The Pennsylvania State University

The Graduate School

College of Agricultural Sciences

SPECIES-SPECIFICITY OF THREE COMMONLY USED AND TWO NOVEL

MOSQUITO FIELD-SAMPLING DEVICES

A Thesis in

Entomology

by

Loyal Philip Hall

© 2012 Loyal Philip Hall

Submitted in Partial Fulfillment

of the Requirements

for the Degree of

Master of Science

May 2012

ii

The thesis of Loyal Philip Hall was reviewed and approved* by the following:

Gary Felton Professor and Department Head of Entomology

Thomas Baker Distinguished Professor of Entomology Thesis Advisor

James Marden Professor of Biology

Michael Saunders Professor of Entomology

Matthew Thomas Professor of Entomology

*Signatures are on file in the Graduate School. iii

Abstract

Effective sampling is a stepping-stone to efficient use of resources, targeted control efforts, and success in nuisance or vector management. Effective sampling to identify locations where disease-vectoring mosquitoes are present and to monitor population levels allows control measures to be targeted towards medically important mosquitoes, and can reduce the environmental and financial costs associated with widespread, indiscriminate pesticide application while also preventing the failure to initiate control in an area due to a perception that there are few important mosquitoes present. A comparative study between the CDC light trap,

ABC light trap, Reiter-Cummings gravid trap, and two traps developed by the author was conducted to test for species-specificity of each trap type. It was found that while no trap was superior over-all, certain species of mosquitoes are more likely to be detected and their populations monitored by some types of traps compared to others and the novel traps were shown to often be as effective in sampling certain important target species of mosquito as the tested commercial mosquito traps. As with the other devices, for some species the novel traps were superior and for others they appeared to be a less effective sampling device. For example,

Co. perturbans tended to prefer CDC light traps, Cx. salinarius tended to prefer Hall light traps, and Cx. pipiens tended to prefer Hall gravid and Reiters-Cummings gravid traps over the other traps included in the study. 14 different species were analyzed for trap preference; results were also analyzed for WNV-infections and variety of species.

iv

Contents

List of Figures v List of Tables vii List of Abbreviations viii Acknowledgements ix Chapter 1, An introduction to mosquito sampling 1

Chapter 2. Species specificity of three commonly used mosquito sampling devices and two novel devices in the field. 9

Introduction 9

Materials and Methods 10

Results 18

Discussion 36

References 43

Appendix A: Number of Mosquitoes In Each Trap Type By Species 46

Appendix B: Minitab Output of Statistical Tests 47

v

Figures

Chapter 1. Figure 1. Larvae. Figure 2. Pupae. Figure 3. Standard dipper Figure 4. New Jersey light trap Figure 5. ABC and CDC light traps Figure 6. Reiter-Cummings gravid trap

Chapter 2. Figure 7. Hall trap base Figure 8. Hall light trap Figure 9. Hall gravid trap Figure 10. ABC light trap Figure 11. CDC light trap Figure 12. RC gravid trap Figure 13. A typical set-up of a randomized complete block Figure 14. Histogram of the number of individuals of each species caught at a wastewater treatment plant site using the data from all 5 trap-types combined Figure 15. Histogram of the number of individuals of each species caught at a human-made wetland site using the data from all 5 trap-types combined Figure 16. Histogram of the number of individuals of each species caught at a wooded site prone to flooding using the data from all 5 traps combined Figure 17. Mean number of Culex pipiens captured per trap in the five different trap-types Figure 18. Mean number of Culex restuans captured per trap in the five different trap types Figure 19. Mean number of Culex salinarius captured per trap in the five different trap types Figure 20. Mean number of WNV-positive sub-samples per trap in the five different trap- types Figure 21. Mean number of Ochlerotatus trivitattus captured per trap in the five different trap-types Figure 22. Mean number of Aedes vexans captured per trap in the five different trap-types Figure 23. Mean number of ferox captured per trap in the five different trap- types Figure 24. Mean number of Anopheles quadrimaculatus captured per trap in the five different trap-types Figure 25. Mean number of Coquillettidia perturbans captured per trap in the five different trap-types Figure 26. Mean number of Anopheles punctipennis captured per trap in the five different trap-types Figure 27. Mean number of Ochlerotatus canadensis captured per trap in the five different trap-types Figure 28. Mean number of Ochlerotatus japonicus captured per trap in the five different trap-types Figure 29. Mean number of Ochlerotatus triseriatus captured per trap in the five different trap-types Figure 30. Mean number of captured per trap in the five different trap-types vi

Figure 31. Mean number of Psorophora horrida captured per trap in the five different trap- types. Figure 32. Mean number of mosquitoes, for all species combined, captured per trap in the five different trap-types Figure 33. Mean number of different species captured per trap in the five different trap-types Figure 34. Mean number of species (non- Culex pipiens/restuans) captured per trap in the five different trap-types

vii

Tables

Table 1. Number of mosquitoes in each trap type by species A Table 2. Number of mosquitoes in each trap type by species B Table 3. Number of mosquitoes in each trap type by species C Table 4. Raw two-way ANOVA test results.

viii

Abbreviations

ABC American Biophysics Corporation Ae. Aedes An. Anopheles CDC Centers for Disease Control Co. Coquillettidia Cx. Culex EEE Eastern Equine Encephalitis Och. Ochlerotatus Ps. Psorophora RC Reiters-Cummings WNV

ix

Acknowledgements.

The Pennsylvania West Nile Program was instrumental to the success of this study. Their laboratory performed all of the identifications with the highest levels of skill and professionalism. The management allowed the use of this valuable Commonwealth resource because they saw the value in this study and the benefit it could bring to disease management and making their own program more effective at protecting human and health.

1

Chapter 1.

An overview concerning mosquito sampling.

Mosquito-vectored diseases, including malaria, yellow fever, eastern equine encephalitis, and West Nile virus, among others, kill millions of people each year and sicken hundreds of millions more. These numbers make mosquitoes the deadliest on earth. According to the Centers for Disease Control, a major priority in combating mosquito-vectored diseases is controlling mosquito populations through integrated pest management practices that incorporate surveillance, habitat modification, and changes in human behavior, along with pesticide applications that employ the least environmentally damaging yet effective control products available (Gubler et al., 2003). Effective sampling to identify locations where disease-vectoring mosquitoes are present and to monitor population levels allows control measures to be targeted towards medically important mosquitoes, and can reduce the environmental and financial costs associated with widespread, indiscriminate pesticide application while also preventing the failure to initiate control in an area due to a perception that there are few medically important mosquitoes present. The goal when sampling mosquitoes is to develop an accurate picture of the population of important species. There may be dozens of species present in a particular area.

However, only certain species are important to a mitigation program or to public health officials.

This study focuses on determining the efficacy of widely used mosquito sampling devices along with experimental devices developed by the author for specific mosquito species.

Pennsylvania is well suited to mosquito studies. It is on the frost line, and here the ranges of northern and southern species overlap. The geologic history of the region has resulted in a wide variety of ecosystems located in close proximity. Human development of large swaths of land has resulted in the fracturing of natural habitats, reducing predators and allowing r- 2 strategists, like mosquitoes, to flourish. These urbanized areas provide prime habitats for many anthropophilic species. This all results in a large variety of disease-vectoring and nuisance species in Pennsylvania, where 59 species of mosquito are recognized by the state’s West Nile

Virus Program. In Pennsylvania and the surrounding areas, several mosquito-vectored maladies are endemic, including West Nile virus (WNV), St. Louis encephalitis (SLE), LaCrosse encephalitis, eastern equine encephalitis (EEE), and dog heartworm. The introduction/reintroduction of other mosquito-vectored diseases is a concern of the state’s WNV program with the increase in international travel and invasive species. Malaria and chikungunya have appeared in sporadic outbreaks in border states. Dengue is a concern for introduction due to the invasive species Aedes albopictus that is now common in many parts of Pennsylvania, although it is currently not found in Lebanon County where the trials reported herein were conducted.

There are many methods for sampling mosquitoes. Because mosquitoes have an aquatic larval stage and a flying adult stage the sampling processes can be divided into adult and larval sampling techniques.

Mosquito larvae are aquatic, but do breathe air. Many genera, Figure 1. Larvae. such as Culex and Aedes have a snorkel-like “siphon” on their dorsal posterior, whereas the genera Anopheles simply have spiracles in the same location through which they breathe (Fig. 1). Pupae of all species have two breathing tubes located dorsally called “trumpets” (Fig. 2). Most Figure 2. Pupae. mosquito larvae and pupae hang on the surface tension of the water, with their breathing apparatus exposed to the air and the rest of the organism under water. Larvae and pupae are most commonly sampled while resting on the water surface using a “dipper”(Fig. 3), (Leisnaham 3 et al., 2005). Standardized dippers consist of a long telescopic or fixed handle with a 1-pint white plastic cup affixed to the end (Hagstrum, 1971). Techniques for using a dipper include scooping out larvae, placing the dipper next to larvae to allow them to flow into the cup, or scraping it against vegetation or the substrate on which larvae may be sheltering. The larvae can be removed into collection vials by pouring them out or, more commonly, by pipetting them into the vials. Another strategy is to use smaller, non-standard dippers, such as a white kitchen ladle, to sample larvae from narrower containers like discarded tires. For even narrower openings, such as tree-holes or human refuse, a

large pipette (a turkey baster is common) can be used. The Figure 3. Standard dipper. larvae of one particular species, Coquillettidia perturbans, insert their siphons into the stems of aquatic plants, using them as extended snorkels, and must be collected by pulling up the plants and agitating them in a container that is then inspected for the larvae. The number of larvae and species content gathered by dipping is highly dependent upon operator skill and technique used

(Miura et al., 1970; Hagstrum, 1971; Leisnham, 2005). Certain species’ habits make them more conspicuous than others, whereas others will hide out on the bottom for a long time when disturbed, out of sight and uncaught while breathing subcutaneously. Early instars can easily be missed due to their size and lack of contrast in the dipper. For all these techniques, the larvae and their medium can be transferred to white sorting pans to increase contrast and thus increase detection capabilities.

