Accepted Manuscript

Title: An Assessment of Detection Canine Alerts using

Flowers that Release Methyl Benzoate, the Cocaine Odorant,
and an Evaluation of their Behavior in Terms of the VOCs
Produced

Author: Michelle M. Cerreta Kenneth G. Furton

PII: S0379-0738(15)00127-9
DOI: http://dx.doi.org/doi:10.1016/j.forsciint.2015.03.021
Reference: FSI 7946

To appear in: FSI

Received date: 12-12-2014

Revised date: 19-3-2015
Accepted date: 23-3-2015

Please cite this article as: M.M. Cerreta, K.G. Furton, An Assessment of Detection
Canine Alerts using Flowers that Release Methyl Benzoate, the Cocaine Odorant, and
an Evaluation of their Behavior in Terms of the VOCs Produced, Forensic Science
International (2015), http://dx.doi.org/10.1016/j.forsciint.2015.03.021

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
An Assessment of Detection Canine Alerts using Flowers that Release Methyl Benzoate, the
Cocaine Odorant, and an Evaluation of their Behavior in Terms of the VOCs Produced

Michelle M. Cerreta and Kenneth G. Furton*

t
ip
cr
International Forensic Research Institute, Department of Chemistry and Biochemistry

Florida International University, Miami, FL, USA

us
an
M
d
p te
ce
Ac

*Corresponding Author
Florida International University
College of Arts and Sciences
11200 SW 8th Street, CP 344
Miami FL 33199
furtonk@fiu.edu
Tel: +1 305 348 2041

Page 1 of 23
Highlights

 Results of this paper validate the reliability of canine evidence in court.

 The odor profiles of snapdragon flowers were found to be statistically different than
cocaine.
 Methyl benzoate is not necessarily the dominant odor produced by snapdragon flowers


t
Narcotic detection canines never falsely alerted to snapdragon flowers

ip
cr
us
an
M
d
p te
ce
Ac

Page 2 of 23
Abstract

In recent years, the high frequency of illicit substance abuse reported in the United States has
made the development of efficient and rapid detection methods important. Biological detectors,
such as canines (Canis familiaris), are valuable tools for rapid, on-site identification of illicit

t
substances. However, research indicates that in many cases canines do not alert to the

ip
contraband, but rather to the volatile organic compounds (VOCs) that are released from the
contraband, referred to as the “active odor”. In 2013, canine accuracy and reliability were

cr
challenged in the Supreme Court case, State of Florida v. Jardines. In this case, it was stated
that if a canine alerts to the active odor, and not the contraband, the canine’s accuracy and

us
selectivity could be questioned, since many of these compounds have been found in common
household products. Specifically, methyl benzoate, the active odor of cocaine, has been found to

an
be the most abundant compound produced by snapdragon flowers. Therefore, the purpose of this
study is to evaluate the odor profiles of various species of snapdragon flowers to assess how
M
significantly methyl benzoate contributes to the total VOC profile or fragrance that is produced.
Particularly, this study examines the VOCs released from newly grown snapdragon flowers and
determines its potential at eliciting a false alert from specially trained detection canines. The
d

ability of detection canines to differentiate between cocaine and snapdragon flowers was
te

determined in order to validate the field accuracy and discrimination power of these detectors.
An optimized method using headspace solid-phase microextraction coupled with gas
p

chromatography-mass spectrometry (HS-SPME/GC-MS) was used to test the different types and
ce

abundances of compounds generated from snapdragon flowers at various stages throughout the
plants’ life cycle. The results indicate that although methyl benzoate is present in the odor
profile of snapdragon flowers, other compounds are present that contribute significantly, if not
Ac

more, than that of methyl benzoate. Canine teams, from various police departments throughout
South Florida, certified for narcotics detection, took part in this study. Two canine trials
involving 21 canines teams were performed by exposing the teams to 4 different species of
snapdragon flowers. Of the 21 canine teams tested, none alerted to the snapdragon flowers
presented, while all (100%) alerted to real cocaine samples, the positive control. Notably, the
results revealed that although methyl benzoate is produced by snapdragon flowers, certified
narcotics detection canines can distinguish cocaine’s odor profile from that of snapdragon
flowers.

