY TECHNOLOGIES IN THEIR CONTROL

The inhabitants of coastal South Carolina have been plagued by biting flies since the time of the first colonists (Duffy 1953). and malaria, two of the most serious diseases vectored by mosquitoes, caused epidemics in the region and remained endemic until the 20th Century (Ellis 1992). The two primary pestiferous dipteran taxa are mosquitoes (Culicidae) and biting midges

(Ceratopogonidae) and both breed prodigiously in the coastal salt marshes surrounding Parris Island (Beaufort County) South Carolina. Deerflies

(Tabanidae) and sandflies (Psychodidae) also reach annoying levels at times.

My study site was located on Marine Corps Recruit Depot (MCRD) at

Parris Island. Parris Island is the United States Marine Corps basic-training facility for male recruits east of the Mississippi River and all female recruits in the

U.S., with over 25,000 recruits trained at the facility each year (Donahue 2008).

During the training process, recruits spend a significant time out of doors and consequently, are repeatedly exposed to local populations of pestiferous biting flies.

The MCRD at Parris Island is approximately 8,000 acres, over half of which is salt marsh. The uplands are composed of open (grass or developed) and forested (pine and pine-hardwood) areas. The salt marshes are dominated by smooth cordgrass (Spartinia alterniflora). The island is bounded by Archers

1 2

Creek to the north, the Beaufort River on the east, Port Royal Sound on the south, and the Broad River on the west. The city of Charleston, South Carolina is 90 km to the north and Savannah, Georgia is 50 km to the south. The surrounding coastal regions are similar in composition and serve as additional breeding areas for biting flies.

Biology of biting midges

Thirty-two species of biting midges (Culicoides Latreille) have been catalogued from South Carolina (Snow et al. 1958, Wirth et al. 1985). Three species present at Parris Island are among the most notorious mammal-feeding biting midges of the Atlantic Coast and a major focus of pest management:

Culicoides furens (Poey), C. hollensis (Melander and Brues), C. melleus

(Coquillett). These midges are generally described as having a seasonal emergence pattern, with species presence overlapping only during the spring and fall. For example, in North Carolina, C. hollensis activity is high in the spring and fall but decreases or disappears during summer months. In contrast, C. furens had abundance peaks during summer with little or no activity in winter

(Kline and Axtell 1976). Kline and Axtell (1975) found C. melleus to have a similar abundance pattern as C. furens but an order of magnitude less prevalent.

Tidal salt marshes are the primary developmental site for larval Culicoides which can be found within the salt marsh substrate or banks of tidal creeks that are sheltered from wave action. Within the salt marsh, Kline and Axtell (1975)

3

identified two sub-habitats: (1) vegetated areas with sandy soil (dominated by smooth cordgrass) and (2) non-vegetated sandy areas. The larvae are generally considered semi-aquatic since they require a saturated substrate to develop but will perish if they remain completely submerged. Eggs are typically deposited singly along the margin of water bodies and hatch within a few days. Culicoides larvae are known to feed on detritus, algae, fungi, and many species are predaceous on nematodes (Linley 1971, Breidenbaugh and Mullens 1999).

There are four instars and a pupal stage prior to emergence as adults. Males form mating swarms over visual markers and females visit these swarms.

Following mating, autogeny may occur in some species but females of all species require a blood meal to complete the development of subsequent egg batches (Kettle 1995). It is this predilection to blood feeding which brings some species of biting midges in contact, and subsequently, conflict with humankind.

Culicoides as nuisance pests and vectors

Worldwide, Oropouche virus (OROV) is the only significant emerging human viral disease vectored by biting midges (Pinheiro et al. 1976, Mercer et al.

2003). The more common symptoms of Oropouche include fever, chills, and headache but in extreme cases include meningitis (Pinheiro et al. 1981).

Between 1955, when it was first isolated, through 2000, OROV was responsible for at least 27 epidemics and many thousands of clinical cases in Brazil, Panama and Peru (Watts et al. 1997).

4

In contrast, biting midges are vectors of important veterinary pathogens.

Primary vectors of bluetongue virus (BTV) and the closely related epizootic hemorrhagic disease virus (Tabachnick 2004), interest in biting midges in Europe has been renewed as BTV outbreaks have increased in frequency there since

1999 (ProMED 2001). Previously unreported in northern Europe, BTV emerged with 2,124 reported cases in 5 months during 2006-2007 (EFSA 2007). Effects on sheep and other ruminants include fever, nasal discharge, excessive salivation, hyperanemia, ulceration of the oral mucosa, muscle weakness, secondary pneumonia, and death (Mellor et al. 2000).

As biting midges are not known to be vectors of human diseases within

North America (Scanlon 1960, Turner et al. 1963), annoyance and secondary infections from bites are the primary concerns in regard to biting midges at Parris

Island. Native Americans were familiar with biting midges. The Delaware tribe called them "Punk", their word for ash, referring to the fact that the painful bites resembled hot ash landing on the skin (Jamnback 1965). The first Europeans recorded that Native Americans would smear bear grease on the body to prevent bites from insects. Biting midges are colloquially referred to as sandflies or sand fleas in the south, and elsewhere in North America as no-see-ums, punkies, and mooseflies.

The direct impact of numerous midge bites is not trivial. Dorsey (1947) related an account during WWII, in the South Pacific, where feeding by

Culicoides peliliouensis Tokunaga, produced numerous skin lesions in troops,

5

lowered morale, and resulted in numerous cases of secondary infections that produced ulcerous conditions requiring hospitalization. In 1976, the Preventive

Medicine Unit, US Naval Hospital, Beaufort, South Carolina recorded 200 skin infection cases from insect bites. The product of these dermatitis cases was

1,500 lost workdays, estimated at a loss of $25,000 (1976 figures) when pay and hospitalization costs were considered (Haile et al. 1984). The battle between completing recruit training while minimizing negative effects from insect bites continues. Recently, an outbreak of skin infections by methicillin-resistant

Staphylococcus aureus (MRSA) at Parris Island affected ~ 1% of recruits

(Zinderman et al. 2004). Although MRSA can be caused by various factors, these outbreaks occurred during periods of high biting midge or mosquito abundance, suggesting that insect bites may have been involved.

Biology of mosquitoes

Sixty-two mosquito species are known from South Carolina (Evans and

Wills 2002). At Parris Island there are five species that reach pestiferous levels.

The most infamous of these are salt marsh species including the black salt marsh mosquito, (Wiedemann), the eastern salt marsh mosquito, Aedes sollicitans (Walker), and Culex salinarius Coquillett (Nayar

1985). Other species develop in freshwater habitats and can also reach high population levels, include the southern house mosquito, Culex quinquefasciatus

Say and the floodwater mosquito, Aedes vexans (Meigen).

6

Tidal salt marshes and low-lying areas as well as industrial and domestic sources serve as larval developmental sites for mosquitoes at Parris Island.

Pestiferous mosquito species deposit their eggs either singly or in groups, directly in water or into moist soil. In contrast to biting midges, mosquito larvae are aquatic and filter feed on algae, bacteria, and detritus. Like biting midges, mosquitoes have four larval instars and a pupal stage. Males will typically emerge first forming swarms that later females visit (Belton 1994). Females of pestiferous species seek a blood meal, using carbon dioxide and other cues to find appropriate hosts.

Mosquitoes as nuisance pests and vectors

Europeans documented the intense biting by salt marsh mosquitoes in

1570 in letters from Fort Santa Elena (modern day Parris Island), implying that cattle were being exsanguinated by their feeding (Connor 1925). The diurnal biting habits of salt marsh mosquitoes put them in direct conflict with human activities, Nayar (1985) states that half of Florida mosquito control operations are directed against Ae. taeniorhynchus and Ae. sollicitans.

Mosquito species found at Parris Island are also potential vectors of several human pathogens (Service 2000). For example, Anopheles quadrimaculatus Say is a competent vector of malaria, representing a potential for a reemergence of a disease that killed 3,500 in South Carolina during 1915-

1930. The 8 cases per year reported between 1990 & 2000 in South Carolina were considered allochthonous malaria cases as stateside transmission has

7

apparently disappeared (Adler and Wills 2003). The arboviruses eastern equine encephalomyelitis virus and West Nile virus pose the biggest contemporary endemic public health threat vectored by mosquitoes at Parris Island (Ortiz et al.

2003, Fonseca et al. 2004). Traditionally, many Marines have viewed mosquitoes and biting midges as just another tribulation in the crucible of recruit training (personal observation). However, with the introduction of West Nile virus into the U.S. in 1999, and its subsequent detection in South Carolina in 2002

(Adler and Wills 2003), excessive mosquito bites are now more likely to be viewed as a public health problem.

Integrated pest management

Considering the threat of arbovirus transmission from mosquitoes coupled with the nuisance effect of biting midges, it is not surprising that many resources are used to reduce the impacts of biting flies. Management methods include physical barriers (e.g., screens, bed nets), cultural controls (e.g., diking, mowing, and draining), biological control agents, and pesticides. When all elements mentioned above are considered and the most appropriate selected, either alone or in concert, to manage pest populations at acceptable levels, the process is referred to as Integrated Pest Management (IPM) (Kogan 1998). This approach is used in all advanced programs relating to vector control.

8

Physical barriers

Physical barriers are the first line of defense used against midges and mosquitoes as part of IPM. At Parris Island, window screens are commonplace and most residences are air conditioned; consequently, few bites are received indoors. However, recruits train outside and many events take place at dusk or at night, increasing recruit exposure to insect bites as many hematophagous species host-seek during those periods. Repellents are another effective physical barrier for preventing exposure to bites from mosquitoes and midges

(Gupta and Rutledge 1994). Unfortunately, use of repellents by recruits may be sporadic for a variety of reasons associated with the basic training environment or dislike of the stickiness of the material on the skin (personal observation), although the latter issue is improving with new formulations (Debboun et al.

2005).

Cultural control of midges and mosquitoes

Cultural controls and habitat manipulation have been shown to be effective at eliminating or controlling mosquito and midge breeding sites (Rogers

1962, de Szalay et al. 1995). However, methods such as wetland drainage, impounding water, and filling are considered too destructive to estuarine environments to offset any benefits and are rarely used (Floore 1985). Thus, most cultural control techniques that involve habitat manipulation are not used on

Parris Island.

9

Biological control

Biological control can be a cost-effective method to control pests while eliminating the need for chemical controls. A broad range of pathogens, parasites, and predators of mosquitoes and relative levels of success associated with their use in pest management are discussed in Floore (2007). Mosquitoes have many predators that help check their populations and the use of natural enemies as a control strategy is commonplace. For example, Walton et al.

(1990) determined that larvicide applications were not necessary against mosquito larvae late in the season, when native natural enemy populations increased. However, mosquito population growth can often outpace a predator’s growth under favorable environmental conditions (e.g. dense vegetation, eutrophic waters) (Jiannino and Walton 2004) and consequently additional pest management methods may be needed to reduce pest populations to acceptable levels.

There is also an array of potential biological control agents for biting midges including protozoa, fungi, nematodes, and viruses. Some of these have appeared promising but have failed to deliver lasting or commercially acceptable control for biting midges (Wirth 1977). For example, parasitism by an obligate mermithid nematode parasite of Culicoides variipennis (Coquillett), is density dependent (Paine and Mullens 1994) and when infection rates are high, the nematodes limit their reproduction. This species does not suppress pest midge populations at acceptable levels and does not infect other Culicoides species

10

(Mullens et al. 1997). Currently, pest managers do not have commercial options for using natural enemies to control biting midges at acceptable population levels

(Blackwell et al. 2004). Furthermore, the use of natural enemies can be controversial because there are also potential harmful effects caused by introduced species (Simberlof and Stiling 1996). For instance, the mosquito fish,

Gambusia affinis Baird & Girard, is often introduced to pools and streams for mosquito control, but can negatively affect nontarget invertebrates and fish.

Chemical control

Because other tiers of IPM for biting midge control are limited, pest control programs (i.e., mosquito control districts) often rely on chemical control methods when intervention is demanded (Axtell 1974, Blanton and Wirth 1979). Following

WWII, the development of DDT revolutionized pest insect control and most applications used chorlinated hydrocarbons or organophosphate formulations.

Thus, effective chemical control of biting midges was carried out with DDT, applied to both adults (Trapido 1947, Bruce and Blakeslee 1948) and into water to control larvae (Dorsey 1947). While DDT was an inexpensive broad-spectrum pesticide and was used extensively with excellent results, the recognition that organochloride insecticides such as DDT, were residual in nature, promoted insecticide resistance, and led to bio-accumulation in nontarget species (Ripper

1956) shifted pesticide selection to compounds such as carbamates and organophosphates which have a shorter duration in the environment (Casida

1963).

11

Within the life cycle of mosquitoes and midges, the larval stage is more susceptible to chemical control efforts because larvae are confined to the aquatic habitat whereas the flying adults are mobile. Currently, effective and relatively narrow-spectrum larvicides (e.g., endotoxins from Bacillus thuringiensis, subsp.

Israelensis and methoprene insect growth regulators) are available for mosquito control (Mulla 1995). However, larvicides are difficult to use against salt marsh mosquitoes because tidal flushing quickly dilutes the application. Similarly, control of Culicoides larvae with larvicides is also problematic because larvae develop in the mud and any potentially insecticidal material must be able to penetrate into this substrate to be most effective. Additionally, materials that show excellent control against mosquitoes may not be effective against biting midges (Lacey and Kline 1983). Therefore, larval chemical control measures, when used at Parris Island, are directed at mosquitoes and limited to standing water not under tidal influence.

Biting fly adulticides applied by aircraft

The use of pressurized nozzles to spray a fine mist is called ultra low volume (ULV). The ULV technique applies small quantities of concentrated chemical at the minimum effective volume. Consequently, application volume is dependent on the toxicity of the insecticide to the target species and the concentration of the insecticide (Mount et al. 1996). This technique began in

1963, as a replacement to thermal fogging and proved to be a successful means of controlling flying insects and was quickly adopted for mosquito control (Lofgren

12

1972). Aerial applications of pesticides are an effective way of rapidly reducing numbers of potential insect vectors across large areas and in a relatively short period of time (Parrish 1962, WHO 2003). Unlike truck mounted spray units or backpack carried spray equipment, aircraft can access developed and undeveloped areas prone to insect outbreaks. Thus, aircraft dispersal using ULV is now widely used to control mosquitoes (Knapp and Pass 1966).

While midge reductions were likely observed during concurrent mosquito control efforts, the first published account of aerial ULV for midge control was in

1980 from the Cayman Islands using the organophosphate insecticide fenitrothion (Giglioli et al. 1980). The authors reported a greater than 95% midge reduction but they also found only marginally comfortable conditions under

“shirtless” conditions the night following the spray. Kline et al. (1981) reported that organophosphate compounds used for adult mosquito control also showed potential for use against biting midges in wind tunnel tests. An operational test was undertaken by Haile et al. (1984) at Parris Island, South Carolina using ULV aerial applications of naled against natural populations of Culicoides. An application rate of 73 ml/ha, each applied on 2 consecutive days produced a 99% reduction in midge numbers for three days following the spray. Based on results presented by Haile et al. (1984), a biting fly pest management program focusing on aerial applications of insecticide was initiated with the goal of reducing midge populations to tolerable levels.

13

CONTENT AND SCOPE OF DISSERTATION

Four interconnected research projects described in subsequent chapters were developed with the objective of improving pest management of biting flies at the MCRD, Parris Island to reduce the risk of vector-borne illness and improve the quality of life for recruits and other military members. To accomplish this, projects were completed to analyze the pest bionomics, efficacy of application methods used by the U.S. Air Force Aerial Spray Unit, and nontarget effects of aerial spraying to control biting flies. Taken together, such information should strengthen the existing IPM plan on the Parris Island training facility.

The primary component of this endeavor was a multi-year trapping study of the biting flies of Parris Island. This project, described in Chapter 2, consisted of weekly insect trapping over a three-year period and relates the seasonal dynamics of these species to annual cycles of temperature and rainfall. Diel trapping studies were also carried out to determine when biting fly species were most actively host-seeking. This knowledge will allow pest managers to project when pests are most active and promote outdoor training when biting flies are less active to reduce the potential of vector-borne illness and dermatitis. Chapter

3 reports on a novel ULV fuselage boom configuration using a U.S. Air Force C-

130H aircraft. Physical characteristics of the droplet spectra produced during aerial spray operations were measured and subsequently used as input parameters in the AGDISP computer model to make predictions concerning droplet fate after release from the aircraft (Teske et al. 1998). Chapter 4 tests

14

the effectiveness of two commonly applied insecticides to control mosquitoes and biting midges when applied with a C-130H aircraft with a fuselage boom configuration. Population densities of pest species were monitored before and after spray applications and caged mosquitoes were used for bioassays. Lastly,

Chapter 5 tested the effects of aerial applications of an organophosphate insecticide on nontarget insect species from the MCRD, Parris Island.

REFERENCES CITED:

Adler PH, Willis W. 2003. The history of -borne human diseases in

coastal South Carolina. Am Entomol 49:216-228.

Axtell R. 1974. Training manual for mosquito and biting fly control in coastal

areas. Raleigh, NC. University of North Carolina Sea Grant Program.

Belton P. 1994. Attraction of male mosquitoes to sound. J. Am. Mosq. Control

Assoc 10:297-301.

Blackwell A, Evans KA, Strang RHC, Cole M. 2004. Toward development of

neem-based repellents against the Scottish Highland biting midge

Culicoides impunctatus. Med Vet Entomol 18:449–452.

Blanton FS, Wirth WW. 1979. The sand flies (Culicoides) of Florida

(Ceratopogonidae: Diptera). Fla Neighb Land Areas 10:1–

204.

15

Breidenbaugh M, Mullens B. 1999. Descriptions of immature stages of six

Culicoides Latreille spp. (Diptera: Ceratopogonidae) from desert mountain

ranges in southern California, with notes on life histories and rearing

technique. Proc Entomol Soc Wash 101:839-867.

Bruce W, Blakeslee E. 1948. Control of salt marsh sandflies and mosquitoes

with DDT insecticides. Mosq News 8:26-27.

Casida JE. 1963. Mode of action of carbamates. Annu Rev Entomol 8:39-58.

Connor, JM. 1925. Colonial records of Spanish Florida: letters & reports of

governors and secular persons, vol. 1. Fla Hist Soc Pub No. 5.

Debboun M, Strickman DA, Klun JA. 2005. Repellents and the military: our first

line of defense. J Am Mosq Control Assoc 10 Suppl:4-6. de Szalay FA, Batzer DP, Schlossberg EB, Resh VH. 1995. A comparison of

small and large scale experiments examining the effects of wetland

management practices on mosquito densities. Proc Calif Mosq Vector

Control Assoc 63:86–90.

Donohue, P. 2008. “Parris Island general ready to go to Iraq”. May 17, 2008.

Beaufort Gazette, Beaufort S.C.

Dorsey C. 1947. Population and control studies of the Palau gnat on Peleliu,

Western Caroline Islands. J Econ Entomol 40:805-814.

Duffy J. 1953. Epidemics in Colonial America. Baton Rouge, LA: Louisiana

State University Press.

16

EFSA [European Food and Safety Authority]. 2007. Bluetongue Serotype 8

Epidemic Bulletin February 2007. Available from European Food and

Safety Authority, Parma, Italy [accessed June 18, 2008].

http://www.efsa.europa.eu/EFSA/1178620925100/efsa_locale-

1178620753812_BluetongueSerotype8EpidemicBulletin.htm.

Ellis JH. 1992. Yellow fever and public health in the new south. Lexington,

KY: The University Press of Kentucky.

Evans CL, Wills W. 2002. Mosquitoes of Oconee County State Park in

Northwestern South Carolina. J. Entomol. Sci 37:275-277.

Floore TG. 1985. Laboratory wind tunnel tests of nine insecticides against adult

Culicoides species (Diptera: Ceratopogonidae). Florida Entomol 68:679-

682.

Floore TG, ed. 2007. Biorational Control of Mosquitoes. J Am Mosq Control

Assoc 23 Suppl:1-330.

Fonseca DM, Keyghobadi N, Malcolm CA, Mehmet C, Schaffner F, Mogi M,

Fleischer RC, Wilkerson RC. 2004. Emerging vectors in the Culex

pipiens complex. Science 303:1535-1538.

Giglioli MC, Gerberg EJ, Todd RG. 1980. Large scale field tests and

environmental assessments of Sumithion® (fenitrothion) against adult biting

midges in Grand Cayman, West lndies. Mosq News 40:7-5.

Gupta RK, Rutledge LC. 1994. Role of repellents in vector control and disease

prevention. Am J Trop Med Hyg Suppl 50:82-86.

17

Haile D, Kline D, Reinert J, Biery T. 1984. Effects of aerial applications of

naled on Culicoides biting midges, mosquitoes and tabanids on Parris

Island, South Carolina. Mosq News 44:178-183.

Jamnback H. 1965. The Culicoides of New York State. NY State Mus Sci Serv

Bul 399:1-154

Jiannino JA, Walton WE. 2004. Evaluation of vegetation management

strategies for controlling mosquitoes in a southern California constructed

wetland. J Am Mosq Control Assoc 20:18-26.

Kettle D. 1995. Medical and Veterinary Entomology, 2nd Ed. CAB International,

United Kingdom.

Kline DL, Axtell RC. 1975. Culicoides melleus (Coq.) (Diptera:

Ceratopogonidae); Seasonal abundance and emergence from sandy

intertidal habitats. Mosq News 35:328-334.

Kline DL, Axtell RC. 1976. Salt marsh Culicoides (Diptera: Ceratopogonidae):

Species, seasonal abundance and comparison of trapping methods.

Mosq News 36:1-10.

Kline D, Haile G, Baldwin K. 1981. Wind tunnel tests with seven insecticides

against adult Culicoides mississippiensis Hoffman. Mosq News 41:745-

747.

Kogan M. 1998. Integrated pest management: Historical perspectives and

contemporary developments. Annu Rev Entomol 43:243-70.

18

Knapp FW, Pass BC. 1966. Low-volume aerial sprays for mosquito control.

Mosq. News 26:22-25.

Lacey LA and DL Kline. 1983. Laboratory Bioassay of Bacillus thuringiensis (H-

14) against Culicoides spp. and Leptoconops spp. (Ceratopogonidae).

Mosq News 43:502-503.

Linley JR, Davies JB. 1971. Sandflies and tourism in Florida and the Bahamas

and Caribbean area. J Econ Entomol 64:264-278.

Lofgren CS. 1972. Ultralow Volume (ULV) application of insecticides. Am J

Trop Med Hyg 21:819-824.

Mellor PS, Boorman J, Baylis, M. 2000. Culicoides biting midges: their role as

arbovirus vectors. Annu Rev Entomol 45:307-340.

Mercer DR, Spineli GR, Watts DM, Tesh RB. 2003. Biting rates and

developmental substrates for biting midges (Diptera: Ceratopogonidae) in

Iquitos, Peru. J Med Ent 40:807-812.

Mount GA, Biery TL, Haile DG. 1996. A review of ultralow-volume aerial sprays

of insecticide for mosquito control. J Am Mosq Control Assoc 12:601-618.

Mulla MS. 1995. The future of insect growth regulators in vector control. J Am

Mosq Control Assoc 11:269-273.

Mullens B, Luhring K, Breidenbaugh M. 1997. Experimental host range studies

with Heleidomermis magnapapula (Mermithidae), a parasite of Culicoides

variipennis (Ceratopogonidae). J Am Mosq Control Assoc 13:398-401.

19

Nayar JK. 1985. Bionomics and physiology of Aedes taeniorhynchus and Aedes

sollicitans, the salt marsh mosquitoes of Florida. JK Nayar [Ed].

Univerisity of Florida Agricultural Experiment Station, Gainesville, FL. Bull

852: 148 p.

Ortiz DI., Wozniak A, Tolson MW, Turner PE, Vaughan DR. 2003. Isolation of

EEE Virus from taeniorhynchus and Culiseta melanura in

coastal South Carolina. J Am Mosq Control Assoc 19:33-38.

Paine EO, Mullens BA. 1994. Distribution, seasonal occurrence, and patterns of

parasitism of Heleidomermis magnapapula (Nematoda: Mermithidae), a

parasite of Culicoides variipennis (Ceratopogonidae: Diptera) in

California. Environ Entomol 23:154-160.

Parrish, D, Hodapp C. 1962. Biological evaluation of the C-47 aerial spray

system for adult mosquito control. Mosq News 22:36-37.

Pinheiro FP, Travossos Da Rosa APA, Travossos Da Rosa JF, Bensabath G.

