Table of Contents

Introduction ........................................................................................................... 1
Review of Literature .............................................................................................. 5
Problem Statement ............................................................................................. 15
Experimental Design .......................................................................................... 16
Data and Observations ....................................................................................... 23
Data Analysis and Interpretation ......................................................................... 33
Conclusion .......................................................................................................... 54
Appendix A ......................................................................................................... 61
Appendix B ......................................................................................................... 67
Works Cited ........................................................................................................ 69












Lidwell Webber 1

Introduction
In 2010 alone, there was an estimate of 219 million malaria cases. Of
these cases, an estimated 660,000 resulted in death, and most of these deaths
were of children five years and younger. Almost all malaria cases occur in third-
world African countries, where resources are scarce and health standards are
low (World Health Organization).

Figure 1. Malaria Regional Deaths
(Boseley)

Shown above in Figure 1 is a map which highlights regions based on the
amount of malaria contraction, where colorless represents no danger of
contraction and red represents high danger. Almost all people in Africa, along
with other areas around the world, such as Burma and Haiti, are in high danger
of contracting malaria. Every part of the map is somewhat colored, meaning no
one is risk-free. This map further proves that malaria is a worldly problem that
must be stopped.
Lidwell Webber 2

Malaria is spread through the bites of mosquitoes, in particular the
Analophes gambiae mosquito. These mosquitoes can become infected with the
malaria parasite, and when they bite a human, the human then gets the malaria
disease. In third-world countries that are not sanitary, the rates of mosquitoes
that carry malaria are high, and therefore humans often contract the disease.
Furthermore, the Analophes gambiae mosquito, when carrying the malaria
parasite, is more likely to bite. The parasite needs nutrients found in blood to
survive, so when the mosquito is infected with this parasite it is more driven to
find a human host (Host Preferences of Blood Feeding Mosquitoes).
Some malaria prevention techniques and cures do exist. There are nets
impregnated with mosquito repellents that can be used to ward off the insect, and
medicine to cure the disease if it is contracted (Nursing Times). But these
methods are neither 100% accurate in curing/preventing malaria, nor are they
affordable to the majority of people who contract the disease. In order to
completely prevent malaria, there needs to be a technique to naturally repel
mosquitoes or prevent the mosquitoes from contracting the parasite in the first
place.
After investigation into this issue, a possible answer was found: skin
bacteria. Every humans skin is teeming with bacteria, and when the bacteria eat
chemical compounds produced on the skin they give off an odor. Mosquitoes are
attracted to odors produced by certain skin bacteria, and this attraction helps
them choose who to bite. If there was a way to get rid of this bacterial odor, or
Lidwell Webber 3

make mosquitoes not attracted to it, their attraction to humans would greatly
decrease, helping to fight malaria around the world.
The idea of using skin bacteria to repel mosquitoes could be applied by
either changing humans bacteria in order to repel mosquitoes, or genetically
engineering mosquitoes so their attraction to skin bacteria, along with other
attractants, are defeated. Neither of these applications have been attempted, but
research in the field suggests both methods may be possible in the future.
Research conducted at John Hopkins University manipulated the bacteria
found in mosquitoes guts to interfere with parasite development. This method
could work to stop malaria by placing cotton balls soaked in the engineered
bacteria and sugar and exposing them to mosquitoes in a heavy malaria-risk
community (Schaffer). Another experiment conducted at Michigan State
University used the bacteria Wolbachia to inhibit the malaria virus in mosquitoes.
Wolbachia was injected into mosquito embryos, and after 34 generations, the
majority of that mosquito heritage was malaria free (Milius). These experiments
show that bacteria may be used externally and internally to fight malaria in
mosquito populations.
The experiment conducted in this research paper focused on using
bacteria as an external repellant. Seven bacteria samples, Staphylococcus
epidermidis, Bacillus subtilis, Micrococcus luteus, combinations of two of these,
and a combination of all three, were tested for mosquitoes attractiveness. Two
Petri dishes of the same bacteria sample, along with two Petri dishes of the plain
agar control, were placed on a heating pad inside a cage of mosquitoes.
Lidwell Webber 4

Attractiveness was determined by the amount of mosquitoes that touched each
bacteria sample. The results will help determine if bacteria can be used to
naturally repel mosquitoes.
This research experiment can build upon to the current knowledge on
mosquito, bacteria interactions. It is important to understand if bacteria
combinations can help naturally repel mosquitoes, because this can help
determine if bacteria can be used as a repellent on human skin, rather than just a
malaria prevention inside mosquitoes guts. Furthermore, knowing what bacteria
mosquitoes are most attracted to can help further research on mosquitoes
olfactory receptors and how they determine their prey.
There is still a long way to go before bacteria can be successfully used to
ward off mosquitoes. In order to make this idea a reality, exploration into
mosquitoes attraction to bacteria is necessary. The insight gained from this
experiment can be directly applied to current research in the field, and eventually
a bacterial way to ward off mosquitoes. Bacteria have the power to permanently
prevent malaria, and this research experiment is one more, vital step to get there.







Lidwell Webber 5


Review of Literature
The purpose of this research experiment was to find which skin bacteria
attract Anopheles gambiae mosquitoes, and if higher concentrations of certain
bacteria are more or less attractive to these mosquitoes. Three different bacteria,
Staphylococcus epidermidis, Bacillus subtilis, and Micrococcus luteus, were
grown in Petri dishes with nutrient agar. Seven different bacteria samples were
cultured: samples of each singular bacterium, samples of each combination of
two bacteria, and a sample of a combination of all three bacteria. Single bacteria
applications samples were used to show mosquitoes attraction to that individual
bacterium. Combination bacteria applications plates were used to find if a mixture
of bacteria was enough to mask the smell of individual bacterium and ward off
mosquitoes.
Next, the bacteria samples were exposed to the mosquitoes. Petri dishes
containing the various bacteria samples were taped onto the wall of a cage
containing mosquitoes. In each trial two of the current bacteria samples being
tested and two control dishes were placed in the cage. A heating pad was
attached to the outside of the mosquito cage to heat up the bacteria plates during
the trial. Each trial lasted ten minutes, and the number of mosquitoes observed to
touch the bacteria sample was used to determine that samples attractiveness.
Lidwell Webber 6


Figure 2. Mosquito Cage Set-Up

Figure 2 above shows how the heating pad was attached to the outside of
the mosquito cage, and how the four Petri dishes were attached the inside of the
mosquito cage. The Petri dishes were placed on the wall so they would be closer
to the flying mosquitoes that tended to gravitate toward the sides and top of the
cage. Attaching the heating pad to the outside of the cage allowed the plates to
be directly heated while the netting of the cage was still exposed for easy
mosquito landings. Though there was guidance from Michael Kaufman at MSU,
the researchers came up with this idea primarily on their own.
Lidwell Webber 7


Figure 3. Life Cycle of Mosquitoes
("Biological Notes on Mosquitoes")

Mosquitoes go through four stages of life, from egg, to larva, to pupa, and
finally to adult. A diagram of this life cycle is shown above in Figure 3. Only
adults are able to fly around, and from these adults, only females need a blood
meal. Mosquitoes actually get their food from plant nectar, but females need the
extra nutrients from blood in order to lay eggs ("Biological Notes on
Mosquitoes"). Female mosquitoes drink enough to fill up their abdomen, which
ranges from 0.001 to 0.01 milliliters in size. This does not seem like much, but
each female feeds approximately every two days, and thousands of mosquitoes
are always looking for hosts. Mosquitoes bite using their proboscis, which
penetrates their hosts skin and sucks blood while injecting their own saliva into
the host. Their saliva can contain deadly viruses, such as West Nile Virus and
Malaria, which can be spread to their hosts when biting (Mosquito FAQs).
Mosquitoes smell with their antenna, and nearly half of their brain is
dedicated to analyzing smells with olfactory receptors (Jarell). Each receptor is
used to pick up a certain scent, and when activated it sends off electrical signals
in the olfactory bulb, which is able to detect what the mosquito smells. When a
Lidwell Webber 8

humans olfactory receptors detect a scent, that smell is associated with memory
and feelings, but in a mosquito the brain signal is used to determine the quality of
their food or prey (How Smell Works). This attraction, or lack-of attraction, to a
blood source tells the mosquito when to hone in and bite (Su).
Certain factors in particular lead to female mosquitoes being attracted to
humans. In the field, it is commonly known that colors such as bright red and
blue, carbon dioxide output, lactic acid output, and drinking beer all attract
mosquitoes, but attraction extends deeper than these factors to the chemicals on
a persons skin (Nierenberg). A professional in the field, Dr. Logan, believes the
key to repelling mosquitoes is recreating the bacterial composition or chemicals
that naturally repel mosquitoes on the skin, rather than trying to mask the smells
that attract them. Current repellents only aim to distract mosquitoes from
attractive scents, rather than targeting the bacterial root of the problem (Wang). If
it was concluded that combinations of bacteria can in fact help to repel
mosquitoes, and a way was found to increase the number of bacterium on a
persons skin, then mosquitoes could be effectively repelled in the way Dr. Logan
described.
For the last 50 years mosquitoes have been repelled with sprays that
include the chemical compound DEET. Until recently it was not understood why
this worked, until an experiment was conducted at Howard Hughes Medical
Institute by Leslie Vosshall. This experiment looked at four olfactory receptors in
fruit flies, which act very similar to mosquitoes. When exposed to DEET the
olfactory receptors being tested did not respond at all, but when exposed to other
Lidwell Webber 9