Page 3 of 23
Key Words: Detection Canines, Canine Accuracy, Methyl Benzoate, Cocaine, Snapdragon, State
of Florida v. Jardines

t
ip
1. Introduction

According to the 2012 National Survey on Drug Use and Health (NSDUH), over 23 million

cr
Americans, were said to be current users of illicit substances, representing 9.2% of the
population. It was also reported that 1.6 million people were current users of cocaine, one of the

us
most commonly abused illicit drugs in the United States [1]. The high frequency of illicit
substance abuse in the United States has made the development of efficient and rapid detection

an
methods exceedingly important. While many forms of instrumentation and technology have
been utilized for on-site detection and screening, biological detectors, more specifically, canines
M
(Canis familiaris), represent one of the most reliable, versatile, rapid, and cost effective real-time
detection tools for contraband [2;3].
d

Ongoing research attempts to understand detector dogs by studying which odors dissipate from
specific targets, which odors induce a response from a trained canine, and the threshold levels
te

associated with said target odor [4;5]. Additionally, a canine’s ability to discriminate between a
p

single odor verses a pool of combined odors containing traces of the single target odor, is a
complex issue that is continuously brought up when a canine’s field validity is debated [6].
ce

Moreover, previous studies have shown that it is not the actual illicit material that induces an
alert by detection canines, but specific components that comprise the substance’s signature odor
Ac

profile [4]. These odor signatures are volatile organic compounds (VOCs) that have been
identified in the headspace of the target substance in question, for example, methyl benzoate has
been identified as an active signature odor of cocaine [2;4]. In other words, the studies indicate
that it is the odor of methyl benzoate, not cocaine, that elicits an alert from the canine [4].
Studies have also shown that a detection canine’s absolute threshold, or limit of detection, for
methyl benzoate was calculated to be 10 parts per billion (ppb), revealing the canine’s high level
of sensitivity [7-9]. While these findings have assisted in improving canine training and
efficiency, as well as assisted in validating a canine’s discrimination power in court, the results

Page 4 of 23
have also raised some questions with regards to canine detection accuracy. Recently, the
reliability of narcotic detection canines was challenged within the forensic community and the
legal system [10].

In 2013, the Supreme Court case, State of Florida v. Joelis Jardines scrutinized the use of

t
detection canines [10]. This case debated whether a warrantless “sniff” from a detection canine,

ip
performed at the doorstep of a private residence, violated the Fourth Amendment of the United
States Constitution, which prohibits unlawful search and seizure. In this argument, the

cr
Respondent held the belief that the use of the canine “sniff” without a warrant, is, in fact,
unconstitutional. The counsel for the respondent used scientific evidence to back up their claim,

us
and in doing so, discredited the use of detection canines under these circumstances. It was stated
that since canines are alerting to the active odor of the substance, and not the contraband itself,

an
the selectivity and accuracy of the canine could be called into question, since many of these
odors are present in common household items [10]. The case specifically highlighted methyl
M
benzoate, which was recently found to be the most abundant compound produced by popular
landscaping flowers, such as snapdragons (Antirrhinums) [10;11]. Thus, the question arose as to
whether snapdragon flowers, contained in flower beds or bouquets, would cause canines to alert
d

and falsely indicate that cocaine is present when it is, in fact, just the flowers producing the
te

methyl benzoate.
p

Snapdragon flowers (Figure 1), belonging to the genus Antirrhinum

and family Scrophulariaceae, are cool weather plants, ranging from
ce

20-100 cm in height. These flowers, owing their name to their

petals resembling that of a dragon’s mouth, are native to southern
Ac

Europe, but the popularity of the flower, has caused it to spread

throughout the United States [12].

Most flowers, including the snapdragon flower, owe their

reproductive success to their emitted floral scent, capable of Figure 1: Snapdragon Flower
(Antirrhinum)
attracting specific pollinators [11;13-15]. The VOCs that constitute
the flower’s fragrance vary from species to species with respect to the number, type, and relative
abundances of compounds. The composition of the flowers’ odor profile and total odor
production changes throughout the plants’ life cycle and emit their maximum levels when the

Page 5 of 23
flower is ready for pollination. Newly opened flowers have under developed anthers (pollen
producing structures) and therefore produce less odor/scent, as they are not ready for pollination.
After pollination, the amount of scent released by the plant decreases, and continues to decrease
until the end of the flower’s life cycle [11;14;16]. Dudareva et al. and Negre et al. conducted
research monitoring the production of methyl benzoate, specifically, from snapdragon flowers.

t
ip
Studies showed that the Maryland True Pink cultivar emitted the most methyl benzoate, making
up 60% of the total volatiles. It was also found that unopened buds released no methyl benzoate,

cr
but began to emit, upon opening, at very low levels. The production of methyl benzoate reached
its peak 5-7 days after anthesis, or after the flower had opened, and then declined thereafter

us
[11;14]. Additionally, methyl benzoate production was found to be highest during the daylight
hours of 9 am and 4 pm, due to the high activity of pollinators during those hours [11;14].

an
While the odors emitted by snapdragon flowers, particularly, the release of methyl benzoate,
have been extensively studied, little is known about whether the flower’s fragrance will elicit a
M
false alert from specially trained narcotics detection canines, as surmised during the Supreme
Court case, State of Florida v. Joelis Jardines (2013). Therefore, the purpose of this study is to
evaluate the odor profiles of various species of snapdragon flowers to assess the significance of
d

methyl benzoate to the overall fragrance produced. In addition, specially trained narcotics
te

detection canines were used to determine if the flower’s fragrance would result in a false alert.
The ability of canines to differentiate between cocaine and snapdragon flowers, that have similar
p

odor pools, was determined in order to validate the field accuracy and discrimination power of
ce

detector canines.