1976. An outbreak of Oropouche virus disease in the vicinity of Santarem,

Para, Brazil. Tropenmed Parasitol 27:213-223.

Pinheiro FP, Travassos Da Rosa APA,Travassos Da Rosa JF, Ishak R, Freitas

RB. 1981. Oropouche virus. I. Overview of clinical, epidemiological and

ecological findings. Am J Trop Med Hyg 30:149-60.

ProMED-mail. Bluetongue upsurge in the Mediterranean basin. ProMED-mail

2001; 22 Nov: 20011122.2864. . Accessed

19 June 2008.

20

Ripper WE. 1956. Effect of pesticides on balance of arthropod populations. Annu

Rev Entomol 1:403-438.

Rogers A.J. 1962. Effects of impounding and filling on the production of sand

flies (Culicoides) in Florida salt marshes. J Econ Entomol 55:521-27.

Scanlon JE. 1960. The relationship of Culicoides (Diptera: Ceratopogonidae) to

the transmission of the virus of eastern equine encephalitis [PhD

dissertation]. University of Maryland.

Service MW. 2000. Medical Entomology for Students, 2nd Ed. Cambridge UK:

University of Cambridge Press.

Simberloff D and P Stiling. 1996. How risky is biological control? Ecology

77:1965-1974.

Snow WE, Pickard E, Moore JB. 1957. The Heleidae of the Tennessee River

Basin. J Tennessee Acad Sci 32:18-36.

Tabachnick WJ. 2004. Culicoides and the global epidemiology of bluetongue

virus infection. Vet Ital 40:45-150.

Teske ME, Thistle HW, Eav B. 1998. New ways to predict aerial spray

deposition and drift. J For 96:25-31.

Trapido, H. 1947. DDT residual spray control of sandflies in Panama. J Econ

Entomol 40:472-5.

Turner EC, Wehrheim NL, Messersmith DH, Davis JW. 1963. Transmission

studies of avian infectious synovitis by selected arthropod vectors. Poultry

Sci 42:1434-1441.

21

Walton WE, Tietze NS, Mulla MS. 1990. Ecology of Culex tarsalis (Diptera:

Culicidae): factors influencing larval abundance in mesocosms in

Southern California. J Med Entomol 27:57-67.

Watts DM, Phillips I, Callahan JD, Griebenow W, Hyams KC, Hayes CG. 1997.

Oropouche virus transmission in the Amazon River basin of Peru. Am J

TropMed Hyg 56:148-52.

WHO [World Health Organization]. 2003. Space spray application of

insecticides for vector and public health pest control: A practitioner’s

guide. Geneva, World Health Organization.

Wirth WW. 1977. A review of the pathogens and parasites of the biting midges

(Diptera: Ceratopogonidae). J Washington Acad Sci 60-75.

Wirth WW, Dyce AL, Peterson BV. 1985. An atlas of wing photographs, with a

summary of the numerical characters of the Nearctic species of Culicoides

(Diptera: Ceratopogonidae). Contrib Am Entomol Inst 22:1-46.

Zinderman CE, Conner B, Malakooti, MA, LaMar JE, Armstrong A, Bohnker BK.

2004. Community-acquired Methicillin-resistant Staphylococcus aureus.

Emerg Infect Dis 10:941-944.

CHAPTER II

SEASONAL AND DIEL PATTERNS OF BITING MIDGES (CERATOPOGONIDAE) AND MOSQUITOES (CULICIDAE) ON THE PARRIS ISLAND MARINE CORPS RECRUIT DEPOT1

ABSTRACT

The Marine Corps Recruit Depot on Parris Island, South Carolina conducts extensive outdoor training throughout the year. It is surrounded by tidal salt marshes, which are breeding habitats for many pestiferous biting flies.

Knowledge of biting fly behavior patterns is needed to develop effective pest management strategies in urban areas adjacent to salt marshes. We measured biting midge (Ceratopogonidae) and mosquito (Culicidae) seasonal abundance and diel activity patterns on Parris Island using CO2-baited suction traps from

November 2001 – November 2004. Of the three biting midge species collected,

Culicoides furens was most abundant (86.2% of total) and was present in high numbers from late March to November. Culicoides hollensis (12.0% of total) was present during spring and fall but absent in summer and winter; and C. melleus

(1.7% of total) was present in spring through fall but absent in winter.

Abundance of C. furens had a positive linear correlation with air temperature (r2=

0.67) and rainfall (r2= 0.29). There were nonlinear correlations between air temperature and C. hollensis and C. melleus numbers (r2= 0.74, r2= 0.41,

1 Submitted to the Journal of Vector Ecology by M.S. Breidenbaugh, J.W. Clark, R.M. Brodeur, F.A. de Szalay.

22

23

respectively), which were most abundant at moderate temperatures. Of 18 mosquito species collected, the most abundant were Aedes taeniorhynchus

(42.7% of total), Ae. sollicitans (26.3% of total), Culex salinarius (15.6% of total),

Cx. quinquefasciatus (7.3% of total), Ae. vexans (5.7% of total); other species comprised <5% of collections. Aedes taeniorhynchus numbers were positively correlated with temperature (r2 = 0.58) and rainfall (r2 = 0.21), and Cx. salinarius was correlated with soil moisture (r2 = 0.59). Activity of most biting midges and mosquitoes was highest the first 2 hours following sunset. Species of biting flies were present in all months, suggesting that year-round control measures are necessary to reduce exposure to potential disease vectors and nuisance biting.

INTRODUCTION Long-term studies of species abundance coupled with an understanding of host-seeking activity are needed by pest managers to plan efficient strategies to control pestiferous biting flies. Potential Integrated Pest Management (IPM) methods for biting flies include physical barriers (e.g., screens, bed nets), cultural controls (e.g., diking and draining), biological control agents, and chemical pesticides. Pest control in salt marshes often relies on adulticides or larvicides

(Blanton and Wirth 1979) because cultural control methods such as wetland drainage have negative environmental impacts (Cilek and Hallmon 2005) and there are no effective biocontrol options (Blackwell et al. 2004). Although narrow-spectrum larvicides (e.g., endotoxins from Bacillus thuringiensis, subsp. israelensis and methoprene insect growth regulators) are available for mosquito

51

24

control (Mulla 1995) in freshwater habitats, these are not effective in salt marshes due to tidal flushing. Information about the seasonal and diel periodicity of biting flies can assist existing programs because applications of adulticides should coincide with times when the target pests are host-seeking (WHO 2003).

In contrast, resting insects are found within dense vegetation (Bidlingmayer

1961) and are unlikely to be exposed to aerial sprays. Knowledge about biting fly behavior can also be used to schedule outdoor activities to avoid peak exposure periods or encourage avoidance technologies such as repellants or permethrin- treated clothing.

At least 32 species of biting midges (Ceratopogonidae) are found in South

Carolina (Snow et al. 1957, Wirth et al. 1985), including some the most notorious pest species along the Atlantic Coast: Culicoides furens (Poey), C. hollensis

(Melander and Brues), C. melleus (Coq.). Peak flight activity of many biting midge species is at dusk and dawn (Kettle and Linley 1969, Barnard and Jones

1980, Linhares and Anderson 1990, Mullens 1995, Breidenbaugh 1997), but others are diurnal or nocturnal (Foulk 1969, Schmidtmann et al. 1980). In general, biting midges are not important human disease vectors, but their painful bites cause severe annoyance when they are present in high numbers. They also can cause skin lesions leading to secondary infections that require hospitalization (Dorsey 1947, Haile et al. 1984).

Sixty-two mosquito (Culicidae) species are found in South Carolina (Evans and Wills 2002). Several important pestiferous species inhabit salt marshes

51

25

including the black salt marsh mosquito, Aedes taeniorhynchus (Wiedemann), the eastern salt marsh mosquito, Ae. sollicitans (Walker), and Culex salinarius

Coq. (Nayar 1985). Mosquitoes display a range of flight activities, including nocturnal, crepuscular, and diurnal (Wright and Knight 1966, Guimarães et al.

2000, Strickman et al. 2000). The host-seeking activity of salt marsh mosquitoes causes annoyance and also potential disease transmission. Eastern equine encephalomyelitis virus is vectored by mosquitoes in South Carolina (Ortiz et al.

2003, Fonseca et al. 2004). West Nile virus (WNV) was detected in the region in

2002 (Adler and Wills 2003), and has led to increased biting fly control efforts.

The United States Marine Corps facility on Parris Island, South Carolina, is used for basic training of over 25,000 recruits each year (Donahue 2008). The facility is located within extensive tidal salt marshes, which provide breeding habitat for significant populations of biting midges and mosquitoes. Spanish explorers in 1570 reported intense biting fly activity in these marshes (Connor

1925), and currently, recruits spend many hours outdoors and are repeatedly exposed to pestiferous biting flies. An extensive aerial spray program is conducted by the US Air Force Aerial Spray Unit at this site, but there is little information on the best times to apply pesticides or when the major species in the region are most numerous. Therefore, information on biting fly populations would assist control programs at Parris Island and other salt marsh sites in the southeastern United States.

51

26

Pest surveillance is a keystone feature of any IPM program (Kogan 1998).

Increased knowledge of biting fly bionomics will allow pest managers to anticipate species prevalence and establish appropriate application schedules for insecticides. Therefore, we conducted a multi-year trapping study to monitor seasonal changes in species and population sizes of biting flies at the Parris

Island MCRD, SC. We also determined the peak diel activities of biting midges and mosquitoes at Parris Island.

METHODS

Research site description

The Marine Corps Recruit Depot at Parris Island is approximately 3,200 hectares, over half of which is tidal salt marsh and dominated by smooth cordgrass (Spartinia alterniflora). Nearby coastal areas are similar in habitat and have extensive salt marshes with breeding areas for biting flies. Upland portions of Parris Island are composed of open, mowed grassy areas, buildings, and wooded (pine and hardwood) areas. Parris Island is bounded by Archers Creek to the North, the Beaufort River to the East, Port Royal Sound to the South, and the Broad River to the West.

Seasonal biting fly abundance

Seasonal occurrence of biting flies was monitored on the Marine Corps

Recruit Depot from November 2001 through November 2004. Four sampling locations were chosen near to military training and recreational areas (Fig. 1): 1)

51

27

Horse Island (HI) (N32°21.218, W80°42.957) was bordered by salt marsh and was densely vegetated with mature live oaks; 2) Provo Marshal’s Office (PMO)

(N32°21.199, W80°40.691) was bordered by salt marsh to the north and the main parade grounds and barracks to the south; 3) Weapons (WEP) (N32°20.015,

W80°42.083) was bordered to the south by salt marsh and the rifle range to the north; 4) Golf Course (GC) (N32°18.452, W80°40.639) was bordered to the south by salt marsh and north by Page Field.

Meteorological data (temperature and rainfall) were recorded at the Parris

Island Water Treatment Facility near the PMO trap site. Times of sunrise/sunset were obtained from the US Navy Observatory Applications Department database

(http://aa.usno.navy.mil/data/docs/RS_OneYear.php). In addition, the Palmer

Drought Severity Index (PDSI) (National Climatic Data Center, http://www.ncdc.noaa.gov/oa/climate/research/monitoring.html#drought), was used to determine moisture conditions during the study. A PDSI < 0 indicates

"dry" conditions and > 0 indicates "moist" conditions (-6 to +6 scale)(Palmer

1965)

Each sampling location was outfitted with a single CO2-baited CDC-style suction trap (Clarke Mosquito Control Products, Chicago, IL.) that was operated one day each week. These traps attract host-seeking female biting flies with CO2

(Sudia and Chamberlain 1962), and flies are captured in a collection chamber by suction with a battery-operated fan. Traps were hung on a tree limb or fence 1.5

51

28

meters above ground. All trap locations were at least 2.5 km apart and were within 1 km of tidal-influenced salt marshes.

Trap contents were collected after 24 hours and processed in the laboratory. Large samples were split using a grid system and a 50% or 25% subset was randomly selected. Specimens within each selected subset were identified to species and counted, and numbers were extrapolated to the entire trap collection. The remainder of the sample was also scanned to avoid missing any rare species. All biting flies were identified to species using taxonomic keys

(Darsie and Ward 1981, Blanton and Wirth 1979).

Monthly trap collection data were checked for normality and were log transformed (X + 1) when needed. Monthly mosquito and biting midge counts were compared with temperature, rainfall, and PDSI using linear and non-linear regression analysis. Statistical analyses were conducted with SPSS for

Windows (Version 13.0, SPSS, Chicago, IL).

Diel activity of biting flies

Diel activity of biting flies was also examined at Parris Island. A standard

CDC-style trap (BioQuip Products, Rancho Dominguez, CA) was used to collect biting midges and mosquitoes at HI, GC, and PMO locations. The CDC-style trap had a suction fan and a rotating motorized turntable with 8 collection chambers filled with 25 ml of water and detergent surfactant. The chambers were sealed as the turntable rotated to take independent collections for 2 hour intervals. A Mosquito Magnet® (American Biophysics, East Greenwich, RI) was

51

29

used to provide CO2 to attract biting flies to the rotating trap. Mesh netting was placed over the Mosquito Magnet intake to assure that only the CDC-style trap collected insects.

Diel trapping was conducted on 1-7, 12,13 October 2003, 13-20 April 2004 and 25 August -1 September 2004. Photophase was 11 hours, 4 minutes on the first day of the study and decreased by approximately 2 minutes each day during

October. During the April collection dates, photophase increased by 9 minutes.

Collection intervals were established in two-hour increments (e.g., 0800-1000 hrs; 1000-1200 hrs, etc.). In this arrangement, the first sampling period began at least 5 hours prior to sunset and trapping was continuous within the study periods. Samples were transferred from collection chambers to 75% ethanol, and flies were identified and counted under a microscope.

RESULTS

Seasonal patterns of biting fly abundance

Monthly temperatures during the 36 month study averaged 1.2 °C above the normal monthly average (1911 – 2004, National Climatic Data Center,

Ashville NC). At the beginning of the study from November 2001 - July 2002, rainfall amounts were much lower than normal (PDSI was about -4 indicating a moderate drought). After this period, rainfall increased and amounts were higher than normal (PDSI was 2 - 3 indicating moist conditions) during May to October

2003. Rainfall declined after November 2003, and conditions remained around

51

30

normal (PDSI 1 to -2) for the remainder of the study. Overall, rainfall was approximately 4% above average rainfall (141 mm) during the study.

A total of 480,811 biting midges were collected during the study, and they were present year-round. All individuals were identified as three species:

Culicoides furens, C. hollensis, and C. melleus. Culicoides furens was the most abundant biting midge at Parris Island, representing 86.2% of total biting midges collected (Table 1). It was also the most abundant taxa at all sites except PMO.

Culicoides hollensis was the next abundant biting midge (12.0% of total).

Culicoides melleus was relatively uncommon (1.7% of total), and its populations reached high levels only briefly.

Culicoides furens numbers had a weakly bimodal seasonal pattern. They appeared in spring in late March or early April and peaked at over 1,000 individuals per trap night in late May. They declined afterwards but remained at relatively high numbers (>150 individuals per trap night) throughout the summer.

They had a second peak in September and disappeared after November (Fig. 2).

The single largest trap collection for a 24-hour sample occurred on 6 May 2003 when >120,000 C. furens were collected at the HI site. Although numbers of C. furens during summer were usually high, there was substantial weekly variation.

Trap collections differed by an average of 1,563 (SE ±517) biting midges from week to week.

Culicoides hollensis also had a bimodal activity pattern with peak numbers in fall and spring (Fig. 3). These biting midges appeared in January and had the

51

31

highest numbers in March and April, and then were absent from June to August.

Numbers during a second peak in fall were highest in November, and declined in

December. The largest numbers of C. hollensis were collected on 20 March

2002 when 6,463 biting midges were collected at the PMO trap site. Weekly variation in numbers was high, and collections differed by an average of 301 (SE

± 81) biting midges from week to week.

Culicoides melleus was collected in relatively low numbers with a spring and fall peak (Fig. 4). The largest collection of C. melleus occurred 8 April 2003 at the WEP site (n=944 midges), which is above levels considered pestiferous

(>150 biting midges / trap night). However, numbers varied between years. For example, collections were never >100 biting midges / trap night in 2004. Weekly variation was less pronounced than for other species and weekly trap collections differed only by an average of 109 biting midges (SE ± 24).

A total of 91,602 individual mosquitoes from 18 species were collected during the study (Table 1). The most abundant species were: Aedes taeniorhynchus, Ae. sollicitans, Cx. salinarius, Cx. quinquefasciatus Say, Ae. vexans (Meigen), Ae. atlanticus Dyar & Knab, Anopheles bradleyi King. The other 11 species were: Aedes albopictus Kuse, Ae. mitchellae (Dyar), Anopheles atrops Dyar & Knab, An. quadrimaculatus Say, Cx. nigripalpus Theobald,

Culiseta melanura (Coq.), signifera (Coq.), ciliata

(Fabr.), Ps. columbiae (Dyar & Knab), Ps. ferox (von Humboldt), and Uranotenia lowii Theobald. These comprised less than 0.3% of collections. Adult host-

51

32

seeking mosquitoes were present at Parris Island year-round. Mosquito populations generally reached 100 individuals per trap night by late-March.

Populations increased to >1000 individuals/night by August but decreased in winter.

Aedes taeniorhynchus was the most abundant mosquito and comprised

42.8% of total mosquitoes collected. This species exhibited a unimodal distribution from mid-March to November (Fig. 5). Populations peaked in mid-September but collections over 100 mosquitoes per trap night occurred on most dates between mid-August and mid-October. The largest sample taken was >5,200 individuals collected on 12 Sept 2002.

Other mosquito species exhibited bimodal distributions. This included Ae. sollicitans (26.4% of collected mosquitoes), which peaked in spring and had a larger peak in fall (Fig. 6). Culex salinarius was present nearly the entire year but had two peaks during spring and November - December (Fig. 7). Aedes vexans was present in March - June and then disappeared until a second peak occurred in November - January (Fig. 8).

Other species populations did not have an obvious seasonal pattern. For example, Cx. quinquefasciatus was common throughout the year, but it occurred in relatively low numbers (Fig. 9). Aedes atlanticus and An. bradleyi were intermittently collected in low numbers throughout spring to early winter.

Several meteorological factors were correlated with biting midge abundance. Culicoides furens abundance was positively correlated with

51

33

temperature (r2 = 0.67, n = 36, P < 0.001) and reached the highest numbers in mid-summer when temperatures were ~30 ºC (Fig. 10). There were also significant non-linear correlations between temperature and C. hollensis and (r2 =

0.74, P < 0.001) and C. melleus (r2 = 0.41, P < 0.001) numbers. These species peaked at air temperatures between 15 to 25 ºC. Culicoides furens was positively correlated with rainfall (r2 = 0.29, n = 36, P = 0.001), but C. hollensis was negatively correlated with rainfall (r2 = 0.20, n = 36, P = 0.006). Culicoides furens was only absent from collections for a single period when rainfall exceeded 125 cm/month while the converse was observed for C. hollensis which was nearly absent from collections when rainfall was above 150 cm/month.

Mosquitoes were also correlated with meteorological factors. Density of

Ae. taeniorhynchus was positively correlated with temperature (r2 = 0.58, n = 36,

P < 0.001) and rainfall (r2 = 0.21, n = 36, P = 0.005). This species became very active above 25 ºC and across a wide range of precipitation. Other correlations were not significant except for Cx. salinarius was positively correlated with PDSI

(r2 = 0.34, n = 36, P < 0.001) and density peaked at a PDSI of 2.0.

Diel activity of biting flies

Diel samples in 2003-2004 collected 1,829 biting midges. The three species collected were C. furens (78.9% of total), C. hollensis (16.7% of total), and C. melleus (4.4% of total). All three species had distinct crepuscular activity.

Most Culicoides furens (73%) and C. hollensis (79%) were collected within

2 hours of sunset, and peak activity occurred just after sunset (Fig. 12).

51

34

Culicoides melleus was not collected in high numbers but its activity also peaked around sunset. Low numbers of all species were active in daytime hours. Some variation was observed between seasons. For example, only 18% of C. furens were collected during the crepuscular period in April 2004, when flight period was observed throughout the night and after sunrise.

Relatively few mosquitoes were collected in diel samples (n=154). Most mosquitoes could only be identified to genus because some morphological characters were obscured after storage in alcohol. The two dominant genera were Culex spp. (mostly Cx. salinarius) and Aedes spp. (mostly Ae. taeniorhynchus). These genera comprised 99% of all mosquitoes collected.

Aedes spp. displayed crepuscular activity peaking in the first 2 hours following sunset (Fig. 13). Culex spp. were present immediately after sunset (31% of individuals collected) but over half were collected from 2400 - 0400 hrs.

Mosquito activity of both species decreased sharply after sunrise.

DISCUSSION

The biting midge and mosquito communities on Parris Island were comprised of species that are common in tidal salt marshes along the Atlantic and Gulf Coasts (Blanton and Wirth 1979, Rueda and Gardner 2003). These species are well adapted to the harsh conditions of fluctuating water levels and saline environments. As a result, they are dominant pest species in similar habitats throughout the region.

51

35

Temporal changes of biting midge abundance on Parris Island were similar to patterns reported in other salt marshes. For example, we found few C. hollensis during the summer months and the winter. Although C. hollensis sometimes is present year-round (e.g. southern Florida, Blanton and Wirth 1979), most studies found the same pattern as we report here (Jamnback 1965, Kline and Axtell 1976). Moreover, the dates of peak numbers of biting midges at Parris

Island fit an expected temporal shift along a geographic ecocline. For example, peak numbers of C. furens on Parris Island occurred in late-May. This is similar to dates of peak numbers reported at similar latitudes in the southeastern United

States (Khalaf 1967, Kline and Axtell 1976, Lillie et al. 1987). In contrast, peak numbers further north in Connecticut and New York occur in July (Lewis 1959,

Jamnback 1965) and peak numbers further south along the Gulf Coast are earlier (Kline and Roberts 1981). Likewise, the seasonal peak of C. melleus numbers on Parris Island also occurred at the time expected at this latitude

(Jamnback 1958; Kline and Axtell 1975, Lillie et al. 1987).

There were also strong correlations between weather conditions and biting fly numbers and this information may help pest managers predict which species should be targeted at different times of the year. Aedes taeniorhynchus and C. furens were positively correlated with increasing temperature. Culicoides furens is widely distributed and reaches as far south as Brazil (Wirth et al. 1988). Aedes taeniorhynchus is also a tropical species that is found in southern Brazil and Peru

(WHO 1989). Therefore, these are adapted to warm temperatures. Other biting

51

36

fly species reached peak numbers at moderate temperatures: C. hollensis at

~17°C and C. melleus at ~21°C. Unlike, C. furens, the distribution of C. hollensis and C. melleus does not extend into Central America, reflecting these species preference for moderate temperatures.

Rainfall also was correlated with biting fly numbers. Some species were more abundant during high rainfall (C. furens, Ae. taeniorhynchus), but others increased during dry periods (C. hollensis, Cx. salinarius). For example, seasonal patterns of C. furens and Ae. taeniorhynchus numbers were almost identical because they were both abundant during the hot and humid months of the year. Culex salinarius and C. hollensis often declined during the wettest months (Janousek and Olson 2006), which indicates they are less dependent on rainfall during larval development.

Peak activity of most species of biting midges and mosquitoes were during the crepuscular period around sunset, which has been reported in other studies

(Service 1971). Furthermore, Cx. salinarius have extended nocturnal activity

(Gladney and Turner 1970, Anderson et al. 2007) and we also found that Culex spp. activity (which were predominantly Cx. salinarius) was high throughout the night. However, some flight activity patterns on Parris Island were different than expected. For example, mosquitoes and biting midges often have two peak activity periods, one immediately at sunset and one at sunrise (Trapido 1947,

Bidlingmayer 1961, Carroll and Bourg 1977, Esbarry 1977). We did not observe this pattern for any species. We observed a minor activity peak of C. furens two

51

37

hours after dawn, but it was only in one month (April 2004). Lillie et al. (1987) reported that diel activity of biting midges can vary with the season but reported a consistently greater peak in the evening similar to our findings at Parris Island.

Management Implications

Pestiferous biting midges were most abundant on Parris Island during spring and autumn. Although there was some temporal segregation between the three species, peak numbers of the most common species, C. furens, overlapped the fall peak of the other two species. This caused extremely intense biting activity at this time and our traps collected >1,000 biting midges per night.

