ArticlePDF Available

Abstract and Figures

Invasive alien species are increasing due to globalization. Their spread has resulted in global economic losses. Asian [Anoplophora glabripennis (Motschulsky)] (ALB) and citrus [A. chinensis (Forster)] (CLB) longhorn beetles are two introduced wood borers which contribute to these economic losses e.g. the destruction of tree plantations. Early detection is key to reduce the ecological influence alongside the detrimental and expensive eradication. Dogs (Canis lupus familiaris) can detect these insects, potentially at an early stage. We trained two privately owned dogs to investigate their use as detection tools. We tested the dog’s ability to discriminate ALB and CLB from native wood borers by carrying out double-blind and randomized experiments in three search conditions; (1) laboratory, (2) semi-field and (3) standardized field. For condition one, a mean sensitivity of 80%, specificity of 95% and accuracy of 92% were achieved. For condition two and three, a mean sensitivity of 88% and 95%, specificity of 94% and 92% and accuracy of 92% and 93% were achieved. We conclude that dogs can detect all types of traces and remains of ALB and CLB and discriminate them from native wood borers and uninfested wood, but further tests on live insects should be initiated.
This content is subject to copyright. Terms and conditions apply.

Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports
Pest detection dogs for wood
boring longhorn beetles
Charlotte Holmstad Arnesen & Frank Rosell*
Invasive alien species are increasing due to globalization. Their spread has resulted in global economic
losses. Asian [Anoplophora glabripennis (Motschulsky)] (ALB) and citrus [A. chinensis (Forster)]
(CLB) longhorn beetles are two introduced wood borers which contribute to these economic losses
e.g. the destruction of tree plantations. Early detection is key to reduce the ecological inuence
alongside the detrimental and expensive eradication. Dogs (Canis lupus familiaris) can detect these
insects, potentially at an early stage. We trained two privately owned dogs to investigate their use as
detection tools. We tested the dog’s ability to discriminate ALB and CLB from native wood borers by
carrying out double-blind and randomized experiments in three search conditions; (1) laboratory, (2)
semi-eld and (3) standardized eld. For condition one, a mean sensitivity of 80%, specicity of 95%
and accuracy of 92% were achieved. For condition two and three, a mean sensitivity of 88% and 95%,
specicity of 94% and 92% and accuracy of 92% and 93% were achieved. We conclude that dogs can
detect all types of traces and remains of ALB and CLB and discriminate them from native wood borers
and uninfested wood, but further tests on live insects should be initiated.
Invasive alien species are species intentionally or unintentionally introduced by humans outside their natural
geographical range1. eir presence is likely to cause damage to the economy and environment as they may
threaten agriculture, forestry, and the original states of long established native ecosystems and native species2,3.
Introduction of alien species are increasing with the eciency, frequency and volume of transportation4,5 result-
ing in global economic losses of hundreds of billions per year3,6,7. Globalization in terms of transportation and the
trade of goods have compromised natural geographical and ecological barriers between countries and continents
through the transportation of species which contributes to their establishment in a new ecosystem4,8. Invading
species may enter as contaminants on or inside containers9, raw logs for saw mills9,10, untreated wood for pack-
age material11 or live plant import12.
Asian longhorn beetle (ALB) [Anoplophora glabripennis (Motschulsky)] and citrus longhorn beetle (CLB)
[A. chinensis (Forster)] are two wood boring beetles which can spread through global trade and transportation
network from their native countries; China, Korea and Japan13,14. While ALB usually spread through untreated
wood packing material11, CLB mainly spread through live plant import15. Usually, the feeding adults do not cause
severe damage. e real harm is caused by the subsequent attacks of the wood boring larvae, which eventually
kills the host tree by tunneling deep into the heartwood, producing structural weakness and compromising water
and nutrient ow16,17. CLB generally attack further down the trunk than ALB, usually at ground level or on the
roots below ground, or exposed18,19. Recently, attacks from these species in European countries such as Italy and
France have shown that they prefer hosts in the genera maple (Acer spp.), birch (Betula spp.), willow (Salix spp.),
horse chestnut (Aesculus spp.), poplar (Populus spp.) and citrus (Citrus spp.) (CLB)19,20.
Both species are considered highly destructive pests due to their polyphagous (non-fastidious)
characteristics14,20. ey are responsible for an estimated loss of billions of dollars in China17 and hundreds of
billions in the US21. Due to their damaging life cycle, the beetles threaten various forestry sectors including urban
forestry16, tree nurseries15, and agriculture such as fruit orchards and poplar plantations11,18. e eradication
program of these beetles are both destructive and expensive, and have accustomed both species to quarantine
status (emergency measures to prevent introduction and spread of the invasive alien species) in Europe2224.
Studies have shown that the time to complete their life cycle depends on the temperature and their chosen
host (in relation to the species and its condition) for ALB25,26. eir life cycle varies between one and two years27,
with two-year cycles oen appearing in colder areas28. Studies on how temperature eects ALB indicates that the
larvae will most likely initiate pupating aer twoyears in northern European countries25,26. Early detection is
therefore key, and could reduce the extent of the damage as the infestation may not be visually detectable when
the larvae is tunneling deep inside of its host29.
OPEN
Faculty of Technology, Natural Sciences and Maritime Sciences, Department of Natural Sciences and Environmental
*
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
Several studies have explored dierent detection methods for these invasive pests. One of these methods
includes the use of lure-baited traps containing plant volatiles30 or a combinations of plant volatiles and beetle
pheromones to attract, detect and trap the beetles31,32. Another explored method is acoustics33 and laser vibrom-
etry detection34, which detects feeding and moving larvae inside its host by using sensors and specialized soware
calibrated to the feeding larvae`s sound frequency. However, the most frequently used detection method is visual
detection, through ground surveys using high contrast binoculars, or tree crown surveys using ladders, hydraulic
lis or trained tree climbers35. ese human detectors look for noticeable signs and symptoms of infestations such
as dieback, adult feeding damage on leaves and branches, exit holes, oviposition pits, oozing sap and frass18,35.
While ground surveys have proven to be a fast, but uncertain method of checking for infestation, the tree crown
surveys are slower and more costly, but with a higher detection rate26,36. Since early detection of infestation is
crucial to reduce the pests ecological and economic inuence, a sensitive and ecient method focusing on non-
visual signs of infestation would be of highly importance.
Dogs (Canis lupus familiaris) have assisted humans for decades in ecological management, conservation
and pest management, through the use of their nose to detect endangered, vulnerable, small and cryptic species
alongside their biological traces like feaces, dens and nests37. Studies have shown that dogs can be used to detect
beetles (Coleoptera) such as red palm weevil [Rhynchophorus ferrugineus (Olivier)]38, hermit beetle [Osmoderma
eremita (Scopoli)]39, bark beetle [Ips typographus (Linnaeus)]40, ALB41,42 and emerald ash beetle [Agrilus pla
nipennis (Fairmaire)]43. ese beetle studies have shown promising results, but some discrepancies that could
potentially confound and bias their result were discovered. Unblind handler and test personnel44,45, insucient
discrimination training46,47, constant target scent presence48, non-random sample positions and number of sam-
ples present49 are all factors that contribute to discrepancies and make the use of detection dogs questionable.
We implemented a testing and training protocol created to minimize bias and confounding results to investi-
gate the potential use of dogs as a tool for the early detection of two invasive alien beetle species. We trained two
privately owned dogs to detect all life stages and traces of ALB and CLB, and to discriminate them from native
wood borers such as ribbed pine borer [Rhagium inquisitor (Linnaeus)], small white-marmorated longicorn
[Monochamus sutor (Linnaeus)], and small poplar borer [Saperda populnea (Linnaeus)]. Aer the training was
completed, we tested the dog`s abilities to detect and discriminate the invasive wood borers from non-target
scents such as native wood borers in three dierent experiments carried out in three dierent search conditions:
(1) laboratory, (2) semi-eld, and (3) standardized eld. We hypothesized that the dogs would both (1) detect
all odors originating from ALB and CLB, and (2) discriminate these from non-target scents such as native wood
borers in all three test conditions.
Results
Experiment one: scent platform. e two dogs detected ALB and CLB with a mean accuracy of 92.2%
3.1 SD) (Fig.1 and Table1). Of the 128 scent samples searched, the two dogs falsely detected ALB and CLB
ve times with a mean, specicity of 95.1 (± 1.9) and falsely rejected ALB and CLB ve times with a mean sensi-
Figure1. Results presented as accuracy (correct indications among all indication) in percent for all three
experiments: laboratory on a scent platform (1), semi-eld (2) and eld (3).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
tivity of 80.4% (± 9.2 SD) (Fig.1 and Table1). e incorrectly identied samples contained frass (three samples),
adult (one sample) and larvae (one sample) from ALB and CLB, and frass (one sample), larvae (one samples),
adult (two samples) and pupae (one sample) from small white-marmorated longicorn and ribbed pine borer.