2. Materials and Methods

Ac

2.1. Snapdragon Planting and Growth

Four different snapdragon flowers (Table 1), varying in species, size, and color were selected for
this study. During a period of cooler climates (January), approximately 3-5 seeds of Maryland
True Pink (PanAmerican Seed, West Chicago, IL), Blank Prince, Twinny Peach, and Rembrandt
(Thompson & Morgan, Ipswich, England) snapdragons were planted. Since snapdragon flowers
require sunlight for germination, seeds were placed on top of fertilized soil (18-6-8 180-d
release; Florikan, Sarasota, FL) in each cell of a 72-cell planting tray; eighteen cells were

Page 6 of 23
designated for each type of snapdragon. The flowers were maintained in a greenhouse located at
Florida International University (FIU, Miami, FL).

Generally, germination began after 1 week and after 3-4 weeks, 2 inch sprouts were present.
One month after planting, the snapdragon sprouts were transplanted into round, plastic flower

t
pots (6.5 in diameter, 5 in. deep; Kord, Toronto, CAN) using only 1 sprout per pot, to allow

ip
ample room for flower growth, as well as to minimize competition for nutrients and sunlight. All
planting supplies, including soil, planting trays, fertilizer, and flower pots, were provided by the

cr
Florida International University Greenhouse and Department of Biological Sciences (FIU,
Miami, FL).

us
Throughout the growth process, the flowers were watered daily and routinely pruned for optimal

an
growth and plant survival. Temperature and relative humidity were continuously monitored
using an environmental meter (Extech, Wilmington, NC), with values ranging from 21.9-28.8°C
and 37.5-78.1% respectively.
M
Table 1: Types of snapdragon flowers used in this study, including name, genus, species, size classification, and color
d
p te
ce

2.2. Sampling and Instrumental Analysis

Ac

2.2.1. Snapdragon Flower and Cocaine Sampling

Each day, the growing snapdragons were inspected for newly bloomed flowers. On the day of
anthesis, or the day the bud opened, which was approximately two months after planting, the
flowers were flagged using Fisherbrand Colored Label Tape (Fisher Scientific, Pittsburg, PA).
Since the study conducted by Dudareva et al. found that odor emission peaked 5-7 days after
anthesis, individual flowers (weighing ~ 0.5 g each) were cut and immediately placed into a 40-
mL clear, screw top vial with PTFE/Silicone septa (Supelco, Bellefonte, PA) 2, 6, and 10 days

Page 7 of 23
after anthesis in order to obtain a snapshot of the odors being produced during each time of the
flower’s life cycle [11].

Preliminary experiments were performed to assess the minimum amount of time needed for
ample VOCs to accumulate in the headspace above the flowers. The results revealed that after

t
ip
30 minutes, equilibrium was achieved for compounds to be released into the headspace while the
amount of time flowers remained capped and deprived of necessary nutrients was minimized

cr
(unpublished data). Immediately following this 30 minute accumulation period, VOCs were
extracted from the headspace for 30 minutes using a polydimethylsiloxane (PDMS), 100 μm

us
solid-phase microextraction (SPME) fiber (Supelco, Bellefonte, PA). The SPME fiber samples
were analyzed using gas chromatography-mass spectrometry (GC-MS). After the flowers were

an
sampled, the vials were opened, filled with water until the bottom stem of the flower was
completely submerged, and then left opened overnight. This was done to simulate the odor
released from flowers cut for floral arrangements/bouquets over an extended period of time. The
M
following day, the vials were again capped for 30 minutes, sampled, and analyzed using
SPME/GC-MS. This process continued until methyl benzoate, the target odor, was no longer
d

present in the odor profiles. For this experiment, triplicates were performed for each type of
snapdragon flower at all designated points of the flower’s life cycle (2, 6, and 10 days after
te

anthesis). To compare the odor profiles of the snapdragon flowers to that of cocaine,
p

approximately 1 gram of cocaine powder (obtained from a local canine trainer with a DEA
license) with a purity level greater than 90%, was transferred into each of three 40-mL clear,
ce

screw top vials with PTFE/Silicone septa. To mimic the sampling conditions of the snapdragon
flowers, the cocaine samples were allowed to sit for 30 minutes; thereafter, headspace-SPME
Ac

extraction was performed for 30 minutes and subsequently analyzed via GC-MS analysis.