Mosquitoes were also the most abundant in autumn, although they were present throughout the year. Ironically, spring and fall are the periods when the region's climate is most enjoyable and levels of outdoor activity by humans are the highest. Therefore, exposure to nuisance biting and epizootic disease can be a severe problem for people living near tidal salt marshes. Moreover, the Marine

Corps conduct extensive outdoor training on Parris Island and our results indicate that some pest management strategies are needed year-round. Our data can help pest managers anticipate peak annual abundance of aggressive human biters and make plans for appropriate control measures during this timeframe.

Knowledge about diel activity patterns of biting flies on Parris Island will also be useful to develop effective pest management strategies. Scheduling nighttime training operations to begin at least two hours after sunset would ease

51

38

some biting fly intensity, although it is important to note that some mosquito vectors of WNV (e.g. Cx. quinquefasciatus) are active throughout the night.

When chemical adulticiding is used, applications between 1 hour prior and 2 hours after sunset should cause maximum mortality of the dominant species in the region. This timing would also reduce exposure of some nontarget insect species, such as honeybees that are active during the day.

Our results demonstrate that there are few calendar dates when biting fly activity is not a potential public health problem. The United States Marine Corps has year-round field training on Parris Island and, subsequently, recruits are more exposed to potential disease vectors and nuisance biting than the general population. Several mosquito species at Parris Island are known vectors of human pathogens. For example, the most abundant species were Ae. taeniorhynchus and Ae. sollicitans, which are severe biters and vector encephalitides (Crans 1977, Ortiz et al. 2003). Other abundant mosquitoes include species from which WNV isolations have been made and may be important vectors locally (e.g., Cx. salinarius, Cx. quinquefasciatus)

(Hayes et al. 2005, Anderson et al. 2007).

Potential public health risks can be examined by looking for patterns between peak pest insect numbers and disease outbreaks. While biting midge species on Parris Island are not vectors of human disease (Blanton and Wirth

1979), they may precipitate dermatitis cases (Haile et al. 1984). For example, the peak Culicoides numbers in 2002 coincided with an outbreak of methicillin-

51

39

resistant Staphylococcus aureus (MRSA) skin infections on Parris Island (August

– November 2002) (Zinderman et al. 2004). MRSA infections start with cuts on the skin and some recruits reported that their MRSA infections started with insect bites. This suggests that there may be an undocumented risk caused by biting flies to recruits on the base. Our data may assist public health organizations in nearby coastal urban areas (e.g. Hilton Head, SC, Charleston, SC, and

Savannah, GA) because the bionomics of biting flies in their salt marshes are likely to be similar to those on Parris Island.

ACKNOWLEDGEMENTS

We gratefully acknowledge the diligent assistance of Charles Pinckney and Cindy Zapotoczny of the Parris Island Natural Resource and Environmental

Affairs Office, who dutifully “put the bugs out” weekly. Joanna Lake provided the

Parris Island weather data. Elizabeth Hager of Beaufort County Mosquito

Control gave technical assistance on low country mosquito identifications. The authors would like to express their gratitude to the leadership at MCRD, Parris

Island for granting us access to the research sites.

REFERENCES CITED

Adler, P.H. and W. Willis. 2003. The history of arthropod-borne human diseases

in coastal South Carolina. Am. Entomol. 49:216-228.

51

40

Anderson J.F., A.J. Main, F.J. Ferrandino, and T.G. Andreadis. 2007. Nocturnal

activity of mosquitoes (Diptera: Culicidae) in a West Nile virus focus in

Connecticut. J. Med. Entomol. 44:1102-1108.

Barnard, D.R. and R.H. Jones. 1980. Diel and seasonal patterns of flight activity

of Ceratopogonidae in northeastern Colorado: Culicoides. Environ.

Entomol. 9:446-451.

Bidlingmayer, W.L. 1961. Field activity studies of adult Culicoides furens. Ann.

Entomol. Soc. Am. 54:149-156.

Blackwell, A., K.A. Evans, R.H.C. Strang, and M. Cole. 2004. Toward

development of neem-based repellents against the Scottish Highland

biting midge Culicoides impunctatus. Med. Vet. Entomol. 18:449–452.

Blanton, F.S. and W. Wirth. 1979. The sand flies (Culicoides) of Florida

(Ceratopogonidae: Diptera). Arthropods Fla. Neighb. Land Areas 10:1–

204.

Breidenbaugh, M. 1997. The morphology and biology of immature stages of

desert biting midges in the genus Culicoides (Diptera: Ceratopogonidae)

[MS thesis]. University of California, Riverside. 179 p.

Carroll, M.K. and J.A. Bourg. 1977. The night-time flight activity and relative

abundance of fifteen species of Louisiana mosquitoes. Mosq. News

37:661-664.

51

41

Cilek, J.E. and C.F. Hallmon. 2005. The effectiveness of the mosquito magnet®

trap for reducing biting midge (Diptera: Ceratopogonidae) populations in

coastal residential backyards. J. Am. Mosq. Control Assoc. 2I:218-221.

Connor, J.M. 1925. Colonial records of Spanish Florida: letters & reports of

governors and secular persons, vol. 1. Fla. Hist. Soc. Pub. No. 5.

Crans, W.J. 1977. A summation of studies pertaining to the migratory behavior of

Aedes sollicitans and their potential value to mosquito control. Proc. NJ

Mosq. Exterm. Assoc. 64: 56-38.

Darsie, R.F. and R.A. Ward. 1981. Identification and geographic distribution of

the mosquitoes of North America, north of Mexico. Mosq. Syst. Suppl. 1:1-

13.

Donohue, P. 2008. “Parris Island general ready to go to Iraq”. May 17, 2008.

Beaufort Gazette, Beaufort S.C.

Dorsey, C. 1947. Population and control studies of the Palau gnat on Peleliu,

Western Caroline Islands. J. Econ. Entomol. 40:805-814.

Esbarry, B.A. and W.J. Crans. 1977. The biting activity of Aedes sollicitans in

New Jersey. Mosq. News 37:721-724.

Evans, C.L. and W. Wills. 2002. Mosquitoes of Oconee County State Park in

Northwestern South Carolina. J. Entomol. Sci. 37:275-277.

Fonseca, D.M., N. Keyghobadi, C.A. Malcolm, C. Mehmet, F. Schaffner, M. Mogi,

R.C. Fleischer and R.C. Wilkerson. 2004. Emerging vectors in the Culex

pipiens complex. Science 303:1535-1538.

51

42

Foulk, J.D. 1969. Attack activity of two species of gnats in southern California.

Ann. Entomol. Soc. Am. 62:112-116.

Gladney, W.J. and F.C. Turner. 1970. Mosquito flight activity studies utilizing

timer-equipped light traps and malaise traps. Mosq. News 30:75-80.

Guimarães, A.E., C. Gentile, C.M. Lopes, and R. Pinto de Mello. 2000. Ecology

of mosquitoes (Diptera: Culicidae) in areas of Serra do Mar State Park, State

of São Paulo, Brazil. III – Daily biting rhythms and lunar cycle influence.

Mem. Inst. Oswaldo Cruz. 95:753-760.

Haile, D., D. Kline, J. Reinert, and T. Biery. 1984. Effects of aerial applications

of naled on Culicoides biting midges, mosquitoes and tabanids on Parris

Island, South Carolina. Mosq. News 44:178-183.

Hayes, E.B., N. Komar, R.S. Nasci, S.P. Montgomery, D.R. O’Leary, and G.L.

Campbell. 2005. Epidemiology and transmission dynamics of West Nile

virus disease. Emerg. Infec. Dis. 11:1167-1173.

Jamnback, H., W.J. Wall, and D.L. Collins. 1958. Control of Culicoides melleus

(Coq.) (Diptera: Heleidae) in small plots, with brief descriptions of the

larvae and pupae of two coastal Culicoides. Mosq. News 18: 64-70.

Jamnback, H. The Culicoides of New York state (Diptera: Ceratopogonidae).

1965. N.Y. State Mus. Sci. Serv., Albany Bull. 399: 1-154.

51

43

Janousek, T.E. and J.K. Olson. 2006. Seasonal variations in activity and size of

adult females and local distribution of larvae for populations of Culex

salinarius in the upper coastal zone of southeastern Texas. J. Am. Mosq.

Control Assoc. 22:47-53.

Khalaf, K.T. 1967. Seasonal incidence and population densities of Culicoides in

the coastal Areas of Louisiana (Diptera: Ceratopogonidae). J. Kansas

Entomol. Soc. 40:472-477.

Kettle, D. and J. Linley. 1969. The biting habits of some Jamaican Culicoides. I.

C. barbosai Wirth and Blanton. Bull. Entomol. Res. 58:729-753.

Kline, D.L. and R.C. Axtell. 1975. Culicoides melleus (Coq.) (Diptera:

Ceratopogonidae): seasonal abundance and emergence from sandy

intertidal habitats. Mosq. News 35:328-334.

Kline, D.L. and R.C. Axtell. 1976. Salt marsh Culicoides (Diptera:

Ceratopogonidae): Species, seasonal abundance and comparisons of

trapping methods. Mosq. News. 36:1-10.

Kline, D.L. and R.H. Roberts. 1981. Daily and seasonal abundance of Culicoides

spp. biting midges (Diptera: Ceratopogonidae) in selected mangrove

areas in Lee County, Florida. Fla. Entomol. 65:126-135.

Kogan, M. 1998. Integrated pest management: Historical perspectives and

contemporary developments. Annu. Rev. Entomol. 43:243-70.

51

44

Lewis, F.B. 1959. Abundance and seasonal distribution of the common species

of Ceratopogonidae (Diptera) occurring in the state of Connecticut. Can.

Ent. 91:15-28.

Lillie, T.H., D.L. Kline, and D.W. Hall. 1987. Diel and seasonal activity of

Culicoides spp. (Diptera: Ceratopongonidae) near Yankeetown, Florida,

monitored with a vehicle-mounted insect trap. J. Med. Entomol. 24:503-

511.

Linhares, A.X. and J.R. Anderson. 1990. The influence of temperature and

moonlight on flight activity of Culicoides variipennis in northern California.

Pan-Pac. Entomol. 66:199- 207.

Mulla, M.S. 1995. The future of insect growth regulators in vector control. J. Am.

Mosq. Control Assoc. 11:269-273.

Mullens, B.A. 1995. Fight activity and response to carbon dioxide of Culicoides

variipennis sonorensis (Diptera: Ceratopogonidae) in southern California.

J. Med. Entomol. 32:310-315.

Nayar, J.K. 1985. Bionomics and physiology of Aedes taeniorhynchus and

Aedes sollicitans, the salt marsh mosquitoes of Florida. JK Nayar [Ed].

Univerisity of Florida Agricultural Experiment Station, Gainesville, FL. Bull

852: 148 p.

51

45

Ortiz, D.I., A. Wozniak, M.W. Tolson, P.E. Turner, and D.R. Vaughan. 2003.

Isolation of EEE virus from Ochlerotatus taeniorhynchus and Culiseta

melanura in coastal South Carolina. J. Am. Mosq. Control Assoc. 19:33-

38.

Palmer, W.C. 1965. Meteorological drought. Research Paper No. 45. U.S.

Weather Bureau. NOAA Library and Information Services Division,

Washington, D.C..

Rueda, L.M. and R.C. Gardner. 2003. Composition and adult activity of salt-

marsh mosquitoes attracted to 1-octen-3-ol, carbon dioxide, and light in

Topsail Island, North Carolina. J. Am. Mosq. Control Assoc. 19:166-169.

Schmidtmann, E.T., J.F. Abend, and M.E. Valla. 1980. Nocturnal blood-feeding

from pastured calves by the ceratopogonid midge, Culicoides venustus, in

New York State. Mosq. News 40:571-576.

Service, M.W. 1971. Adult flight activities of some British Culicoides species. J.

Med. Entomol. 8:605-609.

Snow, W.E., E. Pickard, and J.B. Moore. 1957. The Heleidae of the Tennessee

River Basin. J. Tennessee Acad. Sci. 32:18-36.

Sudia, W.D. and R.W. Chamberlain. 1962. Battery-operated light trap, an

improved model. Mosq. News 22:126-129.

51

46

Strickman, D., M.E. Miller, H. Kim, and K. Lee. 2000. Mosquito surveillance in

the Demilitarized Zone, Republic of Korea, during an outbreak of

Plasmodium vivax malaria in 1996 and 1997. J. Am. Mosq. Control

Assoc. 16:100-113.

Trapido, H. 1947. DDT residual spray control of sandflies in Panama. J. Econ.

Entomol. 40:472-5.

WHO. 1989. Geographical distribution of arthropod borne diseases and their

principle vectors. WHO/VBC/89.967. Geneva, World Health

Organization.

WHO. 2003. Space spray application of insecticides for vector and public health

pest control: A practitioner’s guide. Geneva, World Health Organization.

Wirth, W.W., A.L. Dyce, and B.V. Peterson. 1985. An atlas of wing photographs,

with a summary of the numerical characters of the Nearctic species of

Culicoides (Diptera: Ceratopogonidae). Contrib. Am. Ent. Inst. 22:1-46.

Wirth, W.W., A.L. Dyce, and G.R. Spinelli. 1988. An atlas of wing photographs,

with a summary of the numerical characters of the Neotropical species of

Culicoides (Diptera: Ceratopogonidae). Contrib. Amer. Ent. Inst. 25:2-72.

Wright, R.E. and K.L. Knight. 1966. Effect of environmental factors on biting

activity of Aedes vexans (Meigen) and Aedes trivittatus (Coq.). Mosq.

News 26:565-578.

51

47

Zinderman, C.E., B. Conner, M.A. Malakooti, J.E. LaMar, A. Armstrong, and B.K.

Bohnker. 2004. Community-acquired Methicillin-resistant Staphylococcus

aureus. Emerg. Infect. Dis. 10:941-944.

51

Table 1. Abundance of adult biting midges (Ceratopogonidae) and mosquitoes (Culicidae) collected from

November 2001 to November 2004 at four trapping locations on the Marine Corps Recruit Depot, Parris Island.

Locations are Horse island (HI) Golf course (GC), Provo Marshal’s Office (PMO), Weapons (WEP).

HI GC PMO WEP % of total Ceratopogonidae Culicoides furens 209,611 100,281 16,986 87,758 86.2 C. hollensis 7,609 13,203 26,915 10,051 12.0 C. melleus 864 1,363 2,242 3,928 1.7 Total 218,084 114,847 46,143 101,737 100 Culicidae Aedes taeniorhynchus 18,165 15,915 4,883 157 42.7 Ae. sollicitans 756 1,053 809 21,494 26.3 Culex salinarius 1,955 8,148 811 3,374 15.6 Cx. quinquefasciatus 1,954 219 1,023 3,499 7.3 Ae. vexans 1,506 2,245 108 1,324 5.7 Ae. atlanticus 424 163 26 1,070 1.8 Anopheles bradleyi 79 64 17 118 0.3 Other Culicidae spp. 55 65 57 66 0.3 Total 24,894 27,872 7,734 31,102 100

48

49

1

2 Beaufort River 5

3

Broad River

Broad River

4

Trap locations and weather station:

1. Horse Island (HI) 2. Provo Marshal’s Office (PMO) 3. Weapons (WEP) 4. Golf Course (GC) 5. Water Treatment Facility

Fig. 1. The Marine Corps Recruit Depot on Parris Island, South Carolina,

showing the four trapping locations and the water treatment facility with

weather station.

48

10000

1000

100

10 Midges per trap night trap per Midges

1

Jul

Jun Oct

Jan Apr

Mar Nov

Feb Aug

May Sep Dec Month

Figure 2. Seasonal abundance of Culicoides furens at Parris Island. Daily numbers are an average of samples from November 2001- November 2004. Note counts are on a log scale.

50

1000

100

10 midges per trap night night trap per midges

1

Jul

Jun Oct

Jan Apr

Mar Nov

Feb Aug

May May Sep Dec Month

Figure 3. Seasonal abundance of Culicoides hollensis at Parris Island. Daily numbers are an average of samples from November 2001- November 2004. Note counts are on a log scale.

51

100

10 midges per trap night night trap per midges

1

Jul

Jun Oct

Jan Apr

Mar Nov

Feb Aug

May Sep Dec Month

Figure 4. Seasonal abundance of Culicoides melleus at Parris Island. Daily numbers are an average of samples from November 2001- November 2004. Note counts are on a log scale.

52

1000

100

10

Mosquitoes per trap night trap per Mosquitoes 1

Jul

Jun Oct

Jan Apr

Mar Nov

Feb Aug

May Sep Dec Month

Figure 5. Seasonal abundance of Aedes taeniorhynchus at Parris Island. Daily numbers are an average of samples from November 2001- November 2004. Note counts are on a log scale.

53

1000

100

10

mosquitoes per trap night trap per mosquitoes 1

Jul

Jun Oct

Jan Apr

Mar Nov

Feb Aug

May Sep Dec Month

Figure 6. Seasonal abundance of Aedes sollicitans at Parris Island. Daily numbers are an average of samples from November 2001- November 2004. Note counts are on a log scale.

54

1000

100

10

mosquitoes per trap night trap per mosquitoes 1

Jul

Jun Oct

Jan Apr

Mar Nov

Feb Aug

Sep Dec May Month

Figure 7. Seasonal abundance of Culex salinarius at Parris Island. Daily numbers are an average of samples from

November 2001- November 2004. Note counts are on a log scale.

55

1000

100

10

1

mosquitoes per trap night trap per mosquitoes

Jul

Jun Oct

Jan Apr

Mar Nov

Feb Aug

May Sep Dec Month

Figure 8. Seasonal abundance of Aedes vexans at Parris Island. Daily numbers are an average of samples from

November 2001- November 2004. Note counts are on a log scale.

56

1000

100

10

mosquitoes per trap night trap per mosquitoes 1

Jul

Jun Oct

Jan Apr

Mar Nov

Feb Aug

May Sep Dec Month

Figure 9. Seasonal abundance of Culex quinquefasciatus at Parris Island. Daily numbers are an average of samples from November 2001- November 2004. Note counts are on a log scale.

57

58

Figure 10. Correlations between Culicoides numbers and meteorological data.

A. Culicoides furens vs. air temperature B. C. hollensis vs. air temperature C.

C. melleus vs. air temperature D. Culicoides furens vs. rainfall E. C. hollensis vs. rainfall

59

5 A B

3 4

3 2 2 1 1

0 0

5 10 15 20 25 30 5 10 15 20 25 30 Temperature Temperature

5 D 2.5 C 4 2.0 3 1.5

2

(Log (Log 1) + 1.0

0.5 1

0.0 0 5 10 15 20 25 30 0 100 200 300 400 Temperature (°C)

Rainfall (cm) Number collected Number E 3

2

1

0 0 100 200 300 400 Rainfall (cm)

48

60

A 4 A

3

2

1

0

5 10 15 20 25 30 Temperature (°C)

4 B

3

2

(Log (Log 1) +

1

0

0 100 200 300 400

Rainfall (cm) Number collected Number 3.0 C

2.5

2.0

1.5

1.0

0.5

0.0

-4 -2 0 2 4 PDSI

B Figure 11. Correlations between Culicidae numbers and meteorological data. A.

Aedes taeniorhynchus vs. temperature B. Ae. taeniorhynchus vs. rainfall

C. Culex salinarius vs. Palmer's Drought Severity Index (PDSI).

(Log (Log 1) +

Number collected Number 61

% % ofcollected midges total

Time

Figure 12. Diel activity patterns of Culicoides furens, C. hollensis, and C. melleus from October 2003 to September 2004. Vertical dotted lines indicate the shift in sunset and sunrise times within the study period.

62

40 Aedes spp.

35 Culex spp. 30 25 20 15 10

% % of collected total 5 0

1400-1600 1600-1800 1800-2000 2000-2200 2200-2400 2400-0200 0200-0400 0400-0600 0600-0800 0800-1000 1000-1200 1200-1400 Time

Figure 13. Diel activity patterns of Aedes spp. and Culex spp. from October 2003 to September 2004. Vertical dotted lines indicate the shift in sunset and sunrise times within the study period.

CHAPTER III

CHARACTERIZATION OF A NEW ULTRA-LOW VOLUME FUSELAGE SPRAY CONFIGURATION ON AIR FORCE C-130H AIRCRAFT USED FOR ADULT MOSQUITO CONTROL1

ABSTRACT

The U.S. Air Force (USAF) tested a new fuselage boom configuration on the C-130H aircraft. We used into-the-wind and crosswind field trials to characterize a BVA oil droplet spectra produced by fuselage booms with flat-fan nozzles (8001, 8005) at the Air Force Range at Avon Park, FL. Across all trials, median droplet diameter (DV50) for 8001 and 8005 nozzles were 11.4 µm and

54.3 µm, respectively. For 8005 nozzles 22% of droplets collected were 7-25 µm size range while 75% of droplet from 8001 nozzles were <7 µm. Fuselage configuration parameters and field data were also used as input variables into the

AGDISP computer model to predict aerosol deposition and droplet fate. AGDISP predictions were compared with field data from crosswind tests and the model was found to fit reasonably well to empirical data. However, AGDISP predictions were better correlated with empirical findings for larger droplets than smaller droplets and for locations closer to the release point than further downwind.

1 Submitted to the US Army Medical Department Journal by M. Breidenbaugh, K. Haagsma, and F.A. de Szalay.

63

64

Overall, these field trials indicated that this new fuselage boom configuration creates effective droplet sizes and swath widths (i.e., 610 m) appropriate for military aerial mosquito control operations.

INTRODUCTION

Ultra-low volume (ULV) mosquito spraying has become increasingly technologically advanced since it was first introduced in the 1960’s as a replacement to thermal fogging (Lofgren 1972). Examples of technological advancements are new nozzles designs to produce optimal sized droplets and computer modeling of droplet fate under a variety of metrological conditions.

These technologies help reduce spray volume and are important tools to minimize the use of pesticide.

Aerial ULV sprays are the primary method used to interrupt insect-borne epidemics (Parrish 1962, Fox 1980, Gubler et al. 2000). For example, the U.S.

Air Force (USAF) used aircraft to control mosquito outbreaks after major

Hurricanes, and other public agencies have used aerial ULV sprays to decrease disease transmission and control nuisance mosquitoes (Breidenbaugh and

Haagsma 2008, Carney et al. 2008, Lothrop et al. 2008). ULV applications have been used outside the US to interrupt malaria transmission (Eliason et al. 1975).

A crucial element of ULV technology is to create an effective aerosol cloud that will optimize target pest mortality while reducing pesticide use and non-target species mortality (Brown et al. 2003, Dukes et al. 2004). Maximum efficacy is achieved by dispersing uniform droplets of the correct size. For mosquitoes, the

48

65

optimum droplet size is 7-25 µm in diameter (Mount 1970, Weidhaas et al. 1970,

Brown et al. 2006). However, many aircraft spray systems create a spectrum of droplet sizes between 1-150 µm (Barber et al. 2004). Extremely small droplets may not be lethal and large droplets (>50 um) can settle out too quickly and are less likely to contact flying mosquitoes. Larger droplets are also wasteful because they contain more toxin than needed to kill the pest (Haile et al. 1982).

Large droplets also represent a potential hazard to nontarget pests and thus create unfavorable environment effects (Zhong et al. 2003).

The USAF typically uses a modular aerial spray system (MASS) on the C-

130H aircraft with ULV flat-fan nozzles installed under the aircraft wing (Burkett et al. 1996). The wing boom configuration requires pressurized tubing installed along the length of the wing, which results in residual pesticide waste when the equipment is cleaned after spray operations. Furthermore, the tubing is located in the interior of the wing, which makes installation and repairs more difficult. The

USAF has recently modified the MASS on the C-130H aircraft to use fuselage- mounted spray booms. The fuselage booms do not require axillary equipment for installation and reduce pesticide waste because the pressurized tubing is shorter.

The fate of droplets released from aircraft is strongly affected by aircraft generated turbulence (Mickle 1996). However, it is not known how aircraft vortices affect droplet size and drift when a fuselage boom configuration is used with the C-130. Effective aerial mosquito control operations also need to predict

66

the drift of the aerosol swath. Although keeping pesticide droplets aloft for a longer time is advantageous to increase the chance that droplets contact the target pest, it also makes it more difficult to predict swath drift patterns. The U.S.

Forest Service has developed computer models to predict pesticide drift for agricultural applications (i.e., crop dusting) (Teske 1996). An earlier version, the

Forest Service Cramer Barry Grimm model, has been integrated into the

Agricultural Dispersal model (AGDISP) which can model the lower volumes and smaller droplets used in ULV mosquito adulticide sprays (Teske et al. 1998).