smells at the same time, the receptors behaved differently. Vosshall believes the
compound works as a repellant by momentarily confusing mosquitoes olfactory
receptors, making them not attracted to regular attractants (How Does DEET
Work?).
But DEET repellants may no longer be an effective approach to warding
off the pesky insect. An experiment was conducted to test just how well
repellents work. Mosquitoes were exposed to the main chemical used in bug
sprays, DEET. After three hours of exposure, the effects had worn off completely,
making mosquitoes immune to the repellent effects. Certain species were even
discovered to have a genetic insensitivity to the chemical (Stanczyk). Though
DEET may have been effective when it was discovered 50 years ago,
mosquitoes have evolved to fight against its effects. In order to properly repel
mosquitoes it is necessary to look further, to the root of the problem, which lies in
the chemicals on humans skin.
A study was conducted at Rothamsted University to investigate which
chemicals on the skin would attract mosquitoes most. Human volunteers were
exposed to mosquitoes, and the ones that were attractive to the mosquitoes were
separated from the ones that were not. The two groups were then put in body-
size foil bags for two hours, and the chemicals collected in the bags were
examined afterwards. Subjects who were more likely to be bit had high amounts
of different chemicals on their skin than people who were not. Two chemicals
were found to be significant mosquito repellents, including 6-methyl-5-heptene
Lidwell Webber 10

and geranylacetone. When these chemicals were applied to the mosquito
attracting people, the mosquitoes would no longer bite them (Wang).
Chemicals on a persons skin do not give off their own odor, so how is it
that they can attract mosquitoes? The answer lies in the bacteria that live on
peoples skin. When people sweat, the bacteria colonies living on the skin eat the
sweat, which includes chemicals that are also present on the skin. The waste the
bacteria leave behind is what gives off body odor. Different bacteria types give off
different odors and mosquitoes can detect all of these individually (Smallegange).
Due to this source of odor, the bacteria composition of a persons skin is one of
the main factors that determine their attractiveness to mosquitoes (The Role of
Skin Microbiota). This research aimed to investigate which bacteria would
attract mosquitoes most, and further look into what properties affected this
attraction.
Recently, a study was conducted by Julia Segre from the National Human
Genome Research Institute to understand more about bacteria that occurs on
human skin, and find possible ways of preventing bacterial skin diseases. Skin
samples were taken from human volunteers, then these samples were analyzed
by sequencing the 16S ribosomal RNA genes. This particular type of RNA is
specific to bacteria, and was able to reveal the bacterial compositions of the skin
samples. The study found that there are more than 112,000 bacteria gene
sequences living on human skin, and that the more diversity of bacteria, the
healthier the skin sample. Each persons bacterial make-up is slightly different,
causing difference in mosquito attraction from person to person ("Study finds
Lidwell Webber 11

unexpected bacterial...). Mosquitoes may also prefer certain bacterial odors to
others because of the amount of carbon dioxide emitted by the bacteria colony,
which turns oxygen into carbon dioxide through aerobic respiration (Carbon
Dioxide).
Another experiment conducted at the University of Idaho further suggests
that a balance of bacteria is healthy, but both studies show that too much of one
bacteria can lead to infections (University of Idaho). When investigating which
bacteria repel mosquitoes, and trying to figure out a solution from this
information, it is important to keep in mind bacteria can be dangerous in high
amounts.
The bacteria used in the experiment discussed in this paper were selected
because of their difference in smell, and previous testing in the field. In an
experiment conducted at Wageningen University, in the Netherlands,
mosquitoes attraction to five different bacteria naturally found on the skin was
tested. Shown below, in Figure 4, are the mosquitoes responses to each of the
bacterium.

Figure 4. Mosquitoes Attraction to Certain Bacteria
(Differential Attraction of Malaria Mosquitoes)
Brev. epidermidis
B. subtilits
c. minutissimum
P. aeruginosa
S. epidermidis
Mosquitoes Trapped (%)
Lidwell Webber 12


The graph shows that four of the five bacteria attracted mosquitoes, two of
these being Staphylococcus epidermidis and Bacillus subtilis. This is quantified
by the number of mosquitoes found in the trapping device containing the sample,
displayed on the left axis of the graph. It also shows that bacteria were attractive
in their stationary growth phase, rather than their exponential growth phase,
displayed on the right axis of the graph (Differential Attraction of Malaria
Mosquitoes). During bacterias exponential growth phase, or log phase,
bacteria colonies rapidly expand. Following this phase is the stationary phase,
where the level of bacteria remains near constant (Bacterial Growth Curve). In
this experiment, bacteria samples were allowed to culture in an incubator for two
days, where they reached their exponential growth phase and began to level off
near the end of incubation. When they were removed from the incubator and
used in this study the bacteria samples were in their stationary growth phase.
Staphylococcus epidermidis and Bacillus subtilis were selected or use in
this experiment based upon from the research experiment conducted at
Wageningen University. Micrococcus luteus, though not tested in Wageningen
Universitys experiment, is similar to bacteria that were tested because it
naturally grows on the skin. The use of these bacteria may have led to better
data because they were previously known to be mosquito attractants. The
experiment at conducted at Wageningen University proved that lab grown
bacteria, in agar, can simulate the smell of bacteria on human skin enough to
attract mosquitoes, further justifying this research experiments data.
Lidwell Webber 13

In another similar experiment conducted at Wageningen University glass
beads were rubbed on human volunteers feet, and the bacteria collected was
allowed to grow. The glass beads were then set in a trapping device connected
to a cage of mosquitoes, and the amount of mosquitoes stuck in the trapper after
fifteen minute periods was used to determine the samples attractiveness. The
results of Wageningen Universitys experiment found that especially attractive
human bacteria samples were high in Staphylococcus epidermidis, while the few
samples that actually repelled the mosquitoes were high in Pseudomonas
aeruginosa. The experiment results also suggested that more of a blend of skin
bacteria can help repel mosquitoes (Composition of Human Skin Microbiota).
This is where the hypothesis, that bacteria combinations will be less attractive
than single bacterium samples, was derived.
In both experiments conducted at Wageningen University trapping devices
were used to determine bacteria samples attractiveness to mosquitoes. The way
bacteria samples were exposed to mosquitoes in this experiment mirrors the
basic set-up of these trapping devices, shown below in Figure 5.
Lidwell Webber 14


Figure 5. Trapping Device
(Differential Attraction of Malaria Mosquitoes)

Mosquitoes were released at the top of these devices, and allowed to fly
through tiers of trapping devices. The trapping devices had bacteria samples and
controls, which gave the mosquitoes a choice, helping to determine the real
attraction of the bacteria sample by ruling out random mosquito behavior.
Distilled water was pumped into the trapping devices to warm up the samples,
while the pressurized air and suction kept the air in the trapping device fresh to
eliminate odors besides those of the bacteria samples (The Role of Skin
Microbiota). In this experiment mosquitoes were also given a choice between
different bacteria samples, and the samples were heated. Though the set-up is
cruder, it takes the most important factors from the set-up used in the
Wageningen University experiments.
Aspects were taken from both Wageningen University experiments to help
design this research experiment. Due to this, the data collected in this
Lidwell Webber 15

experiment will expand on previously collected data and give further insight to
mosquitoes relationship to skin bacteria.








Problem Statement

Problem:
The purpose of the experiment is to quantify Anopheles gambiaes
attraction to Staphylococcus epidermidis, Bacillus subtilis, and Micrococcus
luteus, along with combinations of two of these bacteria, and a mixture of all
three. This will show which naturally occurring skin bacteria attract Anopheles
gambiae, and if a healthy balance of bacteria could help to repel the Anopheles
gambiae. This will give insight into future ways to repel Anopheles gambiae, and
help the fight against malaria.

Hypothesis:
The bacteria plates with only one of Staphylococcus epidermidis, Bacillus
subtilis, and Micrococcus luteus will attract the most mosquitoes and the
Lidwell Webber 16

combination of Staphylococcus epidermidis, Bacillus subtilis, and Micrococcus
luteus will attract the least mosquitoes.