2.2.2. Instrumentation Methodology

An optimized GC-MS method was developed using an Agilent 6890 gas chromatograph coupled
to an Agilent 5973 quadrupole mass spectrometer. The GC was fitted with a 0.25 mm x 30 m
HP-5-MS column with a film thickness of 0.25 μm. Helium was used as the carrier gas at a flow
rate of 1.2 mL/min. The injection port was maintained at a temperature of 280°C. The initial
temperature of the GC column was set to 40°C where it increased to 100°C at a rate of 7°C/min,

Page 8 of 23
followed by an increase to 110°C at a rate of 2°C/min, and a final increase to 260°C at a rate of
25°C/min, for a total runtime of 19.57 minutes. The transfer line to the mass spectrometer and
the ion source were maintained at 280°C and 230°C, respectively. The mass spectra were
continuously scanned from 40 to 440 amu. Compounds were identified using the NIST 98 Mass
Spectral Reference Library and confirmed and quantified using reference standards purchased

t
ip
from Sigma-Aldrich (St. Louis, MO).

cr
2.3. Canine Trials

us
2.3.1. General Set-up
Canine teams from various police departments throughout South Florida, certified for narcotics

an
detection, took part in this study. These detection canines were previously trained using narcotics
with a purity level greater than 90%, and have been proofed off of (trained not to alert) all
containment packages, including plastic and metal containers. Two different canine trials were
M
performed to assess the canine’s alert rate to odors emitted by snapdragon flowers; explained in
greater detail in Sections 2.3.2 and 2.3.3. Canine trials were completed in a warehouse,
d

approximately 7,500 square feet in size, where air conditioning and exhaust fans were turned off
te

in order for wind speed and air flow to not disrupt odor dispersion. For both Trial 1 and Trial 2,
15-20 g of cocaine, the positive control, was hidden in a separate room. Handlers and their
p

canines were first asked to search the room containing cocaine, in a single blind scenario. If the
ce

canine alerted to the cocaine, he was rewarded and only then was allowed into a separate
location where the snapdragon flowers and blanks were kept. This was done in order to ensure
that the canine was “working” and would give reliable results during the snapdragon line-up. In
Ac

total, 25 canines were evaluated using the cocaine positive control and of those, 21 canines
positively alerted to the cocaine, and therefore used to complete the experimental trials. Those
canines that did not alert to the positive control did not participate in the snapdragon line-up and
therefore were not included in the results. The size of the warehouse allowed for 10 different
testing locations to be used during the study, allowing for the order of the flowers and blank
samples in the line-up to be varied at each location. While a new test location would be ideal for
each canine, space and the need for consistent testing scenarios was a limitation. In each test
location, the four types of snapdragon flowers, along with one blank soil sample, were placed 3-4

Page 9 of 23
feet apart, as recommended by the canine trainer, in a line-up scenario. These canines have
trained using box line-ups in regular training sessions and therefore were considered a familiar
testing scenario for both canines and their handlers. During all searches involved in this study,
canines and handlers were not given a time limit to search and were permitted to conduct their
search in the same manner they would in the field. In addition, canines were not rewarded

t
ip
during the snapdragon line-up, as requested by the canine trainer, as trained canines are not
rewarded in the field because of uncertainty of the identity of the substances found. The canines’

cr
responses, as indicated by their handlers, were documented.

us
2.3.2. Canine Trial 1
Each type of potted snapdragon plant, consisting of multiple stalks with 15-20 bloomed flowers

an
per stalk, as well as a blank soil sample, were placed in plain sight in a line-up scenario. The
purpose of this trial was to determine if the trained detection canines would alert to growing
snapdragon flowers by presenting, at minimum, 1 μg of methyl benzoate, an amount well above
M
the cited detection limit for methyl benzoate. This study evaluated 14 canine teams resulting in a
total of 30 runs. The “runs” represent the amount of times any canine went through the
d

snapdragon line-up, meaning that there were some canines that were tested more than once. For
te

those canines tested more than once, each test was held on a different day, at a different test
location, within the same warehouse. These canines were required each day to first alert to the
p

positive control in the separate room, before being allowed to participate further in the
snapdragon line-up.
ce

2.3.3. Canine Trial 2

Ac

For this trial each snapdragon flower was hidden to create a single blind testing scenario. The
morning before the trials, twenty flowers from each snapdragon were placed into their own
respective 16-ounce mason jar (Ball, Publix, Miami, FL) to ensure that the canines were
presented with a minimum of 1 μg of methyl benzoate. Jars were sealed for approximately one
hour, which allowed for odor accumulation during transport to the testing site. At the test site,
the jars were opened and hidden under boxes in a line-up. Again, a blank soil sample was hidden
as well. Six canine teams were resulting in a total of 14 runs. For those canines tested more
than once, each test was held on a different day, at a different test location, within the same

Page 10 of 23
warehouse. These canines were required each day to first alert to the positive control in the
separate room, before being allowed to participate further in the snapdragon line-up.