Experiment two: semi-eld. e two dogs detected ALB and CLB when searching on trees and untreated
wooden pallets indoors with a mean accuracy of 92.2% (± 5.5 SD) (Fig.1 and Table2). Of the 274 scent samples
searched, the two dogs falsely detected ALB and CLB 11 times with a mean specicity of 94.1% (± 4.2 SD) and
falsely rejected ALB and CLB 12 times with a mean sensitivity of 87.8% (± 16.5 SD) (Fig.1 and Table2). e
incorrectly identied samples contained frass (six samples), larvae (two samples) and adults (four samples) from
ALB and CLB, and frass (six samples), adult (one sample) and larvae (four samples) from small poplar borer,
ribbed pine borer and small white marmorated longicorn.
Experiment three: standardized eld. e two dogs detected ALB and CLB when searching on trees
and untreated wooden pallets in the eld with a mean accuracy of 93.1% (± 5.8 SD) (Fig.1 and Table3). Of
the 300 scent samples searched, the dogs falsely detected ALB and CLB 17 times with a mean specicity of
91.6% (± 5.8 SD) and falsely rejected ALB and CLB ve with a mean sensitivity of 95.3% (± 10.1 SD) (Fig.1 and
Table 1. Mean results presented as sensitivity, specicity and accuracy in percent, including ± standard
deviation (SD), calculated from 16 randomized and double-blind experimental searches on a scent platform
that was carried out between the 17th and 19th of December 2018 at the University of Southeast Norway
including both invasive and native wood borers. A dog`s response to a scent sample could be either; TP true
positive response, FP false positive response, TN true negative response and FN false negative response. CR
total number of correct responses, ICR total number of incorrect responses, and TOT R total number of
responses.
Dog TP FP TN FN CR ICR TOT R Sensitivity (%) ± SD Specicity (%) ± SD Accuracy (%) ± SD
Akira 11 2 49 2 60 4 64 84.6 ± 1.7 96.1 ± 0.1 93.8 ± 0
Chilli 10 3 48 3 58 6 64 76.9 ± 13.5 94.1 ± 2.6 90.6 ± 4.4
SUM 21 5 97 5 118 10 128
MEAN – 80.4 ± 9.2 95.1 ± 1.9 92.2 ± 3.1
Table 2. Mean results presented as sensitivity, specicity and accuracy in percent, including ± standard
deviation (SD), calculated from 21 randomized and double-blind experimental searches carried out on trees
and untreated wooden pallets in a semi-eld condition between 11 to 20th of May 2020 at the University
of Southeast Norway including both invasive and native wood borers. A dog`s response to a scent sample
could be either; TP true positive response, FP false positive response, TN true negative response and FN false
negative response. CR total number of correct responses, ICR total number of incorrect responses, and TOT R
total number of responses.
Dog TP FP TN FN CR ICR TOT R Sensitivity (%) ± SD Specicity (%) ± SD Accurac y (%) ± SD
Akira 45 7 81 4 126 11 137 92.8 ± 8.3 92.8 ± 6.0 92.6 ± 5.9
Chilli 40 4 85 8 125 12 137 82.8 ± 22.3 95.4 ± 0.9 91.8 ± 5.9
SUM 85 11 166 12 251 23 274
MEAN – – – – 87.8 ± 16.5 94.1 ± 4.2 92.2 ± 5.5
Table 3. Mean results presented as sensitivity, specicity and accuracy in percent, including ± standard
deviation (SD), calculated from 30 randomized and double-blind experimental searches carried out on trees
and untreated wooden pallets in eld condition between 26th of July to 10th of July 2020 at the University of
Southeast Norway and forest areas in close vicinity of the University including both invasive and native wood
borers. A dog`s response to a sample could be either; TP true positive response, FP false positive response, TN
true negative response and FN false negative response. CR total number of correct responses, ICR total number
of incorrect responses, and TOT R total number of responses.
Dog TP FP TN FN CR ICR TOTR Sensitivity (%) ± SD Specicity (%) ± SD Accuracy (%) ± SD
Akira 56 6 87 1 143 7 150 98.5 ± 3.7 93.9 ± 4.9 95.6 ± 2.5
Chilli 52 11 83 4 135 15 150 92.1 ± 13.7 89.3 ± 6.0 90.1 ± 7.2
SUM 108 17 170 5 278 22 300
MEAN – – – – – – 95.3 ± 10.1 91.6 ± 5.8 93.1 ± 5.8
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
Table3). e incorrectly identied samples contained frass (two samples), larvae (two samples) and adult (one
sample) from ALB and CLB, and frass (ten samples), larvae (four samples) and adults (three samples) from small
poplar borer, ribbed pine borer and small white-marmorated longicorn.
Discussion
According to our hypothesis we have shown that with a testing and training protocol in place that minimizes bias
and confounding factors, our dogs were able to detect all traces and remains of ALB and CLB and discriminate
them from non-target scents such as native wood borers and uninfested wood. Our dogs were able to do so in
three dierent test scenarios; laboratory with a scent platform, semi-eld and standardized eld, searching a
total of 702 scents (Appendix2). ey falsely rejected 22 target samples (N = 210) and falsely detected 33 scent
samples originating from native wood borers (N = 152). ey also correctly rejected all scent samples originating
from uninfested wood/wood shavings/natural growing trees (N = 186) and blanks (N = 154).
Our results are similar to those from previous studies carried out on beetle scent. e red palm weevil study
achieved a mean detection accuracy of 78%38 and the hermit beetle study achieved a detection accuracy ranging
from 71 to 93%39. e Swedish bark beetle study tested dogs in laboratory conditions searching for synthetic
bark beetle pheromone. e dogs managed to detect the pheromones 99% of the time and correctly rejected the
control samples 55% of the time40. e previous ALB study from 2016 achieved an overall detection accuracy of
94%41 and the emerald ash borer study achieved a sensitivity ranging from 73.3 to 100% and specicity ranging
from 88.9 to 99.8%43. ese beetle studies have shown promising results, however, some discrepancies that could
potentially confound and bias their result were discovered.
A dog`s exceptional ability to interpret intentional and unintentional cues from humans have proven to have
an extensive eect on problem solving tasks5052. Dog-handlers and test personnel in vicinity to the dog should
be unaware of scent sample ratio49, scent sample order49 and target scent presence48, in order to create a double-
blind test design45. We followed a strictly double-blind test design for all our olfactory tests, indicating that our
dogs did not use human cues to detect ALB and CLB or ignore non-target scents. Unfortunately, the number of
target scents present (and material) were known to both dog-handler and test personnel in the previous ALB
study41 and the emerald ash borer study43. us, if two target scents were approached early in the test trial,
the handler could easily guide the dog to perform TN responses (correctly ignoring non-target scents) for the
remaining scents. is was also the case in the red palm weevil study, though the dog-handler and test personnel
had only knowledge of the sample ratio38. While the dog-handler in the hermit beetle study was blinded, the test
personnel was not (single blind)39. Finally, the bark beetle study carried out both single and double-blind tests40.
Scent generalization is the concept of generalizing between target variations49,53. Target scent variations
(e.g. dierent age, reproductive status, diet, population, sex and concentration) enable dogs to identify a scent
denominator between targets and is therefore important in scent dog detection work53,54. Since ALB and CLB are
quarantine species, all signs of their presence is crucial to detect. We therefore exposed our dogs to scent mate-
rial such as frass (larvae and adult), dead adult beetles, dead larvae, infested wood and eggs. Our results suggest
that the dogs were able to detect scents from specimens and even traces of them, even though the scent material
diered in age (some ALB frass samples were over 10years old and stored at room temperature), provided from
dierent specimens and collected from four dierent laboratories across the world (France, Canada and the USA).