Among the strategies for collecting adult mosquitoes are aspiration, landing counts, and trapping. Adult mosquitoes do , but most species are very weak flyers and unless actively seeking some resource, they prefer stay at rest in some sort of sheltered environment such as 4 vegetation or on the walls of a building. Mosquitoes at rest can be collected by aspiration (Chen et al., 2011). However aspiration of mosquitoes at rest is labor intensive and requires relatively quiescent mosquitoes.

Sampling adult mosquitoes can be performed most easily by offering some resource that serves as an attractant. The oldest of these techniques is a landing count survey in which humans offer themselves as the attractant. The number of mosquitoes landing in a specified period of time can be counted and aspirators can be used to collect the mosquitoes landing on the person for identification or testing, when called for. Landing count surveys work well for anthropophilic species that are host-seeking (Ndiath et al., 2011; Hiwat et al., 2011). However, not all species are so inclined to feed primarily on humans, and humans differ between individuals in mosquito attractiveness (Costantini et al., 1997; Lefevre et al., 2010). There is often a short time-window during which host-seeking behavior is at its zenith, which greatly limits the number of samples that can be taken by an individual or team. Other concerns are possible health risks and violations of worker protection laws (Ndiath et al., 2011). Trapping mosquitoes lured to an effective attractant has become increasingly popular because it is cost- and labor-effective and traps can sample the less anthropophilic species that often play key roles in disease cycles.

One of the earliest and most well known traps is the New Jersey light trap

(Reinert, 1989) (Fig. 4). These traps are baited using a light bulb, and a fan then directs the into a collection chamber containing a killing agent. The trap is

Figure 4. New Jersey Light Trap. 5 powered with an AC electric source. These traps are still in use today. Their fixed locations and constant collections allow for consistent, long-term data collection as well as pre- and post- pesticide application sampling without the variability introduced by changing trap locations.

Many species can be well sampled using light as an attractant. However, some species are not, and many other non-target insects besides mosquitoes are collected and confound the processing of the samples (Reinert, 1989). The traps are also not portable and require AC current.

CO2-baited traps are also commonly called “light traps” because they are based on the

New Jersey light trap design and were historically baited with a light source (Derraik et al.,

2010). DC current (batteries) powers the fan that creates air-flow to draw mosquitoes into a collection chamber, without a killing agent. These features allow for greater portability but also make the traps vulnerable to theft and battery failure. The later introduction of a no-kill collection chamber allowed for better disease detection. CO2 was also added later as an additional attractant (Reeves, 2001). It was found that by removing the light and using solely CO2 the traps collected a “cleaner” sample, i.e., having a larger portion of the sample consisting of mosquitoes and a lower percentage of non-target insects. Although the light is Figure 5. ABC (left) and CDC light traps. still an option for most light traps, it is commonly turned off. Two of the more common CO2- baited light traps are the ABC and the CDC light traps (Fig. 5).

Dry ice or CO2 compressed gas tanks are the most common sources of CO2. There are transportation and supply logistics for both options. Dry ice is the most common CO2 form, due to its portability and availability, in well populated, developed regions. Compressed-gas CO2 tanks are more common when dry ice is not feasible or when CO2flow rates need to be very accurately controlled (Jawara et al., 2011). With dry ice, there is the issue of quality. Dry ice 6 can be of different forms, ranging from powdery, loosely packed blocks, to hard pellets, or else very dense, solid blocks. These different forms of dry ice sublimate at faster or slower rates, respectively, creating highly variable CO2 emission rates and confounding trap performance.

Gravid traps (Fig. 6) are those that attract ovipositing female mosquitoes. These are baited with an “infusion”, consisting of hay, grass, or other vegetation that has been allowed to decompose in water for several days or weeks, imbuing the water with bacteria, decomposing organic matter and volatile emissions that are attractive to the females of many species. The infused water can thus

function as an effective bait for this type of trap. The emitted volatiles Figure 6. Reiter- Cummings gravid trap. attract gravid females looking for quality larval habitat in which to lay eggs. The mosquito species that are attracted are limited to those that lay eggs directly in water rather than those that lay eggs on a dry substrate that is likely to be flooded later (Chen et al 2010). A DC-powered motorized fan draws the females that have arrived to investigate the bait into a collection chamber or net. The battery power for gravid traps imposes the same benefits and drawbacks as light traps, including issues of portability, theft, and battery failure.

Despite the term “hay infusion”, there can be quite a bit of variability in the ingredients used along with duration of brewing, pre-existing bacterial culture, ambient temperature, sunlight, air flow, etc. (Ponnusamy et al., 2009; McPhatter et al., 2009). The creation of the infusion varies, and each person or program often has their own “recipe.” The standard procedure for Pennsylvania’s West Nile program is to stuff a large mesh bag with straw, which is then placed in a 65 gallon black trash can, filled with water and a small amount of milk or milk albumin, and then left in a sunlit area for at least a week. The vegetation used to create the infusion does influence which species are likely to be caught, and many tree-hole species such as 7

Och. triseriatus and Och. hendersoni will avoid the highly organic straw infusions, preferring an oak leaf infusion (Trexler et al., 1998). The amount of infusion used influences sampling efficacy. A study by Michael Hutchinson presented at the 2010 Mid-Atlantic Mosquito

Association Meeting (unpublished) showed that the distance between the water surface and the sample intake pipe, which is a result of using greater or lesser amounts of infusion, directly affected the number of mosquitoes collected (i.e. the greater the distance the greater the number of mosquitoes captured).

Other traps have been created as a result of an inability to collect important disease- transmitting species using any of the above types of traps. Aedes albopictus, a dengue- transmitting and highly anthropophilic species is notoriously disinclined to being sampled by customary traps (Barnard et al., 2011; Farajollahi et al., 2009). For many years, landing counts and larval surveys were the only way to somewhat reliably sample this important mosquito. Of the several novel types of traps created to sample this species, the most well known is the BG

Sentinel, which uses a lure containing 1-octen-3-ol (“octenol”) and lactic acid as a volatile lure, along with visually contrasting colors and a counterflow capturing technique, plus optional CO2

(Rajollahi et al., 2009). Other trap-types targeting Ae. albopictus include the Faye-Prince and the

Zuumba traps. Culiseta melanura is a bird-feeding mosquito species that serves to amplify eastern equine encephalitis in the wild. This species is rarely detected using any kind of trap.

Adding to the degree of difficulty, the larvae use hidden crypts beneath trees. For this and other such challenging species, resting boxes may be used, which are basically boxes with one lateral side open (Howard et al., 2011). Resting boxes provide a humid, dark, sheltered environment for mosquitoes to repose. They can be wooden or cardboard, painted, lined with cloth or unfinished

(Govella et al., 2011). An operator checking the box will aspirate the resting mosquitoes from the walls or else place a cover on it and place the entire resting box into dry ice, a freezer, or use 8 some other method to kill or incapacitate the mosquitoes for collection. Although this does not result in large numbers of mosquitoes being collected, compared to using a gravid trap for Culex restuans for example, it is often the only viable strategy for species such as this, despite its being labor intensive.

Sampling success when using mosquito traps is highly dependent upon operator skill in trap placement and attractant quality. Proximity to refuge habitats, trap height, amount of attractant used, attractant formulation, the juxtapositioning of the dry ice container and the trap, contamination of traps with substances on the operator, and micro-habitats that create fluctuations in wind, lighting, and humidity can all influence the sampling results. This variability demands training and experience for those collecting samples in the field. Despite the inherent variability, the use of mosquito traps is still more reliable than the use of landing counts or personal observations.

9

Chapter 2.

Species specificity of three commonly used mosquito sampling devices and two novel devices in the field.

Introduction.

CO2-baited light traps and gravid traps are two of the most common types of devices used for vector mosquito surveillance in the world. As a West Nile mitigation program manager, I have found myself usually needing to carry a full complement of both gravid and CO2-baited traps. The models that are commonly used are very bulky. I have developed a novel trap, a compact convertible trap that can be used as either a gravid or a CO2 light trap. Comparative studies between multiple models of gravid traps or different models of CO2 trap for particular species efficacies are uncommon, yet would be very useful for improving vector sampling efficacy. Hence, this need formed the impetus for the current study.

Any particular pathogen is normally vectored by certain mosquito species more than others because different mosquito species have different behavior patterns and physiologies.

These behavioral differences also imply that certain types of traps may be more effective than others at capturing particular species (Vaidyanathan and Edman, 1997) because of the traps’ various shapes, sizes, colors, airflows, vibrations, etc. (Allen and Kline, 2004). For example, a particular trap may be more effective at sampling Co. perturbans (vector of eastern equine encephalitis) while another may be more effective for Cx. pipiens (vector of West Nile Virus). It is predicted that through this study, each model tested will have been shown to have different optimal efficacies for particular species, which will inform vector control specialists how to optimize their sampling protocols for different target mosquito species through the use of appropriate traps. 10

Materials and Methods.