Data Measured:
The independent variable was the type of bacteria on the Petri dish, which
determined the dependent variable of the number of the mosquitoes that land on
the plate. The attraction to the bacteria samples was measured in the number of
mosquitoes that land on the plate. Since the sample means of the mosquito
attraction of different combinations of bacteria were being compared, an ANOVA
test and multiple two-sample t tests were the method of statistical analysis.
Bacteria Procedure
Materials:

Carolina Biological LJ-155155 Micrococcus luteus
Carolina Biological LJ-154921 Bacillus subtilis
Carolina Biological LJ-155556 Staphylococcus epidermidis
8 g Carolina Biological Nutrient Agar
20 60 mm x 15 mm Sterile Petri Dishes
Transfer Loop
5 cm Stirring Magnet
1 mL Pipette
100 mL Beaker
330 mL Distilled Water
1000 mL Beaker
500 mL Beaker
(28) 28 mL Test Tubes
Test Tube Rack
Hot Plate
Hot Mitt
Bunsen Burner
Heat-Resistant Glove
Incubator (set at 38C)
Roll of Scotch Masking Tape
Scale (0.0001 g)
Metal Tongs
Weigh Boat
Lidwell Webber 17

TI-nspire Calculator

Procedure:
Labeling the Petri Dishes

1. Label the bottom of eighteen Petri dishes with the trial number and
bacteria type using the permanent marker. Two Petri dishes are used for
each bacterium alone, the mixtures of combinations of two bacteria, the
mixture of all three, and controls of agar not treated with bacteria. See
Figure 7.

Agar Preparation

1. Sterilize 500 mL beaker using tongs by dipping beaker into a 1000 mL
beaker of boiling water. Rinse and repeat.

2. Fill 500 mL beaker with 300 mL of distilled water and set on hot plate.
Place stirring magnet into beaker. Set the hot plate to high for the stirring
speed.

3. Measure 8 g of nutrient agar onto the weigh boat and slowly transfer it to
beaker.

4. Set the hot plate onto high temperature. Heat until the mixture is boiling,
which will signal its completion.

5. Take the beaker off of the hot plate and set it on and pour 8 g agar into
the Petri dishes so that the agar covers the bottom. Set aside for cooling.


Dilutions

1. Sterilize 28 test tubes in the boiling water.

2. Fill the 100 mL beaker with 30 mL of distilled water and fill each test tube
with 1 mL of water using the pipette. Place four of the test tubes aside to
use for the dilution of the bacteria in this trial.

3. Take cap off of the slant of bacteria that is to be used for the specific
combination and immediately sterilize the top edge using the Bunsen
burner.

4. Sterilize transfer loop using the Bunsen burner. Expose all of the wire to
the flame until it turns a bright orange and allow it to cool. Sterilize loop
after each dip.

Lidwell Webber 18

5. Obtain a small sample of bacteria with loop and cover slant. Mix bacteria
sample into the first tubes.

6. Dip transfer loop into the test tube containing the bacteria sample and
obtain a bubble of water. Mix the bubble of it into the next test tube. Do the
same for the remaining test tubes.

7. Repeat for each bacteria and bacteria combination.

Inoculating Petri Dishes

1. Pour the contents of each final diluted test tube into the Petri dish with the
matching trial label, making sure it spreads evenly.

2. Stack and tape Petri dishes according to group with masking tape and
place them in an incubator set at 38 C. Leave for approximately 48 hours.







Diagrams:

Lidwell Webber 19


Figure 6. Materials

Figure 6 above shows a labeled picture of the materials used in the
bacteria procedure. Not included in this picture are the Heat-Resistant Glove,
Incubator (set at 38C), Sharpie Permanent Marker, and TI-nspire Calculator.


Lidwell Webber 20


Figure 7. Petri Dish Labeling

Figure 7 above shows how to label the Petri dishes. The first letter of each
bacteria type is used to label which bacteria is in the plate. The word control is
used to indicate the plates with plain nutrient agar. Also, the plates have the trial
number so that the plate preparation order is randomized.




Lidwell Webber 21

Mosquito Procedure

Materials:

Canon Digital Camera
Anopheles gambiae (approximately 100 on day one, and 75 on day two)
Mosquito Cage (2 ft. x 2 ft. x 3.5 ft.)
Scotch Masking Tape
Scotch Double-Sided Tape
Premium Touch Medical Examination Gloves
Heating Pad
TI Inspire Calculator

Procedures:

1. Randomize the trials, using the random integer function on the TI-nspire
calculator by typing in randInt(1, 7, 7), to determine what bacteria sample
is being used and reduce bias by, to determine which type of bacteria
plates to test.

2. Place heating pad on side of mosquito cage and secure with scotch tape.

3. Retrieve the bacteria covered Petri dishes from the incubator and secure
two controls and two plates of the same bacteria sample on the side of the
cage on the heating pad with scotch double sided tape.

4. Watch the mosquito activity for 10 minutes, keeping track of how many
mosquitoes touch each sample.

5. Repeat for each bacteria combination.

Lidwell Webber 22

Diagram:


Figure 8. Mosquito Procedure Materials

Figure 8 above shows a labeled picture of the materials used in the
mosquito procedure. Not included is the Canon Digital Camera.


Lidwell Webber 23


Figure 9. Mosquito Cage

Shown above in Figure 9 is the cage containing Anopheles gambiae
mosquitoes.

Figure 10. Set-Up
Figure 10 shows how the heating pad and bacteria plates were attached
to the mosquito cage.

Lidwell Webber 24

Data and Observations
Table 1
Bacteria Abbreviation Key
Bacteria
Abbreviation
Bacteria Included in the Combination
B Bacillus subtilis
M Micrococcus luteus
S Staphylococcus epidermidis
BM Bacillus subtilis and Micrococcus luteus
BS Bacillus subtilis and Staphylococcus epidermidis
MS Micrococcus luteus and Staphylococcus epidermidis
BMS Bacillus subtilis, Micrococcus luteus, and Staphylococcus epidermidis

Table 1 above is a key to be used throughout the Data and Observations
and Data Analysis and Interpretation sections. Each bacterium or combination of
bacteria is labeled with a letter or combination of letters. B is for the Bacillus
subtilis trials, M is for the Micrococcus luteus trials, S is for the Staphylococcus
epidermidis trials, BM is for the Bacillus subtilis and Micrococcus luteus trials, BS
is for the Bacillus subtilis and Staphylococcus epidermidis trials, MS is for the
Micrococcus luteus and Staphylococcus epidermidis trials, and BMS is for the
Bacillus subtilis, Micrococcus luteus, and Staphylococcus epidermidis
combination trials.













Lidwell Webber 25

Table 2
Experiment Data
Bacteria
Combination
Day 1 Day 2
Trial
Mosquito Attraction
(# of Touches)
Trial
Mosquito Attraction
(# of Touches)
Plate 1 Plate 2 Plate 1 Plate 2
B 2 6 4 1 7 3
M 4 3 5 7 4 3
S 1 4 5 4 4 4
BM 3 4 2 6 4 1
BS 6 2 4 2 4 2
MS 7 3 2 3 2 2
BMS 5 0 1 5 1 0


Table 2 above shows the data from the bacteria plate trials during the
experiment. The first column states which bacteria combination being tested.
Refer to Table 1 for a key for the abbreviations. The table is divided up into day
one and two since there were two days of trials that were conducted at Michigan
State University. Day 1 was October 30, 2013 and Day 2 was November 13,
2013. The trial columns show the order that each trial was performed due to
randomization. The data measured was mosquito attraction, which is quantified
as the number of times mosquitoes touched the plates of bacteria during the trial.
The mosquito touches were counted visually. During each trial, there were two
plates of bacteria tested inside the mosquito cage, which is why there are two
plate columns per day of trials.






Lidwell Webber 26



Table 3
Combined Results
Bacteria
Combination
Average
(# of Touches)
Standard
Deviation
(# of Touches)
Group
Average
(# of Touches)
Group Standard
Deviation
(# of Touches)
B 5.00 1.83
4.33 1.23 M 3.75 0.96
S 4.25 0.50
BM 2.75 1.50
2.67 1.07 BS 3.00 1.15
MS 2.25 0.50
BMS 0.50 0.58 0.50 0.58

Table 3 above shows the combined results of both days of trials. Refer to
Table 1 for a key for the bacteria combination abbreviations. The average of all
the trials for each of the seven different combinations is shown in the average
column. The standard deviation of all the trials for each of the seven different
combinations is shown in the standard deviation column. The group average and
group standard deviation tables show the averages and standard deviations of all
the bacteria plates with only one bacterium, all the bacteria plates with a
combination of two bacteria, and all the plates with a combination of three
bacteria. The averages and standard were grouped to represent the three
different combinations: a single bacterium, a combination of two bacteria, or a
combination of three bacteria. By using this grouping, the results of how the
number of bacteria affect the mosquito attraction are more clear.