Quarantine restrictions in European countries caused us to use only dead specimens in training and testing.
is may aect our dogs in later eld searches as the dogs may nd it challenging to generalize between live and
dead specimens due to the metabolic processes in live insects. However, an Australian study tested if dogs were
able to identify live insects when only trained on insect extracts or dead specimens, due to problems such as
quarantine restrictions55. ey trained four dogs using either insect extract or dead specimens from the bronze
orange bug [Musgraveia sulciventris (Staal)] to test whether the dogs could detect live insects in the nal olfac-
tory test. e dogs successfully detected live insects in the nal test with a mean sensitivity of 100% and 75%55.
e headspace chromatograms of the scent extraction and the dead specimens showed VOCs similarities of 71%
and 77% with live insects55 indicating that live and dead insect may to some extent excrete similar scents and
dogs may generalize between them. A Taiwanese study carried out on scent detection dogs and the red imported
re ant [Solenopsis invicta (Buren)] used dead specimens in the preliminary training, before further training
and testing with live specimens without any problems56, also indicating scent similarities between live and dead
ants. However, some scent detection dogs do discriminate between live and dead insects to detect only ongoing
infestations, such as bed bug [Cimex lectularius (Linnaeus)] detection dogs57,58. Nevertheless, all signs of the
insect pests (dead or alive) are important to detect since Norway (or the majority of Europeans countries) do
not have an establishment of the species; therefore any detection would be valuable to lessen the expensive and
detrimental eradication programs.
A large sample size goes hand in hand with enabling dogs to detect the target`s scent denominator. A UK
study discussed the memory of dogs, as their dogs possibly remembered 117 unique training scent samples,
their results showed an unmistakable decline when 124 naïve scent samples were used in a nal olfactory test54.
erefore, a large sample size will increase the probability of a detection dog to learn the target`s signature
odor53,54. A limited sample size (e.g. pseudo-replication of individuals, one type of scent material, one population
or one area) will increase the chances of the dog remembering specic samples54. We therefore implemented a
large sample size (N = 420, target = 210, non-target = 210) and used naïve samples for the three olfactory tests to
decrease the chance of the dogs remembering specic samples. We also changed the training samples with new
material/individuals at least twice during the training period. None of the previous beetle studies mentioned the
sample size or if naïve scent samples were used in the nal olfactory tests.
A dog`s highly sensitive nose has an unique ability to discriminate between species, diseases and biologi-
cal and chemical compounds37. e diculty of the discrimination training should be evaluated based on the
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
deployment of the detection dog. In laboratory conditions the researcher can control which scents the detection
dog may encounter and in eld conditions the researcher cannot. Specicity is crucial insituations where the
target`s presence may play an important ecological or economic role in conservation of vulnerable/endangered
species,or the eradication of invasive species. It was therefore of high importance to prepare the dogs for
scents they could potentially encounter in the eld such as frass, larvae, or adult beetles from other longicorns
(Cerambycidae). High quality discrimination training can potentially prevent further species identication in
laboratories as the result inthe eld is immediate. e bark beetle, red palm weevil and the previous ALB study
did not use non-targets scents they could have encountered in eld3941. e previous ALB study used empty
paper sheets or uninfested wood as non-target scents in their two test conditions41. ey did report that in the
old orchard, nine of 14 dogs falsely indicated the presence of ALB, but further discussed that the aim was not
to test their dog`s discrimination abilities, but rather their ability to detect ALB41. Nevertheless, they achieved a
high sensitivity (75–88.1%) and specicity (85.3–95.6%) under standardized eld tests searching for ALB41. In
the emerald ash borer study, they used trees attacked by the fungus ash dieback (Hymenoscyphus fraxineus) as
non-target scents for one of the seven experimental set ups, while in the rest of the experiments they used logs,
re woods, ash trees and other deciduous trees as non-target scents. e hermit beetle study did include non-
target scents from species closely related to the hermit beetle in training and testing, but unfortunately stopped
the preliminary discrimination training as they experienced higher error rates39. However, in the nal tests non-
target scents were included and the dog showed a lower sensitivity (29%) towards ower chafer species compared
to the hermit beetle (69%) and specicity from the tests including natural growing trees ranged from 73 to 92%.
Dogs have proven to be an eective detection tool with both a higher5961 and faster60,62 detection rate than
humans. e revisiting of areas to collect hair, insects or video footage from hair, pheromone or camera traps are
removed since dogs only need one sweep around the search area to disprove or conrm the targets presence63.
e use of dogs can also reduce search bias, as they are highly oriented on olfaction cues, rather than convenient
search paths or visual cues53. While human detectors may need to bring specimens to the laboratory for further
identication and may not be able to discriminate symptoms (e.g., exit-holes, frass or oviposition pits) from other
similar species, dogs with proper discrimination training may potentially identify their presence immediately
as they are trained to ignore other wood borers with a similar ecology and morphology in addition to detect all
life stages from egg to adults and their frass or infested wood.
Successfully training detection dogs can be expensive and time consuming from providing and preparing
samples to the dogs training, testing and maintenance. However, when you have already found a suitable detec-
tion dog, additional target scents can be learned. Williams and Johnston64 showed that dogs could detect and
discriminate 10 target odors from non-targets with a high detection rate within a short time frame in laboratory
conditions. Such dogs can also detect several trained target animals within one search65,66. Furthermore, extinc-
tion training (training on not to indicate the presence of previous conditioned targets67) can also be implemented
on previous targets that has lost its importance for the researcher. Orkin etal.68 suggested that partnership with
public agencies or the police could reduce the cost of training and deploying detection dogs as they have been
previously used for scent detection.
Individual dierences are oen seen between scent detection dogs, as some may be more suitable than oth-
ers in relation to training drive, age or physical condition69. No tests were carried out to compare the results of
the two detection dogs, however, Akira achieved slightly higher results in all three experiments. Several factors
can be the reason for this, such as age and favorable wind/temperature conditions. ere is an age dierence of
eightyears between Chilli and Akira, and Chilli was 13years oldwhen the study was nished. Even though the
olfactory system may not be inuenced by age aer dogs are 14years old70, it may be by physical condition. Fac-
tors such as varying terrain, temperatures and working time may potentially have aected the physical exertion
of Chilli and inuenced her detection abilities due to e.g. excessive panting. Akira may also have experienced
more favorable eld conditions in relation to search strategies, environmental conditions, and detection distance
such as e.g. higher wind speed, lower temperature and higher relative humidity (Savidge etal. 2011; Reed etal.
2011; Glen et.al. 2018). Akira did more frequently use wind currents to navigate to the odor source due to higher
wind speed than Chilli.
When it comes to the numbers of dogs used in the respective study, our numbers were limited. However, we
did score two out of three possible points on the criteria list regarding number of dogs used in scent detection
dog studies, based on a previous study on how to identify bias in scent detection dogs49. It is also important to
state that our dogs were privately-owned and not professional working dogs71. However, the usage of privately-
owned dogs contrary to professional working dogs should imply that majority of dogs should be able to achieve
similar research outcomes following the given training and test protocol.
We show that our dogs were able to detect all types of traces and remains of the ALB and CLB and discrimi-
nate it from native wood borers and other non-target scents such as uninfested wood. However, due to quarantine
restrictions, tests were only carried out on dead specimens, and further testing including live insects should
be initiated. We therefore conclude that dogs can be used as a detection tool for remains of ALB and CLB and
tree symptoms caused by them such as frass and infested wood and be deployed as detection dogs in the eld,
preferably in addition to current detection methods.
Methods
Scent samples. Scent samples (N = 210) from ALB and CLB were provided from Animal and Plant Health
Inspection Service (USDA-APHIS), Insect Production Service (IPS, Canada), Forest Service (FS, USDA) and
European Biological Control Laboratory (USDA-EBCL, France), where the beetles were reared in laboratory
conditions (Appendix1). Scent samples from FS were collected during a quarantine period in 2010, while the
remaining scent samples were collected in September–November 2018. All samples were frozen when killed
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
and/or collected, and were still frozen upon arrival by air mail (except the 10-year old frass samples from FS).
Adults were reared on Acer spp. and Salix spp., while larvae were reared on either similar diets as the adults or
an articial diet. Dead animals were used because of quarantine restrictions and import challenges of invasive
alien species into Norway22,23.
Scent materials were handled with sterilized equipment such as tweezers and disposable gloves and separated
by species, stage of life cycle (egg, larvae and adult) and type of traces (frass from larvae or adult and infested
wood). Frass and infested wood were partitioned into glass jars with teon lids (30ml, 57 × 27.5mm, QORPARK,
USA) by weight (AND Electronic balance FA-2000, AC Adapter DC 12V 0.3A, Norway) (mean = 2.26g ± 0.3
SD), while larvae, adults and eggs were separated into glass jars by individuals (some early developed larvae and
eggs were separated by 1–7 individuals depending on their size) before frozen (− 20°C). Half of the samples
(105) were used in training, whereas the remaining (105) was used in experimental testing to ensure all samples
in the nal testing were naïve to the dogs54. Scent sample material used during training were replaced 1–3 times
during the entire training period to avoid familiarizing to specic specimens.