Because my novel devices are new to mosquito researchers,

further description is warranted. I designed the Hall Trap to be a

compact, convertible trap that can be used as either a CO2 light trap

or a gravid trap. The trap body is a cylinder with a fan at the

exhaust end and a collection chamber between the intake and the

fan (Fig. 7). The lack of fan “pass-through” by this technology Figure 7. Hall trap base. results in better quality samples. Black burlap cloth encircles the intake

opening to create contrast and to absorb aromas, giving the exterior a more

organic appearance. In light-trap mode, the trap base is hung beneath a 2-

liter insulated beverage container having holes drilled in the sides that is

filled with dry ice (Fig. 8). In gravid-trap mode, the trap base is placed on

Figure 8. Hall light trap. its side and inserted through an opening into the side of a two-

gallon black plastic bucket filled to the opening with hay infusion

(Fig. 9). The standard, currently commercially available light-trap

and gravid-trap models that a West Nile mitigation program

Figure 9. Hall gravid trap manager such as myself has heretofore used are bulky, taking up

large volumes of storage space in a vehicle. The Hall Trap is convertible and compact, meaning

more flexibility in the field and leaving more storage space for sprayers, pesticides, dippers,

tripods, tools, BG Sentinels for Ae. albopictus, educational materials and other tools of the trade.

When a professional such as myself is in the field, the trap, battery, and attractant can all be

carried together with one hand using the bucket, allowing a free hand to carry a machete to clear 11

a path or a dipper for larval surveillance. The Hall Trap technology is being offered to potential

manufacturers by the Penn State Intellectual Property Office.

This study compared the American Biophysics Company light trap marketed by Clarke

Mosquito (i.e., the “ABC light trap”; Fig. 10), the Centers for Disease Control light trap

marketed by BioQuip (i.e., the “CDC light trap”; Fig. 11), the Reiter-Cummings gravid trap

marketed by BioQuip (the “RC gravid trap”; Fig. 12), the Hall light trap (Figure 8), and the Hall

gravid trap (Figure 9). Using the protocols employed by Pennsylvania’s West Nile Program,

when operating the three commonly used traps, the light bulbs were removed from the ABC and

the CDC trap and the light sensor for the fan motor was covered on the Reiter-Cummings trap.

Figure 10. ABC light trap Figure 11. CDC light trap Figure 12. RC gravid trap

Over the course of the 2-year study, traps were all stored in the same location, next to

each other, when not in use. All traps were kept in good repair, their motors cleaned and oiled

on a regular basis and replaced when running roughly. 12

Figure 13. A typical set-up of a randomized complete block.

Sites for trials were chosen by selecting high mosquito population locations as ascertained from the West Nile Program’s regular surveillance data. A large variety of sites was required for the greatest diversity in habitat and potential mosquito species inhabiting these locations. Sites that were utilized included wastewater treatment plants, wetlands, flooding terrains, and human-discarded-container-rich sites. Thirty-nine usable blocks were conducted, with a small number of blocks having to be discarded due to battery failures, a tree falling over onto a trap, and passers-by tampering with traps.

13

Nineteen different sites were utilized. At each site a replicate (randomized complete block) was deployed that was comprised of the five trap-treatments evenly spaced approximately

10 m apart within a relatively homogeneous area (c.f., Fig. 13). Dice were used to randomly assign traps to a position within the block each night in order to prevent bias in trap interactions, micro-environments, researcher preferences, or other factors such as wind direction that could unduly impart bias to the treatment means within blocks. Tripods were used for hanging light traps at a uniform height (1 m) and the gravid traps were set on the ground. At the time the traps were set for each night, the dry ice chambers of the light traps were filled to capacity (5 lbs). RC gravid traps received 1 inch of hay infusion in the pan (Hutchinson, 2010) and the Hall gravid- trap (“Hall-G”) was filled to the brim of its side opening with the same hay infusion solution from the same bottle as was used for the RC gravid trap. The infusion was agitated before pouring it into either trap-type. Hay infusions were replaced in the traps each day the traps were sampled and their positions re-randomized. Once all traps were set up, batteries were connected to all the traps in a block. Upon collecting samples, collection chambers were all removed first, in the same order batteries were connected, and then the traps were taken down. Having all traps start and stop as temporally close as possible was important to reducing any possible bias due to one trap collecting mosquitoes while other traps in the block were not. Collection chambers were marked with trap data using a permanent marker as they were removed from traps.

After collection, on the same day, the chambers were placed on dry ice to kill the mosquitoes in the samples. Non-mosquitoes were then picked out of the samples and the mosquitoes were placed into vials, labeled with date, site, and trap type, and delivered on dry ice to Pennsylvania’s West Nile Virus Program’s laboratory to be identified to species and tested for

West Nile virus by their technicians.

14

Testing for WNV by the PA DEP Laboratory was typically performed by IgM ELISA or

PCR. Samples were divided into species-specific sub-samples. Sub-samples with more than 50 mosquitoes were divided into batches of 50 mosquitoes each. Therefore, a sample from a single trap can yield multiple WNV-positive sub-samples. Multiple sub-samples testing positive for

WNV is indicative of high WNV prevalence in that population.

Data was compiled into a spreadsheet and then analyzed using MiniTab 14 statistical analysis software. The data included a total of 39 blocks, each block measuring the number of each species caught in each mosquito trap and the number of WNV-positive sub-samples in each trap. The data was log-transformed normalized after all zeroes were replaced by a value of 0.5.

Block and treatment variances were accounted for using a 2-way Analysis of Variance, and differences among treatment means were determined using Tukey’s w-procedure. The performances of the five trap types were analyzed in terms of their ability to capture different species, to capture individuals carrying the West Nile virus, to collect a wide variety of species, and to capture the greatest number of mosquitoes.

It was understood that not all species were going to be present during each sampling period. For instance, species such as Och. canadensis, are known to be highly seasonal, being found in large number only at certain times of the year. Certain types of trapping habitats were anticipated to be populated with a narrower range of species than others. One example involves

Psorophora horrida, which is dependent on specific weather events, including heavy flooding, to trigger the hatching of its eggs. For a practical example, the reader can refer to Figs. 14, 15 and

16, in which the species demographics from three different sites used in this study are displayed.

15

Wastewater treatment plant 140

120

100

80

60

40 Number ofmosquitoes Number

20

0 Ae. vexans Cx. pipiens Cx. restuans Cx. salinarius Och. triseriatus Mosquito species

Figure 14. Histogram of the number of individuals of each species caught at a wastewater treatment plant site using the data from all 5 trap-types combined.

At the wastewater treatment plant site (Fig. 14) there was a reed bed for the filtration of the wastewater. The primary species present in this location were Cx. pipiens and Cx. restuans.

A few Cx. salinarius were apparently able to colonize the location, or else were possibly migrants from nearby. A small number of Ae. vexans was captured, possibly a result of nearby flooding, and Och. triseriatus likely had emerged from a tire, a bucket or some other container on-site. 16

Wetland habitat 14

12

10

8

6

4

Number ofmosquitoes Number 2

0

Species

Figure 15. Histogram of the number of individuals of each species caught at a human-made wetland site using the data from all 5 trap-types combined. This wetland habitat site (Fig. 15) comprises a human-made wetland used for storm- water management. It exhibited a wider species diversity than did the wastewater treatment plant site. There was a variety of microhabitats within this wetland, including permanent water, flooding terrain, and aquatic, reedy vegetation. The flooding areas under open sky would have produced the Ae. vexans and Och. trivitattus that were captured. The permanent water areas would have been responsible for producing the Anopheles and Culex species. A few Co. perturbans apparently had colonized the reedy areas. 17

Wooded, flooding terrain 45

40 35 30 25 20 15 10 Number ofmosquitoes Number 5 0

Species

Figure 16. Histogram of the number of individuals of each species caught at a wooded site prone to flooding using the data from all 5 traps combined.

The wooded, flooding terrain (Fig. 16) most certainly contained water sources that had persisted for some time, because species of both Culex and Anopheles were present in abundance. Cx. restuans in particular is known to utilize woodland pools. At this site, a greater variety of floodwater species was present than in the wetland site, including Ae. vexans, Och. trivitattus, Ps. ferox, and Ps. horrida, all probably due to the lush tree canopy.

In preparing the experimental design, therefore, it was determined that it was not appropriate to include blocks from habitats or time periods having no individuals of a particular species present. To do otherwise would provide a less accurate estimate of the true experimental error variance. Thus the design anticipated that for each species being analyzed, only blocks that had captured at least one individual of the species of interest were included in the block replication of the analysis. For example, when conducting statistical tests comparing the five 18 traps’ capture performances with regard to An. quadrimaculatus, out of the total 39 blocks there were 16 blocks in which at least one of the traps caught one individual. It was deemed according to the design that An. quadrimaculatus were only present at those 16 times and places such that the traps’ capture abilities in those sample blocks could be appropriately assessed. This same strategy was likewise applied to other comparisons, such as the analysis of traps’ abilities to detect WNV-positive mosquitoes.

Results.

Species-by-species trap performance.

The trapping results showed that the novel trap types (Hall traps) performed as well as, and sometimes better than, the currently used, standard commercial traps for the majority of the disease-vectoring mosquito species and there were differential capture efficacies for the traps used in this study for different species. Thirty-nine replicates were obtained over two years.