Lidwell Webber 27

Table 4
Control Data
Bacteria
Combination
Day 1 Day 2
Trial
Mosquito Attraction
(# of Landings)
Trial
Mosquito Attraction
(# of Landings)
Control 1 Control 2 Control 1 Control 2
B 2 1 2 1 2 1
M 4 0 2 7 1 2
S 1 1 2 4 0 2
BM 3 0 1 6 1 1
BS 6 0 2 2 2 0
MS 7 2 0 3 1 1
BMS 5 2 0 5 1 1

Table 4 above shows the data from the control trials. At the same time the
bacteria trials were being conducted, there were always two control plates of
bacteria being tested as well to show that the mosquitoes were not just randomly
flying up to and touching the bacteria plates. The control plates were Petri dishes
of just plain agar that no bacteria were applied to. However, the plates after being
placed in the incubator for two days did experience some slight growth which can
be seen later in Figures 11 and 12. Still, the growth was minimal, especially
compared to the actual bacteria plates. The table layout is the same as Table 2.
The bacteria combination column shows which bacteria plates were being tested
at the same time as the control plates. Refer to Table 1 for a key for the bacteria
combination abbreviations.





Lidwell Webber 28

Table 5
Control Data Statistics
Bacteria
Combination
Average
(# of
Mosquito
Touches)
Standard
Deviation
B 1.50 0.58
M 1.25 0.96
S 1.25 0.96
BM 0.75 0.50
BS 1.00 1.16
MS 1.00 0.82
BMS 1.00 0.82

Table 5 above shows the average and standard deviations of the
mosquito attraction for the control data during each of the seven different
combinations of bacteria trials. The controls were tested at the same time the
bacteria was being tested in order to provide the mosquitoes with a choice. The
bacteria combination column shows the what bacteria the controls were being
tested with, however, the averages and standard deviations are for the control
data, meaning the plates of plain agar that no bacteria was applied to. Refer to
Table 1 for a key for the bacteria combination abbreviations.













Lidwell Webber 29

Table 6
Day 1 Observations
Trial
Bacteria
Plate
Observations
1 S
On the Control Plate #2, a mosquito landed and stayed on the
control for the rest of the trial.
2 B
Mosquitoes hovered over the plates for awhile before actually
touching the plates. After the trial, Researcher 2 spilled water on
one of the bacteria plates and it was disposed of immediately.
3 BM
Two mosquitoes hovered for awhile over the B1 plate. The
majority of activity for the mosquitoes occurred toward the end of
the ten minute trial.
4 M
Before this trial was conducted, all bacteria plates, including the
controls, were removed from the cage and the mosquitoes were
given a 20 minute period to relax. When placing the M1 plate into
the cage, Researcher 2 accidentally touched the bacteria on it,
leaving a fingerprint. Mosquitoes did not hovering over the
bacteria before touching it in this trial.
5 BMS
Researcher 2 spilled water on the lid of the Petri dish, however,
the bacteria on the plate seemed to be untouched. Mosquitoes
showed no attraction. Mosquitoes did not even land on the hot
pad around the bacteria during this trial.
6 BS
Instead of briefly touching the BS1 plate, a mosquito landed on it
for around 5 seconds.
7 MS
Mosquitoes hovered over the bacteria often during this trial before
landing on the bacteria plates.

Table 6 above shows the observations from Day 1 of testing. Significant
occurrences were noted. On this date, it was determined that the cage contained
approximately 75 mosquitoes, with mainly female mosquitoes. During all the
trials on this day, Researcher 1 carefully monitored when mosquitoes touched
the plates and Researcher 2 placed the plates in and out of the cage for the trial.
Refer to Table 1 for a key for the bacteria combination abbreviations.






Lidwell Webber 30

Table 7
Day 2 Observations
Trial
Bacteria
Plate
Observations
1 B
Mosquitoes appeared to be less active than on the previous day
of trials, as well as stay on the top of the cage rather than the
sides. There was consistent mosquito hovering over the B1 plate.
Though the activity decreased, there was still attraction.
2 BS
Mosquitoes hovered around both the bacteria plates and the
controls. One mosquito landed on the BS1 plate and stayed for a
few seconds.
3 MS Mosquitoes showed slight disinterest in the bacteria plates.
4 S
Mosquito activity was quite low for this trial, so after the trial the
mosquitoes were given a 20 minute resting period before any
more trials were conducted.
5 BMS
Mosquitoes had no activity at all except for three curious
mosquitoes. All the other ones completely avoided the bacteria.
6 BM
This trial had slightly more activity although many mosquitoes
preferred to stay on the wall.
7 M
Mosquitoes seem not as hungry as they were on the previous day
of trials, still they are attracted.

Table 7 above shows the observations from Day 2 of testing. Significant
occurrences were noted. On this date, it was determined that the cage contained
approximately 65 mosquitoes. Along with there being slightly fewer mosquitoes,
there was also a noticeably less activity in the mosquitoes. The cage may have
had had less female mosquitoes than on the Day 1 trials. During all the trials on
this day, Researcher 1 carefully monitored when mosquitoes touched the plates
and Researcher 2 placed the plates in and out of the cage for the trial. Refer to
Table 1 for a key for the bacteria combination abbreviations.



Lidwell Webber 31


Figure 11. Day 1 Bacteria Plates

Figure 11 above shows the bacteria plates from the first day of trials. This
picture was taken after trials, which is why a second Bacillus subtilis plate is
missing from the picture, as explained in the Bacillus subtilis row in the
observation table. Notice the small colonies of bacteria on some of the control
plates. This growth occurred on the second day of incubation. The plates overall
had an even growth of bacteria on them.

Figure 12. Day 2 Bacteria Plates

Figure 12 above shows the bacteria plates from the second day of trials.
Again, there was slight bacteria growth on the controls, but nothing compared to
the actual bacteria plates.

Lidwell Webber 32


Figure 13. Mosquito Landing on Plate

Figure 13 above shows a mosquito that actually landed on a plate, such
as the one described in the Staphylococcus epidermidis trial in Table 6. This was
a rare occurrence since mosquitoes simply flew to the plate, touched it briefly,
and left. This particular plate is a control plate, although landings also occurred
on bacteria plates.

Figure 14. High Activity Mosquito Trial

Figure 14 above shows what the activity looked like during a mosquito trial
with strong attraction and high mosquito activity. This particular photo is from the
Day 1 Bacillus subtilis trials, although similar activity was experienced in other
trials.
Lidwell Webber 33


Figure 15. Low Activity Trial
Figure 15 above shows the mosquito activity during a low activity trial.
This particular trial was the Day 1 Bacillus subtilis, Micrococcus luteus, and
Staphylococcus epidermidis combination trial. In the trials with less mosquito
attraction, rather than multiple mosquitoes visiting the plates, there was only a
couple in the whole ten minute span.











Lidwell Webber 34

Data Analysis and Interpretation
Data was collected using a comparative experiment to find the attraction
of mosquitoes to combinations of the skin bacteria of Bacillus subtilis,
Micrococcus luteus, and Staphylococcus epidermidis. In the experiment, the
mosquitoes that touched the plates of bacteria were counted visually to
determine that samples attraction. In order to ensure accuracy in the data
collected, a control, randomization, and replication were used.
The control was a plate with plain nutrient agar. Two of these control
plates were placed in the mosquito cage during each trial. Since no experiment is
error-free, a control can identify the effect of lurking variables on the results so
that it can be accounted for when analyzing the data. The controls helped to
indicate whether the mosquitoes were attracted to the bacteria or just the plain
plates of agar.
Randomization of the trials was used to reduce bias. To do this, the seven
different variations of bacteria plates were numbered and then randomized using
the random integer function on the TI-Nspire calculator. By doing this, bias is
reduced since the order that the trials are performed is random, so the results are
more reliable.
Replication was also a factor in the experiments to increase the validity of
the data, meaning that there are more results to compare to each other. If the
results are similar from the repetitions, they help to show the results were not just
random occurrences. The experiment was conducted twice, each time being on a
different day. A total of four plates of each of the seven combinations of bacteria
Lidwell Webber 35

were tested. This was the greatest number of trials that was able to be conducted
due to trials being conducted at Michigan State University, having limited access
to the mosquitoes, and having the issue of travel time. Still, by having four trials
per combination of similar results, it showed that the results were not just errors,
and makes it so the consistency of the results can be judged. The more
repetitions, the more dependable the data is. Data has been analyzed by
comparing box plots, using an ANOVA test, and using a various Two Sample t-
test (see Appendix A for the tests).