To assure a correct odor impression and decrease the chance of dogs detecting non-target scents during
search situations, scents naturally occurring in real-search scenarios were added as controls (N = 210)46,47. Envi-
ronmental non-targets assumed to be easy for the dogs to discriminate from target scents were uninfested wood
shavings, bark and human-made sawdust from preferred host trees [goat willow (S. caprea), Norway maple (A
platanoides), white birch (B. pubescence), European ash (Fraxinus excelsior), European aspen (P. tremula), grey
alder (Alnus incana) and European mountain ash (Sorbus aucuparia)]19,20 with no visual signs of infestation of
wood borers. Uninfested wood shavings and bark were collected in broadleaf forestry areas around the University
of South-Eastern Norway, Bø in Midt-Telemark municipality, while sawdust was made by sawing branches o
preferred host species and collectingit. Non-targets assumed to be more challenging to discriminate from target,
were infested wood, frass, larvae, pupae and adults from native wood borers, such as ribbed pine borer, small
white-marmorated longicorn and small poplar borer. ose samples were collected in dierent broadleaf and
coniferous areas within Midt-Telemark municipality. All samples were prepared similarly as the target samples
(sterilized equipment and sample glasses), and half of the non-target samples were used during training (105)
and the remaining (105) were used in the experimental testing. Empty sample glasses (phase one), disposable
gloves (phase one), and blanks (nothing) (all training phases and experiments) were also used as non-targets.
Dog training and olfactory experiments. Two privately owned female dogs, grosspitz and border collie
of three and 11years old at start respectively, were used in this study. Both dogs had been used in previous scent
detection work on beavers (Castor spp.)72,73 and grouse (Lagopus lagopus and L. muta)74. One female handler
with some experience74 trained and handled both dogs.
Training was carried out at the dog laboratory at the University of South-Eastern Norway, Bø in Vestfold
and Telemark county, and in forest areas around Bø. e dogs were trained during three periods instead of one
continuous due to nancial reasons: (1) 1st of October to 21st of December 2018, (2) 8th of January to 13th of
June 2019 and (3) 8th of January to 25th of June 2020. e dogs were periodically trained within the respective
time frames (0–3 sessions a week), and active training time ranged from 10 to 30min for each dog in a session.
We used 100% positive reinforcement training with operant conditioning using a clicker75,76. Food, praise and/
or play was used as a reward. e training program was split into three phases: (1) target imprinting on a scent
platform in the laboratory, (2) semi-eld training and (3) standardized eld training. A double-blind experiment
was carried out aer each training phase (i.e., three experiments) to ensure reliable and consistent detection.
Phase one: target imprinting on scent platform. A standard table platform adapted by Hundcampus Training
Centre (HÄLLEFORS, Sweden) was used in phase one (Fig.2). e rationale behind this scent platform is that
one target scent is always present among three non-target scents (the platform holds a total of seven scent holes,
one target scent and six non-target scents) (Fig.2). us, the dog will learn to discriminate the target scent from
the other non-target scents.
Responses to a scent sample in the scent platform was registered and used to calculate training and test
statistics. Both dogs had been used in previous studies with similar initial training and taught to independently
investigate a platform and lay down in front of a target sample to indicate its position for 3–5s. When a dog was
presented with a scent sample, it could respond in four dierent ways: 1) false positive (FP) response, the dog
is performing a trained response on non-target sample, 2) true positive (TP) response, the dog is performing a
trained response on a target sample, 3) false negative (FN) response, the dog falsely rejects a target sample and
(4) true negative (TN) response, the dog correctly rejects a non-target scent77,78.
Phase one started with target imprinting on the scent platform. We used one glass jar containing a target
sample (frass from larvae, frass from adult, wood shavings, larvae, adult or eggs from ALB or CLB) placed in a
plastic cup among six non-target scents (disposable glove and empty sample glass) also placed inside a plastic cup
or blanks (nothing) (Fig.2). We used disposable gloves and new plastic cups for each scent to avoid scent con-
tamination between target and non-target scents and the platform. To imprint the dogs with the target scent, two
training sessions per target material were carried out and all indications on the target material was immediately
rewarded with a click from the clicker and a food/praise reward. An accuracy level of ≥ 80% in two consecutive
sessions was required before training with all variations of the target. One training session consisted of ve
searches where the handler rapidly changed target position ten times (i.e., a dog searched 200 scent samples per
training session). When the dogs had imprinted on all types of target scents, the next training level was initiated.
e next training level within phase one included discrimination training. First, environmental non-target
scents (uninfested wood shavings, human-made sawdust and bark from preferred host species) were added,
and when the dogs achieved an accuracy of 80% in two consecutive sessions, scents from native wood borers
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
(ribbed pine borer, small white-marmorated longicorn and small poplar borer) were added. Experiment one
was initiated aer achieving an accuracy of 80% in two consecutive sessions in the discrimination training.
Experiment one: standard scent platform. e dogs were tested to see whether they could discriminate all life
stages and traces of ALB and CLB from all types of control species and uninfested wood (non-target scents)
using the scent platform. Experiment one (consisting of two experimental sessions) was carried out the 17th
and 19th of December 2018. Contrary to phase one, the structure of a session was changed, and the dogs had to
search the platform eight times per experimental session (i.e., a dog searched 32 scent samples per experimental
session, thus a total of 64 scent samples was searched per dog in experiment one). To avoid as muchinteraction
and unintentional cueing between the dog and the handler as possible, the handler was placed two meters in
front of the platform (behind the dog), sending the dog to independently smell the platform79.
An experimenter placed the scent samples within the platform to ensure the experiments were blinded45,49.
As intraining phase one, only one target sample could be present in a search, but empty searches were included
to ensure that the handler was unaware of the target presence, hence no unintentional cueing from the handler
to the dog51. Target presence (and its position in the platform) was further randomized each search using a
random number generator so the handler was totally unaware of the sample ratio and unable to cue the dog48,49.
Aer the scent samples were placed within the platform, the experimenter exited the training room, and the
dog-handler team entered the training room (i.e., the experimenter and the dog-handler team were in separate
rooms). In the second room, a video monitor connected to cameras allowed the experimenter to watch the dog-
handler team live, without the ability to cue the dog-handler team, i.e., double blind test design45. e experi-
menter was equipped with a clicker that could be heard from across the two rooms. e clicker was used as a tool
to validate the team’s response, and a click was provided if the response was correct (TP or TN) and withheld
if incorrect response was made (FP or FN). We were therefore able to avoid target confusion (oen associated
with always reinforcing and rewarding each indication regardless of its correctness) and loss of condence (oen
associated with always withholding reinforcement and rewards regardless of the indication`s correctness) in the
experimental design54,80.
Figure2. Table platform used in training and experiments one. All training and experimental trials involved
a random sample layout and could contain additional non-target scents as well. Target: ALB [Anaplophora
glabripennis (Motschulsky)] or CLB [A. chinensis (Forster)]. Non-target: small white-marmorated longicorn
(WML) [Monochamos sutor (Linnaeus)], ribbed pine borer (RPB) [Rhagium inquisitor (Linnaeus) small poplar
borer (SPB) [Saperda populnea (Linnaeus)], and non-infested wood from Salix spp. and Acer spp.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
All experimental searches were recorded with two cameras (SONY HANDYCAM DCR-SR, Norway) lming
the team`s responses. All recordings were watched and responses validated by an independent observer to avoid
observation bias81. When both experimental sessions were completed, phase two was initiated.
Phase two: semi‑eld training. Phase two was designed as an incremental step to prepare the dogs for eld
searches by an environment simulating more closely eld conditions, while still remaining in familiar surround-
ings of the laboratory. Trees and untreated wooden pallets were now used as search object, contrary to the scent
platform used in phase one.
Six truncated trees (height: 1.8m, radius: > 7.5cm, weight: < 20kg, species: European aspen, European moun-
tain ash, European ash, Norway maple, goat willow, white birch and grey alder) were drilled with 7–10 holes
per tree (scent jar measures: circumference = 12cm and depth = 7.5cm) at a varying height (from directly above
ground to 1.8m) (Fig.3a). e scent jar openings were covered with unscented paralm (2 in. × 250 . roll,
BEMIS, USA) that later were perforated to allow scent molecules to exit the paralm lid. Perforated paralm
lids were chosen to minimize the possibility of scent dissipating onto the trees when the jars were placed within
the drilled cavities, which also made it possible to reuse the trees. e truncated trees were hung down from two
railing systems on the roof of the laboratory (Fig.3a). Each tree could be moved throughout the length of the
rail; hence, the trees could be moved to dierent locations alongside the railing system.