Histograms of the mean number of each species and WNV-positive sub-samples caught per trap, along with significance indicators are displayed in Figures 17-34. Means, standard deviations of the means, and the Mini Tab outputs of the 2-way ANOVA with Tukey’s w-procedure are displayed in Tables 1-4 in the Appendix section of this thesis.

Cx. pipiens was captured in significantly higher numbers in the Hall-G and RC traps than in the other trap-types (Fig. 17). It is one of the two primary vectors of West Nile virus in the north/mid-eastern US. Interestingly, it will feed on birds early in the season and then switch to feeding on larger mammals such as humans later in the summer, making it very effective for

WNV transmission to humans. Birds are the reservoir for WNV and humans are dead-end hosts.

Cx. pipiens tends to have its largest populations in late summer and utilizes containers, long-term 19 standing water, and highly organic water such as from waste-water treatment plants and farm run-off. It does not have a long flight range, being typically less than one mile.

Captures of Culex pipiens 45 A 40

35 A 30

25

20

15 B

10 B

Mean numberMeanofmosquitoes pertrap 5 B

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 17. Mean number of Culex pipiens captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w-procedure (P < 0.05; N= 34). This species was captured predominantly near long-term standing water, waste-water treatment facilities and container-rich locations.

Cx. restuans likewise was captured at significantly higher capture rates in the Hall-G and

RC traps than in the other trap-types (Fig. 18). This species is the other primary vector of WNV in the north/mid-eastern US. It feeds mostly on birds, but will feed occasionally on other animals and humans and serves as an amplification vector for WNV among birds. It tends to have its peak population in spring and early summer. Cx. restuans lays eggs on long-term standing water such as ephemeral spring pools, in containers, and moderately organic water such as is found in off-line waste-water treatment plants and unused manure lagoons. Its flight range is also relatively short, typically less than a mile. 20

Captures of Culex restuans

120 A

100

A 80

60

40 B B 20 Mean numberMeanofmosquitoes pertrap B

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 18. Mean number of Culex restuans captured per trap in the five different trap types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w-procedure (P < 0.05; N= 35). This species was captured predominantly near long-term standing water, waste-water treatment facilities and container-rich locations.

Cx. salinarius exhibited the highest levels of capture in the Hall-L trap, which outperformed the ABC, RC and Hall-G traps (Fig. 19). The CDC trap outperformed the 2 gravid traps (Fig. 19). Cx. salinarius is a major West Nile virus bridge vector, feeding readily on humans and birds. This species’ larvae are found primarily in mildly brackish waters, such as coastal wetlands and reclaimed salt-marsh habitats. However, it can utilize the same habitats as

Cx. restuans, but with much less success. Phosphates can also create favorable larval conditions

(Slaff and Haeffner, 1985) and so an agricultural field that has been fertilized and subsequently flooded for several weeks or storm run-off from a fertilized lawn has the potential to become an extremely productive habitat hundreds of miles from the nearest naturally occurring brackish water. 21

Captures of Culex salinarius 12

A 10

8 AB 6

BC 4

2 Mean numberMeanofmosquitoes pertrap C C

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 19. Mean number of Culex salinarius captured per trap in the five different trap types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w-procedure (P < 0.05; N= 23). This species was captured predominantly near long-term standing water locations.

In the 8 blocks where WNV was detected in the mosquitoes, the RC (Reiter-Cummings) gravid trap captured a significantly greater number of WNV-positive mosquitoes than the Hall-G gravid trap, which also outperformed the other trap-types in this regard (Fig. 20). Because the vast majority of WNV positive samples detected since the implementation of regular surveillance in PA have been either Cx. pipiens or Cx. restuans, it is not surprising that the gravid traps, which were most successful with these species (Figs. 17, 18) , should also be most successful at detecting the presence of this virus in mosquitoes.

22

Number of West Nile-positive sub-samples 1.6

1.4 A

1.2

1

0.8 B 0.6

0.4

Mean numberMeanofspecies per trap C C 0.2 C

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 20. Mean number of WNV-positive sub-samples per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w-procedure (P < 0.05; N= 8). These positive samples were found exclusively in Cx. pipiens and Cx. restuans; other species were also tested but WNV was not detected in those other species.

23

Och. trivitattus was captured in statistically insignificantly different numbers by the Hall-

L, the CDC and the ABC traps. All three of these light trap types were more effective than the two gravid traps (Fig. 21). This is a floodwater species that lays its eggs on dry land, the larvae hatching after being flooded. The larvae are very wary and will spend long periods of time submerged when disturbed. The mosquito adults are major human pests, and are a bridge vector of WNV and dog heartworm

Captures of Ochlerotatus trivitattus

25

A 20

A 15

10

A 5

Meannumber of mosquitoesper trap B B

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 21. Mean number of Ochlerotatus trivitattus captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 15). This species was captured predominantly near flooding terrain.

24

Ae. vexans was captured at as high a rate in the Hall-L traps as for the ABC and CDC traps (Fig. 22). All these light traps captured significantly greater numbers of this species than the RC and Hall-G gravid traps. This is a floodwater species, laying its eggs on dry land and then hatching following flooding. Ae. vexans does not require the water to be present for long before the eggs hatch. The adults have a relatively long flight range, often being found several miles from where they emerge from their pupae. It is a common vector of dog heartworm and a bridge vector for West Nile virus and EEE.

Captures of Aedes vexans 30

A

25 A

20

15 A

10

5 Meannumber of mosquitoesper trap B B 0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 22. Mean number of Aedes vexans captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w-procedure (P < 0.05; N= 30). This species was captured predominantly at sites near flooding terrain.

25

In some cases, the commercial traps performed better than the novel trap-types. One such instance occurred in trapping Ps. ferox, another floodwater species. The adults prefer wooded habitats and can be a bridge vector for WNV. Only 6 blocks contained traps that captured this species, and the CDC and ABC traps outperformed the RC and Hall-G traps. (Fig. 23)

Captures of 18 A

16

14 AB 12

10

8

6 ABCD 4

Mean numberMeanofmosquitoes pertrap 2 D CD 0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 23. Mean number of Psorophora ferox captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 6). This species was captured predominantly near flooding terrain.

26

Another case of commercial traps outperforming the novel trap-types occurred in the trapping of An. quadrimaculatus, a vector for human malaria and a bridge vector for WNV and

EEE. Adults of this species were captured in the CDC trap at significantly higher numbers than in other trap-types (Fig. 24). An. quadrimaculatus has a wide range of habitats. The larvae thrive in clean, clear water such as pristine wetlands, but they can also be found in containers, off-line wastewater treatment plant tanks, and at the edges of moving water.

Captures of Anopheles quadrimaculatus 14 A 12

10

8 AB 6

4

B B B

Mean numberMeanofmosquitoes pertrap 2

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 24. Mean number of Anopheles quadrimaculatus captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 16). This species was captured predominantly near wetland and container-rich locations.

27

Co. perturbans, a vector for Eastern Equine Encephalitis, also was captured in significantly greater numbers by the CDC trap than in other traps (Fig. 25). Eastern Equine

Encephalitis is a deadly, though rare, disease that is amplified by another species, Culesita melanura, which is limited to acidic (usually red maple) wetlands, through birds and then Cq. perturbans (and a few other species) acts as a bridge vector, transmitting the disease to humans.

The larvae of Cq. perturbans are found in clean water with reedy plants. The larvae insert their siphons into the stems of those plants, effectively using them as snorkels, allowing the larvae to stay submerged indefinitely and making them difficult to detect.

Captures of Coquillettidia perturbans 3

A 2.5

2

1.5 B 1

B

0.5 Meannumber of mosquitoesper trap B B 0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 25. Mean number of Coquillettidia perturbans captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 10). This species was captured predominantly near wetland locations.

28

An. punctipennis adults were captured at insignificantly different levels by all five types of traps (Fig. 26). This species has the ability to transmit human malaria and its larvae thrive in clean water habitats, including stream edges, wetlands, and ephemeral pools. The number of blocks in which at least one adult was captured (14) would seem to have been sufficient to assess the differential capture rates of the traps, yet no such differences arose, even between the gravid traps and the light traps (Fig. 26).

Captures of Anopheles punctipennis 1.8 A 1.6

1.4

1.2 A 1

0.8 A A 0.6 A

0.4

Meannumber of mosquitoesper trap 0.2

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 26. Mean number of Anopheles punctipennis captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 14). This species was captured predominantly at sites near wetlands.

29

Several other species were captured in numbers of blocks that were apparently too low to allow a determination of any significant differences among the trap types’ capture abilities. Och. canadensis is an example of this, and is a very early season species. The larvae can be found in ephemeral pools formed by snow melt, and even in melt water in the snow. It is a bridge vector for EEE. Only two blocks contained traps that captured individuals of this species, with all of them being light traps (Fig. 27).

Captures of Ochlerotatus canadensis 4.5

4 A

3.5

3

2.5 A 2

1.5 A 1 A A Mean numberMeanofmosquitoes pertrap 0.5

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 27. Mean number of Ochlerotatus canadensis captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 2). This species was captured predominantly near wooded areas with early spring ephemeral pools.

30

Another example is Och. japonicus, which is an invasive species that was first detected in the US in 1998, and whose larvae utilize containers. It is an efficient vector of WNV. Only 7 blocks contained traps that captured individuals of this species, all of them being gravid traps

(Fig. 28).