2.4. Statistical Analysis

t
Each of the four snapdragon flower’s VOCs, as well as their abundances, were statistically

ip
compared against those of cocaine through Principle Components Analysis (PCA) in order to

cr
determine the similarity or difference between each of the odor profiles. PCA is a multivariate,
data reducing technique used to highlight the correlation or variation in a data set. To do this,

us
principle components, or linear combinations of the original data values, are obtained such that
the first principle component has the highest variation in the data set, while the second accounts

an
for the second highest. Therefore, when plotting the principle components, correlations and
variations are visualized, with correlations representing the lowest amounts of principle
components [17]. In addition to PCA, a box-and-whisker plot was used to highlight the percent
M
differences of methyl benzoate contained in both the snapdragon and cocaine samples. A box-
and-whisker plot represents the data through their upper and lower limits, as well as their median
d

values, thereby showing a spread of the collected data values [17]. To analyze the results of the
canine trials, the negative predictive value (NPV) was calculated. The NPV gives the likelihood
te

that any negative response is a “true negative,” or the likelihood that a canine’s non-response to a
p

snapdragon flower is because it is not cocaine [18]. Thus, when the canine fails to alert to a
snapdragon flower, the NPV value will be closer to 100%. The higher the NPV, the better the
ce

canine’s ability to differentiate between snapdragon and cocaine odor profiles, and therefore the
more likely that the non-responses are not by chance. The NPV can be calculated by the
Ac

following formula:

NPV =

3. Results and Discussion

3.1. Volatile Organic Compounds Released from Snapdragon Flowers

Page 11 of 23
The odors released from Black Prince, Maryland True Pink, Rembrandt, and Twinny Peach
snapdragon flowers were evaluated in triplicate at 2, 6, and 10 days post anthesis via HS-
SPME/GC-MS. Figure 2 represents the 9 compounds identified for each trial at the three
designated days of the flowers’ life cycle. Each pattern/color represents a different VOC
detected while the length of each bar segment represents the abundance extracted. The odor

t
ip
profiles varied with each type of snapdragon flower, with the Maryland True Pink producing the
highest abundance of compounds and Twinny Peach producing the least abundant. Additionally,

cr
the Maryland True Pink produced the highest abundance of methyl benzoate which was in
agreement with Dudareva et al., but it was not the most abundant compound produced by each

us
flower [11]. This observation is most apparent in Figure 3 which shows the distribution of
VOCs released from each snapdragon flower on Day 2 of anthesis. Day 2 was selected for this

an
comparison because it was the time at which the highest abundance of methyl benzoate was
produced, based on the results from Figure 2. Figure 3 revealed that, in most cases, other
compounds contributed more to the flowers’ odor profile than that of methyl benzoate. For
M
instance, the most abundant compound released from the Black Prince snapdragon was β-cis
ocimene (63%), while both β-Myrcene (11%) and 2,6-dimethyl-2,4,6-octatriene (14%)
d

contributed more than methyl benzoate (8%). The Rembrandt snapdragon’s odor profile showed
te

that acetophenone (42%) and β-cis ocimene (30%) were most dominant, with methyl benzoate
contributing 26%. For the Maryland True Pink, methyl benzoate and β-cis ocimene contributed
p

relatively similar amounts to the odor profile, at 41% and 45%, respectively. Conversely, methyl
ce

benzoate was the most dominant odor at 67% for Twinny Peach, but in general also the least
odiferous type of snapdragon flower, as seen in Figure 2.
Ac

Page 12 of 23
 t
ip
cr
us
an
M
d

Figure 2: The odor profiles of each snapdragon (flower cut 2, 6, 10 days after blooming)
p te
ce
Ac

Page 13 of 23
 t
ip
cr
us
an
M
d
te

Figure 3: The distribution of the volatile organic compounds released from each snapdragon flower 2 days after anthesis
p

3.2. Persistence of Methyl Benzoate from Cut Flowers

ce

To simulate a floral bouquet, as well as to assess the persistence of methyl benzoate, Black
Prince, Maryland True Pink, Rembrandt, and Twinny Peach snapdragon flowers were cut 2, 6,
Ac

and 10 Days after anthesis and analyzed daily; the results are presented in Figure 4, where Time0
represents the initial flower sampling, and each additional time point represents sampling 24
hours after the previous point. Initial sampling of each flower showed that the Maryland True
Pink produced the highest abundance of methyl benzoate (at Time0), while the Black Prince
produced the least. Although Figure 2 showed the Twinny Peach to be the least odiferous, it
produced more methyl benzoate than the Black Prince. Those flowers that were cut at Days 6
and 10 after anthesis ceased odor production at a faster rate than those that were cut at Day 2.
For instance, methyl benzoate was no longer detectable for the Black Prince when it was
sampled on Time1 for both Days 6 and 10 after anthesis; however, for those that were cut 2 Days