Phase three started with training the dogs to pinpoint target positions on dierent objects than the scent
platform, i.e. trees. First, target samples were placed inside cavities at a similar height to the dog´s head. e dogs
were told to search while the handler pointed at the target position, clicking with the clicker immediately when
they snied the target. Soon they understood that pinpointing the target position with their nose would result
in a click and a reward. Distance from the dog to the target scent gradually increased, which enabled the dogs
to follow the odor source40. Target samples were now placed at varying height, from ground level up to 1.8m,
corresponding with the ALBs and CLBs general attack pattern20,25,82.
When the dogs were able to pinpoint target positions at varying heights, the semi-eld training continued. A
training session consisted of six searches, i.e., a dog searched 36 trees per training session. Non-target samples
and empty searches (searchers with no target sample present) were included so the dogs would still train on
discriminating target scents from non-target scents. is should increase their detection reliability, as no indica-
tions should be given if the target scent was absent48. Number of target and non-target scents per search varied
and sample position was always randomized (MICROSOFT EXCEL, version: 16.16.3, 2016)49.
Every third training session, the dogs searched stacks of wooden pallets (three stacks of three pallets on top
of each other) or pallet bedded walls (seven pallets; three and four pallets per long side of the wall) (Fig.3b). e
paralm-covered scent jars were placed within the pallets in dierent hiding places to imitate beetle-attack in
untreated wood packing material11. Six searches were carried out per training session, either on the three pallet
stacks, i.e., a dog searched 18 stacks of wooden pallets per training session, or the seven wall-bedded pallets,
i.e., a dog searched 42 wooden pallets per training session. As with the tree searches, the number of target and
non-target scents varied, as did sample position49. When the dogs achieved an accuracy of 80% experiment
two was carried out.
Figure3. (a) e laboratory during the semi-eld condition and experiment two. Six truncated trees
[European aspen (Populus tremula), European mountain ash (Sorbus excelsior), European ash (Fraxinus
excelsior), Norway maple (Acer platanoides), goat willow (Salix caprea), white birch (Betula pubescence) and grey
alder (Alnus incana)] were drilled with 7–10 scent holes and hanged down from two railing systems. e trees
could be moved alongside the railing systems. (b) e laboratory during semi-eld condition and experiment
two. Seven pallets stacked against the wall or three stacks of three pallets.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
Experiment two: semi‑eld. e dogs were tested to see whether they could discriminate all life stages and
traces of ALB and CLB from all types of control species and uninfested wood (non-target scents) in a semi-eld
condition searching on trees and untreated wooden pallets indoor. Experiment two (consisting of four experi-
mental sessions; two sessions searching on pallet bedded walls and two sessions searching on trees) was carried
out between 11 and 20th of May 2020.
Pallet and tree searches were carried out in the same manner as in phase two. Originally, an experimental
session consisted of six searches of either the seven pallets or the six trees, but because both dogs experienced
fatigue and lack of motivation during the last search, we readjusted the number of searches in the remaining
experimental sessions to consist of ve searches instead of six. Accordingly, a dog searched 35 pallets or 30 trees
per experimental session (except for the rst experimental session where the dogs searched 42 pallets each).
We further followed the same procedures as in experiment one. When the four experimental sessions were
completed, phase three was initiated.
Phase three: standardized eld training. Training was carried out in forest areas searching on trees, or in close
vicinity to the dog laboratory searching on stacks of wooden pallets. e time used to initiate this phase was
controlled by factors such as nance (three training periods), detection accuracy (≥ 80% accuracy in two con-
secutive training sessions) and environmental factors (wind speed (m/s), temperature (C°), precipitation (mm)
and snow). Training was only carried out with wind speed < 5.0m/s, temperatures between 0 and 29°C, pre-
cipitation < 1.5mm and in addition the absence of snow. ese weather factors were based on previous ndings
on search strategies, environmental factors and detection distance for detection dogs8388 and measured and
registered before each training session. Terrain could vary from at to very steep (at, moderate, or very steep),
temperature from 0 to 29°C, precipitation from 0 to 1.5mm and wind speed from 0 to 5.0m/s.
Tree searches were carried out in forest areas with a heterogeneous species compositions due to the ALBs and
CLBs polyphagous characteristics14,20. Forest areas were chosen based on tree availability and search scenario
(operative industry areas, urban areas, or busy roads). All scent materials were inserted into small individual
lter bags (tea bags, 50 × 70mm) to ease placement and hiding on branches, inside cracks in the bark at dierent
heights or under exposed roots to imitate beetle attacks11,18,19. As the lter bags were hidden, no visual cues could
be followed from neither the dog nor the handler52. All trees searched were walked up to, in a random order,
and touched by the person placing the scent samples before each search, so no scent trails could be followed
to a specic scent sample49,89. One training session consisted of six searches on six trees (i.e., a dog searched
36 trees per training session), and to avoid familiarization, residual scent and scent trails, the same areas were
never revisited before at least 3 weeks39. Dogs were on a tracking leash while searching, and the handler did no
corrections using the leash, but simply guided the search so the dogs snied each tree in the trial while looking
for indication behaviors.
Every third training session, the dog searched stacks of pallets outside near the laboratory. e training design
was equal to the indoor training on pallets, but the dogs were now only searching on stacks of pallets (two stacks
of ve pallets), i.e., each training sessions consisted of six searches. Contrary to the tree search, all scent material
was kept inside the sample jars as in the previous indoor training, using perforated paralm lids, so the same
pallets could be reused with minimal chance of residual scent. Experiment three began when the eld training
was completed (accuracy of 80% in three consecutive sessions).
Experiment three: standardized eld. e dogs were tested to see whether they could discriminate all life stages
and traces of ALB and CLB from all types of control species and uninfested wood (non-target scents) in eld
condition searching on trees and untreated wooden pallets outside. Experiment three (consisting of six experi-
mental sessions; three sessions searching on stacks of pallets and three sessions searching on trees) was carried
out between 30th of June and 20th of July 2020.
e experiments took place at areas previously used in training, but to decrease the chance of residual target/
non-target scent, we used at least 3weeks in between a search.
Experiment three was carried out in the same manner as phase three. Either two stacks of wooden pallets
or six naturally growing trees were searched ve times per experimental session, i.e., a dog searched 10 stacks
of pallets or 30 trees per experimental session. Standardization of the search areas were done by marking the
six trees that were to be searched with colored ribbons, so the dog handler knew which trees the dog needed to
investigate for each search. e experimenter and dog-handler went through the order of trees together before
a search. Further, the same experimental design as experiment one and two was used.
A search object (a tree or a stack of pallet) could in this experiment contain either one target scent, one non-
target scent or no scent (natural growing tree as non-target scent) or two target scents, two non-target scents or
two blanks (one stack of pallet was divided into two (le and right side of the stack)). Aer the scent samples
were placed within the trees/pallets, the experimenter visually exited the search area and the dog-handler team`s
responses were now veried by telephone between the dog handler and the experimenter. A chest-mounted video
camera (GOPRO HERO 5, 445222, Norway) recorded each search.
Environmental conditions were registered in the beginning and end of each experimental session, taking the
average as measurements of temperature (TESTO SAVERIS 2, Norway), wind speed (EXTECH AN25, Norway)
and relative humidity (%) (TESTO SAVERIS 2, Norway). Data on how many days since precipitation (0–1days)
was retrieved from a local weather station90. All the eld experiments were carried out in either sunny and cloud-
less weather, cloudy weather, or light rain (< 3mm). Light rain was experienced in a total of nine experimental
sessions (three for Chilli andsix sessions for Akira) (Table4). Only Chilli experienced experimental sessions
without precipitation. Wind speed ranged from 0 to 4m/s with a mean wind speed of 0.75 (± 1.26 SD) m/s
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Vol:.(1234567890)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
(Table4). Temperature ranged from 14.4 to 25.5°C with a mean temperature of 18.7°C (± 3.4 SD) (Table4).
Relative humidity ranged from 45.1 to 95.2% with a mean percentage of 65.8% (± 17.5 SD) (Table4).
Data analysis. Of the four responses TP, FP, TN and FN, three parameters were calculated to evaluate
the dogs detection performance in the nal experiments91: correct responses among all responses, accuracy:
(TP + TN) / (TP + TN + FP + FN), true positive responses among all target samples, sensitivity: TP / (TP + FN)
and correct rejections of non-targets among all non-target samples: specicity: TN / (TN + FP). Accuracy express
the percentage of correct responses (correctly identied targets and correctly rejected non-targets) among all
possible responses (all scent samples in the experiment), while sensitivity express the percentage of correctly
identied target scents among all target scents in the experiment and specicity express the percentage of cor-
rectly rejected non-targets among all non-target scents in the experiment.