Figure 16. Bacteria and Control Box Plot
Figure 16 shows the comparison of the data of the bacteria trials and the
control trials. The median of the controls is 1 and the median of the bacteria
samples is 3. Both plots are relatively symmetric, but due to lack of variation in
the data, the control has no tails on the box plot. The range of the control box plot
is two and the range of the bacteria box plot is 7, meaning the bacteria box plot
has a much larger spread than the control box plot. Though there is overlap
between the box plots, the control box plot only overlaps with 25% of the bacteria
Mosquito Attraction (# of Mosquitoes Touches)
A
l
l

b
a
c
t
e
r
i
a

C
o
n
t
r
o
l
s

Lidwell Webber 36

plot data, meaning that there still could be a significant difference among all the
bacteria groups. The overlap is due to the bacteria plates with lower mosquito
attraction. Further statistical analysis may prove to have no significant difference
from the control, therefore suggesting that the bacteria combination is effective at
helping to repel mosquitoes.

Figure 17. Histograms of All the Bacteria Combinations
Shown above in Figure 17 are histograms of all seven of the different
bacteria samples: a combination of Bacillus subtilis, Micrococcus luteus, and
Staphylococcus epidermidis, a combination of Bacillus subtilis and Micrococcus
luteus, a combination of Bacillus subtilis and Staphylococcus epidermidis, a
combination of Micrococcus luteus and Staphylococcus epidermidis, Bacillus
subtilis, Micrococcus luteus, and Staphylococcus epidermidis. A histogram had to
be used to compare data point since 4 data points are not enough to create a
Mosquito Attraction (# of Mosquitoes Touches)
F
r
e
q
u
e
n
c
y

b
m


b

m

s

b
s

m
s

b
m
s

2
0
2
0
2
0
2
0
2
0
2
0
2
0
b m s bm bs ms bms
Lidwell Webber 37

quality box plot. All the histograms overlap at least one histogram, but there is a
large range between the single and combination of three bacteria samples. The
plates with only one bacterium have the highest mosquito attraction. Their
medians are 5 for Bacillus subtilis, 3.5 for Micrococcus luteus, and 4 for
Staphylococcus epidermidis, whereas all the other bacteria plates have medians
3 or below, with the combination of all three bacteria having a median of only 0.5.
The histograms suggest that the more bacteria that is combined, the less
mosquito attraction there was. The Bacillus subtilis histogram has the largest
range of 4, meaning it has the largest spread. The combination of Bacillus
subtilis, Micrococcus luteus, and Staphylococcus epidermidis, the combination of
Micrococcus luteus and Staphylococcus epidermidis, and the singular
Staphylococcus epidermidis all have the smallest range of 1, meaning they have
the smallest spreads and smallest variation in the data. All statistics considered,
the histograms could mean that there is a significant difference between all the
different bacteria samples.
ANOVA Test
First, an ANOVA test was run because an ANOVA Test is used to
compare the means or three or more populations to test how far apart sample
means are with how much variation there is within the samples. Since there are
seven different bacteria combinations being compared, an ANOVA test is
appropriate.




Lidwell Webber 38

Assumptions:

1. There are I Independent SRSs, one from each of I populations.
2. Each population has a normal distribution.
3. All populations have the same standard deviation, o, whose value is
unknown.
4. The largest sample standard deviation is no more than twice as large as
the smallest sample deviation.

Assumptions are used to ensure that the test is valid and that there are
not any underlying factors affecting the test. The first assumption was met since
the randomization used made each trial a simple random sample. Also, each trial
performed did not affect another trials results in any way, so the trials were
independent. The second assumption was met since the data was indeed
relatively normal, shown in the normal probability plots below in Figures 18, 19,
20, 21, 22, 23, and 24. An ANOVA test is robust, meaning the test will still
provide quality insight even if the conditions of the test are not perfectly ideal, so
the small sample sizes in the experiment do not affect the results when the
distribution is normal. The third assumption was met because the data used is
real sample data, which means the population mean and standard deviations are
unknown. The fourth assumption, however, was not met because the largest
sample standard deviation was 1.826, and the smallest was 0.5. Two times 0.5 is
1.0, so the largest standard deviation is more than twice as large as the smallest
standard deviation. This means that the results of the ANOVA test many not be
trustworthy, making it inconclusive.



Lidwell Webber 39


Figure 18. Normal Probability Plot for Bacillus subtilis
Figure 18 above shows the normal probability plot for the Bacillus subtilis
trials. The data is almost completely normal since it does not vary very far from
the line at all. Because of this, it is usable data for the ANOVA test.

Figure 19. Normal Probability Plot for Micrococcus luteus
Figure 19 above shows the normal probability plot for the Micrococcus
luteus trials. The vertically aligned points are due to the repeated points in the
data. The data is relatively normal since it does not vary too far from the line.
Because of this, it is usable data for the ANOVA test.
Mosquito Attraction (# of Mosquitoes Touches)
Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 40


Figure 20. Normal Probability Plot for Staphylococcus epidermidis
Figure 20 above shows the normal probability plot for the Staphylococcus
epidermidis trials. The vertically aligned points are due to the lack of variation in
the data. The data is relatively normal so it is usable for the ANOVA test.

Figure 21. Normal Probability Plot for Bacillus subtilis and Micrococcus luteus
Figure 21 above shows the normal probability plot for the trials with a
combination of Bacillus subtilis and Micrococcus luteus. The vertical points are
due to repeated data point values. The data is relatively normal since it does not
vary too far from the line. Because of this, it is usable data for the ANOVA test.
Mosquito Attraction (# of Mosquitoes Touches)
Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 41


Figure 22. Plot for Bacillus subtilis and Staphylococcus epidermidis
Figure 22 above shows the normal probability plot for the trials with a
combination of Bacillus subtilis and Staphylococcus epidermidis. The pairs of
vertical aligned points are due to lack of variation in the data. The data is
relatively normal so it is usable data for the ANOVA test.

Figure 23. Plot for Micrococcus luteus and Staphylococcus epidermidis
Figure 23 above shows the normal probability plot for the trials with a
combination of Micrococcus luteus and Staphylococcus epidermidis. The
Mosquito Attraction (# of Mosquitoes Touches)
Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 42

vertically aligned points are due to the lack of variation in the data. The data is
relatively normal so it is usable data for the ANOVA test

Figure 24. Normal Probability Plot for Bacillus subtilis, Micrococcus luteus, and
Staphylococcus epidermidis
Figure 24 above shows the normal probability plot for the trials with a
combination of Bacillus subtilis, Micrococcus luteus, and Staphylococcus
epidermidis. The data is relatively normal since it does not vary too far from the
line. The vertical points are due to the lack of variation in the data. Because of
this, it is usable data for the ANOVA test.
Hypotheses:
H
o
:
b
=
m
=
s
=
bm
=
bs
=
ms
=
bms

H
a
: Not all
b
,

m
,
s
,
bm
,
bs
,
ms
, and
bms
are equal.

The hypotheses shown above are the null hypothesis and alternate
hypothesis of the ANOVA test conducted. The null hypothesis states that the
average all the bacteria combinations are equal. This hypothesis is either
rejected or failed to be rejected. The alternate hypothesis states that not all the
Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 43

averages of the bacteria combinations are equal, meaning that there is a
significant difference between them.
ANOVA Test Results
F statistic: 6.98077
p-value: 0.000347


Figure 25. Distribution from the ANOVA Test
Figure 25 above shows the distribution of the F statistic from the ANOVA
test. Notice there is no visible shading under the curve because the p-value is
nearly zero.
Reject H
0
because the p-value of 0.000347 is below the alpha level of
0.05. There is significant evidence that there is a difference in mosquito attraction
between the different bacteria combinations. There is a 0.03% chance of getting
these results by chance alone, assuming H
0
is true.




Lidwell Webber 44

Two-Sample t Tests
Next, three two-sample t tests were run since this test is used to
determine whether the average difference between two groups is significant on
the alpha level of 0.05. In these tests, different combinations of two groups are
compared, making two-sample t tests appropriate.
Two-Sample t Test of Singular Bacteria Plates and Double Bacteria Plates
A test was conducted to compare the singular plates of Bacillus subtilis,
Micrococcus luteus, and Staphylococcus epidermidis to the two bacteria
combinations of Bacillus subtilis and Micrococcus luteus, Bacillus subtilis and
Staphylococcus epidermidis, and Micrococcus luteus and Staphylococcus
epidermidis. This test was conducted to determine if one skin bacteria was more
attractive than a mixture of two.
Assumptions:
1. Two simple random samples from two distinct populations
2. Samples are independent
3. Both populations are normally distributed
4. The population means and standard deviations are not known

Assumptions are used to ensure that the test is valid and that there are
not any underlying factors affecting the test. The randomization used made each
trial a simple random sample. Each trial performed did not affect another trials
results in any way, so the trials were independent. The normality of the data is
shown below in Figures 26 and 27. The data used is real sample data, which
makes it so that the population mean and standard deviations are unknown.
Since the assumptions above were met, a two-sample t test was able to be
Lidwell Webber 45

performed. Note that the data was not perfectly normally distributed, but it was
close enough for the test to still be performed.