Captures of Ochlerotatus japonicus 3.5

A

3

2.5

2

1.5 A 1

Mean numberMeanofmosquitoes pertrap 0.5 A A A 0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 28. Mean number of Ochlerotatus japonicus captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 7). This species was captured predominantly near container-rich sites.

31

Och. triseriatus (Fig. 29) is a container-utilizing mosquito species. It normally lays eggs in tree holes, but will also use tires and other artificial containers. It prefers larval habitats with low organic loads, usually opting for water with only decaying leaves to provide sustenance for the larvae. This species is a vector for St. Louis Encephalis (SLE) and can be a bridge vector for

WNV. Only five blocks contained traps that captured this species, comprising a mixture of gravid traps and light traps.

Captures of Ochlerotatus triseriatus 0.7 A A 0.6

0.5

0.4 A

0.3

0.2

Mean numberMeanofmosquitoes pertrap 0.1 A A 0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 29. Mean number of Ochlerotatus triseriatus captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 5). This species was captured predominantly near areas with containers.

32

Only 6 blocks contained traps that captured Ps. columbiae (Fig. 30), with all five trap- types contributing to mosquito captures. It is a floodwater mosquito and can occur in great numbers and act as aggressive pests of humans and cattle. This species also can be a bridge vector for WNV.

Captures of Psorophora columbiae 18

16 A 14

12 A

10

8

6 A 4

Mean numberMeanofmosquitoes pertrap 2 A A 0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 30. Mean number of Psorophora columbiae captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 6). This species was captured predominantly near flooding terrain.

33

Ps. horrida is a relatively rare woodland floodwater species. It is similar to Ps. ferox in both habitat and morphology. In the entire state of Pennsylvania only 27 traps since 2001 from the PA WNV Program have managed to trap individuals of this species. In the current study, however, three of the light-traps from two different blocks did manage to capture these rare mosquitoes, now making a total of 30 trap-capture detections (Fig. 31).

Captures of Psorophora horrida

6

5 A

4

3

A 2

1 Mean numberMeanofmosquitoes pertrap A A A 0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 31. Mean number of Psorophora horrida captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 2). This species was captured predominantly near flooding terrain.

Across-species trap-type performances.

It is well known that gravid traps can capture very large numbers of mosquitoes, particularly Cx. pipiens and Cx. restuans. Capturing over 1000 in a gravid trap, although not common, is not highly unusual. Although while not in the course of conducting this study, rather 34 during regular surveillance for WNV in 2011, an RC gravid trap captured 2,578 mosquitoes and a Hall-G captured 2,275 mosquitoes. Out of the 541 gravid traps deployed in Lebanon County,

PA in 2011, 10 had catches over 1000. Extraordinarily high populations of mosquitoes are necessary to capture more than a few hundred mosquitoes in a light trap. Thus it was expected that the gravid traps should outperform the light traps over-all (Fig. 32).

Captures of all species combined 160

140 A

120 A 100

80

60

40 B B B

Mean numberMeanofmosquitoes pertrap 20

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 32. Mean number of mosquitoes, for all species combined, captured per trap in the five different trap- types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w-procedure (P < 0.05; N= 39).

In terms of variety of species caught, the gravid traps caught, on average, the greatest variety (Fig. 33). However, I believe this to be because they were extremely effective at catching Cx. pipiens and Cx. restuans. These mosquitoes were ubiquitous, being found together in almost every block (Figs. 17, 18). The other species were not so common, only Ae. vexans being found in more than half of the blocks. The gravid traps were not very effective at catching 35 other species and so when the Cx. pipiens and Cx. restuans were excluded, the light traps were shown to be statistically equivalent with each other and more effective than the gravid traps for detecting the variety of species present (Fig. 34).

Variety of species captured 3 AB A B B AB 2.5

2

1.5

1

Mean numberMeanofspecies per trap 0.5

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 33. Mean number of different species captured per trap in the five different trap-types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w- procedure (P < 0.05; N= 39). 36

Variety of non- Culex pipiens/restuans species 3 A A A 2.5

2

1.5 B B 1

0.5 Mean numberMeanofspecies per trap

0 ABC-L CDC-L Hall-G Hall-L RC-G Trap type

Figure 34. Mean number of species (non- Culex pipiens/restuans) captured per trap in the five different trap- types. Means having no letters in common are statistically different according to a 2-way ANOVA followed by Tukey's w-procedure (P < 0.05; N= 39).

Discussion.

The two novel, highly portable types of trap that I developed and tested were shown to often be as effective in sampling certain important target species of mosquito as the standard, currently used commercial mosquito traps. As with the other devices, for some species the novel traps were superior and for others they appeared to be a less effective sampling device. Due to the compactness and versatility of these traps, however, they should prove to be a valuable addition to a sampling arsenal. The results also reiterate that the type of trap used can greatly influence the species that are able to be detected.

For many of the most important disease-vectoring mosquito species, the novel Hall traps performed as well as, if not better than, the standard commercial traps in capturing certain 37 species. Individuals of Ae. vexans, vector of dog heartworm and bridge vector for WNV and

EEE, were captured by the Hall-L, CDC, and ABC at statistically equal rates, with these light traps being more effective than the gravid traps (Fig. 22). Cx. pipiens and Cx. restuans, vectors of WNV, were captured by Hall-G and RC traps at statistically similar rates and at much greater levels than in the light traps (Figs 17, 18). Cx. salinarius, vector of WNV, was captured in the

Hall-L traps at greater rates than in the ABC, Hall-G, and RC traps. The CDC trap, although capturing somewhat lower numbers, still was better at sampling this species than the gravid traps

(Fig. 19). Och. trivitattus, a bridge vector of WNV and dog heartworm, was captured at statistically equivalent rates by Hall-L, CDC and ABC traps; all three of these light traps were more effective than the two gravid traps (Fig. 21).

For some species, the traditionally used commercial traps performed better. An. quadrimaculatus, vector of malaria and bridge vector for WNV, was most effectively caught by the CDC light trap (Fig. 24). Co. perturbans, vector of EEE and bridge vector for WNV, was also most effectively caught by the CDC trap (Fig. 25).

Several species were not as widely caught as others during the course of the study

(Figs.23, 26-31). Thus, trapping results for species having small sample sizes should be viewed as preliminary. For example, Ps. ferox (Fig. 23), vector of dog heartworm and bridge vector for

WNV, was captured by Hall-L traps at statistically the same rate as the other two light traps, but not better than the gravid traps. The ABC and the CDC light traps were, however, shown to capture Ps. ferox more effectively than the gravid traps. However with only 6 blocks, one should be reticent to draw strong conclusions. In another example, there were only 8 blocks in which

WNV-positive mosquitoes were captured. In two of these blocks, more than one mosquito tested positive in the same RC trap. Therefore, the RC was indicated to be more effective at capturing 38

WNV-infected mosquitoes than the Hall-G, with both being more effective than the light traps

(Fig. 20). If one were to simply measure whether the traps caught any WNV-positive mosquitoes, the Hall-G and the RC would be statistically equivalent. An. punctipennis, although detected in 14 blocks, showed no preference for light-traps compared to gravid traps (Fig. 26), with insignificantly different mean capture levels across all five trap-types. Further studies in locations having higher populations of this and the other species mentioned above, or having known disease presence in a certain trap deployment location, would be useful.

It was not surprising that certain species such as Ae. vexans, An. quadrimaculatus, Co. perturbans, Cx. salinarius, Och. trivitattus, and Ps. ferox were captured at higher rates in the light traps because it was already well known among professionals in the field that these species are more likely to be caught using this trap-type. Additionally, it was not surprising that Cx. pipiens and Cx. restuans were captured at higher rates in the gravid traps because this tendency, too, had already been well established among professionals in the field. What is surprising is that there were species preferences among the light traps and, although not statistically significant, patterns also emerged with regard to species preferences among the gravid traps

(Figs. 17, 18, 28) that seem to suggest that with further sampling one might find significant differences there as well.

All of the traps I tested used the same motor and the same fan. The color schemes were similar. Even the containers holding the dry ice were primarily blue for all traps and the infusion trays or buckets were black. The attractant for the two types of gravid traps, a straw infusion with some milk/milk albumin added, was the same. The attractant for the CO2 light traps was the same for all 3 traps, being comprised of pelleted dry ice purchased from the same manufacturer each time. 39

There were some slight differences between the traps that were tested, however, that might account for some of the capture rates of some species compared to others. For instance, the location between the CO2 release point and the intake opening for the light traps was different in the Hall trap compared to the CDC or ABC traps. The CDC trap had a CO2 release point to the side and below the trap intake opening, whereas the release points for the ABC and Hall-L were above the intake point. The Hall-L had a greater distance between the release point and the trap intake than the ABC.

Some of the light traps had differences in their rain shields. The ABC and the CDC both had black rain shields that also served to funnel mosquitoes approaching from below towards the intake, but also directed away those approaching from above. The Hall-L had no rain shield.

The color schemes were slightly different, although white/clear, black and blue were involved for all three light traps. The ABC had a white plastic intake set below a black rain shield below a blue dry ice container. The CDC had a clear plastic intake set below a black rain shield and next to a blue-and-silver dry ice container. The Hall-L had black burlap wrapped around a white plastic intake set beneath a blue and silver dry ice container.