Ethical note. We carried out all work in accordance with the relevant guidelines and regulations of our
university. e experimental protocol was approved by our animal ethical committee at the Department of
Natural Sciences and Environmental Health,University of South-Eastern Norway. e study was also carried
out in compliance with the ARRIVE guidelines92. Approvals from other ethics committees or ethics boards were
not required. A written consent was not obtained from dog owners since both dogs are owned by the authors.
Animals used in this study did not experience anesthesia, euthanasia or any kind of sacrice.
Data availability
Data used in this study will be available through the USN Research Data Archive.
Received: 20 January 2021; Accepted: 15 July 2021
References
1. IUCN. Guidlines for the prevention of biodiversity loss caused by alien invasive species. 1–24 (International Union for the Con-
servation of Nature, Switzerland, 2000).
2. Wilcove, D. S., Rothstein, D., Dubow, J., Phillips, A. & Losos, E. Quantifying threats to imperiled species in the United States. Bio
Sci. 48, 607–615 (1998).
3. Pimentel, D. et al. Economic and environmental threats of alien plant, animal, and microbe invasions. Agr. Ecosyst. Environ. 84,
1–20 (2001).
4. Perrings, C., Dehnen-Schmutz, K., Touza, J. & Williamson, M. How to manage biological invasions under globalization. Tren ds
Ecol. Evol. 20, 212–215 (2005).
5. Baier, S. L. & Bergstrand, J. H. e growth of world trade: taris, transport costs, and income similarity. J. Int. Econ. 53, 1–27
(2001).
6. Pimentel, D., Zuniga, R. & Morrison, D. Update on the environmental and economic costs associated with alien-invasive species
in the United States. Ecol. Econ. 52, 273–288 (2005).
7. Xu, H. et al. e distribution and economic losses of alien species invasion to China. Biol. Invasions 8, 1495–1500 (2006).
8. Meyerson, L. A. & Mooney, H. A. Invasive alien species in an era of globalization. Front. Ecol. Environ. 5, 199–208 (2007).
9. Hulme, P. E. et al. Grasping at the routes of biological invasions: a framework for integrating pathways into policy. J. Appl. Ecol.
45, 403–414 (2008).
10. Piel, F., Gilbert, M., De Cannière, C. & Grégoire, J. C. Coniferous round wood imports from Russia and Baltic countries to Belgium.
A pathway analysis for assessing risks of exotic pest insect introductions. Divers. Distrib. 14, 318–328 (2008).
11. MacLeod, A., Evans, H. F. & Baker, R. H. A. An analysis of pest risk from an Asian longhorn beetle (Anoplophora glabripennis) to
hardwood trees in the European community. Crop Prot. 21, 635–645 (2002).
Table 4. Environmental variables during 30 randomized and double-blind experimental searches carried out
on trees and untreated wooden pallets in eld. e environmental variables include temperature (°C), wind
speed (m/s), relative humidity (%) and days since precipitation (> 3mm), and represents mean (± standard
deviation, (SD)) and range of measures for all the 30 experimental searches.
Dog Winds speed (m/s) Temperature (°C) Relative humidity (%) Weather Days since precipitation
Akira
1.2 21.0 48.9 Light rain 0
0 22.4 45.1 Light rain 0
4 21.1 95.2 Light rain 0
2.3 18.2 78.9 Light rain 0
1.3 25.5 64.7 Light rain 0
0 16.6 93.4 Light rain 0
Chilli
1.1 15.2 75 Cloudy 1
0 14.4 61.3 Cloudy 0
0 15.2 73.9 Cloudy 0
0 15.7 54.7 Cloudy 1
0.1 19.8 50.9 Sunny 1
0.1 19.1 47.2 Sunny 0
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Vol.:(0123456789)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
12. Liebhold, A. M., Brockerho, E. G., Garrett, L. J., Parke, J. L. & Britton, K. O. Live plant imports: the major pathway for forest insect
and pathogen invasions of the US. Front. Ecol. Environ. 10, 135–143 (2012).
13. Pan, H. Y. Review of the Asian longhorned beetle: research, biology, distribution and management in China. (e General Station
of Forest Pest Control e State Administration of Forestry, Shenyang, China, 2005).
14. Lingafelter, S. W. & Hoebeke, E. R. Revision of the genus Anoplophora (Coleoptera: Cerambycidae). (Entomological Society of
Washington Washington, DC, 2002).
15. E schen, R. et al. Trade patterns of the tree nursery industry in Europe and changes following ndings of citrus longhorn beetle,
Anoplophora chinensis Forster. NeoBiota 26, 1–20 (2015).
16. Sjöman, H., Östberg, J. & Nilsson, J. Review of host trees for the wood-boring pests Anoplophora glabripennis and Anoplophora
chinensis: an urban forest perspective. Arboric. Urban For. 40, 143–164 (2014).
17. Cavey, J. F., Hoebeke, E. R., Passoa, S. & Lingafelter, S. W. A new exotic threat to North American hardwood forests: an Asian long
horned beetle, Anoplophora glabripennis (Motschulsky) (Coleoptera: Cerambycidae) (Entomological Society of Washington (USA),
1998).
18. Haack, R. A., Hérard, F., Sun, J. & Turgeon, J. J. Managing invasive populations of Asian longhorned beetle and Citrus longhorned
beetle: a worldwide perspective. Annu. Rev. Entomol. 55, 521–546 (2010).
19. Hérard, F. et al. Anoplophora species in Europe: infestations and management processes. EPPO Bull. 36, 470–474 (2006).
20. Hérard, F. et al. Anoplophora glabripennis infestation (col: cerambycidae) in Italy. EPPO Bull. 39, 146–152 (2009).
21. Nowak, D. J., Pasek, J. E., Sequeira, R. A., Crane, D. E. & Mastro, V. C. Potential eect of Anoplophora glabripennis (Coleoptera:
Cerambycidae) on urban trees in the United States. Econ. Entomol. 94, 116–122 (2001).
22. European-Comission. Comission implementing decision of 1 of March 2012 as regards emergency measures to preven the intro-
duction into and the spread within the Union of Anoplophora chinensis (Forster) Ocial Journal of the European Union 64, 38–47
(2012).
23. European-Comission. Comission implementing decision 2015/893 of 9 June 2015 as regards measures to prevent the introduc-
tion into and the spread within the Union of Anoplophora glabirpennis (Motschulsky). Ocial Journal of the European Union 146,
16–28 (2015).
24. USDA-APHIS. Asian Longhorned Beetle Eradication Program: Final Programmatic Environmental Impact Statement—September
2015. (USDA-APHIS, U.S, 2015).
25. Keena, M. Anoplophora glabr ipennis (Coleoptera: Cerambycidae) fecundity and longevity under laboratory conditions: comparison
of populations from New York and Illinois on Acer saccharum. Environ. Entomol. 31, 490–498 (2002).
26. Meng, P. S., Hoover, K. & Keena, M. A. Asian longhorned beetle (Coleoptera: Cerambycidae), an introduced pest of maple and
other hardwood trees in North America and Europe. J. Integ. Pest Manag. 6, 1–15 (2015).
27. Haack, R. A., Law, K. R., Mastro, V. C., Ossenburgen, H. S. & Raimo, B. J. New York’s battle with the Asian long-horned beetle. J.
Forest. 2, 11–15 (1997).
28. Hua, L., Li, S. & Zhang, X. in Iconography of forest insects Hunan, China (eds. J Peng & Y Liu) 467–524 (Hunan Scientifc and
Technical Publishing House, 1992).
29. Smith, M. J., Turgeon, J. J., De Groot, P. & Gasman, B. Asian longhorned beetle Anoplophora glabripennis (Motschulsky): lessons
learned and opportunities to improve the process of eradication and management. Am. Entomol. 55, 21 (2009).
30. Jin, Y., Li, J., Li, J., Luo, Y. & Teale, A. S. Olfactory response of Anoplophora glabripennis to volatile compounds from ash-leaf maple
(Acer negundo) under drought stress. Scientia Silvae Sinicae 40, 99–105 (2004).
31. Zhang, A., Oliver, J. E., Aldrich, J. R., Wang, B. & Mastro, V. C. Stimulatory beetle volatiles for the Asian longhorned beetle, Ano
plophora glabripennis (Motschulsky). Zeitschri für Naturforschung C 57, 553–558 (2002).
32. Meng, P. S. et al. Eects of pheromone and plant volatile release rates and ratios on trapping Anoplophora glabripennis (Coleoptera:
Cerambycidae) in China. Environ. Entomol. 43, 1379–1388 (2014).
33. Mankin, R. W., Smith, M. T., Tropp, J. M., Atkinson, E. B. & Jong, D. Y. Detection of Anoplophora glabripennis (Coleoptera: Cer
ambycidae) larvae in dierent host trees and tissues by automated analyses of sound-impulse frequency and temporal patterns.