Page 14 of 23
after anthesis, a detectable amount of methyl benzoate was observed at Time2. These findings
could be attributed to the time of pollination, as those flowers that were cut later in the flowers’
life cycle had already pollinated, resulting in a decrease in the production of odor, while those
cut earlier may not have been pollinated yet and are still producing odors at higher abundances
[14]. Moreover, results showed that flowers collected at Day 2 after anthesis generally produced

t
ip
the highest abundance of methyl benzoate. A previous study found that snapdragon flowers
generated the most odor when they were close to or at full blossom; thus the flowers collected at

cr
Day 2 most likely fully opened close to or after collection [19]. This could also explain the
initial increase before the decline in methyl benzoate at Time2 for the Rembrandt cut at Day 2

us
after anthesis. Overall, the results indicate that, on average, the odor of methyl benzoate persists
for approximately 4 days after cutting, with newly bloomed flowers lasting as long as 7 days,

an
which was the case for the Rembrandt and Mary True Pink snapdragons. To shed some light on
whether the persistence of methyl benzoate of cut flowers could cause an alert from a narcotics
detection canine, two flower wholesale companies were contacted and questioned regarding the
M
time frame in which snapdragon flowers are received and incorporated into their floral
bouquets/arrangements. Both companies stated that when a floral bouquet/arrangement is
d

ordered, the recipient will receive the flowers, if they are in season, one day after they have been
cut. However, if the specific flowers are out of season or are unattainable by a local farm, they
te

would be imported from an outside source, extending the amount of time for the
p

bouquet/arrangement to reach the recipient to approximately four days. Based upon this
information, and the results gathered from this study, there is a high possibility that the amount
ce

of methyl benzoate, if any, released from the floral bouquet/arrangements containing

snapdragons would be minimal, especially when receiving flowers from an outside source.
Ac

Contrarily, those bouquets/arrangements that were cut in the early stages of their life cycle or
delivered within 24 hours of cutting will likely release methyl benzoate upon delivery.

Page 15 of 23
 t
ip
cr
us
an
Figure 4: Abundance of methyl benzoate over time after flowers have been cut at 2, 6, and 10 days after anthesis for (A)
M
Black Prince, (B) Maryland True Pink, (C) Rembrandt, (D) Twinny Peach
d

3.3. Canine Trials

Previous studies have found that methyl benzoate was the active odorant released from cocaine
te

and, along with Dudareva et al., this study found that methyl benzoate was one of the compounds
p

released from snapdragon flowers [4;11;14]. This raises the question: will narcotics detection
canines positively alert to snapdragon flowers? To answer this question, both potted and cut
ce

snapdragon flowers were presented to specially trained detection canines in order to assess their
accuracy. Fourteen canine teams were evaluated using potted snapdragon flowers that contained
Ac

a minimum of 15-20 flowers, or at least 1 μg of methyl benzoate. The amount of flowers needed
to reach 1 μg of methyl benzoate was estimated based on the abundances obtained from Figure 2.
In order to assess the alert rate to cut snapdragon flowers, six canine teams were evaluated,
single blind. Presented in Figure 5 are the canines’ alert rates to each of the four snapdragon
flowers, a blank (soil), and a positive control (15-20 g cocaine). Graph 5(A) is representative of
the results for the potted flowers, while Graph 5(B) is for the cut flowers. Significantly, the
detection canines did not alert to any of the four snapdragon flowers or the blank samples, while
all canines (100%) alerted to the positive control (cocaine), keeping in mind that only canines

Page 16 of 23
that alerted to the positive control were allowed to participate in the snapdragon line-up. The
negative predictive value (NPV), which represents the likelihood that the non-response exhibited
during the canine trials was not by chance. For both trials, the canines did not alert to any of the
snapdragon flowers, and thus, the NPV was determined to be 100%. Therefore, the fact that the
canines did not alert to the snapdragon flowers was not by chance, and the non-response was a

t
ip
correct response. These results are in agreement with similar studies completed in 2002 when
various perfume samples found to contain methyl benzoate were presented to narcotics detection

cr
canines to determine canine false-alert rates [4]. As with this study, canines did not alert to the
perfume samples, stating that canines can “readily distinguish these odors” from cocaine samples

us
[4].

an
M
d
te

Figure 5: Canine trials for the determination of snapdragon alert rate for (A) potted flowers and (B) cut flowers, where N represents the
p

number of runs completed for each trial

ce

3.4. Statistical Analysis

The results from Figure 5 show that the canines did not associate the odor profile emitted by the
Ac

snapdragon flowers with that of cocaine. An explanation for the results of the canine trial was
sought through a Box-and-Whisker Plot and Principal Components Analysis (PCA). Figure 6
shows a Box-and-Whisker plot, in which the percentage of methyl benzoate making up the odor
profile of cocaine is compared to that of the snapdragon flowers. This chart highlights the range
or spread of data points, as well as the median of each sample [17]. The results reveal that the
percentage of methyl benzoate within a cocaine sample does not overlap with any of the
snapdragon flowers that were evaluated. According to Schoon and Haak, differences in the
amount of odors may be perceived as qualitative differences by detection canines [6]. Notably,