Standard parameters for over 173 spray aircraft are available in the AGDISP model’s library, including the Lockheed C-130H with wing booms and flat-fan nozzles. However, parameters for the new fuselage spray boom configuration are not available. Computer model predictions validated by field testing are helpful to estimate the effects of changing parameters (e.g., altitude, nozzle type, etc.) on droplet drift. While the model has been well-validated for agricultural applications (Bird et al. 2002), few field trials have been conducted to confirm its predictive ability for the small droplets used in ULV mosquito sprays.

This study reports on the characterization of droplet size and drift generated by USAF C-130H aircraft with the newly developed ULV fuselage boom configuration. Droplet size and downwind dispersion was measured using into-the-wind and crosswind field trials, respectively. The resulting field data was then input into the AGDISP computer model to compare the model’s predictions with the empirical field results.

67

METHODS

Study site description

Testing of the fuselage booms was performed from December 2004 to

February 2005 on the Avon Park Air Force Range (APAFR), a ~42,000 ha facility in Highland and Polk counties, Florida. The field site was chosen to minimize disruption of drifting droplets by vegetation (Barber et al. 2008), and vegetation was dominated by shrubs or open woodlands (Fig. 1). Primary roads on the

APAFR were aligned along cardinal directions, which facilitated re-alignment of sampling stations when wind direction changed between different test dates.

Field characterization of droplet size

Applications were made by USAF C-130H aircraft with fuselage mounted booms that were directed towards the ground. The aircraft was equipped with a

Satloc GPS Agricultural Navigation System (Hemisphere GPS, Calgary, Canada) to record aircraft position and time when the spray system was turned on.

Fuselage boom configurations were tested with flat-fan TeeJet® nozzles

(Spraying Systems Co., Wheaton, IL) sizes 8001 and 8005, which were rated by the manufacturer to deliver 0.4 liters/min and 1.9 liters/min per nozzle, respectively.

The aircraft flew at 370.4 km/hr and spray sprays continued from 30 seconds prior to reaching the sampling transect until 30 seconds after passing

68

the transect (60 seconds total). Wind speed, direction, air temperature, and humidity were recorded 2.5 m above ground surface using a Swath Kit Weather

Station (Droplet Technologies, College Station, PA) and at spray altitude using the aircraft’s Self-Contained Navigation System.

The MASS delivered an application rate of 45.3 ml/ha that was based on standard operational practices of public health agencies using a common mosquito control pesticide, Anvil® 10+10 (Clarke® Mosquito Control Products,

Inc. Roselle, IL; hereafter Anvil). We conducted multiple trials in the same location, which would have caused excessive pesticide accumulation. Therefore, our sprays used only the pesticide carrier, BVA spray oil 13 (BVA Inc., Wixom,

MI), without the pesticide.

On 4, 6, 8 December 2004 and 15 February 2005, the aircraft flew directly into the prevailing wind (into-the-wind trials) at 46 m above ground. To quantify the droplet spectra, 9 sampling stations were positioned every 61 m along a

488 m transect perpendicular to the prevailing wind (Fig. 1). The aircraft flew directly over the center station. Slide rotator devices (John Hock Company,

Gainesville, FL) held spinning Teflon® coated glass microscope slides (25 X 75 mm, ~420 rpm) that collected the droplet cloud as it passed over the station.

Microscope slides were collected 30 minutes after the aircraft passed to allow enough time for airborne droplets to drift through the transect. Droplet data from all stations were pooled to determine the characteristics of the droplet spectra.

69

In addition to characterizing the droplet size spectra with “into-the-wind” tests, crosswind trials were used to compare actual drift in field trials with predictions by the AGDISP model. On 16 February 2005, we conducted a crosswind trial at a release height of 46 m above ground, which is standard for the USAF mosquito control operations. Although we ran two trials, a wind shift during the second trial disrupted the application and the data was lost. Also, we only used the 8005 nozzles for the crosswind trials because 8001 nozzles clogged repeatedly during the into-the-wind trials. On 17 February 2005, two trials measured spray drift released at 91 m, which is an altitude proposed for potential nighttime mosquito control operations with the C-130 aircraft. In all crosswind trials, droplets were collected with rotating microscope slides at 8 stations arranged along the prevailing wind direction. On 16 February 2005, stations were set 154 m apart (154-1,219 m) on a transect downwind from the release point (Fig. 1). On 17 February 2005, we spread the stations 457 m apart

(457-3,658 m) along the transect in anticipation of greater drift from the higher release altitude.

All slides were processed within 6 hours after the trial was conducted.

Droplets on slides were measured under a compound microscope equipped with a reticule. A total of 100 droplets were measured or the entire slide was scanned, whichever came first. Measured drop diameter was converted to airborne drop diameter with a 0.59 correction factor (Anvil 10+10 Resource guide) that accounted for the spread of droplets when they impacted the glass

70

slide (Anderson and Schulte 1971). These data were used to determine volume median diameter (DV50) and droplet density at each sampling station after

Yeomans (1949). The DV50 represents the droplet size which divides the droplet spectrum in half, or in other words, where 50% of the droplet spectra is contained in droplets smaller than the DV50. Also of interest are the DV10 and DV90 values which are the points in the droplet distribution where 10% and 90% of the spray is in drops smaller than this size (Dukes et al. 2004).

AGDISP computer model

Droplet size data obtained from the into-the-wind field trials with 8005 nozzles were input into the AGDISP computer model (version 8.08, USDA Forest

Service) to predict crosswind droplet trajectories and deposition. Operational parameters and meteorological conditions recorded during the trials at Avon Park were used as input values for the AGDISP model (Table 1). While the AGDISP model has a library of aircraft, nozzle types, and placements, the C-130 fuselage configuration is not included so we used values measured in our field trials (e.g., boom placement, nozzle position, and DV10, DV50, and DV90 droplet sizes) to test the model’s accuracy. Predictions for downwind deposition of BVA oil released from spray altitudes of 46 m and 91 m were modeled. Output values regarding droplet deposition and trajectories for droplets sized 22.3, 54.3, and 104.7 µm were plotted. Predictions of droplet trajectories made by AGDISP were then compared to empirical data derived from crosswind field trials.

71

RESULTS

Flight parameters and meteorological conditions during the trials are given in Table 2. Meteorological conditions were acceptable during the into-the wind and crosswind trials. Humidity and temperature were typical for ULV mosquito control operations. Boom pressure was relatively constant during the into-the- wind trials (621 – 641 KPa) in December, but was slightly lower during the

February trials (472 -486 KPa). Wind speeds were within acceptable ranges

(1.6–6.4 km/hr) during the into-the-wind trials. However, lower wind speeds increased variability in drift (personal observation) and may have reduced the accuracy of the data.

Over 7,000 droplets were measured in samples collected during the into- the-wind and crosswind field trials. The two nozzle types produced different cumulative volume curves (Fig. 2). Average volume median diameter (DV50) was

11.4 µm (SE ± 1.0) for 8001 flat-fan nozzles and 54.3 µm (SE ± 2.2) for 8005 nozzles. Droplets from 8001 nozzles produced a narrow range of relatively small droplets: DV10 was 1.7 (SE ± 0.6) µm and DV90 was 30.8 µm (SE ± 1.6). The larger orifices of the 8005 nozzles delivered a wider range of droplet sizes: DV10 was 22.3 µm (SE ± 2.1), and DV90 was 104.7 µm (SE ± 0.6). Thus, the cumulative volume curve of 8005 nozzles generated a lower slope than the 8001 nozzles.

The distribution of drop sizes in spray clouds produced during the into-the- wind trials was also different between the 8001 and 8005 nozzles. The 8001

72

nozzles produced smaller droplets, with 40% of drops in the smallest class size and 75% were <7 µm (Fig. 3). In contrast, 8005 nozzles generated a wide range of size classes with 1.2% of drops <7 µm and 22% of droplets in the 7-25 µm size range (Fig. 3).

On 16 February, the crosswind trial with a spray release height of 46 m had droplet sizes ranging from 42.5 µm to 10.1 µm (Table 3). The swath drifted much further when the spray was released at 91 m above ground (Fig.4). In both trials the largest droplets were deposited at the first collection station and the mean droplet size decreased downwind. There was some variability between trials on 17 February, and droplets drifted further during the second trial due to higher wind speeds (7.5 km/hr).

AGDISP computer model predictions and field-trial results

Droplet spectra data for 8005 nozzles and meteorological data from into- the-wind field trials were input into the AGDISP model to predict droplet fate

(Table 1). The model predicted heavy deposition near the aircraft when spraying at 46 m above ground with the fuselage booms (Fig. 5). Deposition was predicted to reach a maximum of 17.1 ml/ha at 10 m from the release point and drop to 4.9 ml/ha at 182 m downwind. There was a second smaller peak of

7.3 ml/ha at 306 m before deposition dropped off again to <1 ml/ha at 1,582 m downwind. Average deposition over 1,582 m was predicted to be 3.5 ml/ha.

73

Between the release point and the standard operational swath width of the USAF

C-130 (610 m) the model predicted an average deposition of 7.0 ml/ha.

When release height was increased to 91 m, the predicted deposition was more spread out (Fig. 5). Deposition was low (< 2.0 ml/ha) until 270 m downwind of the release point, when deposition increased rapidly peaking at

5.2 ml/ha at 394 m. Between 290 m and 600 m deposition decreased and rose again, after which it gradually tapered off. Average deposition over the model’s predictive range of 22 -1,582 m was 2.6 ml/ha.

When we modeled droplets sprayed from the four 8005 nozzles on C-130 fuselage booms, the AGDISP model predicted different trajectories for small (set at 22.3 µm, which is our DV10) medium (54.3 µm, our DV50) and large (104.7 µm, our DV90) droplets and for sprays released at 46 m or 91 m above ground.

Overall, AGDISP predictions followed standard ballistics such that smaller droplets released at greater altitudes drifted further than larger droplets released at lower heights. However, the model also predicted droplet trajectories would be affected by aircraft vortices. For example, all droplets released on the windward side at 46 m were affected by aircraft induced vortices for ~125 m downwind, which first lifted them and then allowed them to drift downwards (Fig.

6). In contrast, droplets released on the leeward side of the aircraft were entrained in downward vortices that pushed them toward the ground until the vortices broke apart. The model predicted that most small droplets reached the ground ~450 m downwind of the release point, and most medium droplets

74

reached the ground by 200 m. The large droplets also fit the previous pattern but reached the ground faster (~100 m downwind) than the other size classes (Fig.

6).

When the droplet trajectories were modeled from a release height of 91 m, trajectories were different than when released at 46 m (Fig. 7). Overall, an increase in spiraling caused by vortices was predicted in all drop sizes. Most notable was that droplets from the leeward side of the aircraft were not pushed to the ground but were either brought back in contact with droplets from the windward side (small droplets) or crossed paths with these droplets’ trajectories

(medium or large droplets) (Fig. 7).

DISCUSSION

These trials characterized droplet spectra and dispersal to evaluate if the

USAF C-130H fuselage boom configuration is effective for adult mosquito control operations. Under optimal conditions, aerial spray operations produce droplets that adhere to mosquitoes but do not drift beyond the intended spray area (Latta

1947). We found that both flat-fan nozzles (8001, 8005) produced droplets within the optimal range for mosquito control (7-25 µm), but they had very different droplet size distributions.

In general, droplet distributions from flat-fan nozzles show that the largest numbers of drops are found in the smallest size classes while the greatest volume is found in the relatively scarce but larger size classes (Ekblad and Barry

1983). In this study, the 8001 nozzles produced a narrow spectrum of smaller

75

droplets, and consequently, the majority of the volume sprayed was comprised of small droplets (i.e., DV90 = 30.8 µm). Although most droplets were within the most effective size range (7-25 µm), droplets <7 µm comprised 35% of spray volume. In contrast, 8005 nozzles produced larger drops but a more even size class distribution (Fig. 3).

We tested the fuselage boom configuration for its potential usefulness for

ULV adult mosquito control with the C-130 aircraft. The narrow droplet spectra and ideal DV50 of the 8001 nozzles would appear to make these a better choice for mosquito adulticiding. However, these nozzles produce many droplets considered too small for effective mosquito control. In addition, the small orifice size required 5 times more nozzles than 8005s to produce the desired flow-rate, and they often became clogged during field trials. The 8005 nozzles produced relatively large drops which equates to additional chemical waste if they deposit without contacting the target pest. Characterization tests with intermediate nozzle sizes (e.g., 8002, 8003) would be useful to further examine the efficacy of fuselage booms for mosquito control.

We used rotating microscope slides to measure droplet sizes for 2 flat-fan nozzles, which is a widely used method to characterize droplet size (Carroll and

Bourg 1979, Brown et al. 1993, Meisch et al. 2005, Lothrop et al. 2007). A more precise method to determine droplet size is a wind tunnel equipped with laser- diffraction equipment. Recent wind tunnel studies with 8001 nozzles at wind speeds of 225 km/hr produced a DV50 of 55.2 µm, which is nearly 5-times larger

76

than in our study (Hornby et al. 2006). Volume median diameter for 8005 nozzles at wind speeds of 225 km/hr were 87.68 µm, which were also larger than in our results. However Hornby et al.'s (2006) data suggests that increased wind shear at faster speeds can create smaller droplets. Since the C-130 aircraft flew at 370 km/hr, this might have caused the smaller droplet size we measured with the 8005 nozzles. However, wind shear alone probably cannot account for the very small DV50 that we measured with the 8001 nozzles. A possible explanation is that minute oil droplets from other sources (e.g., engine exhaust) were trapped on the microscope slides and were counted as sprayed material. This would be more likely to affect the data from the 8001 nozzles because they made smaller droplets than the 8005 nozzles. Adding florescent dye to the spray material (e.g.

Barber et al. 2008) might be help distinguish been sprayed material and environmental contaminants in future studies. Also, we may have undercounted the largest possible droplets because these would be rare and might have dropped out of the air column before they impacted our sampling stations.

Making accurate predictions of pesticide droplet dispersal is desirable

(Teske 1996) and, therefore, we compared AGDISP model predictions to our downwind field data. The overall AGDISP predictions and empirical data followed a similar and expected pattern of larger drops falling first and smaller drops drifting further. For example, in crosswind field trials with a 46 m release height, the average droplet size measured at the first collection station (154 m) was 42.5 µm (Table 3), while AGDISP predicted that 42.5 µm droplets would

77

reach the ground around 220 m downwind (unpublished data). Taking into consideration the influence that wind speed and direction can have on these medium-sized drops, the disparity of 66 m between the model and the empirical data may be considered reasonably similar at this level of resolution. AGDISP predictions regarding trajectories of large droplets were also fairly closely confirmed by field trials. AGDISP predicted 104.7 µm droplets to reach ground level 100 m downwind (Fig. 6) and many droplets of this size were collected at the 150 m collection station although droplets from the leeward side are not predicted to reach the ground until 280 m.

However, we found that some AGDISP predictions were different than observed in our field data. For instance, a greater disparity existed between

AGDISP predicted trajectories and field data for droplets released from 90 m.

The AGDISP model could not predict droplet fate past 400 m, but many droplets were still airborne at this distance. The AGDISP model also predicted further drift than we observed in the crosswind trials. For example, the large droplets

(104.7 µm) were predicted to be aloft at 390 m (Fig.7). If the trajectory path of a large droplet is extrapolated from the predicted drop, these would fall to the ground ~500 m downwind from release. However, the largest sized droplet we collected at the 500 m sampling station during the field trials was 90 µm, and the average size was less than 13 µm (Fig. 4).

Our comparisons between predicted and empirical data indicate that the

AGDISP model is more accurate at assessing the environmental fate of larger

78

droplets and their movement closer to their release points. In general, the

AGDISP predictions made the most accurate predictions of droplet fate at 46 m release height but was less accurate for sprays released at 91 m above ground.

Fuselage sprays from 8005 nozzles produced wide pesticide swaths that suggest they would be appropriate for military aerial ULV operations where a minimum 600 m swath width is required (Breidenbaugh et al. 2008). Additional operational evaluations will be necessary to determine effective swath width for

C-130 fuselage booms using sentinel mosquito mortality and various pesticides.

A potential reason for inaccuracies in AGDISP models is that the model is limited to using average weather conditions as input values. Obviously, modest but significant changes in meteorological conditions (e.g., wind direction) could have major effects on droplet fate. Modeling ULV sprays at high altitude sprays is also difficult because the model does not calculate downwind drift past 3,600 seconds. This is an artifact from the model’s origins in depositional spraying

(Teske 1996) and, subsequently, the development of better algorithms that accurately incorporate field conditions would improve the AGDISP model.

In conclusion, the fuselage boom configuration we tested would be desirable for use in military operations because setup and maintenance is simple and it produces less pesticide waste than the wing boom configuration. Our trials suggest that 8001 and 8005 flat-fan nozzles produce a droplet spectra and swath dispersal that would be effective for ULV mosquito control operations. However, we also found that small changes in wind speed and direction substantially affect

79

droplet dispersion and deposition behavior. Therefore, continual monitoring of current meteorological conditions should be an ongoing consideration of ULV mosquito control operations. Increased wind speed and directional variability also make it difficult to predict insecticide dispersal characteristics with the currently available AGDISP modeling software.

REFERENCES CITED:

Anderson CH, W Schulte. 1971. Teflon as a surface for deposition on aerosol

droplets. Mosq News 31:499-504.

Barber JAS, Greer M, Hewitt A. 2004. A field measurement device for aerosols

used in mosquito control. ASAE Paper No. AA04-0010. St. Joseph, Mich.:

ASAE.

Barber JAS, Greer M, Latham M, Stout G. 2008. Canopy effects droplet size

distribution and meteorological change. J Am Mosq Control Assoc

24:177-181.

Bird SL, Perry SG, Ray SL, Teske ME. 2002. Evaluation of the AgDISP aerial

spray algorithms in the AgDRIFT model. Environ Toxicol Chem 21: 672–

681.

Breidenbaugh M, Haagsma K. 2008. The US Air Force Aerial Spray Unit:

A history of large area disease vector control operations, WWII through

Katrina. Army Med Dept J. April-June 2008:54-61.

80

Breidenbaugh M, Haagsma K, Walker W, Sanders, D. 2008. Post-Hurricane

Rita mosquito surveillance and the efficacy of Air Force aerial applications

for mosquito control in east Texas. J Am Mosq Control Assoc 24:327-330.

Brown JR, Mickle RE, Yates M, Zhai J. 2003. Optimizing an aerial spray for

mosquito control. J Am Mosq Control Assoc 19:243-250.

Brown JR, Dukes JC, Beidler EJ, Chew V, Ruff J. 1993. A comparison of Teflon

slides and the Army insecticide measuring system for sampling aerosol

clouds. J Am Mosq Control Assoc 9:32-35.

Brown JR, Rutledge CR, Reynolds W, Dame DA. 2006. Impact of low aerial

application rates of Dibrom 14 on potential vectors. J Am Mosq Control

Assoc 22:87-92.

Burkett DA, TL Biery, DG Haile. 1996. An operational perspective on measuring

aerosol cloud dynamics. J Am Mosq Control Assoc 12:380-383.

Carney RM, Husted S, Jean C, Glaser C, Kramer V. 2008. Efficacy of aerial

spraying of mosquito adulticide in reducing incidence of West Nile virus,

California, 2005. Emerg Infect Dis 14:747-754.

Carroll MK, Bourg JA. 1977. Methods of ULV droplet sampling and analysis:

effects on the size and spectrum of the droplets collected. Mosq News

39:645-656.

Dukes JC, Zhong HE, Greer M, Hester PG, Hogan D, Barber JS. 2004. A

81

comparison of two ultra-low volume spray nozzle systems by using a

multiple swath scenario for the aerial application of fenthion against caged

mosquitoes. J Am Mosq Control Assoc 20:36-44.

Ekblad RB, Barry JW. 1983. A review of progress in technology of aerial

application of pesticides. USDA-Forest Service, Equipment and

Development Center, Missoula, Montana.

Eliason DA, Joseph VR, Karam JS. 1975. A prospective study of the effects of

aerial ultralow volume (ULV) application of malathion on epidemic

Plasmodium falciparum malaria. I. Study design and perspective. Am J

Trop Med Hyg 24: 183-187.

Fox I. 1980. Evaluation of ultra-low volume aerial and ground applications of

malathion against natural populations of Aedes aegypti in Puerto Rico.

Mosq News 40:280-283.

Gubler DJ, Campbell GL, Nasci R, Komar N, Petersen L, Roehrig JT. 2000. West

Nile virus in the United States: guidelines for detection, prevention, and

control. Viral Irnmunol 13:469-475.

Haile DG, Mount GA, Pierce NW. 1982. Effect of droplet size of malathion

aerosols on kill of caged adult mosquitoes. Mosq News 42:576-583.

82

Hornby JA, Robinson J, Opp W, Sterling M. 2006. Laser-diffraction

Characterization of flat-fan nozzles used to develop aerosol clouds of

aerially applied mosquito adulticides. J Am Mosq Control Assoc 22:702-

706.

Latta R, Anderson LD, Rogers EE, LaMer VK, Hochberg S, Lauterbach H,

Johnson I. 1947. The effect of particle size and velocity of movement of

DDT aerosols in a wind tunnel on the mortality of mosquitoes. J Wash

Acad Sci 37:397-407.

Lofgren CS. 1972. Ultralow Volume (ULV) application of insecticides. Am J

Trop Med Hyg 21:819-824.

Lothrop HD, Huang HZ, Lothrop BB, Gee S, Gomsi OE, Reisen WK. 2007.

Deposition of pyrethrins and piperonyl butoxide following aerial ultra-low

volume applications in the Coachella Valley, California. J Am Mosq

Control Assoc 23:213-219.

Lothrop HD, Lothrop BB, Gomsi DE, Reisen WK. 2008. Intensive early season

adulticide applications decrease arbovirus transmission throughout the

Coachella Valley, Riverside County, California. Vector-Borne Zoonot 8:

475-490.

Meisch MV, Dame DA, Brown JR. 2005. Aerial ultra-low-volume assessment of

Anvil 10+ 10® against Anopheles quadrimaculatus. J Am Mosq Control

Assoc 21:301-304.

83

Mickle RE. 1996. Influence of aircraft vortices on spray cloud behavior. J Am

Mosq Control Assoc 12:372-379.

Mount GA. 1970. Optimum droplet size for adult mosquito control with space

sprays or aerosols of insecticides. Mosq News 30:70-75.

Parrish DW, Hodapp CJ. 1962. Biological evaluation of the C-47 aerial spray

system for adult mosquito control. Mosq News 22:36-37.

Teske ME. 1996. An introduction to aerial spray modeling with FSCBG. J Am

Mosq Control Assoc 12:353-358.

Teske ME, Thistle HW, Eav B. 1998. New ways to predict aerial spray

deposition and drift. J For 96:25-31.

Weidhaas DE, Bowman MC, Mount GA, Lofgren CS, Ford HR. 1970.

Relationship of minimum lethal dose to the optimum size of insecticides

for mosquito control. Mosq News 30:195-200.

Yeomans AH. 1949. Directions for determining particle size of aerosols and fine

sprays. USDA Bur Ent and Plant Quar ET-267:1-7.

Zhong H, Dukes J, Greer M, Hester P, Shirley M, Anderson B. 2003. Ground

deposition impact of aerially applied fenthion on the fiddler crabs, Uca

pugilator. J Am Mosq Control Assoc 19:47-52.

Wind direction (into the wind) Wind direction (crosswind)

Transect line

Aircraft path (into the wind) Aircraft path (crosswind) 100 meters

Fig. 1. View of field trial location at Avon Park Air Force Range, Florida.

84

1

0.9 8005 trial 1

0.8 8005 trial 2

8001 trial 1

.

0.7 8001 trial 2 0.6

DV50 0.5

0.4

0.3

Cumulative volume fraction fraction volume Cumulative 0.2

0.1

0 1 16 31 46 60 75 90 105 119 134 149 Droplet diameter (µm) Fig. 2. Drop size distribution for droplet spectra produced by 8001 and 8005 flat-fan nozzles used with C-130H

Air Force fuselage boom characterization trials.

85

Fig. 3. Distribution of drop sizes in spray cloud from A) 8001 nozzles; B) 8005 nozzles.

86

Fig. 4. Average droplet size recovered from USAF C-130 fuselage sprays using 8005 flat-fan nozzles (91 m release height). Bars are ±1 standard error. 17 February 2005, Avon Park, FL.

87

88

A

B

Fig. 5. Predicted deposition from AGDISP model of droplets sprayed with C-130 fuselage booms released at (A) 46 m and (B) 91 m above ground.