Figure 26. Normal Probability Plot for the Singular Bacteria Plates
Figure 26 above shows the normal probability plot for the bacteria plates
with only one bacterium. The distribution is relatively normal since the data
points are near the normal line. There are two grouping of vertically aligned
points, but even so, they do not deviate far from the normal line. From this it can
be concluded the data is normal enough to conduct the two-sample t test.

Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 46


Figure 27. Normal Probability Plot for the Double Bacteria Plates
Figure 27 above shows the normal probability plot for the bacteria plates
with only one bacterium. The distribution is relatively normal since the data
points are moderately near the normal line. There are two grouping of vertically
aligned points, but even so, they do not deviate far from the normal line. From
this it can be concluded the data is normal enough to conduct the two-sample t
test.
Hypotheses:
H
o
:
one
=
two

H
a
:
one
>
two

The null hypothesis states that the average mosquito attraction of the
bacteria plates with one bacterium is equal to that of bacteria plates with a
combination of two bacteria. The alternate hypothesis states that the mosquito
attraction of the bacteria plates with one bacterium is greater than the mosquito
attraction of bacteria plates with a combination of two bacteria.

Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 47


Figure 28. Distribution from Singular and Double Bacteria Plate test
Figure 28 above shows the distribution of the t value from the two-sample
t test. Note that none of the shaded part can be seen, since the p-value is
extremely close to zero.
Two-Sample t Test Results
t value: 3.5355
p-value: 0.0009

Reject H
0,
because the p-value 0.0009 is below the alpha level 0.05. There
is significant evidence that the mosquito attraction of the single bacteria plates is
greater than the attraction of the double bacteria plates. There is about a 0.09%
chance of getting a difference in mosquito attractions this extreme by chance
alone if no difference is assumed.




Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 48

Two-Sample t Test of Double Bacteria Plates and Triple Bacteria Plates
A test was conducted to compare the two bacteria combinations of
Bacillus subtilis and Micrococcus luteus, Bacillus subtilis and Staphylococcus
epidermidis, and Micrococcus luteus and Staphylococcus epidermidis to the
combination of all three bacteria Bacillus subtilis, Micrococcus luteus, and
Staphylococcus epidermidis. This test was conducted to determine if a
combination of two skin bacteria was more attractive than a mixture of three.
The assumptions were met for this t test. The randomization used made
each trial a simple random sample. Each trial performed did not affect another
trials results in any way, so the trials were independent. The normality of the
data is shown above in Figures 27 (in the previous t test) and below in Figure 29.
The data used is real sample data, which makes it so that the population mean
and standard deviations are unknown. Since the assumptions above were
basically met, a Two-Sample t test was able to be performed. Note that the data
was not perfectly normally distributed, but it was close enough for the test to still
be performed.

Lidwell Webber 49


Figure 29. Normal Probability Plot for Triple Bacteria Plates
Figure 29 above shows the known triple bacteria plates normal probability
plot. The distribution is relatively normal since the data points are moderately
near the normal line. The vertically aligned points represent that data point
occurring twice. From this it can be concluded the data is normal enough to
conduct the two-sample t test.
Hypotheses:
H
o
:
two
=
three

H
a
:
two
>
three


The null hypothesis states that the average mosquito attraction of the
plates of bacteria with a combination of two bacteria is equal to the mosquito
attraction of bacteria plates with a combination of three bacteria. The alternate
hypothesis states that the average mosquito attraction of the plates of bacteria
with a combination of two bacteria is greater than the mosquito attraction of
bacteria plates with a combination of three bacteria.
Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 50


Figure 30. Distribution of Double Bacteria Plates and Triple Bacteria Plates Test
Figure 30 above shows the distribution of the t value from the Two-Sample
t test. Note that none of the shaded part can be seen, since the p-value is very
close to zero.
Two-Sample t Test Results
t value: 5.1169
p-value: 0.0002

Reject H
0,
because the p-value 0.0002 is below the alpha level of 0.05.
There is significant evidence that average mosquito attraction of the plates with a
combination of two bacteria is greater than the average mosquito attraction of
plates with a combination of three bacteria. There is a 0.02% chance of getting a
difference this extreme by chance alone if no difference is assumed.





Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 51

Two-Sample t Test of Combination of All Three Bacteria and Controls
A test was conducted to compare the combination of all three bacteria
Bacillus subtilis, Micrococcus luteus, and Staphylococcus epidermidis to the
controls. This test was conducted to determine if there was a significant
difference between the combination for all three bacteria and the control. No
significant difference would help to prove that a balance of multiple bacteria does
not attract mosquitoes.
The randomization used made each trial a simple random sample. Each
trial performed did not affect another trials results in any way, so the trials were
independent. The normality of the data is shown above in Figure 29 (in the
previous t test) and below in Figure 31. The data used is real sample data, which
makes it so that the population mean and standard deviations are unknown.
Since the assumptions above were basically met, a two-sample t test was able to
be performed. Note that the data was not perfectly normally distributed, but it was
close enough for the test to still be performed.


Lidwell Webber 52


Figure 31. Normal Probability Plot for Control Data
Figure 31 above shows the normal probability plot for the control data. The
data is relatively normal, although the points are separated into three vertically
aligned groupings. This is just due to the lack of variability in the control data, and
even still, the values do not deviate far from the line, making the data usable in
the test.
Hypotheses:
H
o
:
three
=
control

H
a
:
three

control


The null hypothesis states that the average mosquito attraction of the
three bacteria mixture plates is equal to the average mosquito attraction of the
control plates. The alternate hypothesis states that the average mosquito
attraction of the three bacteria mixture plates and the average mosquito
attraction of the control plates are not equal.
Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 53


Figure 32. Distribution from Combination of All Three Bacteria and Controls Test
Figure 32 above shows the distribution of the t value from the two-sample
t test. There is obvious shading on both ends of the distribution, revealing that
the p-value is rather high.
Two-Sample t Test Results
t value: -1.5090
p-value: 0.1900

Fail to reject H
0,
because the p-value 0.19 is above the alpha level 0.05.
There is no significant evidence that there is a difference in mosquito attraction
between the plates with a combination of Bacillus subtilis, Micrococcus luteus,
and Staphylococcus epidermidis and the controls. There is about a 19% chance
of getting this difference between the combination of three bacteria and the
controls by chance alone if no difference is assumed.

Mosquito Attraction (# of Mosquitoes Touches)
Lidwell Webber 54

To recapitulate the statistical tests, the ANOVA test yielded a p-value of
0.000347 providing significant evidence that there is a difference in mosquito
attraction between the different bacteria combinations. The two-sample t test of
the single and double bacteria plates yielded a p-value of 0.0009, providing
significant evidence that the mosquito attraction of the single bacteria plates is
greater than the attraction of the double bacteria plates. The two sample t test of
the double and triple bacteria plates yielded a p-value of 0.0002, providing
significant evidence that average mosquito attraction of the plates with a
combination of two bacteria is greater than the average mosquito attraction of
plates with a combination of three bacteria. The two sample t test of the controls
and the bacteria plates with a combination of two bacteria yielded a p-value of
0.1900, providing no significant evidence that there is a difference in mosquito
attraction between the plates with a combination of Bacillus subtilis, Micrococcus
luteus, and Staphylococcus epidermidis and the controls.






Lidwell Webber 55

Conclusion
The purpose of this experiment was to determine if the attraction of
mosquitoes to bacteria combinations will be less than the attraction of
mosquitoes to individual bacterium. A total of 14 trials were conducted. The trials
took place in an entomology lab at Michigan State University, where mosquitoes
were provided. During each trial, two Petri dishes with a control of plain agar, and
two Petri dishes with the selected bacteria sample were exposed to the
mosquitoes. Giving the mosquitoes a choice among the plates eliminated the
uncontrollable factor of random mosquito behavior and flight patterns. A heating
pad, that was about 30 C, was taped on an outside side of the cage containing
mosquitoes. Using double sided tape, the Petri dishes containing the various
bacteria samples were secured to the inside side of the cage so that they were
exposed to the mosquitoes and being directly heated. The bacteria samples were
heated to help simulate an actual host and therefore attract more mosquitoes.
Each trial lasted ten minutes, and the attraction of the mosquitoes was measured
by how many landed on the bacteria sample in that time span.
It was hypothesized that cultures of an individual bacterium Bacillus
subtilus, Micrococcus luteus, and Staphylococcus epidermidis would attract more
mosquitoes than cultures of mixtures of two of these bacteria, and a mixture of all
three. At the end of experimentation this hypothesis was accepted.
The results from the trials were analyzed using multiple statistical tests.
First, an ANOVA test was used to compare all seven of the different bacteria
samples. The results were deemed significant on a 0.05 alpha level with a
Lidwell Webber 56