For the gravid traps, the position of the intake opening in relation to the water level of the grass infusion was different. The Hall-G’s intake was located laterally, allowing uninhibited access to the entire surface of the attractant and capturing mosquitoes from the side with a vertical intake opening. The RC trap’s intake was located ventrally and mosquitoes had to pass under the trap to explore the entire surface of the attractant and capturing mosquitoes from above with a horizontal intake opening.

The colors of the intakes were similar, black, but the Hall-G had black burlap encircling white plastic while the RC’s intake was solid black plastic. The body of the hall trap was clear, 40 white and metallic while the body of the RC was gray. It seems clear from these results that mosquitoes are sensitive to these sorts of nuances, and these differential sensitivities can possibly translate into different behavioral reactions to the traps, depending on the species. Further research will be needed to clarify and define any behavioral reasons for these differences.

All light traps were set at a uniform height using tripods. All gravid traps were placed on the ground in near-identical settings. However, the placement of the gravid traps on the ground, as well as the use of this one particular type of hay infusion perhaps favored the trapping of certain species compared to others. Likewise, the uniform height at which the light-traps were deployed may also have preferentially trapped particular species over others. For the light-traps, some species may have preferred to feed or rest at higher or lower heights than this particular trap level (Swanson and Adler, 2010; Votypka and Svobodova, 2010), and a change in trap height could possibly have increased the performance of the light traps for such species.

For the gravid traps as well, the vertical distribution of individuals of a species in the environment due to their reproductive behaviors would seem to be able to influence their capture levels in the current study. For example, Och. hendersonii is often found in the same locale as

Och. triseriatus, but Och. hendersonii lay eggs in tree-holes at higher elevations with Och. triseriatus laying eggs nearer the ground. I captured no specimens of Och. hendersonii in this study using the on-ground gravid traps, but did catch significant numbers of Och. triseriatus.

There are species, for example Och. triseriatus, Och. hendersonii, Och. japonicus, Tox. rutilus, and other tree-hole-utilizing species, that would not be optimally attracted to the highly organic hay infusion that I used. These species all prefer an oak leaf infusion (Trexler et al., 1998).

Thus, a different infusion would have improved the gravid traps in capturing some species, but of course this would have made it worse for capturing others, particularly species like Cx. 41 pipiens and Cx. restuans that thrive in more organic environments and were captured in large numbers over the course of this study (Figs. 17, 18)

With this study I have provided comparative data that show that these novel trap-types can confidently be added to the mosquito sampling arsenal when appropriate. When transportation of equipment is an issue, switching to a different trap type that is able to be more easily deployed and at higher trap densities to enable detection of the offending species, can now be done so with confidence. When there is a concern that samples might become degraded due to contact with the fans of standard commercial traps, the novel trap types tested here now present themselves as an appropriate alternative. In addition, for certain species these novel traps have now been shown to be more effective at acquiring a larger sample size.

This study testing possible improvements on the currently available traps to create a more versatile, highly portable novel trap type shows that these traps are as good as, and sometimes better than, the commercially available traps for the majority of species studied. Moreover, the increased versatility and utility of these traps may make them preferable to other types under certain sampling protocols and public health pressures. If a mosquito control district is getting repeated complaints from a particular area, but sampling is not indicating large mosquito populations or else a disease management program has a cluster of cases yet sampling indicates that there are few medically important mosquitoes present, then selecting a different trap type to use may be a solution.

The implications from this study for mosquito control districts and mosquito-vectored disease programs are that trap selection is important. Effective sampling is a stepping stone to efficient use of resources, targeted control efforts, and success in nuisance or vector mosquito management. Moreover, optimizing sampling techniques to identify locations where disease- 42 vectoring mosquitoes are present and to monitor population levels should allow control measures to be targeted towards medically important mosquitoes while reducing any environmental and financial costs associated with widespread, indiscriminate pesticide applications. Having effective sampling tools available to be easily deployed at sufficiently high trap densities also can prevent faulty decisions that result in a failure to initiate control in some locations when

“standard” sampling strategies might falsely indicate that there are too few target mosquitoes present to be concerned about.

Having proper knowledge such as I have tried to supply in this study, in which I have compared mosquitoes’ responses to trap types and lures for the species that are most likely to vector a particular disease of concern to the public, I hope will serve to better protect the public health. It is my view that in general it would be advisable to conduct a comparative trap study in each locale that has concern about impending mosquito-vectored disease introduction or invasive species threat in advance of such a situation. This small investment might pay large dividends with regard to better data, more efficient use of resources, and better protection of the public health.

43

References

Allan SA, Kline D (2004) Evaluation of various attributes of gravid female traps for collecting Culex in Florida. Journal of Vector Ecology 29:285-294 Barnard DR, Knue GJ, Dickerson CZ, Bernier UR, Kline DL (2010) Relationship between mosquito landing rates on a human subject and numbers captured using CO2-baited light traps. Bulletin of Entomological Research 101:277-285 Bidlingmayer WL (1994) How mosquitoes see traps: role of visual responses. Journal of the American Mosquito Control Association 10:272-279 Burkett-Cadena ND, Mullen GR (2007) Field comparison of Bermuda-hay infusion of emergent aquatic vegetation for collecting female mosquitoes. Journal of the American Mosquito Control Association 23:117-123 Chen YC, Wang CY, Teng HJ, Chen CF, Chang MC, Lu LC, Lin C, Jian SW, Wu HS (2011) Comparison of the efficacy of CO2-baited and unbaited light traps, gravid traps, backpack aspirators, and sweep net collections for sampling mosquitoes infected with Japanese encephalitis virus. Journal of Vector Ecology 36:68-74 Chevalier V, Lecollinet S, Durand B (2011) West Nile virus in Europe: a comparison of surveillance system designs in a changing epidemiological context. Vector-Borne and Zoonotic Diseases 11:1085-1091 Costantini C, Sagnon N, Gibson G, Coluzzi M, Brady J (1997) The role of body odours in the relative attractiveness of different men to malarial vectors in Burkina Faso. Annals of Tropical Medicine and Parasitology 91:121-122 Cooperband MF, Carde RT (2006) Comparison of plume structures of carbon dioxide emitted from different mosquito traps. Medical and Veterinary Entomology 20:1-10 Darsie RF, Hutchinson ML (2009) The Mosquitoes of Pennsylvania. Technical Bulletin #2009- 001 of the Pennsylvania Vector Control Association 191pp. Derraik RGB, Barraclough RK (2011) A trial using rechargeable batteries to run mosquito light traps. Journal of Vector Ecology 36:221-223 DiMenna MA, Bueno R Jr, Parmenter RR, Norris DE, Sheyka JM, Molina JL, LaBeau EM, Hatton ES, Glass GE (2006) Comparison of mosquito trapping method efficacy for West Nile virus surveillance in New Mexico. Journal of the American Mosquito Control Association 22:246-253 Farajollahi A, Kesavaraju B, Price DC, Williams GM, Healy SP, Gaugler R, Nelder MP (2009) Field efficacy of BG-Sentinel and industry standard traps for Aedes albopictus and West Nile surveillance. Journal of Medical Entomology 46:919-925 Govella NJ, Chaki PP, Mpangile JM, Killeen GF (2011) Monitoring mosquitoes in urban Dar es Salaam: evaluation of resting boxes, window exit traps, CDC light traps, Ifakara tent traps and human landing catches. Parasites and Vectors 4:40 Gubler DJ et al., (2003) Centers for Disease Control and Prevention: Epidemic/Epizootic West Nile Virus in the United States: Guidelines for Surveillance, Prevention, and Control. U.S. Department of Health and Human Services, Division of Vector-Borne Infectious Diseases. Fort Collins, Colorado 77pp. Hagstrum DW (1971) Evaluation of the standard pint dipper as a quantitative sampling device for mosquito larvae. Annals of the Entomological Society of America 64:537-540 Harris C, Kihonda J, Lwetoijera D, Dongus S, Devine G, Majambere S (2011) A simple and efficient tool for trapping gravid Anopheles at breeding sites. Parasites & Vectors 4:125 44