89

Fig. 6. AGDISP model predicted trajectory of an A) 22.3 µm droplet (DV10 );

B) 54.3 µm droplet (DV50); and C) 104.7 µm droplet (DV90) from 46 m release height by USAF C-130. W and L are windward and leeward sides of aircraft.

90

A W

L

B W

(m) Altitude L

C W

L

Distance from aircraft (m)

91

Fig. 7. AGDISP model predicted trajectory of an A) 22.3 µm droplet (DV10 );

B) 54.3 µm droplet (DV50); and C) 104.7 µm droplet (DV90) from 91 m release height by USAF C-130. W and L are windward and leeward sides of aircraft.

92

A W

L

B W

(m)

Altitude L

C W

L

Distance from aircraft (m)

Table 1. Parameters input into the AGDISP model to predict droplet fate from fuselage boom spray applications.

Drop size distribution used field collected data.

Aircraft = Lockheed C-130H, weight 63,047 kg; speed = 271 km/h Release height = 46 m, 91 m; flight lines = 1 Nozzles = 8, 8005 flat-fan, positioned at 3.61, 3.71, 3.81, 3.91 m from aircraft centerline, both sides Drop size distribution = DV10 = 22.3 µm; DV50 = 54.3 µm; DV90 = 104.7 µm Material = BVA oil (specific gravity = 0.85; nonvolatile fractions = 1; active = 1; rate = 45.3 ml/ha Swath width = 152 m (maximum allowed), Swath displacement = 0 Wind speed = 6.4 or 7.5 km at 2 m above ground level (speed depended on trial modeled) Temperature and relative humidity = 17.7 °C and 81% Stability = overcast (sunset to 1 h after sunrise, stable) Canopy = none Surface roughness = 0.0075 m

93

Table 2. Parameters during BVA-13 spray-oil field characterization trials of USAF ULV fuselage boom configurations December 2004 – February 2005 at Avon Park Range, FL.

Average wind Nozzle Application Boom Date of Flow rate speed km/hr configuration Relative rate Pressure Temperature trial Liters/min at Size Humidity ml/hectare kPa ground/altitude (number)

4 Dec 041 45.3 8.5 634 3.2/13.7 8005 (8) 16.1°C 51%

6 Dec 041 45.3 8.5 641 3.2/17.7 8005 (8) 25.6°C 59%

8 Dec 041 45.3 8.5 621 6.4/12.9 8001 (40) 24.4°C 79%

15 Feb 051 45.3 8.5 483 1.6/6.4 8001 (40) 20.6°C 70%

16 Feb 052 45.3 8.5 486 6.4/13.0 8005 (8) 18.3°C 70%

17 Feb 052 45.3 8.5 472 7.5/12.1 8005 (8) 17.1°C 92%

1Into-the-wind trials

2Crosswind trials

94

Table 3. Average droplet size (µm) collected at stations downwind of release point. C-130 fuselage configuration used 8005 nozzles in a 7.5 km/hr crosswind and BVA oil from 46 m release height.

Distance downwind 154 m 305 m 457 m 610 m 762 m 914 m 1,066 m 1,219 m Mean ±SE 42.5 ±2.5 24.9 ±2.6 13.9±2.8 7.8 ±0.5 8.4±1.2 5.7 ±0.8 6.0±0.3 6.2 ±0.4

95

CHAPTER IV

EFFICACY OF AERIAL SPRAY APPLICATIONS WITH FUSELAGE BOOMS ON AIR FORCE C-130 AIRCRAFT AGAINST MOSQUITOES (CULICIDAE) AND BITING MIDGES (CERATOPOGONIDAE)1

ABSTRACT

We tested the effectiveness of a new fuselage boom configuration on a

US Air Force C-130 aircraft to create aerial ultra-low volume (ULV) sprays to control mosquitoes (Culicidae) and biting midges (Ceratopogonidae). We measured mosquito mortality when using two public health insecticides, an organophosphate (Dibrom) and a pyrethroid (Anvil 10+10). We observed 100% mosquito mortality 639 m downwind in single pass trials using bioassay cages and Dibrom. The volume median diameter of droplets collected from sprays was

44 µm at 213 m and decreased in size to 11 µm at 2,130 m downwind of the release point. Droplet density ranged from a maximum of 18.4 drops/cm2 at

213m to 2 drops/cm2 at 2,130m. In wide-area applications of Dibrom, we measured an 86% reduction of Culicoides spp. and an 83% reduction in mosquitoes from natural populations. The efficacy of large scale applications of

Anvil was less clear due to naturally decreasing mosquito and midge densities.

We found mortality averaged 37% in caged mosquitoes at the maximum Anvil label rate. Results indicate that the new fuselage boom configuration is effective for ULV applications with Dibrom to control mosquito and biting midges, but further testing is needed with Anvil.

1 To be submitted to the Journal of American Mosquito Control Association by M.Breidenbaugh, K. Haagsma, G. Wojick, and F. de Szalay.

96

97

INTRODUCTION

Mosquito control via aerial applications of pesticides is the primary approach to disrupt insect disease transmission across large areas (Parrish

1962, Moore et al. 1993, Service 2000, WHO 2003, Carney et al. 2008). The US

Air Force (USAF) developed a modular aerial spray system (MASS) for use with the C-130H aircraft, which can be loaded on the airframe in less than one hour.

The MASS has a 7,571 L capacity and is available for worldwide deployment for control of insects of medical importance (Breidenbaugh and Haagsma 2008).

Historically, this system used a wing-mounted ULV configuration (Burkett et al.

1996). A new fuselage boom configuration was recently developed that has significant benefits over the wing-mounted booms. First, this new configuration is more easily deployable since it does not require additional aircraft ground equipment for installation. Second, the shorter delivery lines result in reduced chemical waste and maintenance costs. Preliminary tests indicated droplet sizes and swath widths produced by the fuselage configuration are acceptable for use by the USAF for aerial mosquito and biting midge control (Breidenbaugh et al., in press). However, the system has not been field tested with pesticides for use in controlling biting fly populations such as mosquitoes and biting midges.

There are currently only two major classes of insecticides approved for aerial use in controlling adult mosquitoes: pyrethroids and organophosphates

(OPs) (Wata 1989, Rose 2001). Public concerns over the potential negative effects of pesticides on human health and the environment have eliminated the use of many older compounds (e.g., DDT, carbamates) (Pedigo 1989, Wardman

98 and Thomas 2000, Davis et al. 2007). At the same time, development of new pesticides has become a lengthy and expensive process (Rose 2001). Public health pesticides represent a small portion of the pesticide market and therefore, few companies have financial incentive to develop new materials. Thus, it is important to test a variety of spray configurations using existing compounds for control of mosquitoes and other biting flies.

Pyrethroids and OPs are both contact poisons that affect the ability of a nerve to transmit impulses. The available OPs (e.g., naled & malathion) are cholinesterase inhibitors, which prevent the formation of the enzyme required to metabolize acetylcholine from between nerve synapses (Niemczyk and Sheltar

2000). This results in neurons that continually fire, incapacitating the musculature and leading to death by respiratory failure. Pyrethroid insecticides

(e.g., sumithrin & resmethrin) bind to a protein in the nerve axon that controls the sodium channel (Stenersen 2004). Normally, this protein opens to initiate stimulation and closes to terminate the nerve signal. Pyrethroids bind to this protein gate and prevent it from closing normally, which results in continuous nerve stimulation. The lower toxicity of pyrethroid insecticides to mammals is due in large part to their rapid metabolism by non-specific carboxyl esterases

(WHO 1990).

While these two insecticide classes elicit similar physiological effects, their efficacy will vary by physical property (i.e., density, viscosity, and volatility) or susceptibility of the target species population. Additionally, cross-resistance, can lead to separate chemical classes performing with lowered efficacies because a

99 general detoxification mechanism has been selected in a pest population

(Hardstone et al. 2007).

In order to evaluate the efficacy of the USAF C-130 fuselage boom configuration, we applied two insecticides (Anvil and Dibrom) representing the two public health adulticide classes used against mosquitoes and biting midges.

We used large-scale field trials at two federally owned sites to test how well the

C-130 aircraft with the fuselage booms created an insecticide swath that was effective over broad areas. We held mosquitoes in bioassay cages to determine efficacy of aerial spray applications at one site and we also compared changes in natural abundances of mosquitoes and biting midges at a test and control site.

METHODS

Study site descriptions

One test site was Craney Island, VA (N36°54.617, W76°22.144), which is located north of Portsmouth, VA. It is a US Army Corps of Engineers dredge spoil site in the Newport News Channel. The perimeter dike of Craney Island has a utility road that was used as a transect parallel to the prevailing wind direction (Fig. 1). The island vegetation is largely cattails and grasses, except for deciduous and coniferous trees on the southern portion.

Additional testing was conducted at the Marine Corps Recruit Depot

(MCRD), Parris Island, SC. (N32°21.199, W80°40.691). This 3,240 hectare facility is approximately half salt marsh, dominated by smooth cordgrass

(Spartinia alterniflora) and half upland open (grass or developed) and forested

100

(pine and pine-hardwood) areas. Parris Island is bounded by Archers Creek to the north, the Beaufort River on the east, Port Royal Sound on the south, and the

Broad River on the west.

We also used the Spring Island Planned Community (32°20’42 N,80°50’06

N) on Spring Island as a Control site. Spring Island is located 7 miles west of

Parris Island and is also surrounded by salt marsh and upland pine and oak communities. Spring Island Community does not use adulticides for mosquito and biting midge control (B. Lampwright, personal communication), housing development has been minimized and integrated into the natural environment; making it a reasonable location to contrast with changes to insect populations on

Parris Island.

Bioassays

To determine pesticide effects, wild mosquitoes were field-collected at

Craney Island and Parris Island the evening and morning prior to insecticide applications using CO2- baited traps and backpack aspirators. Captured mosquitoes were transferred to cylindrical screened cardboard ring cages (16.5 x

3.8 cm) (Townzen and Natvig 1973). Approximately 25 mosquitoes were placed in each bioassay cage.

To test for incidental mortality, control cages were deployed for 30 minutes prior to the spray and then returned to the laboratory just prior to the application.

The other cages were clamped to 1.2 m wooden stakes, 15 minutes prior to the application along a transect parallel with the prominent wind direction. Cages remained in place until 30 minutes post-treatment and afterwards were quickly

101 transferred to clean holding containers and returned to the laboratory where a sugar water solution (10%) was provided. Percent mortality in test and control cages was measured at 24 hrs. Reported mortality during pesticide treatment was corrected for incidental mortality that occurred in the control groups using

Abbott’s formula (Abbott 1925). Bioassay cages could not be used to monitor mortality effects on biting midges because their small size would have required a fine mesh that would have excluded larger droplets (Boobar et al. 1988).

Aircraft applications

Applications were made with an USAF C-130H aircraft modified to make aerial spray applications with fuselage booms configured with TeeJet 8005 flat- fan nozzles (TeeJet Wheaton, IL) directed towards the ground (90°). The aircraft was equipped with a Satloc GPS Agricultural Navigation System (Hemisphere

GPS, Calgary, Canada) capable of recording aircraft position and time when the spray system was turned on. Aircraft altitude was 46 m above ground level at

370.4 km/hr. Meteorology (wind speed, direction, air temperature, and humidity) was recorded at 2.5 m above ground surface using a Swath Kit Weather Station

(Droplet Technologies, College Station, PA) and at spray altitude via the C-130H

Self-Contained Navigation System.

Craney Island trials

Single pass applications were made against caged field-collected adult mosquitoes with an organophosphate insecticide, Dibrom Concentrate®

(AMVAC, Los Angeles, CA; hereafter Dibrom), at Craney Island on 3, 6 June and

102

6, 7 October 2004 with a pesticide flow rate of 36.5 ml/ha. On each date, we established an east-west transect with ten sampling stations at 213 m intervals across the northern edge of Craney Island We also established a second transect across the southern edge on October 7 (Fig. 1). These transects were parallel to the prevailing wind and perpendicular to the aircraft flight path, which passed 213 m upwind from the first collection station. The spray system was turned on 30 seconds prior to reaching the transect and continued until 30 seconds after passing the transect (60 seconds total). On June 3, we ran a second trial two hours after the first trial. To compensate for changes in wind direction and reduce the chance of the spray cloud missing the sampling stations, the spray-on and spray-off locations of the aircraft were at least as long as the sampling transect (Dumbauld and Rafferty 1977).

Sampling stations had one bioassay cage and one slide rotator device.

Slide rotators (John Hock Company, Gainesville, FL) held 2 Teflon®-coated glass microscope slides (25 X 75 mm; rotated at 420 rpm) that collected the droplet cloud as it passed through the sampling transect. Slides were collected 30 minutes after the application to allow time for the aerosol cloud to completely traverse the sampling transect, and slides were processed the same day as the trial. Spray droplets on the slides were measured and counted under a compound microscope equipped with a reticule. We measured the first 100 droplets or the entire slide was scanned when there were less than 100 droplets.

Volume median diameter (VMD) along with droplet density were determined at each sampling station (Yeomans 1949, Mount et al. 1970).

103

Parris Island trials

Efficacy of aerial sprays with fuselage booms was tested at Parris Island against natural biting fly populations (mosquitoes and biting midges). Biting fly densities were monitored weekly using CO2 -baited CDC-style traps at four locations on Parris Island described in detail in Chapter 2. Wind speed, direction, temperature, and humidity were recorded on the ground and at spray release altitude during all applications.

Dibrom applications

Aerial sprays using Dibrom were conducted April 2002 – April 2005.

Dibrom was applied using the same aircraft configuration as at Craney Island but with an application rate of 74 ml/ha and a flight path (swath width) separation of

305 m. The entire installation was sprayed, except for a 100 ha no-spray area located on the northwest corner of the facility. All sprays were made within two hours of sunset. Percent reduction of biting midge and mosquito densities were calculated from total trap collections and averaged over two weeks before and one week after Dibrom applications.

Anvil applications

Efficacy of fuselage sprays were carried out with a synergized synthetic pyrethroid, Anvil 10+10® (Clarke Mosquito Control, Roselle, IL, hereafter Anvil).

Applications were made on April 18 and May 3, 2006 using the same aircraft configuration as described for Dibrom, except the application rate was 45.3 ml/ha.

104

We conducted additional monitoring during the May 3, 2006 spray to test for spray effectiveness. Daily trapping at paired locations on Parris Island and the control location on the Spring Island Planned Community were used to compare changes in biting fly densities during the May 3, 2006 Anvil application.

Daily trapping was done with Mosquito Magnet traps, contents were emptied every 24 hours and biting flies were counted and identified from April 25 – May 7.

We also examined mortality in caged mosquitoes during this spray using the methods described above.

Statistical analyses

Numbers of midges and mosquitoes collected in CO2-baited traps were log + 1 transformed to equalize variances (Zar 1984). Mean number of midges and mosquitoes before and after Dibrom sprays at Parris Island were analyzed using paired t-tests. We also used a Before/After Control Interaction Design

(BACI) design with paired t-tests to compare numbers in treatment (Parris Island) and control (Spring Island) locations before and after Anvil sprays. Statistical analyses were conducted with SPSS for Windows (Version 13.0, SPSS,

Chicago, IL).

Bioassay cage data was examined to determine the relationships between mosquito mortality vs. distance from the spray point, mosquito mortality vs. droplet size and mosquito mortality vs. density. We used both linear and polynomial regressions and the best fit curves were selected by comparing the F- statistic from the regression ANOVA table. A higher F-value indicated a closer approximation of the model (Zar 1984).

105

RESULTS

Craney Island trials

Prevailing wind direction during the June and October trials at Craney

Island was steady, and the aircraft maintained a north to south vector during all applications (Fig. 2). On June 3 wind direction was from 78° and speed was 9.3 km/hr. Winds at aircraft altitude were from the same direction at 14.8 km/hr.

During the second test, ground wind direction was 122° at 11.1 km/hr and wind direction at aircraft altitude was 140° at 20.4 km/hr. On October 6, the average wind speed and direction during the test were 30° at 10.4 km/hr on the ground and 16.7 km/hr at altitude. During the October 7 test, wind direction was 43° at

7.2 km/hr on the ground and 10.7 km/hr at application release height.

Composition of field-collected mosquitoes varied by date with Aedes sollicitans (Walker), Ae. vexans (Meigen), and Culex salinarius (Coq.) dominating collections (Table 1). On each spray date, mosquito mortality after the Dibrom spray ranged from 11% to 100% between sampling stations (Table 2). Mosquito mortality was 100% from 213 m to 639 m downwind of the release point and most stations had 90% mortality up to 1491 m downwind. The October 6 trial had the lowest mortalities of the 4 applications, but mortality was still at least

79% from 0-852 m downwind from release point.

Droplet size and density decreased with distance from the release point.

Volume median diameter was 44 µm at the first station and decreased in size to

11 µm at the station furthest from the release point (2,130 m) (Table 2). Droplet

106 density ranged from a maximum of 18.4 drops/cm2 at the first station to a low of 2 drops/cm2 at the final sampling station.

Regression analysis determined that droplet density and mortality had a positive linear correlation (r2 = 0.37, n = 40, P < 0.001). VMD and mortality were also positively correlated (r2 = 0.14, n = 40, P = 0.017) (Fig. 3). There was also a negatively linear correlation between distance downwind from the insecticide release point and mortality (r2 = 0.38, n = 40, P < 0.001) (Fig. 4). Although polynomial regressions were also significant, linear regression models had a better fit for all data.

Parris Island trials

Dibrom applications

Three species of biting midges were collected during the test periods:

Culicoides furens (Poey), C. hollensis (Melander and Brues), and C. melleus

(Coq.). One week following Dibrom applications biting midge densities were 86%

(SE±0.1) lower than pre-spray densities. Two weeks after sprays, biting midge numbers had increased somewhat but were still 55.8% (SE±0.2) lower than pre- spray densities. However, comparisons between pre-spray and post-spray midges densities were marginally not significant at both 1-week (t = 2.56, df = 5,

P = 0.051) or 2-weeks (t =2.16, df = 4, P = 0.097) post-spray.

Aedes taeniorhynchus (Weidemann) and Ae. sollicitans made up 70% of the mosquito fauna in the trap collections. Mosquito numbers decreased by

83.2% (SE±0.1) the one week post spray and 44.7% (SE±0.2) two weeks post-

107 spray, relative to pre-spray densities. Mosquito numbers were different between pre-spray and 1-week post spray (t = 2.62, df = 5, P = 0.047), but not after 2- weeks (t =1.25, df = 4, P = 0.279).

Anvil applications

The 18 April Anvil application was made under favorable environmental conditions. Temperature during the application was 20.6-17.2 °C, relative humidity was 60-67%; wind direction was between 080-110° and wind speed was

6.5 km/hr on the ground and 11.1 km/hr at spray altitude.

Acceptable environmental conditions were also present during the 3 May

Anvil application. Wind direction was 310° at 10.2 km/hr on the ground and 310° at 22.2 km/hr at spray release altitude. Temperature and relative humidity were

26.1-25.6°C and 45-52%, respectively.

Biting midges collected during the Anvil study were C. furens (71%), C. hollensis (20%), and C. melleus (9%). After Anvil sprays on 18 April 2006, trap collections showed a 19% increase in biting midges numbers at Parris Island while numbers at the control site on Spring Island decreased by 44% (Fig. 5).

On the 3 May 2006 spray, the aircraft covered the entire Parris Island installation (Fig. 6). Biting fly numbers had already started to decline on the treatment and control areas after 25 April through the spray date (3 May) (Fig. 7).

Biting fly densities were significantly decreased after the 3 May Anvil spray on

Parris Island (t = 5.59, df = 3, P = 0.011) (Fig. 7). However, biting midge numbers were also lower post-spray at the control site on Spring Island (t = 4.20, df = 3, P = 0.025). No difference in biting midge densities were found between

108

Parris and Spring Islands prior to the 3 May spray (t = 1.31, df = 8, P = 0.225) or after the application (t = 0.47, df = 3, P = 0.673).

Mosquito species on Parris Island were dominated by Culex salinarius

(78%) and Ae. vexans (11%). There was a significant decrease in mosquito numbers after the 3 May 2006 spray on Parris Island (t = 7.13, df = 3, P = 0.006) but also on Spring Island (t = 4.91, df = 3, P = 0.016).

One of the pre-spray mosquito bioassay cages on 3 May had a high mortality (57%) for an unknown reason. When we corrected the post spray data, this caused in a strong reduction in estimated mortality of the post spray bioassay cages. Mortality rates were highly variable, and the greatest mortality was observed on the downwind section of the treatment area (Fig. 6). Corrected mortality of mosquitoes in all bioassay cages ranged from 0-87% and averaged

37.3% (SE±16.0).

DISCUSSION

Craney Island trials

Our study tested the functionality of the C-130 aircraft fuselage boom configuration for ULV sprays to control mosquitoes and biting midges. We found that fuselage booms created high insecticide droplet densities (18-10 droplets/cm2) that were correlated with high mosquito mortality rates (>90%) in bioassay cages during single pass trials. It is interesting to note that others reported mortality rates decreased when droplet densities were below ~ 30

109 drops/cm2 (Brown et al. 2003, Meisch et al. 2005), and it is not clear why our trials had a high mortality at lower droplet densities.

Our findings indicate that the new fuselage boom effectiveness was comparable to the USAF wing boom configuration in creating an effective swath width of 609 m (Haile et al. 1982, Burkett et al. 1996). A wide swath is useful to promote rapid coverage of large areas (Pinkovsky 1972) during combat areas and domestic emergencies (DOD 2008). The field trials indicate that single pass applications of Dibrom can produce high mosquito mortality (e.g., > 95%) from

213 m up to 1491 m downwind of the release point. Our findings imply that under ideal conditions, a swath width greater than the standard 609 m used by the USAF for their spray operations could be employed with acceptable results.

Furthermore, this study measured mortality from single pass spray applications. During normal public health spray operations, the spray area is crossed several times by the aircraft and fractions of the resultant insecticide aerosol is able to drift sometimes more than 2 km downwind (Mount et al. 1996).

Field trials often use multiple passes over a single release point or a series of passes upwind to simulate the additive effects observed downwind during actual operations (Brown et al. 2003; Meisch et al. 2005). Our trials took the more conservative approach of testing a single pass to determine the effective swath width delivered from fuselage booms, and field operations may have an even wider effective swath width.

Meteorological conditions often determine the outcome of aerial spray operations (Thistle 1996). Wind direction during the October 6 test had the

110 smallest crosswind component of the 4 trials. Three of 4 trials showed 100% mortality in bioassay cages downwind to 852 m but mortality dropped off steeply at this same distance on the October 6 trial. Since a direct crosswind is optimum for proper insecticide dispersal (Barber et al. 2007), the departure from optimum wind conditions observed during the October 6 trial likely accounts for decreased mortality. However, the October 6 VMD and droplet densities were still similar to other trials at most sampling locations.

There were some other aspects of our experimental design that might influence our results. Caged mosquitoes are useful for measuring the efficacy of a single pass but likely represent a best-case scenario and, consequently, may not accurately assess the mortality of natural populations (Rathburn et al. 1971).

Furthermore, the open vegetation of Craney Island study does not reflect the heterogeneous landscape found in some spray operations. For example, heavily forested areas could negatively effect pest mortality (Barber et al. 2007; Lothrop et al. 2007). However, these would also affect spray characteristics from other aircraft spray systems, and therefore they do not affect our conclusions that the new fuselage boom configuration was effective at Craney Island.

Parris Island trials

The Parris Island trials tested large scale applications with the fuselage boom configuration using both Dibrom and Anvil. At Parris Island, midge numbers were reduced 86% relative to pre-spray collections, one week after

Dibrom applications. This level of reduction is considered acceptable control but

>95% reductions would be optimum. The numbers increased again at two weeks

111 post-spray, but this was not unexpected because biting midge densities at Parris

Island typically increase quickly by immigration and newly emerging adults

(Breidenbaugh et al., in review).

Although Dibrom was effective, the Anvil spray results were not as conclusive. The first application (18 April) was ineffective because Culicoides spp. numbers at Parris Island increased, while they decreased at the control site,

Spring Island. The trials on 3 May were more rigorously tested using both short and long term population changes and bioassay cages. Both of these methods indicated that Anvil was ineffective against the biting flies species tested. These results contradict some favorable reports of Anvil for mosquito control (Meisch et al. 2005).

There are several potential explanations for the apparent lower performance of Anvil. Insecticides can have variable efficacies against different biting fly species (Pridgeon et al. 2008) and while the efficacy of both insecticides used in these trials is well documented against mosquitoes with ground ULV and aerial sprays (Linley and Jordan 1992, Mount et al.1996, Breidenbaugh et al.