Econ. Entomol. 101, 838–849 (2008).
34. Zorović, M. & Čokl, A. Laser vibrometry as a diagnostic tool for detecting wood-boring beetle larvae. Pest of Sci. 88, 107–112
(2015).
35. Ric, J. et al. Detecting Signs and Symptons of Asian Longhorned Beetle Injury: Training Guide (Natural Resources Canada Canadian
Forest Service, 2006).
36. Hu, J., Angeli, S., Schuetz, S., Luo, Y. & Hajek, A. E. Ecology and management of exotic and endemic Asian longhorned beetle
Anoplophora glabripennis. Agric. For. Entomol. 11, 359–375 (2009).
37. R osell, F. Secrets of the Snout: e Dog’s Incredible Nose (University of Chicago Press, 2018).
38. Suma, P., La Pergola, A., Longo, S. & Soroker, V. e use of sning dogs for the detection of Rhynchophorus ferrugineus. Phy
toparasitica 42, 269–274 (2014).
39. Mosconi, F. et al. Training of a dog for the monitoring of Osmoderma eremita. Nat. Conserv. 20, 237–264 (2017).
40. Johansson, A., Birgersson, G. & Schlyter, F. Using synthetic semiochemicals to train canines to detect bark beetle–infested trees.
Ann. For. Sci. 76, 58. https:// doi. org/ 10. 1007/ s13595- 019- 0841-z (2019).
41. Hoyer-Tomiczek, U., Sauseng, G. & Hoch, G. Scent detection dogs for the Asian longhorn beetle, Anoplophora glabripennis. EPPO
Bull. 46, 148–155 (2016).
42. Hoyer-Tomiczek, U. & Sauseng, G. in Anoplophora chinensis & A. glabrinpennis: new tools for predicting, detecting and ghting.
How to save our forests and our urban green spaces. J. Entomolog. Acarolog. Res. 45, 10–12 (2013).
43. Hoyer-Tomiczek, U. & Hoch, G. Progress in the use of detection dogs for emerald ash borer monitoring. Forestry 93, 326–330
(2020).
44. DeChant, M. T., Ford, C. & Hall, N. J. Eect of handler knowledge of the detection task on canine search behavior and performance.
Front. Vet. Sci. 7, 11 (2020).
45. Kardish, M. R. et al. Blind trust in unblinded observation in ecology, evolution, and behavior. Front. Ecol. Evol. 3, 1–4 (2015).
46. Gadbois, S. & Reeve, C. in Domestic dog dognition and behavior: the scientic study of Canis familiaris (ed. Horowitz, A.) 3–29
(Springer, 2014).
47. Minhinnick, S. in Canine Olaction Science and Law: Advances in Forensic Science, Medicine, Conservation and Environmental
Remediation (eds. Jezierski, T., Ensminger, J., & Papet, E.) (CRC Press Taylor and Francis Group, 2016).
48. Schoon, G. A. A. Inuence of experimental setup parameters of scent identication linups on their reliability. Problemy Krymi
nalistyki 226, 43–49 (2002).
49. Johnen, D., Heuwieser, W. & Fischer-Tenhagen, C. An approach to identify bias in scent detection dog testing. Appl. Anim. Behav.
Sci. 189, 1–12 (2017).
50. Miklösi, Á., Polgárdi, R., Topál, J. & Csányi, V. Use of experimenter-given cues in dogs. Anim. Cogn. 1, 113–121 (1998).
51. Lit, L., Schweitzer, J. B. & Oberbauer, A. M. Handler beliefs aect scent detection dog outcomes. Anim. Cogn. 14, 387–394 (2011).
52. Cooper, J. J. et al. Clever hounds: social cognition in the domestic dog (Canis familiaris). Appl. Anim. Behav. Sci. 81, 229–244
(2003).
53. Wasser, S. K. et al. Scat detection dogs in wildlife research and management: application to grizzly and black bears in the Yellowhead
Ecosystem, Alberta, Canada. Can. J. Zool. 82, 475–492 (2004).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Vol:.(1234567890)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
54. Elliker, K. R. et al. Key considerations for the experimental training and evaluation of cancer odour detection dogs: lessons learnt
from a double-blind, controlled trial of prostate cancer detection. BMC Urol. 14, 22–31 (2014).
55. Moser, A., Brown, W., Bizo, L., Andrew, N. R. & Taylor, M. Biosecurity dogs detect live insects aer training with odor-proxy
training aids: scent extract and dead specimens. Chem. Senses 45, 179–186 (2020).
56. L in, H.-M. et al. Fire ant-detecting canines: a complementary method in detecting red imported re ants. J. Econ. Entomol. 104,
225–231 (2011).
57. Cooper, R., Wang, C. & Singh, N. Accuracy of trained canines for detecting bed bugs (Hemiptera: Cimicidae). J. Econ. Entomol.
107, 2171–2181 (2014).
58. Pester, M., Koehler, P. & Pereira, R. Ability of bed bug-detecting canines to locate live bed bugs and viable bed bug eggs. J. Econ.
Entomol. 101, 1389–1396 (2008).
59. Harrison, R. L. From the eld: a comparison of survey methods for detecting bobcats. Wildl. Soc. Bull. 34, 548–552 (2006).
60. Reindl-ompson, S. A., Shivik, J. A., Whitelaw, A., Hurt, A. & Higgins, K. F. Ecacy of scent dogs in detecting black-footed ferrets
at a reintroduction site in South Dakota. Wildl. Soc. Bull. 34, 1435–1439 (2006).
61. Cristescu, R. H. et al. Accuracy and eciency of detection dogs: a powerful new tool for koala conservation and management. Sci.
Rep. 5, 8349 (2015).
62. Mathews, F. et al. Eectiveness of search dogs compared with human observers in locating bat carcasses at wind-turbine sites: a
blinded randomized trial. Wildl. Soc. Bull. 37, 34–40 (2013).
63. Long, R. A., Donovan, T. M., Mackay, P., Zielinski, W. J. & Buzas, J. S. Comparing scat detection dogs, cameras, and hair snares
for surveying carnivores. J. Wildl. Manag. 71, 2018–2025 (2007).
64. Williams, M. & Johnston, J. M. Training and maintaining the performance of dogs (Canis familiaris) on an increasing number of
odor discriminations in a controlled setting. Appl. Anim. Behav. Sci. 78, 55–65 (2002).
65. Vynne, C. et al. Eectiveness of scat-detection dogs in determining species presence in a tropical savanna landscape. Conserv. Biol.
25, 154–162 (2011).
66. Long, R. A., Donovan, T. M., Mackay, P., Zielinski, W. J. & Buzas, J. S. Eectiveness of scat detection dogs for detecting forest
carnivores. J. Wildl. Manag. 71, 2007–2017 (2007).
67. Mazur, J. E. Learning & Behavior 8th edn. (Routledge, 2016).
68. Orkin, J. D., Yang, Y., Yang, C., Douglas, W. Y. & Jiang, X. Cost-eective scat-detection dogs: unleashing a powerful new tool for
international mammalian conservation biology. Sci. Rep. 6, 1–10 (2016).
69. Beebe, S. C., Howell, T. J. & Bennett, P. C. Using scent detection dogs in conservation settings: a review of scientic literature
regarding their selection. Front. Vet. Sci. 3, 1–13 (2016).
70. Hirai, T. et al. Age-related changes in the olfactory system of dogs. Neuropathol. Apll. Neurobiol. 22, 531–539 (1996).
71. Gadbois, S. & Reeve, C. e semiotic canine: scent processing dogs as research assistants in biomedical and environmental research.
Dog Behav. 2, 26–32 (2016).
72. Rosell, F., Cross, H. B., Johnsen, B. C., Sundell, J. & Zedrosser, A. Scent-sning dogs can discriminate between native Eurasian
and invasive North American beavers. Sci. Rep. 9, e15952 (2019).
73. Rosell, F., Havier, M. & Kniha, D. Dogs can discriminate individual beavers from their anal gland secretion. Wildlife Biol. 2, 1–7
(2020).
74. Arnesen, H. C., Johnsen, B. C., Costanzi, J.-M. & Rosell, F. Canines (Canis lupus familiaris) as biodetectors for conservation work:
can they discriminate the rock ptarmigan (Lagopus muta) from the willow grouse (L. lagopus) in a yes/no task?. PLoS ONE 15,
143 (2020).
75. Deldalle, S. & Gaunet, F. Eects of 2 training methods on stress-related behaviors of the dog (Canis familiaris) and on the dog–
owner relationship. J. Vet. Beh.: Clin. Appl. Res. 9, 58–65 (2014).