Henderson JP, Westwood R, Galloway T (2006) An assessment of the effectiveness of the Mosquito Magnet Pro Model for suppression of nuisance mosquitoes. Journal of the American Mosquito Control Association 22:401-407 Hiwat-van Laar H, Rijk M, Andriessen R, Koenraadt CJM, Takken W (2011) Evaluation of methods for sampling the malaria vector Anopheles darlingi (Diptera: Culicidae) in Suriname and the relation with its biting behavior. Journal of Medical Entomology 48:1039-1046 Howard JJ, Oliver J, Kramer LD (2009) Assessing the use of diurnal resting shelters by Culiseta melanura. Journal of Medical Entomology 48:909-913 Huang S, Hamer GL, Molaei G, Walker ED, Goldberg TL, Kitron UD, Andreadis TG (2009) Genetic Variation Associated with Mammalian Feeding in from a West Nile Virus Epidemic Region in Chicago, Illinois. Vector Borne and Zoonotic Diseases 9:637-667 Jackson BT, Paulson SL, Youngman RR, Scheffel SL, Hawkins B (2006) Oviposition preferences of Culex restuans and Culex pipiens for selected infusions in oviposition traps and gravid traps. Journal of the American Mosquito Control Association 22:360-365 Jawara M, Awolola TS, Pinder M, Jeffries D, Smallegange RC, Takken W, Conway DJ (2011) Field testing of different chemical combinations as odour baits for trapping wild mosquitoes in The Gambia. PloS One 6:e19676 Lefèvre T, Gouagna LC, Dabiré KR, Elguero E, Fontenille D, Renaud F, Costantini C, Thomas F (2010) Beer consumption increases human attractiveness to malaria mosquitoes. PLoS One 5:e9546 Leisnaham PT, Slaney DP, Lester PJ, Weinstein F (2005) Evaluation of two dipping methods for sampling immature Culex and Ochlerotatus mosquitoes (Diptera: Culicidae) from artificial containers. New Zealand Journal of Marine and Freshwater Research 39:1233-1241 McPhatter LP, Debboun M (2009) Attractiveness of botanical infusions to ovipositing Culex quiquefasciatus, Cx. nigripalpus, and Cx. erraticus in San Antonio Texas. Journal of the American Mosquito Control Association 25:508-510 Miura T, Husbands RC, Reed DE (1970) Field evaluation of the concentrator-dipper technique for sampling mosquito larvae. Mosquito News 30:448-453 Mohr RM, Mullens BA, Gerry AC (2011) Evaluation of ammonia, human sweat, and bovine blood as attractants for the female canyon fly, Fannia conspicua (Diptera: Muscidae), in southern California. Journal of Vector Ecology 36:55-58 Müller GC, Xue RD, Schlein Y (2011) Differential attraction of Aedes albopictus in the field to flowers, fruits and honeydew. Acta Tropica 118:45-49 Ndiath MO, Mazenot C, Gaye A, Konate L, Bouganali C, Faye O, Sokhna C, Trape JF (2011) Methods to collect Anopheles mosquitoes and evaluate malaria transmission: a comparative study in two villages in Senegal. Malaria Journal 10:270 Ponnusamy L, Wesson DM, Arellano C, Schal C, Apperson CS (2009) Species composition of bacterial communities influences attraction of mosquitoes to experimental plant infusions. Microbiology Ecology 59:158-173 Reeves WC (2001) Partners: serendipity in arbovirus research. Journal of Vector Ecology 26:1-6 Reinert WC (1989) The New Jersey light trap: an old standard for most mosquito control programs. Proceedings of the Seventy-Sixth Annual Meeting of the New Jersey Mosquito Control Association, Inc. 1989, pp 17-25 Shone SM, Glass GE, Norris DE (2006) Targeted trapping of mosquito vectors in the Chesapeake Bay area of Maryland. Journal of Medical Entomology 43:151-158 Slaff M, Haeffner JD (1985) The impact of phosphate mining on Culex nigripalpus and Culex salinarius populations in central Florida. The Florida Entomologist 68:444-450 45

Swanson DA, Adler PH (2010) Vertical distribution of haematopagous diptera in temperate forests of southeastern USA. Journal of Medical and Veterinary Entomology 24: 182-188 Trexler JD, Apperson CS, Schal C (1998) Laboratory and field evaluations of Oviposition responses of Aedes albopictus and Aedes triseriatus (Diptera : Culicidae) to oak leaf infusions. Journal of Medical Entomology 35:967-976 Vaidyanathan R, Edman JD (1997) Sampling methods for potential epidemic vectors of eastern equine encephalomyelitis virus in Massachusetts. Journal of the American Mosquito Control Association 13:342-347 Votýpka J, Svobodová M, Cerný O (2010) Spacial feeding preferences of ornithophilic mosquitoes, blackflies and biting midges. Journal of Medical and Veterinary Entomology. 25: 104-108

46

Appendix.

Table1. Number of mosquitoes in each trap type by species A

Anopheles Anopheles Coquillettidia Culex Culex Culex Culex Aedes vexans punctipen- quadrimacu- perturbans Pipiens Restuans Salinarius Species trap value (N=30) nis (N=14) latus (N=16) (N=10) (N=35) (N=35) (N=23) (N=39) ABC Mean 14.37 0.50 3.81 0.70 0.60 0.57 2.87 2.77 SD 39.35 0.76 10.22 1.06 1.75 2.40 4.56 4.46 CDC Mean 8.87 0.71 8.56 1.90 0.69 0.00 4.83 3.54 SD 15.26 1.14 16.86 2.08 2.11 0.00 7.41 6.23 Hall -G Mean 0.50 0.93 0.44 0.00 20.00 55.26 0.52 80.44 SD 0.97 2.67 0.89 0.00 51.37 77.24 0.99 111.69 Hall -L Mean 18.97 0.43 0.69 0.30 0.49 0.11 7.35 4.95 SD 43.38 0.65 1.25 0.48 1.15 0.40 11.70 9.49 RC Mean 0.33 0.36 0.81 0.10 29.11 88.60 0.39 110.00 SD 0.61 0.74 2.51 0.32 55.50 151.63 0.78 165.10 Table2. Number of mosquitoes in each trap type by species B

Ochlerotatus Ochlerotatus Ochlerotatus Ochlerotatus Psorophora Psorophora canadensis japonicus triseriatus trivitattus columbiae Psorophora horrida trap value (N=2) (N=7) (N=5) (N=15) (N=6) ferox (N=6) (N=2) ABC Mean 1.50 0.00 0.00 10.80 2.50 6.83 1.00 SD 0.71 0.00 0.00 14.48 3.56 13.38 1.41 CDC Mean 2.00 0.00 0.00 15.33 2.00 8.83 0.00 SD 2.83 0.00 0.00 22.05 2.45 18.76 0.00 Hall -G Mean 0.00 2.00 0.40 0.20 0.00 0.17 0.00 SD 0.00 3.06 0.55 0.56 0.00 0.41 0.00 Hall -L Mean 0.50 0.00 0.20 3.60 1.00 2.00 3.00 SD 0.71 0.00 0.45 6.43 1.55 4.90 2.83 RC Mean 0.00 0.86 0.40 0.00 0.00 0.33 0.00 SD 0.00 0.69 0.55 0.00 0.00 0.82 0.00 Table 3. Number of mosquitoes in each trap type by species C

Variety: # # of Non- Culex Total catch of Species pipiens/restuans WNV-Positive sub- trap value (N=39) (N=39) species (N=39) samples (N=8) ABC Mean 21.49 2.18 1.82 0.00 SD 45.27 2.02 1.96 0.00 CDC Mean 22.31 2.21 2.00 0.00 SD 35.32 2.05 1.92 0.00 Hall -G Mean 81.77 2.36 0.77 0.50 SD 111.23 1.14 0.87 0.53 Hall -L Mean 22.13 2.05 1.77 0.00 SD 45.80 1.70 1.74 0.00 RC Mean 111.05 2.49 0.79 1.13 SD 164.89 1.12 0.83 0.64

47

Table 4. MiniTab output of statistical tests of traps by species, etc. Two-way ANOVA: Aedes vexans plus .5 log versus Trial, Trap Two-way ANOVA: Och tri log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 29 30.6091 1.05549 5.30 0.000 Trial 4 0.000000 0.0000000 0.00 1.000 Trap 4 12.7940 3.19849 16.05 0.000 Trap 4 0.182116 0.0455289 1.00 0.436 Error 116 23.1229 0.19934 Error 16 0.728463 0.0455289 Total 149 66.5260 Total 24 0.910579

S = 0.4465 R-Sq = 65.24% R-Sq(adj) = 55.35% S = 0.2134 R-Sq = 20.00% R-Sq(adj) = 0.00%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled StDev Pooled StDev Trap Mean -+------+------+------+------Trap Mean ----+------+------+------+----- ABC-L 0.420181 (----*----) ABC-L -0.301030 (------*------) CDC-L 0.394975 (----*-----) CDC-L -0.301030 (------*------) Hall-G -0.122329 (----*----) Hall-G -0.110181 (------*------) Hall-L 0.528585 (-----*----) Hall-L -0.205606 (------*------) RC-G -0.159008 (-----*----) RC-G -0.110181 (------*------) -+------+------+------+------+------+------+------+------0.30 0.00 0.30 0.60 -0.45 -0.30 -0.15 0.00

Two-way ANOVA: An. Punct log versus Trial, Trap Two-way ANOVA: Och triv log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 13 0.63796 0.049074 0.43 0.951 Trial 14 11.8970 0.84978 4.01 0.000 Trap 4 0.08118 0.020294 0.18 0.948 Trap 4 10.2817 2.57041 12.13 0.000 Error 52 5.90987 0.113651 Error 56 11.8660 0.21189 Total 69 6.62900 Total 74 34.0447

S = 0.3371 R-Sq = 10.85% R-Sq(adj) = 0.00% S = 0.4603 R-Sq = 65.15% R-Sq(adj) = 53.94%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled StDev Pooled StDev Trap Mean ------+------+------+------+ Trap Mean -----+------+------+------+---- ABC-L -0.098937 (------*------) ABC-L 0.555895 (------*------) CDC-L -0.062215 (------*------) CDC-L 0.566221 (------*------) Hall-G -0.122579 (------*------) Hall-G -0.222624 (------*-----) Hall-L -0.114783 (------*------) Hall-L 0.230060 (------*-----) RC-G -0.167097 (------*------) RC-G -0.301030 (-----*------) ------+------+------+------+ -----+------+------+------+---- -0.24 -0.12 0.00 0.12 -0.35 0.00 0.35 0.70

Two-way ANOVA: An. Quad log versus Trial, Trap Two-way ANOVA: Ps col log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 15 9.0140 0.600932 4.80 0.000 Trial 5 0.89962 0.179924 1.38 0.272 Trap 4 3.5877 0.896933 7.16 0.000 Trap 4 1.45305 0.363263 2.80 0.054 Error 60 7.5173 0.125288 Error 20 2.59912 0.129956 Total 79 20.1190 Total 29 4.95179