p-value of 0.000347. This test stated that there is a significant difference between
the seven different bacteria combinations that were tested. A significant
difference leads to the conclusion that the combinations of bacteria do indeed
affect the attraction of mosquitoes. The assumptions, which are the conditions
statistical tests need to be to yield reliable results, were not fully met for this
ANOVA test, so the results did not lead to a concrete conclusion. However, the
assumptions were met for the other statistical tests conducted, which were two-
sample t tests, making the results of those tests reliable and allowing a
dependable conclusion to be made from them. Since the results of the ANOVA
test are supported by the reliable results of the two-sample t tests conducted, it is
likely that the results of the ANOVA test can the trusted.
To further understand how the bacteria combinations affect the mosquito
attraction, three two-sample t tests were conducted. These compared each
grouping of bacteria samples, the individuals, combinations of two, and
combinations of three, and the combination of three bacteria to the control. The
two-sample t test between the single bacteria plates and the combination of two
bacteria plates yielded a p-value of 0.0009. This means that there is significant
evidence that plates of a singular bacterium are more attractive to mosquitoes
than plates of a combination of two bacteria. This supports the original
hypothesis that singular bacteria plates would be the most attractive. The more
concentrated a singular bacterium is, the greater its mosquito attraction.
The second two-sample t test of the combinations of two and three
bacteria yielded a p-value of 0.0002. This test supports the idea that the greater
Lidwell Webber 57

composition of bacteria, the less attractive the bacteria sample is. The average
mosquito attraction for the plates with a combination of two bacteria is
significantly greater than the average attraction for plates with a combination of
three bacteria. The final two-sample t test between the plates with three bacteria
and the controls yielded a p-value of 0.19. Since the difference between the
bacteria and controls was not significant, it suggests that there was not much of a
difference between the three bacteria combinations and controls. These results
show that a combination of bacteria effectively works to repel mosquitoes,
because it yields the same attraction as no bacteria at all. Because the results of
these statistical tests supported the claims made in the hypothesis, the
hypothesis was accepted.
The original hypothesis of this experiment came from another experiment,
conducted at Wageningen University. In the experiment, glass beads were
rubbed on human volunteers feet, and the bacteria was then cultured and
exposed to mosquitoes. Certain samples attracted a significant amount more of
mosquitoes than others. When the bacteria was analyzed it was found that
samples high in a singular bacterium, such as Staphylococcus epidermidis,
attracted more mosquitoes, and samples that had more bacterial diversity were
less attractive (Composition of Human Skin Microbiota). Five types of
bacteria that were found in the glass bead samples were then tested individually
for their attraction to mosquitoes, in another experiment conducted at
Wageningen University. This experiment found that bacteria such as
Lidwell Webber 58

Staphylococcus epidermidis and Bacillus subtilis do attract mosquitoes
(Differential Attraction of Malaria Mosquitoes).
The experiment conducted in this research paper was motivated by these
two experiments, and aimed to expand on their data and prove that a greater
bacterial diversity can help repel mosquitoes. For the average person, this
means that having a more diverse mixture of skin bacteria, rather than a
prominent amount of one kind, will prevent getting mosquito bites.
Research in this field is somewhat new, and it is unknown why exactly
mosquitoes are attracted to certain bacterial odors. Only female mosquitoes bite
humans, and it is not because they are hungry. The nutrients in blood allow
female mosquitoes to lay their eggs. Certain odors given off by humans, along
with other factors that affect mosquito attraction, couldt be a female mosquitoes
way of determining what human will give her the most nutrients for her eggs
(Nierenberg).
The odors of a singular bacterium may be the best attractants because the
mosquitoes only smell that one odor. By mixing the bacteria together, the smell
of each individual bacterium is masked by the odors of the other bacteria, making
the sample less attractive. This distraction would work in a way similar to the
current chemical active in mosquito repellents, DEET. This chemical confuses
mosquitoes olfactory sensors so they no longer pick up attractive scents, and
are then less likely to bite a host (How Does DEET Work?). Combinations of
bacteria may cause more excitement of the neurons in the mosquitoes olfactory
bulb, which would lead to less response of the individual bacteria that attract the
Lidwell Webber 59

mosquitoes. Mosquitoes could be repelled by over-exciting their senses with a
mixture of bacterial odors (Su). This method would repel mosquitoes more
effectively than DEET, because it would stop the attraction to bacterial odors
completely, rather than just momentarily distracting the mosquitoes from them.
Another reason why the bacteria tested in this experiment might have
attracted mosquitoes is because of the amount of carbon dioxide it gave off.
Carbon dioxide is a main mosquito attractant (Nierenberg), and if different
bacteria samples produced more carbon dioxide than others, this could have
affected the experiments data.
Though the hypothesis was accepted, some design flaws could have also
affected the experiments results. The main flaw was how the bacteria was
exposed to the mosquitoes. In other, previous experiments, the bacteria being
tested was placed in a wind tunnel including multiple trapping devices. The wind
tunnel fanned the bacterial odors towards the mosquitoes while reducing other
outside odors. An apparatus such as this was not available in testing. In other
experiments trials were often conducted for longer periods of time too. Due to
time constraints, trials could only last ten minutes and a trapping tunnel could not
be built. Also, it was hard to make the bacteria seem like a real nutrient source.
Placing the bacteria on a heating pad helped, but it still was not enough to
simulate a human. These design flaws led to the mosquitoes being interested in
the samples when they flew close by, but not being too interested from far away.
The set up was also flawed because the heating pad covered more space
than just under the heating pad. It was observed that some mosquitoes just
Lidwell Webber 60

stayed on the heating pad instead of the bacteria samples. The attraction to the
general heat of the area may have been a lurking variable that affected data.
Along with this, the trials themselves also could have been conducted
farther apart. Though the mosquitoes were given time to rest between trials, they
still seemed less interested near the end of experimentation. If more time was
available, the trials could have been conducted farther apart or at the same time
every day to avoid the varying activity amounts of the mosquitoes.
Finally, some errors occurred involving the bacteria. The Staphylococcus
epidermidis and Bacillus subtilis were used in the experiment conducted at
Wageningen University, and they were both found to be mosquito attractants.
The other two bacteria that were found to be attractants, Brevibacterium
epidermidis and Corynebacterium minitissimum, could not be used due to high-
school lab safety constraints. Because of this, Micrococcus luteus, was selected
instead. Though this ended up attracting mosquitoes, it wasnt included in any
past experiments. Along with this, small bacteria colonies were found on the
control plates after two days of incubation. This shows that the growth on the
bacteria samples may have included other unknown bacteria, which may have
affected the mosquitoes attraction. Data may have been more conclusive if
these design flaws had not occurred.
Further testing could be done to correct the design flaws in this
experiment. If given more time and resources, another, similar experiment could
be set up that uses more advanced methods found in past experiments, which
could produce more conclusive data. The experiment proved that a mixture of
Lidwell Webber 61

bacteria does in fact help to repel mosquitoes. Using this fact, a mosquito
repellent could be designed. Current mosquito repellents work by masking
natural human odors that attract mosquitoes, with the chemical compound DEET
(Stanczyk). The problem is that too soon the smell of the repellent fades, leaving
humans exposed. By finding a way to balance a persons skin bacteria, one of
the major roots of the problem would be solved. This potential repellent would not
just fade away like current repellents on the market.
The results of this experiment are another step forward in preventing
pesky mosquito bites, and more importantly, stopping the spread diseases that
mosquitoes carry. In many places around the world, malaria is still a large cause
of death. Recall that 660,000 people die a year from malaria, and most of these
are children. It is vitally important to stop this horrible disease, and warding off
the mosquitoes that spread it is a large part of this. The results of this experiment
can be used to find a mosquito repellent that will save lives.







Lidwell Webber 62

Appendix A - Calculations
ANOVA Test:
An ANOVA Test is used to compare the means or three or more
populations to test how far apart sample means are with how much variation
there is within the samples. This is done by finding the F statistic. The F statistic
is the mean square group, which is the variation among sample means between
each population, divided by the mean square error, which is the variation among
individuals in all the samples within each population. In Figure 33 on the next
page, the mean square group is found, where MSG is the mean square group,
is the number of observations in each sample times the mean of each sample, N
is the total number of trials in all sample populations, I is the number of
populations, n
n
is the sample size for each of the 7 populations, and the

is the
sample mean for each of the seven populations.