2000, Lesser 2002, Meisch et al. 2005, Meisch et al. 2007, Breidenbaugh and

Haagsma 2008), some trials have found that pyrethroids were more effective against biting midges than organophosphates (Kline et al. 1981, Floore 1985). In contrast, others found Dibrom more effective than pyrethroids against biting midges and mosquitoes (Linley and Jordon 1992). The causes for this are not fully understood.

112

Furthermore, Anvil is lighter (specific gravity = 0.884) than Dibrom

(specific gravity = 1.8), and may have been dispersed by wind and may not have settled out of the air column as quickly. In addition, Anvil is applied at a lower concentration (0.004 kg/ha) of active ingredient than Dibrom (0.11 kg/ha of active ingredient) because pyrethroids are highly toxic to aquatic organisms (Siegfried

1993). Therefore, the lower application rate of Anvil and its physical properties may have limited its effectiveness to reduce biting midges.

Other experimental conditions also affected the experiment during the

Parris Island trials. First, an aircraft malfunction cancelled one scheduled application on May 2. We collected low numbers of wild mosquitoes on May 3, and therefore had to use others collected for the previous day’s trial. These were likely stressed from holding conditions and this resulted in the pre-spray control group having a high mortality (29%). Most studies will discard trial data when control mortality exceeds 20% (WHOPES 1996, WHO 1998), but we could not reschedule the C-130 aircraft based on this issue. The problems we encountered during the trials at Parris Island suggest further investigation is needed to thoroughly examine the efficacy of large scale applications of Anvil.

In conclusion, fuselage boom configuration using flat-fan nozzles on the

C-130 aircraft appear to be an effective configuration for biting midge and mosquito control. Our experimental design of combining medium and large-scale field trials presented a realistic picture of the potential efficacies of this method as well as showing the inherent challenges associated with real world variation in field trial conditions.

113

REFERENCES CITED

Abbott WS. 1925. A method for computing the effectiveness of an insecticide. J

Econ Entomol 18:265-267.

Barber JA, Greer M, Coughlin J. 2007. Field tests of malathion and permethrin

applied via a truck-mounted cold fogger to both open and vegetated

habitats. J Am Mosq Control Assoc 23:55-59.

Boobar LR, Dobson SE, Perich MJ, Darby WM, Nelson JH. 1988. Effects of

screen materials on droplet size frequency distribution of aerosols entering

sentinel mosquito exposure tubes. Med Vet Entomol 2:379-384.

Breidenbaugh M, Olson S, Spears B, Teig D. 2000. Emergency Mosquito

Control after Hurricane Flovd-U. S. Air Force Reserve. Wing Beats Fla

Mosq Control Assoc 11:37-38.

Breidenbaugh M, Haagsma K. 2008. The US Air Force Aerial Spray Unit:

A history of large area disease vector control operations, WWII through

Katrina. Army Med Dept J. April-June 2008:54-61.

Brown JR, Mickle RE, Yates M, Zhai J. 2003. Optimizing an aerial spray for

mosquito control. J Am Mosq Control Assoc 19:243-250.

Burkett DA, TL Biery, DG Haile. 1996. An operational perspective on

measuring aerosol cloud dynamics. J Am Mosq Control Assoc 12:380-

383.

Carney RM, Husted S, Jean C, Glaser C, Kramer V. 2008. Efficacy of aerial

spraying of mosquito adulticide in reducing incidence of West Nile virus,

California, 2005. Emerg Infect Dis 14:747-754.

114

Davis RS, Peterson RKD, Macedo PA. 2007. An ecological risk assessment for

insecticides used in adult mosquito management. Integr Environ Assess

Manag 3:373–382.

DoD [US Department of Defense]. 2008. Department of Defense Instruction

4150.07. DoD Pest Management Program, May 29, 2008 [Internet].

Silver Spring, MD [accessed July 27, 2008]. Available from:

http://www.afpmb.org/pubs/dir_inst/dod4150.7-i.pdf.

Dumbauld RD, Rafferty JE. 1977. Field manual for characterizing spray from

small aircraft. Davis, CA: USDA Forest Insect and Disease Management

Methods Application Group.

Floore TG. 1985. Laboratory wind tunnel tests of nine insecticides against adult

Culicoides species (Diptera: Ceratopogonidae). Florida Entomol 68:678-

682.

Haile DG, Biery T, Reinert JF, Pierce NW. 1982. Aerial applications of naled

diluted in HAN with UC-123K aircraft for adult mosquito control. Mosq

News 42:41-48.

Hardstone MC, Leichter C, Harrington LC, Kasai S, Tomita T, Scott JG. 2007.

Cytochrome P450 monooxygenase-mediated permethrin resistance

confers limited and larval specific cross-resistance in the southern house

mosquito, Culex pipiens quinquefasciatus. Pestic Biochem Phys 89:175-

184.

115

Kline DL, Haile DG, Baldwin KF. 1981. Wind tunnel tests with seven

insecticides against adult Culicoides mississippiensis Hoffman. Mosq

News 41:745-747.

Lesser CR. 2002. Field trial efficacy of Anvil 10+10 and Biomist 31:66 against

Ochlerotatus sollicitans in Delaware. J Am Mosq Control Assoc 18:36-39.

Lindley JR, Jordan S. 1992. Effects of ultra-low volume and thermal fog

malathion, Scourge® and naled applied against caged adult Culicoides

furens and Culex quinquefasciatus in open and vegetated terrain. J Am

Mosq Control Assoc 8:69-76.

Lothrop HD, Huang HZ, Lothrop BB, Gee S, Gomsi OE, Reisen WK. 2007.

Deposition of pyrethrins and piperonyl butoxide following aerial ultra-low

volume applications in the Coachella Valley, California. J Am Mosq

Control Assoc 23:213-219.

Meisch MV, Dame DA, Brown JR. 2005. Aerial ultra-low-volume assessment of

Anvil 10+ 10® against Anopheles quadrimaculatus. J Am Mosq Control

Assoc 21:301-304.

Meisch, MV, Dame DA, Lewis CN. 2007. Comparison of dilute and neat

formulations of Anvil by ultra-low volume ground application against

Anopheles quadrimaculatus. J Am Mosq Control Assoc 23:312-314.

Moore CG, McLean RG, Mitchell CJ, Nasci RS, Tsai TF, Calisher CH, Martin AA,

Moore PS, Gubler DJ. 1993. Guidelines for arbovirus surveillance

programs in the United States. Fort Collins, CO: Centers for Disease

Control and Prevention.

116

Mount GA, Lofgren CS, Pierece NW, Baldwin DJ, Adams CT. 1970. Droplet

size, density, distribution and effictiveness in ultra-low volume aerial

sprays dispersed with teejet nozzles. Mosq News 30:589-599.

Mount GA, Biery TI, Haile DG. 1996. A review of ultralow-volume aerial sprays

of insecticide for mosquito control. J Am Mosq Control Assoc 12:601-618.

Niemczyk HD, Shetlar DJ. 2000. Destructive Turf Insects, 2nd edition. HDN

Books, Wooster, OH.

Parrish DW, Hodapp CJ. 1962. Biological evaluation of the C-47 aerial spray

system for adult mosquito control. Mosq News 22:36-7.

Pedigo P. 1989. Entomology and pest management. New York: Macmillan

Publishing.

Pinkovsky DD. 1972. United States Air Force aerial spray activities in Operation

Combat VEE. Mosq News 32:332-334.

Pridgeon JW, Pereira RM, Becnel JJ, Allan SA, Clark GG, Linthicum KJ. 2008.

Susceptibility of Aedes aegypti, Culex quinquefasciatus Say, and

Anopheles quadrimaculatus Say to 19 pesticides with different modes of

action. J Med Entomol 45:82-87.

Rathburn CB, Rogers AJ, Boike AH, Lee RM. 1971. The effectiveness of aerial

sprays for the control of adult mosquitoes in Florida as assessed by three

methods. Mosq News 31:52-55.

Rose RI. 2001. Pesticides and public health: Integrated methods of mosquito

management. Emerg Infect Dis 7:17-23.

117

Siegfried BD. 1993. Comparative toxicity of pyrethroid insecticides to terrestrial

and aquatic insects. Environ Toxicol Chem 12:1683–1689.

Service MW. 2000. Medical Entomology for Students, 2nd Ed. Cambridge UK:

University of Cambridge Press.

Stenersen J. 2004. Chemical pesticides: mode of action and toxicology Boca

Raton, FL: CRC Publishing.

Thistle HW. 1996. Atmospheric stability and the dispersion of pesticides. J Am

Mosq Control Assoc 12:359-363.

Townzen KR, Natvig HL. 1973. A disposable adult mosquito bioassay cage.

Mosq News 33:113-114.

Wardman OL, Thomas MR. 2000. Aerial applications of pesticides in the United

Kingdom: 1978 to 1998. Pest Manag Sci 56:237-244.

Wata Y. 1989. Control of Japanese encephalitis vectors. S.E. Asian J. Trop.

Med. Public Health 20:623-26

WHO [World Health Organization]. 1990. Environmental Health Criteria, No. 94.

Permethrin. Geneva, World Health Organization.

WHO [World Health Organization]. 1998. Test procedures for insecticide

resistance monitoring in malaria vectors. Bio-efficiency and persistence of

insecticides on treated surfaces. WHO/CDS/CPC/MAL/91.12.Geneva,

Switzerland: World Health Organization.

WHO [World Health Organization]. 2003. Space spray application of

insecticides for vector and public health pest control: A practitioner’s

guide. Geneva, World Health Organization.

118

WHOPES [World Health Organization Pesticide Evaluation Scheme]. 1996. Test

methods. CTD/WHOPES/IC/96.1. Geneva, Switzerland: World Health

Organization.

Yeomans AH. 1949. Directions for determining particle size of aerosols and fine

sprays. USDA Bul Ent and Plant Quar ET-267:1-7.

Zar JH. 1984. Biostatistical Analysis. Prentice–Hall,Englewood Cliffs, NJ.

119

Fig. 1. Overview of experimental design for USAF C-130 single-pass Dibrom sprays using a fuselage boom configuration on Craney Island, VA in 2004.

120

360° 360°

A. B.

D. C. 180° 180°

wind direction aircraft path sampling line

Fig. 2. Diagrammatic representation of flight path, sampling line, and wind direction on a magnetic grid for A. June 3-trial 1. B. June 3-trial 2. C. October 6

D. October 7, 2004 .

121

Table 1. Percent species composition of field-collected mosquitoes used in caged bioassays at Craney Island, VA in 2004. Other species included: Culiseta melanura (Coq.), Psorophora ciliata (Fab.), and Ps. ferox (von Humboldt).

Craney Island June 3 October 6 October 7 Species % % % Aedes atlanticus Dyar & Knab - 11 - Ae. sollicitans (Walker) 73 13 91 Ae. taeniorhynchus (Weidemann) - 13 2 Ae. vexans (Meigen) 9 32 4

Anopheles bradleyi King 2 10 - Coquillettidia perturbans (Walker) 2 - - Culex salinarius (Coq.) 12 19 1 Other spp. 2 2 2

122

Table 2. Values for 24 hr mortality of mosquitoes, volume median diameter

(VMD), and droplet density at sampling stations downwind from Dibrom applications in June and October 2004 on Craney Island, VA.

Distance downwind 213 m 426 m 639 m 852 m 1065 m 1278 m 1491 m 1704 m 1917 m 2130 m

24 hr mortality of caged mosquitoes June 3 (1) 100 100 100 100 61.7 92.1 91.5 58.6 71.3 42.5 June 3 (2) 100 100 100 100 100 100 91.8 91.8 77.9 48.9 October 6 100 100 100 77.8 29.4 65.5 27.8 12 14.3 11.1 October 7 100 100 100 100 100 47.4 100 36.8 89.5 89.5 Mean 100 100 100 94.5 72.8 76.3 77.8 49.8 63.3 48.0 (±SE) (±0.0) ±0.0 (±0.0) (±5.6) (±17.0) (±12.1) (±16.8) (±16.9) (±16.7) (±16.1)

VMD of droplets June 3 (1) 43 39.7 43.7 33.9 29.9 20.1 22.3 19 15.6 15.5 June 3 (2) 45.7 20.6 24.7 19.8 18.3 10.2 9.5 10.2 8.6 8.5 October 6 43.5 20.4 20.4 17 15.5 11.5 12.8 15 10.5 12 October 7 45.2 18.1 10.3 9.5 8 9.6 9 9.1 9.2 8.3 Mean 44.4 24.7 24.8 20.1 17.9 12.9 13.4 13.3 11.0 11.1 (±SE) (±0.7) ±5.0 (±7.0) (±5.1) (±4.5) (±2.4) (±3.1) (±2.3) (±1.6) (±1.7)

Droplet density per cm2 June 3 (1) 18.4 17.3 13.5 6.3 5.5 6.7 11 6.4 4.1 2.2 June 3 (2) 15.2 15.2 13.2 3.7 4.7 6.4 10.9 7.3 4.2 2.4 October 6 16.4 15.7 13.1 4.3 3.3 6.1 3.8 4.1 3.9 2 October 7 15.3 15.7 13.9 5.2 5.7 7.5 11.1 7 5.1 2.8 Mean 16.3 16.0 13.4 4.9 4.8 6.7 9.2 6.2 4.3 2.4 (±SE) (±0.7) ±.05 (±0.2) (±0.6) (±0.5) (±0.3) (±1.8) (±0.7) (±0.3) (±0.2)

123

124

Fig. 3. Relationship between (A) droplet density and mortality (B) droplet size

(volume median diameter) and mortality (B) determined from bioassay cages from sampling stations 213 m - 2130 m downwind from release point following

Dibrom applications in June and October 2004 on Craney Island, VA .

125

A

100

80

60

40

20

0

0 5 10 15 20

Density (drops/cm2)

B Mortality Mortality (%)

100

80 Morta lity (%)

60

40

20

0

10 20 30 40 50 Volume median diameter

126

100

80

60

40 (%) Mortality

20

0

0 500 1000 1500 2000 2500 Distance (meters)

Fig. 4. Relationship between mosquito mortality and distance downwind from pesticide release point following Dibrom applications in June and October 2004 on Craney Island, VA .

127

16000 Parris Island 14000 12000 Spring Island 10000 8000 6000 4000

Total midges collectedTotal midges 2000 0 5 April 12 April 19 April 28 April

Fig 5. Relative numbers of Culicoides spp. before and after Anvil sprays on Parris

Island and a no-spray area (Spring Island). Vertical dashed line indicates spray date (April 18, 2006).

60% 15% 87% 60% 0% 59% 27%

128

Fig. 6. Flight path of aircraft (black lines) while making an Anvil insecticide application, May 3, 2006. Yellow shaded area is the treatment area. Circular shaded area is a no fly/no spray area. Percent markers indicate percent mortality of caged mosquitoes exposed during the application.

129

130

6000 Parris Island Spring Island 5000

4000

3000

2000 Total midges collected midges Total

1000

0

1 May 2 May 3 May 4 May 5 May 6 May 7 May 25 April 26 April 27 April 28 April 29 April 30 April

Fig. 7. Daily biting midge density monitored on Parris Island and a no-spray area

(Spring Island). Vertical dashed line indicates spray date (May 3, 2006).

CHAPTER V

EFFECTS OF AERIAL APPLICATIONS OF NALED ON NONTARGET INSECTS AT PARRIS ISLAND, SOUTH CAROLINA1

ABSTRACT

Responses of nontarget insects to aerial applications of an organophosphate insecticide (naled), used for biting fly control, were studied on the Parris Island Marine Corps Recruit Depot, SC. We used Malaise traps and yellow pan traps to determine nontarget terrestrial insect diversity and abundance with before and after impact analysis (BACI). Two naled applications were made at dusk in October 2003 and a single application was made in April

2005. Four of 12 major taxa (Syrphidae, Dolichopodidae, Sarcophagidae,

Tachinidae) in Malaise trap samples and 3 of 12 major taxa (Chironomidae,

Microhymenoptera, Ceratopogonidae) in pan traps had lower numbers after sprays. The taxa which declined were not consistent between years or between treatment sites. Total nontarget insect abundance was lower after sprays in

Malaise trap collections in 2003 (P <0.025), with these numbers decreasing by about 50%. However, there were no differences in total numbers after sprays in pan traps in 2005 (P = 0.756). Shannon diversity indices were not different after sprays in either year indicating that sprays had minimal impact on overall

1 To be submitted to the Journal of Environmental Entomology by M.S. Breidenbaugh and F.A. de Szalay.

131

132

community biodiversity. In contrast, populations of the target biting midges

(Culicoides spp.) collected in CDC style light traps were reduced by >98.0% after spraying and mosquitoes numbers also declined after sprays. Our results indicate that there are some impacts on non-target species from aerial sprays but applying sprays at dusk helps minimize these impacts while still controlling biting flies.

INTRODUCTION

Control of adult mosquitoes and other biting flies over large areas usually requires aerial applications of insecticides (Blanton et al. 1950, Mount et al.

1996). Aerial mosquito adulticides have been used to interrupt enzootic outbreaks, lower the threat of mosquito-borne disease, and reduce nuisance biting to tolerable levels (Pinkovsky 1972, Breidenbaugh et al. 2008, Carney et al. 2008). However, there are ongoing public concerns over acute and chronic hazards to the environment caused by aerial spray treatments (Their 2001).

Some have noted that large-scale field trials under natural conditions are needed to conduct environmentally relevant pesticide exposure studies (Jensen et al. 1999). This is because the complexity of the natural world can interact in ways not predicted in carefully controlled laboratory studies. For example, insect populations already stressed by natural or human induced factors might have very pronounced responses to pesticides (Coats et al. 1989). Also, behavioral mechanisms (i.e., diel activity patterns, movement and resting behavior) can moderate exposure in field applications. Most studies on lethal dose

133

concentrations tested these on a small number of nontarget species that are laboratory model organisms such as invertebrates (e.g., Daphnia sp.), fish

(rainbow trout, Oncorhynchus mykiss; fathead minnow, Pimephales notatus), and mammals (e.g., house mice, Mus musculus; and Norway rat, Rattus norvegicus)

(Mulla et al. 1978, USEPA 2002). Furthermore, field studies testing effects of aerial spraying on nontarget insect populations have focused mostly on impacts on honey bee populations (Apis mellifera), because they are economically important pollinators and sensitive to organophosphate pesticides (Caron 1979,

Johansen and Mayer 1990, Delaplane and Mayer 2001). However, there are data indicating that insect species vary in their tolerance to insecticides, even within the superfamily Apoidea (Johansen and Mayer 1990). Research on mosquito control impacts to nontarget insect species is scarce, in spite of a growing concern over declines of natural pollinators and impacts on agriculture and biodiversity (Allen-Wardell 1998). Consequently, testing effects of mosquito adulticide applications on a broad range of nontarget species is useful in order to gauge the actual environmental cost of pest management practices.

Naled is an organophosphate insecticide commonly used by state and federal public health agencies as an adulticide for biting flies (Breidenbaugh et al.

2000). This compound was recently reviewed and re-registered as a restricted use public health insecticide (USEPA 2002). Peterson et al. (2006) conducted risk assessments for naled and several other commonly used mosquito insecticides and concluded that human-health risks from residential exposure

134

was low and that the risks from West Nile virus exceed the risks from exposure to insecticides. Extensive testing has been conducted to check for effects of naled on key nontarget and invertebrates, but little is known about its impacts on overall structure of invertebrate communities (Swanson et al. 1996,

Zhong et al. 2003). Davis et al. (2007) conducted risk assessments for naled on aquatic nontarget invertebrates and projected low risk to these ecological receptors.

The US Air Force (USAF) uses specially modified C-130 aircraft when conducting ultra-low volume (ULV) applications to protect troops from vector- borne illness. Following natural disasters when mosquito populations may surge, areas as large as 1.1 million ha have been sprayed (Breidenbaugh and Haagsma

2008). Domestically, this group trains for its war-time mission with vector control operations on military installations and associated municipalities, approaching

500,000 ha treated annually. For example, the USAF uses aerial naled applications at the Parris Island Marine Corps Recruit Depot (Parris Island),

South Carolina to control mosquitoes (Culicidae) and biting midges

(Ceratopogonidae). However, nontarget impacts of the spray methods and equipment used by the USAF have not been tested.

The project objectives were to examine how USAF aerial spray operations on Parris Island impact nontarget insect populations. Our null-hypothesis was that no effects from aerial applications of insecticides would be seen on nontarget insects. However, naled has a mode of action that is not species

135

specific, and it is not unreasonable that applications would cause mortality of both target and nontarget insect species. Because it is difficult to create true replicates, we used a Before-After Control Impact (BACI) experimental design

(Eberhardt 1976) to compare changes in nontarget insects. We sampled insects with two methods to account for the wide range of insects expected to come in contact with pesticides during aerial sprays. These data should be useful to examine impacts of aerial sprays on nontarget insects in field situations and also improve IPM programs for similar habitats.

METHODS

Evaluations of nontarget effects were conducted at Parris Island during routine USAF spray operations for mosquito and biting midge control in October

2003 and April 2005. Parris Island is approximately 3,200 ha, over half of which is salt marsh (Fig. 1). Upland regions are composed of open (grassed or developed) and forested (pine and pine-hardwood) areas. The salt marshes are dominated by smooth cordgrass (Spartinia alterniflora). Parris Island is bounded by Archers Creek to the north, the Beaufort River on the east, Port Royal Sound on the south, and the Broad River on the west. The city of Charleston, South

Carolina is 90 km to the north and Savannah, Georgia is 50 km to the south. The surrounding coastal regions are similar in composition and serve as additional breeding areas for biting flies.

136

Insect trapping methods

Pest populations of mosquitoes and midges were monitored weekly with

CDC-style CO2-baited traps at 4 locations during the 2003 sprays (Breidenbaugh

& de Szalay, in review). During the 2005 application, biting fly monitoring was done the day prior and day after to determine the relative success of aerial sprays in controlling target species (Fig. 1). During both 2003 and 2005, trap location catches were pooled and averaged to give a single mean mosquito and midge density before and after the naled application.

Malaise traps (BioQuip Products, Rancho Dominguez, CA) and pan traps were used to compare insect diversity and abundance before and after aerial applications at Parris Island. Malaise traps collect Lepidoptera, Coleoptera,

Diptera, Hymenoptera, and other orders of insects that fly into the mesh sides and then move upward into an apical collection container (Marston 1965,

Mathews and Mathews 1970, Townes and Townes 1981, Darling and Packer

1988). Because these traps do not use attractants, they passively collect flying insects to give an unbiased estimate of insect species composition (Gunstream and Chew 1967). Malaise trap contents were collected every 24 hours and identified to order, family, or species.

Pan traps were used to collect insects in 2005. These were 26 cm disposable yellow plastic plates filled with a 5% detergent and water solution.

Pan traps attract diurnal flying insects that are attracted to reflected light and land on the surface of the water and drown. The yellow color of the pan increases

137

positive phototaxis of many insect species (Deansfield et al. 1982). All pan trap contents were collected every 24 hours and stored in alcohol until the insects were identified in the laboratory.

Experimental design

Aerial spraying was conducted on 6 and 8 October 2003 (1700-1900 hrs,

1750-1950 hrs, respectively) at the maximum label rate for mosquito control of 74 ml/ha of Dibrom (AMVAC, Los Angeles, CA), which is 87.4% naled by weight.

Spraying was conducted with a USAF C-130H modified to spray insecticides flying at 46 m above the ground. Application data and weather conditions associated with sprays by date are given in Table 1.

In 2003, insects were sampled between 2-14 October in 2 treatment areas. One Malaise trap was set on the northern section of the study area on

Horse Island (HI) and another Malaise trap in the interior of the spray area near the Veterinarian Clinic (VC) (Fig. 1). At each site, trapping began 5 days prior to the first application (6 October) to establish pre-spray insect abundance and continued for 5 days after the second application (8 October) to test if the populations changed.

In April 2005, we sampled insects using yellow pan traps at 3 sprayed locations (Elliot Beach (EB), Horse island (HI), and Page Field (PF)) and one unsprayed control site (Spring Island Planned Community (SI)) (Fig. 1). All sites were similar in appearance and near to developmental habitats for salt marsh mosquitoes and biting midges. The sprayed locations were all on Parris Island,

138

and the untreated site was located on adjacent privately-owned land

(32°20’42 N, 80°50’06 N), 7 miles west of Parris Island. Spring Island

Community does not use adulticides for mosquito and biting midge control (B.

Lampwright, personal communication), which makes it ideally suited to contrast with changes to insect populations on Parris Island.