S = 0.3540 R-Sq = 62.64% R-Sq(adj) = 50.80% S = 0.3605 R-Sq = 47.51% R-Sq(adj) = 23.89%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled Pooled StDev StDev Trap Mean ---+------+------+------+------Trap Mean +------+------+------+------ABC-L 0.147156 (------*------) ABC-L 0.169222 (------*------) CDC-L 0.398979 (------*------) CDC-L 0.202708 (------*------) Hall-G -0.144886 (------*------) Hall-G -0.301030 (------*------) Hall-L -0.085245 (------*------) Hall-L 0.017051 (------*------) RC-G -0.144886 (------*------) RC-G -0.301030 (------*------) ---+------+------+------+------+------+------+------+------0.25 0.00 0.25 0.50 -0.60 -0.30 0.00 0.30

Two-way ANOVA: coq pert log versus Trial, Trap Two-way ANOVA: Ps ferox log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 9 0.78114 0.086793 1.69 0.126 Trial 5 5.28211 1.05642 10.23 0.000 Trap 4 2.05200 0.512999 10.02 0.000 Trap 4 1.98555 0.49639 4.81 0.007 Error 36 1.84395 0.051221 Error 20 2.06591 0.10330 Total 49 4.67709 Total 29 9.33357

S = 0.2263 R-Sq = 60.57% R-Sq(adj) = 46.34% S = 0.3214 R-Sq = 77.87% R-Sq(adj) = 67.91%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled StDev Pooled StDev Trap Mean ------+------+------+------+- Trap Mean ------+------+------+------+-- ABC-L -0.051199 (-----*-----) ABC-L 0.366664 (------*------) CDC-L 0.268173 (-----*-----) CDC-L 0.346671 (------*------) Hall-G -0.301030 (-----*-----) Hall-G -0.221510 (------*------) Hall-L -0.157894 (-----*----) Hall-L -0.068040 (------*------) RC-G -0.253318 (-----*-----) RC-G -0.184535 (------*------) ------+------+------+------+------+------+------+------+-- -0.25 0.00 0.25 0.50 -0.30 0.00 0.30 0.60

Two-way ANOVA: cx pip log versus Trial, Trap Two-way ANOVA: Ps horrida log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 33 12.3296 0.3736 1.89 0.006 Trial 1 0.15958 0.159578 2.62 0.181 Trap 4 48.3536 12.0884 61.11 0.000 Trap 4 0.90550 0.226375 3.71 0.116 Error 132 26.1105 0.1978 Error 4 0.24390 0.060976 Total 169 86.7936 Total 9 1.30898

S = 0.4448 R-Sq = 69.92% R-Sq(adj) = 61.48% S = 0.2469 R-Sq = 81.37% R-Sq(adj) = 58.08%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled StDev Pooled StDev Trap Mean ------+------+------+------+-- Trap Mean ------+------+------+------+--- ABC-L -0.13030 (---*---) ABC-L 0.048455 (------*------) CDC-L -0.13129 (---*--) CDC-L -0.301030 (------*------) Hall-G 0.81364 (--*---) Hall-G -0.301030 (------*------) Hall-L -0.14509 (--*---) Hall-L 0.458227 (------*------) RC-G 1.06759 (---*--) RC-G -0.301030 (------*------) ------+------+------+------+------+------+------+------+--- 0.00 0.40 0.80 1.20 -0.50 0.00 0.50 1.00 48

Two-way ANOVA: cx res log versus Trial, Trap Two-way ANOVA: Total catch log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 34 15.425 0.4537 2.01 0.003 Trial 38 54.059 1.42262 4.48 0.000 Trap 4 109.480 27.3700 121.17 0.000 Trap 4 39.326 9.83159 30.98 0.000 Error 136 30.720 0.2259 Error 152 48.231 0.31731 Total 174 155.625 Total 194 141.617

S = 0.4753 R-Sq = 80.26% R-Sq(adj) = 74.74% S = 0.5633 R-Sq = 65.94% R-Sq(adj) = 56.53%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled StDev Pooled StDev Trap Mean ------+------+------+------+- Trap Mean ------+------+------+------+--- ABC-L -0.19421 (--*-) ABC-L 0.66355 (----*----) CDC-L -0.30103 (--*--) CDC-L 0.69786 (----*----) Hall-G 1.26949 (-*--) Hall-G 1.55405 (----*----) Hall-L -0.25380 (--*-) Hall-L 0.74948 (----*-----) RC-G 1.44891 (-*--) RC-G 1.67634 (----*----) ------+------+------+------+------+------+------+------+--- 0.00 0.60 1.20 1.80 0.70 1.05 1.40 1.75

Two-way ANOVA: cx sal log versus Trial, Trap Two-way ANOVA: Positives plus .5 log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 22 12.3271 0.56032 3.25 0.000 Trial 7 0.11072 0.015817 0.66 0.703 Trap 4 7.1884 1.79711 10.41 0.000 Trap 4 1.43471 0.358678 14.98 0.000 Error 88 15.1884 0.17260 Error 28 0.67051 0.023947 Total 114 34.7040 Total 39 2.21594

S = 0.4154 R-Sq = 56.23% R-Sq(adj) = 43.30% S = 0.1547 R-Sq = 69.74% R-Sq(adj) = 57.85%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled StDev Pooled StDev Trap Mean ---+------+------+------+------Trap Mean -+------+------+------+------ABC-L 0.163801 (------*-----) ABC-L -0.301030 (-----*-----) CDC-L 0.306000 (------*------) CDC-L -0.301030 (-----*-----) Hall-G -0.125274 (------*------) Hall-G -0.062469 (-----*----) Hall-L 0.499753 (------*------) Hall-L -0.301030 (-----*-----) RC-G -0.150919 (------*------) RC-G 0.171913 (-----*----) ---+------+------+------+------+------+------+------+------0.25 0.00 0.25 0.50 -0.40 -0.20 -0.00 0.20

Two-way ANOVA: cx spp log versus Trial, Trap Two-way ANOVA: species count log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 38 37.279 0.9810 4.76 0.000 Trial 38 10.5139 0.276682 4.42 0.000 Trap 4 84.419 21.1047 102.39 0.000 Trap 4 0.9696 0.242390 3.87 0.005 Error 152 31.329 0.2061 Error 152 9.5215 0.062642 Total 194 153.026 Total 194 21.0050

S = 0.4540 R-Sq = 79.53% R-Sq(adj) = 73.87% S = 0.2503 R-Sq = 54.67% R-Sq(adj) = 42.14%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled StDev Pooled StDev Trap Mean ------+------+------+------+ Trap Mean -+------+------+------+------ABC-L 0.19564 (--*--) ABC-L 0.270487 (------*------) CDC-L 0.20142 (--*--) CDC-L 0.279277 (------*------) Hall-G 1.50499 (--*--) Hall-G 0.410688 (------*------) Hall-L 0.33962 (--*--) Hall-L 0.299076 (------*------) RC-G 1.65682 (--*--) RC-G 0.438338 (------*------) ------+------+------+------+ -+------+------+------+------0.50 1.00 1.50 2.00 0.20 0.30 0.40 0.50

Two-way ANOVA: Och can log versus Trial, Trap Two-way ANOVA: non-pip/rest spp log versus Trial, Trap

Source DF SS MS F P Source DF SS MS F P Trial 1 0.27331 0.273311 3.41 0.138 Trial 32 7.9160 0.247376 3.23 0.000 Trap 4 0.58082 0.145206 1.81 0.289 Trap 4 2.3765 0.594125 7.75 0.000 Error 4 0.32041 0.080102 Error 128 9.8155 0.076683 Total 9 1.17455 Total 164 20.1080

S = 0.2830 R-Sq = 72.72% R-Sq(adj) = 38.62% S = 0.2769 R-Sq = 51.19% R-Sq(adj) = 37.46%

Individual 95% CIs For Mean Based on Individual 95% CIs For Mean Based on Pooled StDev Pooled StDev Trap Mean ------+------+------+------+-- Trap Mean ---+------+------+------+------ABC-L 0.287016 (------*------) ABC-L 0.280164 (------*-----) CDC-L 0.176091 (------*------) CDC-L 0.344847 (-----*-----) Hall-G -0.301030 (------*------) Hall-G 0.053067 (------*-----) Hall-L -0.062469 (------*------) Hall-L 0.296999 (------*-----) RC-G -0.301030 (------*------) RC-G 0.080701 (-----*------) ------+------+------+------+-- ---+------+------+------+------0.50 0.00 0.50 1.00 0.00 0.15 0.30 0.45

Two-way ANOVA: Och jap log versus Trial, Trap

Source DF SS MS F P Trial 6 0.34498 0.057496 0.76 0.606 Trap 4 1.29769 0.324423 4.31 0.009 Error 24 1.80811 0.075338 Total 34 3.45078

S = 0.2745 R-Sq = 47.60% R-Sq(adj) = 25.77%

Individual 95% CIs For Mean Based on Pooled StDev Trap Mean -+------+------+------+------ABC-L -0.301030 (------*------) CDC-L -0.301030 (------*------) Hall-G 0.110922 (------*------) Hall-L -0.301030 (------*------) RC-G 0.071464 (------*------) -+------+------+------+------0.50 -0.25 0.00 0.25