Page 17 of 23
the results of this study showed that the percentage of methyl benzoate was exceedingly higher
in the snapdragon flowers, than in the cocaine samples. Therefore, this finding suggests that the
canines may perceive the snapdragon flowers as an unfamiliar odor, and therefore, not alert.

t
ip
cr
us
an
Figure 6: Box-and-Whisker Plot of the percentage of methyl benzoate within samples
M
Principal components analysis was performed using the type and abundances of VOCs detected
d

from the Black Prince, Maryland True Pink, Rembrandt, and Twinny Peach snapdragon flowers
te

represented in Figure 2, as well as for cocaine. Each set of triplicates from Figure 2 was
averaged and the data was normalized before conducting PCA analysis to account for error and
p

give equal weight to each variable. Principal components analysis (Figure 7) revealed three
ce

distinct clusters which are representative of the snapdragon flowers sampled at 2, 6, and 10 days
after anthesis, the Twinny Peach at Day 10, and the cocaine sample. The tight cluster formations
reveal the similarities within the data set, while any distance from clusters shows its
Ac

dissimilarity. Thus, the cluster of snapdragon flowers represents how statistically similar the
odor profiles are to one another, but statistically dissimilar from that of cocaine. Although the
odor profile of Twinny Peach at Day 10 was not similar to the other snapdragon flowers, it still
differed from that of cocaine. The statistical dissimilarity of Twinny Peach at Day 10 is also
represented in Figure 2, as the Twinny Peach flower was the least odiferous compared to the
other three, as the Twinny Peach flower at Day 10, particularly, produced the lowest abundance
of VOCs. This observation is possibly due to early pollination which would cause a change or
decrease in odor production to discourage pollinators or it could also be the result of an

Page 18 of 23
environmental stressor, such as climate or water changes, which could lead to accelerated
senescence, or flower death [20;21]. The significant difference between the odor profiles of the
snapdragon flowers and cocaine suggests that the pool of odor released from the flower aids in
the canines’ ability to differentiate between the snapdragon flowers and cocaine, even if the
active odor is present in both samples. Since the active odor of illicit materials depends on the

t
canines’ olfactory receptor response, and is not necessarily the most prominent odor present, it is

ip
possible that multiple odors produced by the snapdragons induce a canine’s olfactory receptor

cr
response, while for cocaine, it is only methyl benzoate. This would allow for canines to interpret
the odors produced by snapdragon flowers as entirely different than that of cocaine [5;22].

us
an
M
d
p te
ce

Figure 7: Principal Components Analysis (PCA) of snapdragon flowers and cocaine

Ac

4. Conclusion

The purpose of this study was to determine whether snapdragon flowers, which were reported to
produce methyl benzoate [11], the odorant of cocaine, could elicit an alert from specially
certified narcotics detection canines [10]. Overall, the study sought to assess a detection canine’s
ability to distinguish between the odor of cocaine and the odor of snapdragon flowers, and the
possibility of snapdragon flowers eliciting an alert by canines trained to alert to cocaine.

Page 19 of 23
To complete this research, the VOCs produced from four different snapdragon flowers, Black
Prince, Maryland True Pink, Rembrandt, and Twinny Peach, were evaluated using HS-
SPME/GC-MS and compared to those of cocaine. Results showed that methyl benzoate was
produced by each snapdragon flower, as was previously cited, but was not necessarily the most
significant contributor to the odor profile.

t
ip
Snapdragon flowers were presented to 21 canine teams to determine if it would result in a
positive alert. When presenting canines with both potted and cut flowers, canines alerted at a

cr
rate of 0% to both potted and cut flowers. Principle Components Analysis revealed that the odor
profiles of the snapdragon flowers were statistically different from that of cocaine, giving a

us
justification for canines not alerting to the snapdragon flowers, even though methyl benzoate is
one of the odors produced. This study demonstrates the reliability of narcotic detection canines

an
by revealing that they will not falsely alert to snapdragon flowers, even when there is a
component of the cocaine odor present, namely methyl benzoate.
M
d
p te
ce
Ac

Page 20 of 23
 Reference List

[1] US Departmen of Health and Human Services. Results from the 2012 National Survey on
Drug Use and Health: Summary of National Findings. 2013. Rockville, Substance Abuse
and Mental Health Services Administration.

t
ip
[2] N. Lorenzo, T. Wan, R.J. Harper, Y.L. Hsu, M. Chow, S. Rose, K.G. Furton, Laboratory
and field experiments used to identify Canis lupus var. familiaris active odor signature
chemicals from drugs, explosives, and humans. Analytical and bioanalytical chemistry 376

cr
(2003) 1212-1224.