In 2005, pan trap collections at all sites began 4 days before the single spray application (19 April) and continued for 5 days after the application (16-24

April).

Statistical analyses

All insects were identified to order, family or species in the laboratory, except Lepidoptera <4 mm long were grouped as microlepidoptera and

Hymenoptera <3 mm long were grouped as microhymenoptera. Abundance data were also summed by family for analysis. All abundance data were first tested for normality and log (X+1) transformed when needed. Shannon diversity indices

(H′ = −∑Pi (ln Pi); where P is the proportion of individuals in the i-th taxon) were calculated to measure nontarget insect diversity (Shannon & Weaver 1949).

Target fauna collected (mosquitoes and biting midges in the genus Culicoides) were excluded from nontarget abundance and nontarget diversity analyses.

The 2003 Malaise trap data were analyzed using a before and after design

(BA) (Green 1979) that compares differences before and after a specific event

(i.e., insecticide application). One limitation to the BA analysis is that observed differences were attributed to adulticide applications but in the absence of a

139

Control site, definitive causal inference is problematic as other contributing factors are possible. Before and after insect abundance and Shannon diversity in each location were compared with independent sample t-tests, with samples collected on different dates treated as replicates. Because we ran a separate t- tests at two locations, we used a Bonferroni correction (i.e. significant p value was P <0.05/2 = 0.025) to maintain a procedure-wise alpha error of 0.05.

Because mosquito and biting midge CDC collections were only made for a single collection period statistical analyses were not made. Mosquito and midge abundance was averaged from the 4 trap locations to give before and after comparisons.

The 2005 pan trap family abundance and Shannon diversity data were analyzed with a Before - After Control Impact design (BACI). Samples collected on each date in control and treatment areas were considered paired replicates.

The BACI method compares the mean differences between paired samples before and after spraying. A t-test was used to test if there was a significant change in the size of the differences before and after the treatment periods, which would indicate an impact caused by spraying (Smith 2002). Because we had multiple treatment sites but did not employ a gradient analysis (see Ellis and

Schneider 1997), we ran separate t-tests for each Treatment site to Control comparison. Because we ran three t-tests, we used a Bonferroni correction (e.g., significant P value was P <0.05/3 = 0.017) to maintain a procedure-wise alpha error of 0.05. An ANOVA was used to compare abundance before and after

140

sprays as well as investigate the interaction between period and location. All analyses were conducted with SPSS for Windows (Version 13.0, SPSS,

Chicago, IL).

RESULTS

Nontarget species

In 2003, a total of 2,879 insects in 7 orders (Hymenoptera, Diptera,

Lepidoptera, Hemiptera, Homoptera, Coleoptera, and Orthoptera) and 35 families were collected in Malaise traps. Hymenoptera was most abundant and accounted for 47% of total collected, followed by Diptera accounting for 34% of total. We termed those taxa that were >2% of total collected as our major taxa.

Dolichopodidae (23.5% of total; Diptera), Cicadellidae (8.9% of total; Homoptera) and Sarcophagidae (8.9% of total; Diptera) were the three most abundant major taxa. All other families accounted for less than 8% of collections by taxon

(Table 2).

Four of the 12 major taxa showed a significant change after aerial sprays at the Horse Island site and three of 12 taxa changed after spraying at the

Veterinarian Clinic site. All significantly affected taxa were dipterans. Total numbers of nontarget insects decreased after spraying at both sites. The

Shannon index (H′) between the two sites was not different prior to sprays (t-test,

P = 0.425) indicating the community compositions were similar. Shannon

141

diversity indices were not significantly different from before and after sprays at either site.

In 2005, pan traps collected 2,313 insects in 9 orders (Diptera,

Hymenoptera, Homoptera, Lepidoptera, Coleoptera, Orthoptera, Blattodea,

Psocoptera, and Thysanoptera) and 49 families. Diptera was the most abundant order (79% of total) followed by Hymenoptera (16.4% of total). The most common family in pan traps was Dolichopodidae (Table 3).

In contrast to Malaise traps findings in 2003, total numbers of nontarget insects collected in pan traps were not significantly different before and after the single spray application in 2005 (Fig. 2). However, 3 of 12 major taxa decreased in numbers after spraying. The major taxa that changed were nonbiting midges

(Chironomidae) and microhymenoptera at the EB site and nontarget biting midges (Ceratopogonidae) at the PF site (Table 3). There was also a significant interaction (period X location) observed with dance flies (Empididae) at the PF site (P = 0.039). Shannon diversity indices (H′) in pan traps at all sites were statistically unchanged following the spray (Table 4).

Target species

Target insects (mosquitoes and Culicoides spp.) were present at Parris

Island during both 2003 and 2005 insecticide applications and were subsequently collected in large numbers by CDC-style traps. However, both groups were poorly collected by Malaise traps with mosquitoes making up only 1.7% of collections and biting midges completely absent. Pan traps were also poor

142

collectors of mosquitoes (1.3% of collections) but relatively effective at collecting biting midges (i.e., target = Culicoides spp. and nontarget Dasyhelea spp.), with nontarget Ceratopogonidae making up 6.5% of the total insects collected. In

2005, abundance of the target pest Culicoides hollensis (Melander and Brues) collected in pan traps was asymmetrical between the Control site and Treatment sites and no effects from spray were observed (Table 5). In contrast to Malaise and pan traps, populations of target insects were markedly lower in CDC-style traps after naled applications in October 2003 and April 2005. In 2003,

Culicoides furens (Poey) abundance in CDC-style trap collections on Parris

Island were reduced by 94.3% 1-day after spraying relative to average population levels 5 days preceding the sprays (Fig. 4). In April 2005, the primary pests were the spring and summer active biting midges Culicoides melleus (Coquillett) and

C. hollensis. The day following the application, C. melleus collections fell 99.6 % on average. Likewise, C. hollensis numbers dropped 98.5 % (Fig. 5). Overall biting midge reduction averaged 99.0%. Mosquito numbers were relatively low during the study but showed marked reductions as well. Target insect abundance did not increase following sprays in either year or with any trap type

(Malaise, pan, or CDC-style).

DISCUSSION

Our results indicate that there was higher mortality of biting flies than nontarget species associated with aerial applications of naled. The mosquito and biting midge numbers in CDC-style traps showed major reductions following

143

sprays. While some nontarget species showed significant changes after spraying in 2003 and 2005, many taxa did not. In particular, it should be noted that the four dipteran families (Cecidomyiidae, Sciaridae, Phoridae, and

Chironomidae) that made up 41% of collections, share a similar size-class to both mosquitoes and biting midges and were not significantly affected by sprays.

Shannon diversity was not significantly different following naled applications in the Malaise traps during October 2003 or pan traps in April 2005, but mean values of the indices were never higher after sprays. Although both

PIMCRD sites sampled with Malaise traps had similar H’ in 2003, the control site

(SI) was significantly different (P = 0.899) than the 3 treatment locations during

2005 when sampled with pan traps. This indicates that the control location may have had fundamental differences in overall community structure. Despite the immediate effects (or perceived effects) from pesticide is it likely that such effects on the entire macroinvertebrate community is minimal. For example, Bond et al.

(2007) found that year to year variation in river discharge was more influential in determining insect abundance and community diversity than corollary effects associated with filamentous algae removal from river pools in southern Mexico.

Exposure to insecticides will vary by taxa based on application time and insect diel activity patterns. Therefore, insecticide applications have the highest efficacy when they are timed to correspond with the peak activity period of the target pests (WHO 2003). Salt marsh biting midges such as C. furens and C. hollensis are most active from 1 hour before to 2 hours after sunset (Lillie et al.

144

1987, Breidenbaugh et al., in review), which partially overlaps with the application times in this study and explain the considerable reduction in Culicoides spp. observed following sprays. Likewise, some salt marsh mosquitoes (e.g., Aedes taeniorhynchus (Wiedemann), Ae. sollicitans [Walker]) become active before sunset (Bidlingmayer 1967, Ebsary and Crans 1977). Other Ceratopogonidae are not hematophagous but may exhibit crepuscular flight patterns. Nontarget biting midges (e.g., Atrichopogon sp., Forcipomyia sp.) were collected in pan traps and decreased after sprays at the PF site. These nontargets, which have a similar size-class and may exhibit crepuscular behavior, are likely candidates for experiencing negative effects from aerial applications made near sunset.

However, no effects on nontarget Ceratopogonidae were observed at 2 of the 3

Treatment sites. Exposure of some key diurnal nontarget species such as long- legged flies (Dolichopodidae), would likely be lower if pesticides were applied after dark. However, the USAF Aerial Spray Unit does not currently have an operational profile for making night applications (Breidenbaugh, personal communication). Insecticide applications during this study were made within the final 2 hours prior to sunset, therefore, it is not unexpected that some insecticide effects on nontarget insect numbers after sprays were observed.

The Malaise and pan traps collected both diurnal and nocturnal insects, and therefore examined the impact on a large section of the total insect community. In 2003, four families had lower abundances after naled applications. These families (Dolichopodidae, Sarcophagidae, Tachinidae, and

145

Syrphidae) all are day-active dipterans and contain species that may exhibit resting behaviors (e.g., open canopy) that expose these families to aerial applications. Naled is effective in managing filth fly numbers (DuBose 1970,

Breidenbaugh & Haagsma 2008). Although some pestiferous filth fly families were collected during 2003 & 2005 (Calliphoridae, Muscidae, Sarcophagidae) only the flesh flies (Sarcophagidae) were susceptible to application parameters used at Parris Island.

The long-legged flies (Dolichopodidae) were the most abundant nontarget family collected in both trapping methods. Some have proposed this group as a bioindicator of good quality habitat conditions in salt marsh systems (Pollet

2001). These flies are visual predators and use visual mating behaviors making them largely day-active (Peterson 1960). Therefore, they are presumably less exposed than target species during evening pesticide applications. However, their numbers were significantly lower in Malaise traps after spraying in 2003 but not in pan traps in 2005. Two successive applications were made in 2003, and suggest that these diurnal flies were exposed to a greater degree than during the single application in 2005. Another explanation for varying effects observed on long-legged flies between years is the temporal population’s age such that during fall, older flies may have been more susceptible to naled applications.

Relatively large diurnal Lepidoptera such as skippers (Hesperiidae) and yellows (Pieridae) were collected during the study but in low numbers and were not statically analyzed but were consistently collected both before and after

146

sprays. The artificial group, microlepidoptera, which accounted for 5-6% of the collections in both trapping methods, was not affected by sprays.

Domestic honeybee (Hymenoptera: Apidae) populations are declining and subsequently, the importance of native bees as agricultural pollinators is increasing (Winfree et al. 2007). Pesticides have been listed as one potential factor contributing to the honey bee decline (Stokstad 2007). Although honey bees were not collected in either year, another pollinator bee family, Halictidae, made up 5% of collections in 2003. These were not affected by naled applications, and no other hymenopteran pollinator was affected. However,

Syrphidae and Dolichopodidae, which are dipteran pollinators, were lower after sprays in 2003. These families have been suggested as primary pollinators of mudflat plants (Sawyer et al. 2005) and could function in a similar role at Parris

Island. Other hymenopterans are ecologically important as key predators, parasites, or herbivores. Large wasps (Pompilidae and Sphecidae) and medium sized wasps (Diapriidae and Braconidae) were fairly common in both years, and were not affected by sprays. In contrast, microhymenoptera were significantly affected in 2005.

Species capable of producing multiple generations a season have a stronger potential to have populations recover more quickly from pesticide applications than univoltine species. Additionally, larval stages may escape pesticide effects if their developmental sites offer refuge from insecticide exposure. Many nontarget insects have important roles in the salt marsh food

147

chain both directly or indirectly. For example, breeding birds could potentially be impacted if invertebrate numbers were suddenly lowered from pesticide applications. Conversely, vector control operations may benefit birds by reducing arbovirus transmission within and among populations (Boyce et al. 2007).

The two trapping approaches yielded somewhat different results and underscore the importance of multiple assessment methods in analyzing biological impacts. The pan traps were intended to give a neutral response to surface active diurnal insects while most likely attracting diurnal insects with a phototactically positive response to reflected light; therefore, absence of taxa such as hover flies (Syrphidae) in pan traps was a behavioral function rather than an indication of seasonal absence. These results are not surprising because

Malaise and pan traps were deployed in different seasons and consequently sampled different species. Excluding the artificial group microlepidoptera, the only taxon found in both trap types at greater than 2% was the Dolichopodidae.

Another factor helping to explain differences in observed effects between trapping methods was the additional application of naled that occurred in 2003 which increased nontargets to pesticide exposure. Neither Malaise traps nor pan traps proved effective for measuring mosquito and biting midge abundance in comparison to CDC-style traps. This suggests that baiting traps with attractants can improve impact analyses over passive trapping. The methods used here can be easily adopted to compare nontarget insect effects with other spray regimes.

148

Aerial applications at Parris Island are used to manage populations of biting midges and mosquitoes to protect public health and improve quality of life for military members and civilians working outdoors. Recent studies analyzing effects of pyrethroid insecticides used in mosquito control reported results similar to our findings with naled. Boyce et al. (2007) saw no effect on caged sentinel insects (dragonflies, spider, butterflies, and honey bees) but found effects on small-bodied insects collected on tarps. Likewise, Davis and Peterson (2008) found few and irregular examples of pyrethroid effects on terrestrial nontarget insects. Jensen et al. (1999) investigated both pyrethroids and an organophosphate insecticide and concluded that flying insect abundance in treated and control areas decreased but rebounded within 48-hours. Findings from this study support the idea that aerial applications for biting fly control with naled provide an acceptable level of success against target pests without creating undue harmful effects on nontarget insect abundance, even when applied at maximum label rates. Impacts on nontarget insects, appeared to be greatest in the Diptera but did not follow a recognized pattern (e.g., affecting smaller or larger insects, calypterate or acalypterate flies) and was not consistent among treatment sites. We conclude that responsible mosquito management techniques such as using the appropriate application time relative to the target pest’s activity period and lowest effective quantity of pesticide will limit potential negative effects on nontarget insects.

149

ACKNOWLEDGEMENTS

We would like to thank Bruce Lampwright, of the Spring Island Trust.

Technical assistance was given by Gregg Hunt and Elizabeth Hager of Beaufort

County Mosquito Control, as well as Jim Clark and Robert Brodeur of Parris

Island MCRD. C. Breidenbaugh, K. Haagsma, and S. Olson helped collect and process samples. Special thanks to the aircrews and maintenance personnel of the Air Force Aerial Spray Unit.

REFERENCES CITED

Allen-Wardell G. 1998. The potential consequences of pollinator declines on the

conservation of biodiversity and stability of crop yields. Conserv Biol 12:8-

17.

Bidlingmayer WL. 1967. The influence of environmental factors and

physiological stage on flight patterns of mosquitoes taken in the vehicle

aspirator and truck, suction, bait and New Jersey light traps. J Med Ent

11:119-146.

Blanton FS, Travis BV, Smith N, Husman CN. 1950. Control of adult mosquitoes

in Alaska with aerial sprays. J Econ Entomol 43:347-350.

Bond JG, Quiroz-Martínez H, Rojas JC, Valle J, Ulloa J, Williams T. 2007.

Impact of environmental manipulation for Anopheles pseudopunctipennis

Theobald control on aquatic insect communities in southern Mexico. J

Vector Ecol 32:41-53.

150

Boyce WM, Lawler SP, Schultz JM, McCauley SJ, Kimsey LS, Niemela MK,

Nielsen CF, Reisen WK. 2007. Nontarget effects of the mosquito

adulticide pyrethrin applied aerially during a West Nile virus outbreak in an

urban California environment. J Am Mosq Control Assoc 23:335–339.

Breidenbaugh M, Haagsma K, Olson S, Teig D, Spears B, McHugh C, Walker W,

Sanders D. 2000. Air Force aerial spray operation to control adult

mosquitoes following Hurricanes Katrina and Rita. Wing Beats Fla Mosq

Control Assoc 17:7-15.

Breidenbaugh M, Haagsma K. 2008. The US Air Force Aerial Spray Unit:

A history of large area disease vector control operations, WWII through

Katrina. Army Med Dept J. April-June:1-8.

Breidenbaugh M, Haagsma K, Walker W, Sanders, D. 2008. Post-Hurricane

Rita mosquito surveillance and the efficacy of Air Force aerial applications

for mosquito control in east Texas. J Am Mosq Control Assoc 24:327–330.

Carney RM, Husted S, Jean C, Glaser C, Kramer V. 2008. Efficacy of aerial

spraying of mosquito adulticide in reducing incidence of West Nile virus,

California, 2005. Emerg Infect Dis 14:747-754.

Caron DM. 1979. Effects of some ultra low volume mosquito abatement

insecticides on honey bees. J Econ Entomol 72:148–151.

Coats JR, Symonik DM, Bradbury SR, Dyer DD, Timson LK, Atchison GJ. 1989.

Toxicology of synthetic pyrethroids in aquatic organisms: an overview.

Environ Toxicol Chem 8:671-679

151

Darling DC, Packer L. 1988. Effectiveness of Malaise traps in collecting

Hymenoptera: The influence of trap design, mesh size and location. Can

Entomo. 120:787- 796.

Davis RS, Peterson RKD, Macedo PA. 2007. An ecological risk assessment for

insecticides used in adult mosquito management. Integr Environ Assess

Manag 3:373–382.

Davis RS, Peterson RKD. 2008. Effects of single and multiple applications of

mosquito insecticides on nontarget arthropods. J Am Mosq Control Assoc

24:270–280.

Delaplane KS, Mayer DF. 2001. Crop Pollination by Bees. New York: CABI

Publishing.

Deansfield RD, Brightwell R, Onah J, Okolo Cj. 1982. Populations dynamics of

Glossina morsitans submorsitans, Newstead, and G. tachinoides

Westwood (Diptera: Glossinidae) in sub-Sudan savanna in northern

Nigeria. I. Sampling methodology for adults and seasonal changes in

numbers caught in different vegetations types. Bull Ent Res 72:175-192.

Dubose WP. 1970. Aerial spray tests with caged filth flies. Mosq News 30:266.

Eberhardt LL. 1976. Quantitative ecology and impact assessment. J Environ

manage 4:27–70.

Ellis JI, Schneider DC. 1997. Evaluation of a gradient sampling design for

environmental impact assessment. Environ Monit Assess 48:157–172.

152

Ebsary BA, Crans WJ. 1977. The biting activity of Aedes sollicitans in New

Jersey. Mosq News 37:721-724.

Green RH. 1979. Sampling Design and Statistical Methods for Environmental

Biologists. Chichester, UK: Wiley.

Gunstream SE, Chew RM. 1967. A comparison of mosquito collection by

Malaise and miniature light traps. J Med Entomol 4:495-496.

Jensen T, Lawler SP, Dritz DA. 1999. Effects of ultralow volume pyrethrin,

malathion, and permethrin on nontarget invertebrates, sentinel

mosquitoes, and mosquitofish in seasonally impounded wetlands. J Am

Mosq Control Assoc 15:330–338.

Johansen CA, Mayer DF. 1990. Pollinator protection: a bee and pesticide

handbook. Cheshire, CT: Wicwas Press.

Lillie TH, Kline DL, Hall DW. 1987. Diel and seasonal activity of Culicoides spp.

(Diptera: Ceratopogonidae) near Yankeetown, Florida, monitored with a

vehicle-mounted insect trap. J Med Entomol 24:503-511.

Marston N. 1965. Recent modifications in the design of Malaise traps with a

summary of the insects represented in the collections. J Kans Entomo.

Soc. 38: 154-162.

Mount GA, Biery TL, Haile DG. 1996. A review of ultralow-volume aerial sprays

of insecticide for mosquito control. J Am Mosq Control Assoc 12:601-618.

153

Matthews RW, Matthews JR. 1970. Malaise trap studies of flying insects in a

New York mesic forest. I. Ordinal composition and seasonal abundance. J

N Y Entomol Soc 78:52-59.

Mulla MS, Navvab-Gojrati HA, Darwazeh HA. 1978. Biological activity and

longevity of new synthetic pyrethroids against mosquitoes and some

nontarget insects. Mosq News 38:90-96.

Pinkovsky DD. 1972. United States Air Force aerial spray activities in operation

combat VEE. Mosq News 32:332-334.

Peterson BV. 1960. Notes on some natural enemies of Utah black flies (Diptera:

Simuliidae). Can Entomol 92:266-74.

Peterson RKD, Macedo PA, Davis RS. 2006. A human-health risk assessment

for West Nile virus and insecticides used in mosquito management.

Environ Health Perspect 114:366-372.

Pollet, M. 2001. Dolichopodid biodiversity and site quality assessment of reed

marshes and grasslands in Belgium (Diptera: Dolichopodidae). J Insect

Conserv 5:99-116.

Sawyer NW, Mertins DS, Schuster LA. 2005. Pollination biology of Eriocaulon

parkeri in Connecticut. Aquat Bot 82:113-120.

Shannon CE, Weaver W. 1949. The mathematical theory of communication.

Urbana, IL, USA, University of Illinois Press, 117 pp.

154

Smith EP. 2002. BACI design. In El-Shaarawi AH, Piegorsch WW, eds.

Encyclopedia of Environmetrics. Volume 1. Chinchester, UK: John Wiley

and Sons. p. 141-148.

Stokstad E. 2007. The case of the empty hives. Science 316:970–972.

Swanson C, Cech JJ Jr., Piedrahita RH. 1996. Mosquitofish: biology, culture,

and use in mosquito control. Sacramento: Mosq Vector Control Assoc

Calif and Univ Calif Mosq Research Program.

Their A. 2001. Balancing the risks: vector control and pesticide use in response

to emerging illness. J Urban Health 78:372–381.

Townes H, Townes M. 1981. Species diversity of Ichneumonidae and

Serphidae (Hymenoptera) in an English suburban garden. Bio J Linn Soc

16:315-336.

USEPA [United States Environmental Protection Agency]. 2002. Interim

Reregistration Eligibility Decision for Naled. United States Environmental

Protection Agency, Washington DC.

Winfree R, Williams NM, Dushoff J, Kremen C. 2007. Native bees provide

insurance against ongoing honey bee losses. Ecol Lett 10:1105–1113.

WHO. 2003. Space spray application of insecticides for vector and public health

pest control: A practitioner’s guide. Geneva, World Health Organization.

155

Zhong HE, Latham M, Payne S, Brock C. 2003. Minimizing the impact of the

mosquito adulticide naled on honey bees, Apis mellifera (Hymenoptera:

Apidae): aerial ultra-low-volume application using a high-pressure nozzle

system. J Econ Entomol 97:1-7.

156

Fig. 1. Map of study area showing Malaise, pan, and CDC trap locations on the

Parris Island Marine Corps Recruit Depot, South Carolina. Malaise traps were at

HI and VC. Pan traps were at HI, VC, PF, EB, and SI. CDC-style traps were at

HI, VC, W, and GC.

157

1 Beaufort River

2

7 (7 miles west)

3

Broad River

4 5

Parris Island Marine Corps Recruit 6 Depot, South Carolina 4 1. Horse Island (HI) 2. Veterinary Clinic (VC) 3. Weapons (W) 4. Page Field (PF) 5. Elliot Beach (EB) 6. Golf Course (GC) 7. Spring Island (SI)

Table 1. Parameters measured during naled application at Parris Island October 2003 and April 2005.

Date of trial Application Flow rate Boom Average wind Area Time of Temperature Relative rate Liters per Pressure speed at 2 m treated application Humidity ml/hectare min kPa (psi) & 46 m hectares (sunset) ( km/hr) 6 October 74.0 8.5 207-379 9.7/15.0 3,008 1705-1845 24.4°C 58 – 64% 2003 (1901) 74.0 8.5 207-379 3.2/13.0 1,675 1810-1905 22.2°C 77% 8 October (1859) 2003 74.0 8.5 207-379 13.0/18.5 3,076 1750-1950 22.8°C 43% (1955) 19 April 2005

158

Table 2. Means (1 SE), Shannon's diversity (H') and P-values of nontarget insect abundance collected in malaise traps before and after naled applications in 2003. Sites are Horse Island (HI) and Veterinary Clinic (VC). Significant P-values

(P<0.025, with Bonferroni correction) are in bold.