Lidwell Webber 63

MSG =

)

(

)

(

)

(

)

(

)

(

)

(

)

(

)

(

)

(

)

(

)

(

n
1
= 4
n
2
= 4
n
3
= 4
n
4
= 4
n
5
= 4
n
6
= 4
n
7
= 4
I = 7
N =28

= 5

= 3.75

= 4.25

= 2.75

= 3

= 2.25

= 0.5


() () () () () () ()

= 3.07143

MSG =
()()()()()()()

MSG= 8.6428
Figure 33. Calculation of the Mean Square Group Value




Lidwell Webber 64

In Figure 34 below, the mean square error is found, where MSE is the
mean square error, s
n
is the sample standard deviation for each of the 7
populations, and N, I, n
n
are the same as when the mean square group was
found.
MSE =
(

s
1
= 1.826
s
2
= 0.957
s
3
= 0.500
s
4
= 1.500
s
5
= 1.155
s
6
= 0.500
s
7
= 0.577

MSE =
(

MSE=
()

()

()

()

()

()

()

MSE = 1.2381
Figure 34. Calculation of the Mean Square Error Value




Lidwell Webber 65

After calculating both the mean square group and mean square error, the
F statistic can be calculated, as shown below in Figure 35.
F =

F =

F = 6.98077
Figure 35. Calculation of the F Statistic
Two-Sample t Tests
A two-sample t test is used to determine whether the average difference
between groups is significant on an alpha level. In this experiment, the alpha
level was 0.05. The t value is found by subtracting the average of group 2,
2
x , by
the average of group 1,
1
x , then dividing by the square root of the standard
deviation of group 1 squared divided by the number of trials in group 1 plus the
standard deviation of group 2 squared divided by the number of trials in group 2.
t=
( ) ( )
2
2
2
1
2
1
2 1
n
s
n
s
x x
+

Lidwell Webber 66

Shown in below in Figure 36 is the two-sample t test for the singular
bacteria plates and double bacteria plates.
t=
( ) ( )
2
2
2
1
2
1
2 1
n
s
n
s
x x
+

t=

t = 3.5355
Figure 36. Singular Bacteria Plates and Double Bacteria Plates two-sample t test
Calculation
Shown below in Figure 37 is the two-sample t test for the double bacteria
plates and triple bacteria plates.
t=
( ) ( )
2
2
2
1
2
1
2 1
n
s
n
s
x x
+

t=

t= 5.1169
Figure 37. Double Bacteria Plates and Triple Bacteria Plates Two-Sample t Test
Calculation




Lidwell Webber 67

Shown below in Figure 38 is the Two-Sample t Test for combination of all
three bacteria and controls.
t=
( ) ( )
2
2
2
1
2
1
2 1
n
s
n
s
x x
+

t=

t= -1.5090
Figure 38. Combination of Three Bacteria and Controls Two-Sample t Test
Calculation















Lidwell Webber 68


Appendix B

Senior Research Professional Consultant Contact Form

Names: Kristen Lidwell, Jessa Webber

Research Topic: The Attraction of Micrococcus luteus, Bacillus subitilis,
Staphylococcus epidermidis, combinations of two of these, and a combination of
all three, to Anolophes gambiae.

Professional Contact Information
Name: Michael Kaufman

Title: Associate Professor

Organization: Michigan State Universitys Department of Entomology

Phone (area code and extension): (517) 881-0556

Email: kaufma15@msu.edu

Mailing Address:

Natural Science Bldg., room 243
288 Farm Lane
Michigan State University
East Lansing, MI 48824


Dialogue Information

1. Contact Goal:

Our goal when first emailing our contact was to get information on raising
mosquitoes and some advice about our experiments set up. Dr. Kaufman helped
us design a practical experiment, and actually brought us to his lab at Michigan
State University to conduct trials there with his mosquitoes, which was very
helpful to us.
Lidwell Webber 69

2. Questions we asked:

A. Could my partner and I come visit your lab sometime in the near future to get
a better understanding about current research in the mosquito field?

B. How do you contain your mosquitoes, and trap them when necessary?

C. What can we do to make the Petri dishes of bacteria simulate real humans?






















Lidwell Webber 70

Works Cited
"2-Methyl-2-hepten-6-ol." ChemSpider. Royal Society of Chemistry, n.d. Web. 4
Dec. 2013. <http://www.chemspider.com/Chemical-Structure.19533.html>.
"6-Methyl-5-hepten-2-one." Sigma-Aldrich. Sigma-Aldrich Co., n.d. Web. 4 Dec.
2013.<http://www.sigmaaldrich.com/catalog/product/aldrich/w270733?lang
=enion=US>.
"Biological Notes on Mosquitoes." Mosquitoes.org. Alameda County Mosquito
Abatement District, n.d. Web. 22 Sept. 2013.
<http://www.mosquitoes.org/LifeCycle.html>.
Dowdey, Sarah. "How Smell Works." How Stuff Works. HowStuffWorks, Inc, n.d.
Web. 5 Dec. 2013. <http://science.howstuffworks.com/life/human-
biology/smell3.htm>.
"How Does DEET Work?." Phys.org. Phys.org, 21 Sept. 2011. Web. 4 Dec.
2013. <http://phys.org/news/2011-09-deet-insects.html>.
"Information on Carbon Dioxide Amounts and Water Quality." Water/Wastewater
Distance Learning Website. Mountain Empire Community College, n.d.
Web. 4 Dec. 2013.
<http://water.me.vccs.edu/exam_prep/carbondioxide.html>.
Jarrell, Allison. "Smelling the Way to Better Mosquito Management." Inside
Science. American Institute of Physics, 20 July 2012. Web. 5 Dec. 2013.
<http://www.insidescience.org/content/smelling-way-better-mosquito-
management/744>.
Kaufman, Michael. Professional." E-mail interview. 9 Sept. 2013.

Lidwell Webber 71

Kay, Leslie M., Catherine A. Lowry, and Heather A. Jacobs. "Receptor
Contributions to Configural and Elemental Odor Mixture Perception."
Behavioral Neuroscience 117.5 (2003): 1108-114. American
Psychological Association Inc. PDF file. 3 Dec. 2013.
"Mosquito FAQ's." Mosquito World. Mosquito World, n.d. Web. 4 Dec. 2013.
<http://www.mosquitoworld.net/mosquitofaqs.php>.
Nierenberg, Cari. "Why Some People Are Mosquito Magnets." NBC News
Health.NBC News, 29 May 2011. Web. 22 Sept. 2013.
<http://www.nbcnews.com/health/why-some-people-are-mosquito-
magnets-1C6437380>.
Smallegange, R., and Niels O. Verhulst. "Sweaty Skin: An Invitation to Bite?"
Trendsin Parasitology (2011): 143-48. Wageningen UR. Wageningen
University. Web. 20 Sept. 2013.
<http://www.wageningenur.nl/en/Publication-
details.htm?publicationId=publication-way-343034353433>.
Stanczyk, Nina M., John Brookfield, Linda M. Field, and James G. Logan. "Aedes
Aegypti Mosquitoes Exhibit Decreased Repellency by DEET following
Previous Exposure." (n.d.): n. pag. PLoS ONE. Creative Commons
Attribution License, 20 Feb. 2013. Web. 22 Sept. 2013.
<http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.
0054438>.


Lidwell Webber 72

"Stationary Phase." Encyclopdia Britannica. Encyclopdia Britannica, Inc., n.d.
Web. 4 Dec. 2013.
<http://www.britannica.com/EBchecked/topic/564117/stationary-phase>.
"Study finds unexpected bacterial diversity on human skin." Dermatology Nursing
July-Aug. 2010: 41+. Academic OneFile. Web. 22 Sept. 2013.
University of Idaho. Bacterial Pathogens and Normal Flora of Human. 5
Aug. 2011. Laboratory Experiment. University of Idaho Department of
Biological Sciences.
Su, Chih-Ying. "Questions About Mosquitoes Olfactory Receptors." E-mail
interview. 25 Oct. 2013.
Verhulst, Niels O., Y. Qiu, and H. Beijleveld. "Composition of Human Skin
Microbiota Affects Attractiveness to Malaria Mosquitoes." PLOS ONE
(2011): 7. Wageningen UR. Wageningen University. Web. 20 Sept. 2013.
<http://www.wageningenur.nl/en/Publication-
details.htm?publicationId=publication-way-343138353339>.
Verhulst, Niels O., Rob Andriessen, and Gabriella Bukovinszkine-Kiss.
"Differential Attraction of Malaria Mosquitoes to Volatile Blends Produced
by Human Skin Bacteria." (2010): n. pag. PLOS ONE. Web. 20 Sept.
2013.<http://www.plosone.org/article/info:doi/10.1371/journal
.pone.0015829>.
Verhulst, Niels O. The Role of Skin Microbiota in the Attractiveness of Humans to
the Malaria Mosquito Anopheles Gambiae Giles. Thesis. Wigeningen
University, 2010. N.p.: n.p., n.d. Print.
Lidwell Webber 73

Wang, Shirley. "Finding Smells That Repel." The Wall Street Journal. Dow Jones
& Company, 1 Sept. 2009. Web. 22 Sept.
2013.<http://online.wsj.com/article/SB10001424052970204660604574378
933761528214.html>.