[3] K. Furton, J. Greb, H. Holness, The Scientific Working Group on Dog and Orthogonal

us
Detector Guidelines (SWGDOG). National Criminal Justice Reference Service, US Dept of
Justice, Rockville, MD (2010) 155.

an
[4] K.G. Furton, Y. Hong, Y.L. Hsu, T. Luo, S. Rose, J. Walton, Identification of Odor
Signature Chemicals in Cocaine Using Solid-Phase Microextraction-Gas Chromatography
and Detector-Dog Response to Isolated Compounds Spiked on US Paper Currency. Journal
of chromatographic science 40 (2002) 147-155.
M
[5] K.G. Furton, L.J. Myers, The scientific foundation and efficacy of the use of canines as
chemical detectors for explosives. Talanta 54 (2001) 487-500.
d

[6] A. Schoon, R. Haak K9 Suspect Discrimination, Detselig Enterprises Ltd., Calgary, 2002.
te

[7] M. Marshall, J.C. Oxley Aspects of explosives detection, Elsevier, 2011.

[8] J.M. Johnston, Canine detection capabilities: Operational implications of recent R & D
p

findings. Institute for Biological Detection Systems, Auburn University (1999) 1-7.
ce

[9] J.X.J. Zhang, K. Hoshino Molecular Sensors and Nanodevices: Principles, Designs, and
Applications in Biomedical Engineering, Elsevier Inc, Waltham, 2014.

[10] L.A. Shoebotham. State of Florida v. Joelis Jardines: Brief of Amici Curiae Fourth
Ac

Amendment in Support of Respondent. 2012. Cuckle Law Brief Printing.

[11] N. Dudareva, L.M. Murfitt, C.J. Mann, N. Gorenstein, N. Kolosova, C.M. Kish, C.
Bonham, K. Wood, Developmental regulation of methyl benzoate biosynthesis and
emission in snapdragon flowers. The Plant Cell Online 12 (2000) 949-961.

[12] A.K. Singh Snapdragon. Flower Crops: Cultivation and Management, New India
Publishing Agency, Pitam Pura, 2006, pp. 337-346.

[13] L. Van der Pijl, Ecological aspects of flower evolution. I. Phyletic evolution. Evolution
(1960) 403-416.

Page 21 of 23
[14] F. Negre, C.M. Kish, J. Boatright, B. Underwood, K. Shibuya, C. Wagner, D.G. Clark, N.
Dudareva, Regulation of methylbenzoate emission after pollination in snapdragon and
petunia flowers. The Plant Cell Online 15 (2003) 2992-3006.

[15] J.T. Knudsen, L. Tollsten, L.G. Bergstrom, Floral scents - a checklist of volatile
compounds isolated by head-space techniques. Phytochemistry 33 (1993) 253-280.

t
[16] K.N. Jones, J.S. Reithel, R.E. Irwin, A trade-off between the frequency and duration of

ip
bumblebee visits to flowers. Oecologia 117 (1998) 161-168.

[17] J. Miller, J. Miller Statistics and chemometrics for analytical chemistry, Pearson Education

cr
Limited, Harlow, 2005.

[18] R. Parikh, A. Mathai, S. Parikh, G.C. Sekhar, R. Thomas, Understanding and using

us
sensitivity, specificity and predictive values. Indian journal of ophthalmology 56 (2008) 45.

[19] A.M.B. Gutierrez. Studies on fragrance, vase life and ethylene regulation of volatile

an
production in rose flowers. 2009. University of Florida.

[20] B.B. Casper, T.R. La Pine, Changes in corolla color and other floral characteristics in
Cryptantha humilis (Boraginaceae): cues to discourage pollinators? Evolution (1984) 128-
M
141.

[21] S.K. Tripathi, N. Tuteja, Integrated signaling in flower senescence: an overview. Plant
Signal Behav 2 (2007) 437-445.
d

[22] M.S. Macias, R.J. Harper, K.G. Furton, A Comparison of Real Versus Simulated
te

Contraband VOCs for Reliable Detector Dog Training Utilizing SPME-GC-MS. American
laboratory 40 (2008) 16.
p
ce
Ac

Page 22 of 23
Acknowledgements

The authors would like to thank Peter Nunez, director of the US K-9 Academy, for coordinating
the testing of the drug detection canines. Additionally, we would like to thank all of the handlers
and canines in South Florida who participated in the field testing. We would like to acknowledge

t
ip
Scott Zona, the greenhouse coordinator of the Department of Biological Sciences at Florida
International University, for assisting with planting and growing the snapdragon flowers, as well

cr
as for providing the necessary supplies for the plants’ survival. Finally, a special thanks to Dr.
Abuzar Kabir, Dr. Howard Holness, Dr. Katylynn Beltz, Alexandra Palacios, and Dr. Norma Iris

us
Caraballo for their guidance and advice throughout this study.

an
M
d
p te
ce
Ac

Page 23 of 23