Taxa % of total HI-before HI-after P VC - before VC-after P Dolichopodidae 23.5% 47.2 ± 5.4 7.5 ± 2.6 0.001 34.2 ± 6.1 15.6 ± 2.6 0.008 Cicadellidae 8.9% 7.8 ± 3.3 20.3 ± 10.3 0.377 1.8 ± 1.8 5.3 ± 2.6 0.356 Sarcophagidae 8.9% 22.6 ± 4.2 7.4 ± 1.8 0.003 9.2 ± 1.7 4.1 ± 1.0 0.018 Anthomyiidae 7.7% 17.8 ± 7.2 4.5 ± 1.2 0.139 17.4 ± 8.4 0.62 ± 0.3 0.117 Bibionidae 6.2% 20.8 ± 7.7 5.6 ± 1.2 0.118 2.2 ± 1.5 2.1 ±1.3 0.972 microlepidoptera 6.1% 6.6 ± 2.3 6.4 ± 1.5 0.933 2.8 ± 1.7 9.5 ± 5.4 0.370 Sphecidae 5.8% 4.2 ± 2.1 4.5 ± 1.0 0.886 8.4 ± 3.5 8.0 ± 1.5 0.906 Halictidae 5.2% 7.4 ± 6.6 2.1 ± 1.0 0.153 9.8 ± 5.2 5.4 ± 1.5 0.452 Tachinidae 4.3% 8.2 ± 1.2 2.1 ± 0.8 0.002 6.4 ± 2.3 3.1 ± 1.1 0.174 Muscidae 4.1% 5.4 ± 2.3 3.8 ± 0.8 0.530 7.4 ± 3.1 2.8 ± 1.0 0.209 Syrphidae 3.2% 4.6 ± 1.8 1.3 ± 0.3 0.041 10.4 ± 2.8 0.63 ± 0.3 0.001 Pompilidae 2.1% 3.0 ± 0.8 2.0 ± 0.5 0.311 3.4 ± 0.9 1.3 ± 0.3 0.080 Total Numbers 100.0% 184.6 ± 53.2 87.6 ± 36.0 0.002 132.0 ± 23.2 69.0 ± 11.6 0.020 H’ 2.50 ± 0.07 2.40 ± 0.14 0.602 2.40 ± 0.10 2.13 ± 0.09 0.086

159

160

Table 3. Means (±1 SE) and P-values of nontarget insect abundance collected in pan traps before and after an aerial naled application in 2005. Treatment sites

(Elliot Beach, Horse Island, Page Field) were compared to the Control site

(Spring Island). Significant P-values (P<0.05, ANOVA, P<0.017, t-test) are in bold and indicate there was a change before and after spraying between the

Treatment sites and the Control site

P-value Mean No. Percent Mean No. P-value Associated insects P-value of Taxa Location insects Period X with t-statistic post-spray B/A Total pre-spray ±SE Location before and ±SE taxa after spray Dolichopodidae Elliot Beach 10.8 ± 4.2 11.0 ± 2.9 .643 .706 15.9% .724 Horse Island 0.5 ± 0.3 4.8 ± 3.6 .193 .714 .704

Page Field 8.8 ± 4.5 16.2 ± 1.9 .117 .223 .318

Spring Island (Control) 6.8 ± 2.4 9.2 ± 1.7 Cecidomyiidae Elliot Beach 5.3 ± 1.6 4.6 ± 1.4 .370 .453 12.7% .687

Cecidomyiidae Horse Island 16.0 ± 6.5 10.2 ± 2.4 .322 .473 .587

Page Field 5.8 ± 1.9 7.8 ± 1.8 .734 .361 .205 Spring Island (Control) 3.8 ± 1.5 2.8 ± 1.0

Sciaridae Elliot Beach 7.0 ± 2.9 4.6 ± 1.4 .460 .441 11.3% .661 Horse Island 5.3 ± 4.3 12.6 ± 3.2 .180 .185 .495 Cecidomyiidae Page Field 8.3 ± 2.1 9.2 ± 4.1 .845 .860 .572 Spring Island (Control) 0.75 ± 0.5 0.8 ± 0.3

Phoridae Elliot Beach 7.0 ± 2.3 3.4 ± 1.4 .257 .121 8.7% .340 Horse Island 6.5 ± 5.3 8.2 ± 4.2 .735 .871 .802 Page Field 4.0 ± 1.1 8.0 ± 3.1 .228 .367 .487 Spring Island (Control) 0 0.6 ± 0.2

161

Table 3. Continued. Ceratopogonidae Elliot Beach 2.0 ± 1.2 1.8 ± 0.8 .437 .577 6.5% .863 Horse Island 2.0 ± 0.9 10 ± 9.0 .521 .388 382 Page Field 2.8 ± 0.8 0.4 ± 0.2 .014 .377 .495 Spring Island (Control) 2.0 ± 0.9 0.8 ± 0.6 Chironomidae Elliot Beach 4.0 ± 0.8 1.2 ± 0.7 .046 .376 5.8% .476 Horse Island 4.0 ± 1.7 1.0 ± 0.3 .067 .391 .622 Page Field 6.8 ± 1.1 7.8 ± 1.8 .971 .429 .604 Spring Island (Control) 2.8 ± 1.1 1.6 ± 0.9 microhymenoptera Elliot Beach 3.3 ± 0.3 1.6 ± 0.7 .043 .858 5.6% .827 Horse Island 4.8 ± 2.8 0.8 ± 0.8 .078 .379 .686 Page Field 7.5 ± 1.7 5.8 ± 2.0 .307 .920 .840 Spring Island (Control) 2.0 ± 1.2 0.6 ± 0.4 microlepidoptera Elliot Beach 1.0 ± 1.0 2.0 ± 0.9 .202 .632 5.1% .836 Horse Island 8.0 ± 2.1 9.2 ± 3.5 .492 .844 .854 Page Field 0 0 .160 .596 .638 Spring Island (Control) 0.3 ± 0.3 2.4 ± 1.7 Diapriidae Elliot Beach 5.3 ± 4.6 0 .233 .200 4.6% .337 Horse Island 5.5 ± 3.0 1.6 ± 1.0 .213 .170 .307 Page Field 4.5 ± 2.5 4.0 ± 3.3 .946 .874 .783 Spring Island (Control) 0 0.2 ± 0.2 Calliphoridae Elliot Beach 6.0 ± 3.0 0.6 ± 0.4 .065 .065 3.4% .188 Horse Island 0.8 ± 0.5 0 .096 .096 .215 Page Field 5.5 ± 3.2 3.0 ± 0.8 .417 .417 .611 Spring Island (Control) 0 0

162

Table 3. Continued.

Braconidae Elliot Beach 0.8 ± 0.5 0.2 ± 0.2 .303 .471 2.6% .718 Horse Island 6.0 ± 2.7 2.8 ± 1.5 .275 .303 .491 Page Field 1.0 ± 0.4 0.2 ± 0.2 .135 .237 .058 Spring Island (Control) 0.5 ± 0.3 0.4 ± 0.2 Empididae Elliot Beach 4.0 ± 1.2 3.6 ± 1.1 .415 .724 2.3% .798 Horse Island 0.5 ± 0.3 0.4 ± 0.4 .103 .176 .391 Page Field 0.3 ± 0.3 0.4 ± 0.2 .114 .039 .080 Spring Island (Control) 1.0 ± 0.4 0 All taxa Elliot Beach 68.5 ± 18.5 40.8 ± 8.0 .185 .230 100% .594 Horse Island 74.8 ± 28.1 73.8 ± 27.6 .954 .990 .931 Page Field 66.8 ± 14.5 67.8 ± 4.7 .981 .884 .843 Spring Island (Control) 24.3 ± 8.3 22.8 ± 5.4

163

Table 4. Means (±1 SE) and P-values of Shannon Diversity Indices (H’) for all insects collected in pan traps before and after an aerial naled application in 2005. Treatment sites (Elliot Beach, Horse Island, Page Field) were compared to the Control site (Spring Island). Significant P-values (P<0.05, ANOVA, P<0.017, t-test) are in bold and indicate there was a change before and after spraying between the Treatment sites and the Control site.

P-value Associated Average H’ Average H’ P-value P-value with t- Location pre-spray post-spray Period X B/A statistic ±SE ±SE Location before and after spray Elliot Beach 2.4 ± 0.1 2.1 ± 0.1 .204 .618 .818 Horse Island 2.3 ± 0.2 2.1 ± 0.1 .332 .800 .902 Page Field 2.2 ± 0.1 2.0 ± 0.0 .135 .489 .977 Spring Island (Control) 1.9 ± 0.2 1.7 ± 0.2

164

Table 5. Means (±1 SE) and P-values of primary target insect (Culicoides hollensis) abundance collected in pan traps before and after an aerial naled application in 2005. Treatment sites (Elliot Beach, Horse Island, Page Field) were compared to the Control site (Spring Island). Significant P-values (P<0.05, ANOVA, P<0.017, t-test) are in bold and indicate there was a change before and after spraying between the Treatment sites and the Control site.

P-value Mean No. Mean No. associated P-value insects insects P-value with t- Taxa Location Period X pre-spray post-spray B/A statistic Location ±SE ±SE before and after spray Culicoides Elliot Beach 1.0 ± 0.6 0.6 ± 0.6 .398 .412 .235 ppsphollensishollensis Horse Island 3.0 ± 1.0 1.2 ± 0.8 .365 .453 .300 Page Field 0 0.2 ± 0.2 .412 .402 .247 Spring Island (Control) 36.5 ± 22.6 17.8 ± 8.1

165

166

120 before spray 100 after spray

80

60

40

Average No. ofAverage insects . 20

0 Elliot Beach Horse Island Page Field Spring Island

Fig. 2. Mean (+/- 1 SE) total nontarget insect abundance collected in pan traps before and after a single naled application in 2005.

167

250 before spray (30 Sept) after spray (9 Oct) 200

150

100

50 number of individuals of . number

0

C. furens C. melleus C. hollensis Ae. sollicitans

Ae. taeniorhynchus Cx. quinquefasciatus

Fig. 3. Number of target pests (biting midges and mosquitoes) 5 days before and

1 day after naled sprays in 2003 at Parris Island MCRD, SC. Data was averaged from 4 CO2-baited, CDC-style traps, operated for 24-hrs and located inside of the spray area. [C. = Culicoides; Ae. = Aedes; Cx. = Culex].

168

1800 before spray (18 Apr) 1600 after spray (21 Apr) 1400 1200 1000 800 600 400

number of individuals of . number 200 0

C. furens C. melleus C. hollensis Ae. vexans Ae. atlanticusCx. salinarius

Ae. taeniorhynchus

Fig. 4. Number of target pests (biting midges and mosquitoes) 1 day before and

2 days after a naled spray in 2005 at Parris Island MCRD, SC. Data was averaged from 4 CO2-baited, CDC-style traps, operated for 24-hrs and located inside of the spray area. [C. = Culicoides; Ae. = Aedes; Cx. = Culex].

CHAPTER VI

GENERAL DISSCUSSION AND CONCLUSIONS

The overarching goal of the interconnected research projects described in previous chapters was, simply put, to improve the integrated pest management of biting flies at the Marine Corps Recruit Depot, Parris Island, SC in order to reduce the risk of vector-borne illness and improve the quality of life for recruits and other military personnel. To accomplish this, the composition of pest species and their activity patterns were studied in combination with the primary mechanism used for temporarily lowering pest densities at Parris Island: aerial applications of pesticides. Generalized findings from all chapters include the following:

1. Three species of biting midges and 5 species of mosquitoes were

found to be seasonally abundant at Parris Island.

2. Abundance of the most common species was positively correlated with

temperature.

3. Biting flies were most active the within 2 hours of sunset.

4. Due to the potential of some biting flies to transmit disease or promote

dermatitis, high densities observed during some seasons (i.e., spring)

suggest the need for intervention using pest control measures.

169

170

5. Droplet size produced from US Air Force (USAF) fuselage boom

configurations using flat-fan nozzles were 11.4 µm for 8001 and

54.3 µm for 8005 sized nozzles.

6. AGDISP predictions were better correlated with larger drops than

smaller ones when using empirical data.

7. Single pass trials with the USAF fuselage booms and 8005 nozzles

demonstrated an effective swath width of 639 m. This was determined

using caged mosquitoes and an organophosphate pesticide (naled).

8. Applications with fuselage booms were also effective in reducing biting

midge and mosquito numbers in large area (~3,000 ha) applications

under field conditions.

9. Applications of the pesticide naled at Parris Island did not create

biological vacuums following sprays. However, negative effects were

found in approximately 29% of the families that were captured using

pan traps and malaise traps. Diversity was not found to be affected

following pesticide sprays.

The nine generalized findings are further expanded upon below:

1. Seasonal abundance of biting midges and mosquitoes

Biting midges and mosquitoes are troublesome pests that have impacted human behavior along the Atlantic seaboard for years (Blanton and Wirth 1979).

The Culicoides species found at Parris Island are most likely to be a problem

171

during warm wet periods but since all 3 species can seasonally reach pest proportions, recruits and other military members can be exposed throughout the year. Nonetheless, late spring is the most intense period for biting flies at Parris

Island as all 3 Culicoides species are present and a significant amount of mosquito activity occurs then as well.

2. Most abundant species correlated with temperature

The biting midge C. furens and the mosquito, Ae. taeniorhynchus were the most abundant species encountered during 2001-2004. In the nearly subtropical environment of South Carolina, some measurable biting fly activity was possible any day of the year. However, the most abundant species were found to be positively correlated with temperature. This means that in summer months, C. furens and Ae. taeniorhynchus are likely to be the primary pests. This was the case when densities were consistently high enough for the Department of

Defense to complete 8 consecutive days of repellent formula field tests in 2004.

This finding was in direct contradiction to local lore. For example, one pest controller stated at the beginning of the project, that biting midges were not active during the summer months. This is likely a reflection of individual work behavior and the fact that good weather in spring results in a disproportionate amount of time spent outside relative to the time spent in the field during hot summer months.

172

3. Diel activity of biting flies

Using rotating traps to identify activity periods of biting flies yielded useful information. Perception of pest pressure is often relative to an individual’s habit patterns and statements such as “the sandflies are worst on cool mornings” may be directly related to experiences gathered during a group run occurring regularly just after sunrise. The person making this statement may not have any experience with biting pressure at any other time of day, but still reaches such a conclusion and, potentially, may pass this information on to newly arriving individuals. The routine sampling method here found minor variability between seasons but when all collections were combined, showed a clear crepuscular activity period for the most abundance biting fly species at Parris Island.

4. Biting flies of Parris Island as vectors

In addition to their annoying biting habits, mosquitoes create an additional concern as vectors of human pathogens (Service 2000). The most likely pathogen encountered at Parris Island is West Nile virus and of the 18 mosquito species found, Culex quinquefasciatus is the most likely vector of this arbovirus

(Hayes et al 2005). This species is not associated with salt marsh ecology and control measures should focus on eliminating sources of standing eutrophic fresh water (Metzger et al. 2008). Other prevalent mosquitoes such as Ae. taeniorhynchus, Ae. sollicitans, and Cx. salinarius are potential vectors of encephalitides (Ortiz et al. 2003). Surveillance trapping should be expanded to

173

include traps targeting gravid females as the most important vector component since vertical transmission is low if it occurs at all (Kramer et al. 2008).

5. Droplet sizes produced by USAF fuselage booms

Physical characteristics of the droplet spectra produced during aerial spray operations were measured for 2 nozzle types. Droplet VMD was found to be at the low end (11.4 µm) and above (54.3 µm) the industry accepted optimal size of approximately 15 µm for mosquito control for the respective nozzle sizes of 8001, 8005 (Mount 1970, Weidhaas et al. 1970). These droplet sizes are smaller than what has been reported in wind tunnel tests using similar nozzles but at lower wind speeds (Hornby et al. 2006). This means that either the glass slide method is inaccurate or that at faster wind speed the droplets are further broken into smaller droplets by wind shear (Lakhamraju 2005). Laser diffraction measurement of the USAF spray configuration is vital to solving the question of droplet size but currently no facility is available which meets both the wind speed and use of pesticide requirements. Determining droplet sizes with laser diffraction technology remains a research priority for the USAF while use of flat- fan nozzles continues.

6. AGDISP predictions

Droplet characterization studies in Chapter 3 generated specific input parameters for use in the AGDISP computer model beyond what is available in the model’s library. In addition, moving the placement of the booms further

174

challenged the model’s predictive strength. The AGDISP model was originally created for use in agricultural systems and the US Forest Service further developed the Gaussian diffusion portion of the model at the request of vector control aerial spray operators (Teske et al. 1996). Surprisingly few tests were accomplished to confirm its downwind predictive ability for small drops (<50 µm), most likely because of the difficulty of working with small drops and because the near-wake Lagrangian portion of the model worked well (Rafferty and Bowers

1993). Nonetheless, this model has been incorporated into some commercially available flight navigation products to predict droplet fate since the model is available in the public domain (Bilanin et al. 1989). For these reasons, AGDISP model testing with the new fuselage boom configuration in the current study was important. If laser diffraction analysis is accomplished for the USAF fuselage configuration in the future, it will be interesting to see if the AGDISP model predictions more closely approximate empirical data for small drops.

7. Single pass trials

The single pass trials at Craney Island, demonstrated that the USAF fuselage boom configuration was as effective as the previous wing boom configuration in the swath width it generated (Burkett et al.1996). These open area trials are obviously a “best case scenario” because mosquito control aerial sprays are often conducted over various degrees of canopy rather than grasses and bare ground; subsequently, a thick canopy can filter out drifting pesticide

(Brown et al. 2005). After recognizing that a 610 m swath width generated

175

acceptable results, this configuration was quickly tested in operational settings and was used to spray nearly 1.2 million hectares following Hurricanes Katrina and Rita in Louisiana and Texas (Breidenbaugh et al. 2006, Breidenbaugh et al.

2008).

8. Large area applications under field conditions

The small size of biting midges would require a fine mesh screening to hold them in bioassay cages. It has been shown that such fine mesh screens block pesticide drops from passing through the cage (Boobar et al. 1988). Thus, before and after spray comparisons must be made instead. One advantage of comparing field populations is that they are indicative of what takes place in an operational setting, however, the disadvantage is that the addition of variables such as weather, insect emergence, and immigration patterns can make analysis more difficult. To improve the strength of these comparisons, a control site was also chosen. The Spring Island location appeared to be ecologically similar to

Parris Island and preliminary sampling revealed the same pest species were present there. Likewise, during the April and May 2006 trials no significant differences were found between midge densities at Parris and Spring Islands.

Unfortunately, pest populations began to drop during the 2 week experimental period. Normally, operational applications would not be made under these conditions but as it is extremely difficult to budget for these types of trials (e.g., airplane, pesticide, staff, and lodging expenses) the test went forward. In particular, the Anvil trials did not receive a fair shake because caged mosquitoes

176

were stressed after being held overnight. The Anvil trials should be repeated when possible.

9. Nontarget effects of naled

Perhaps the most interesting project in the dissertation was testing nontarget effects of naled when applied to control biting flies. The null hypothesis was that aerial sprays would not have significant effects. However, naled is a broad spectrum pesticide with a well-documented efficacy on a variety of pests, so I expected to find an array of significant effects on various taxa (Dubose 1970, Haile et al. 1984). Post-spray effects were observed following naled sprays but only a third of the taxa encountered in the study were negatively affected while target pest species were simultaneously reduced. The use of an appropriate pesticide droplet size and timing applications to coincide with the primary activity period of target pest species undoubtedly will serve to reduce the potential for nontarget effects associated with aerial vector control.

GENERAL CONCLUSIONS

The mantra of the great research machine is, of course, further studies needed. This project is not the exception. For example, additional field work with pyrethroids for controlling biting midges is required and it would be prudent to determine if lower rates of naled are still effective (Brown et al. 2006).

Likewise, laser diffraction analysis of the droplet spectra generated by the USAF fuselage boom configurations is another near future project.

177

Apart from planning future research projects, it can also be noted that the primary goal of the current project was accomplished: improve pest management practices at the MCRD, Parris Island. The incorporation of surveillance methods has been a technology transfer to the pest control program at Parris Island and pest managers can anticipate seasons when biting flies are most active. At the same time, pest managers are strictly charged with not over- treating with pesticides or placing an undue burden on nontarget species. The effects of aerial applications on nontarget insects are now better understood. A major tenent of integrated mosquito management programs is to reduce vectors to below the injury threshold (Kogan 1998). This threshold varies among individuals but is likely to be quite high for the military community at Parris Island.

Since this group is potentially exposed to vector and nuisance biting, to a greater degree than the general public, monitoring for vector species of mosquitoes and dermatitis inducing midges is a critical aspect of the pest control program and appears to now be firmly in place.

REFERENCES CITED:

Bilanin AJ, Teske ME, Barry JW, Ekblad RB. 1989. AgDISP: the aircraft. Spray

dispersion model code development and experiment validation. Trans ASAE

32:327-334.

Blanton FS, Wirth WW. 1979. The sand flies (Culicoides) of Florida

(Ceratopogonidae: Diptera). Arthropods Fla Neighb Land Areas 10:1–

204.

178

Boobar LR, Dobson SE, Perich MJ, Darby WM, Nelson JH. 1988. Effects of

screen materials on droplet size frequency distribution of aerosols entering

sentinel mosquito exposure tubes. Med.Vet. Entomol. 2:379-384.

Breidenbaugh M, Haagsma K, Olson S, Teig D, Spears B, McHugh C, Walker W,

Sanders D. 2006. Air Force aerial spray operation to control adult

mosquitoes following Hurricanes Katrina and Rita. Wing Beats Fla Mosq

Control Assoc. 17:7-15.

Breidenbaugh M, Haagsma K, Walker W, Sanders, D. 2008. Post-Hurricane

Rita mosquito surveillance and the efficacy of Air Force aerial applications

for mosquito control in east Texas. J Am Mosq Control Assoc 24:327-330.

Brown JR, Reynolds WH, Palmisano C, Taylor V. 2005. Aerial optimization and

canopy penetration study of Dibrom 14 Concentrate. J Am Mosq Control

Assoc 21:106-113.

Brown JR, Rutledge CR, Reynolds W, Dame DA. 2006. Impact of low aerial

application rates of Dibrom 14 on potential vectors. J Am Mosq Control

Assoc 22:87-92.

Burkett DA, TL Biery, DG Haile. 1996. An operational perspective on measuring

aerosol cloud dynamics. J Am Mosq Control Assoc 12:380-383.

Dubose WP. 1970. Aerial spray tests with caged filth flies. Mosq News 30:266.

179

Haile D, Kline D, Reinert J, Biery T. 1984. Effects of aerial applications of

naled on Culicoides biting midges, mosquitoes and tabanids on Parris

Island, South Carolina. Mosq News 44:178-183.

Hayes EB, Komar N, Nasci RS, Montgomery SP, O’Leary DR, Campbell GL.

2005. Epidemiology and transmission dynamics of West Nile virus

disease. Emerg Infec Dis 11:1167-1173.

Hornby JA, Robinson J, Opp W, Sterling M. 2006. Laser-diffraction

characterization of flat-fan nozzles used to develop aerosol clouds of

aerially applied mosquito adulticides. J Am Mosq Control Assoc 22:702-

706.

Kogan M. 1998. Integrated pest management: Historical perspectives and

contemporary developments. Annu Rev Entomol 43:243-70.

Kramer LD, Styer LM, Ebel GD. 2008. A global perspective on the epidemiology

of West Nile virus. Annu Rev Entomol 53:61-81.

Lakhamraju, RR. 2005. Liquid jet breakup studies in subsonic airstream at

elevated temperatures [MS thesis]. University of Cincinnati. 86 p.

Metzger ME, Myers CM, Kluh S, Wakoli JW, Hu R, Kramer VL. 2008. An

assessment of mosquito production and nonchemical control measures in

structural stormwater best management practices in Southern California.

J Am Mosq Control Assoc 24:70-81.

Mount GA. 1970. Optimum droplet size for adult mosquito control with space

sprays or aerosols of insecticides. Mosq News 30:70-75.

180

Ortiz DI., Wozniak A, Tolson MW, Turner PE, Vaughan DR. 2003. Isolation of

EEE Virus from Ochlerotatus taeniorhynchus and Culiseta melanura in

coastal South Carolina. J Am Mosq Control Assoc 19:33-38.

Rafferty, J. E. and J. F. Bowers. 1993. Comparison of FSCBG spray model

predictions with field measurements. Environ Toxicol Chem 12:465-480.

Service MW. 2000. Medical Entomology for Students, 2nd Ed. Cambridge UK:

University of Cambridge Press.

Teske ME. 1996. An introduction to aerial spray modeling with FSCBG. J Am

Mosq Control Assoc 12:353-358.

Weidhaas DE, Bowman MC, Mount GA, Lofgren CS, Ford HR. 1970.

Relationship of minimum lethal dose to the optimum size of insecticides

for mosquito control. Mosq News 30:195-200.