76. Chiandetti, C., Avella, S., Fongaro, E. & Cerri, F. Can clicker training facilitate conditioning in dogs?. Appl. Anim. Behav. Sci. 184,
109–116 (2016).
77. Fjellanger, R., Andersen, E. K. & McLean, I. A training program for lter-search mine detection dogs. Int. J. Comp. Psychol. 15,
278–287 (2002).
78. Jezierski, T. A. in Canine olfaction sceince and law: advances in forensic science, medicine, conservation, and environmental remedia
tion (eds T.A. Jezierski, J.J. Ensminger, & L.E. Papet) Ch. 19, (CRC Press, 2016).
79. Pfungst, O. & Rahn, C. L. Clever Hans (the horse of Mr. Von Osten.): a Contribution to Experimental Animal and Human Psychology
(Holt and Company, 1911).
80. Jezierski, T. et al. Ecacy of drug detection by fully-trained police dogs varies by breed, training level, type of drug and search
environment. Forensic Sci. Int. 237, 112–118 (2014).
81. Traniello, J. F. & Bakker, T. C. Minimizing observer bias in behavioral research: blinded methods reporting requirements for
behavioral ccology and sociobiology. Behav. Ecol. Sociobiol. 69, 1573–1574 (2015).
82. Haack, R. A. et al. Anoplophora glabripennis within-tree distribution, seasonal development, and host suitability in China and
Chicago. e Great Lakes Entomologist 39, 15 (2006).
83. Smith, D. A. et al. Detection and accuracy rates of dogs trained to nd scats of San Joaquin kit foxes (Vulpes macrotis mutica).
Anim. Conserv. Forum 6, 339–346 (2003).
84. Gazit, I. & Terkel, J. Domination of olfaction over vision in explosives detection by dogs. Appl. Anim. Behav. Sci. 82, 65–73 (2003).
85. Glen, A. S. & Veltman, C. J. Search strategies for conservation detection dogs. Wildlife Biology. 1, 1–9 (2018).
86. Dahlgren, D. K. et al. in Wildlife Techniques Manual Vol. 1 (ed. N. Silvy) 140–153 (Wildlife Society Inc, 2012).
87. Savidge, J. A., Stanford, J. W., Reed, R. N., Haddock, G. R. & Adams, A. A. Y. Canine detection of free-ranging brown treesnakes
on Guam. N. Z. J. Ecol. 35, 174–181 (2011).
88. Reed, S. E., Bidlack, A. L., Hurt, A. & Getz, W. M. Detection distance and environmental factors in conservation detection dog
surveys. J. Wildl. Manag. 75, 243–251 (2011).
89. Glen, A. S., Russell, J. C., Veltman, C. J. & Fewster, R. M. I smell a rat! Estimating eective sweep width for searches using wildlife-
detector dogs. Wildl. Res. 45, 500–504 (2018).
90. Jensen, I. S., Skaalin, R. & Eriksen, T. G. Yr.no, <https:// www. yr. n o/ place/ Norway/ Vestf old_ og_ Telem ark/ Midt- Telem ark/B% C3%
B8~43270/> (2020).
91. Jezierski, T., Walczak, M., Ligor, T., Rudnicka, J. & Buszewski, B. Study of the art: canine olfaction used for cancer detection on
the basis of breath odour. Perspectives and limitations. J. Breath Res. 9, 1–12 (2015).
92. Percie du Sert, N. et al. Reporting animal research: explanation and elaboration for the ARRIVE guidelines 2.0. PLOS Biol. 18,
411. https:// doi. org/ 10. 1371/ journ al. pbio. 30004 11 (2020).
Acknowledgements
We would like to thank Dr. Melody Keena (USDA-Forest Service, Northern Research Station, US), Hannah
Nadel and Carrie L. Crook (USDA-APHIS-PSDEL), John Dedes and Jacob St.Amour (IPS, Canada) and Nathalie
Remualde and Gaylord Desurmont (USDA-EBCL, France) for providing us with scent material. We would also
like to thank Christin Beate Johnsen, Elin Holmstad Arnesen, Viktoria Holmstad Arnesen and Rachel Hinds
Content courtesy of Springer Nature, terms of use apply. Rights reserved

Vol.:(0123456789)
Scientic Reports | (2021) 11:16887 | 
www.nature.com/scientificreports/
for helping out with the experiments. Also, a thank to Hanna Lodberg-Holm and Lydia Samuel for commenting
on the manuscript. is study was funded by the Norwegian Food Safety Authority.
Author contributions
C.H.A. and F.R. developed the study design. C.H.A. contributed to the data collection and performed the data
analyses, and C.H.A. and F.R. wrote the manuscript.
Competing interests
e authors declare no competing interests.
Additional information
Supplementary Information e online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 021- 96450-0.
Correspondence and requests for materials should be addressed to F.R.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2021
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
... This ten-year period is marked by an explosion of canine detection research resulting in a growing list of detectable human diseases by BMDDs and BMDDs able to detect virus [bovine viral diarrhea virus (10)], bacteria [C. difficile (7), Escherichia coli, Klebsiella pneumoniae, Enterococcus faecalis, and Staphylococcus aureus (9)], pests (brown tree snakes (22), palm weevils (23), gypsy moths (24), longhorn beetles (25), termites (26), bed bugs (27), and quagga and zebra mussels (28), fouling agents [catfight offflavoring compounds (29), microbial growth in buildings (30)], animals important to conservation efforts [grizzly and black bears (31), brown bears (32), geckos and tuataras (33), tortoises (34), quolls (35), jackals (36), giant bullfrogs (37), wolves (38), rabbits (39), rock ptarmigans (40), bats (41), koalas (42), kit foxes (43), tigers (44), cougars (45), cheetahs (46), bobcats (47), and gorillas (48)], and disease odor directly on humans [Parkinson's (49), epilepsy (50), diabetes (16,51)]. ...
Article
Full-text available
Biomedical detection dogs offer incredible advantages during disease outbreaks that are presently unmatched by current technologies, however, dogs still face hurdles of implementation due to lack of inter-governmental cooperation and acceptance by the public health community. Here, we refine the definition of a biomedical detection dog, discuss the potential applications, capabilities, and limitations of biomedical detection dogs in disease outbreak scenarios, and the safety measures that must be considered before and during deployment. Finally, we provide recommendations on how to address and overcome the barriers to acceptance of biomedical detection dogs through a dedicated research and development investment in olfactory sciences.
Article
Full-text available
Improving the reproducibility of biomedical research is a major challenge. Transparent and accurate reporting is vital to this process; it allows readers to assess the reliability of the findings and repeat or build upon the work of other researchers. The ARRIVE guidelines (Animal Research: Reporting In Vivo Experiments) were developed in 2010 to help authors and journals identify the minimum information necessary to report in publications describing in vivo experiments. Despite widespread endorsement by the scientific community, the impact of ARRIVE on the transparency of reporting in animal research publications has been limited. We have revised the ARRIVE guidelines to update them and facilitate their use in practice. The revised guidelines are published alongside this paper. This explanation and elaboration document was developed as part of the revision. It provides further information about each of the 21 items in ARRIVE 2.0, including the rationale and supporting evidence for their inclusion in the guidelines, elaboration of details to report, and examples of good reporting from the published literature. This document also covers advice and best practice in the design and conduct of animal studies to support researchers in improving standards from the start of the experimental design process through to publication.
Article
Full-text available
Detection dogs are commonly trained and tested under conditions in which the handler or the evaluator knows the true presence or absence of a target odor. Previous research has demonstrated that when handlers are deceived and led to believe that a target odor is present, more false alerts occur. However, many detection teams operate under unknown conditions, and it remains unclear how handler knowledge (or lack thereof) of odor presence/absence influences the dog's behavior. The aim of this study was to evaluate if knowing the number of hides placed influenced detection dog performance in an applied search environment. Professional (n = 20) and sport (n = 39) detection handler-dog teams were asked to search three separate areas (area 1 had one hide, area 2 had one hide, area 3 was blank). Handlers in the Unknown Group were not told any information on the number of hides whereas the Known Group were told there was a total of two hides in the three areas. The sport Unknown Group spent a longer duration (69.04 s) searching in area 3 compared to the sport Known Group (p = 0.004). Further, sport dogs in the Unknown group looked back to the handler more frequently. When a miss did occur, dogs of both sport and professional handlers showed an increase interest in the location of the target odor compared to a comparison location. Critically, however, there was no difference in false alerts between the Known Group and Unknown Group for sport or professional handlers. In a second experiment, fourteen professional, and thirty-nine sport teams from Experiment 1 conducted an additional search double-blind and an additional search single-blind. Both sport and professional-handler dog teams had statistically similar accuracy rate under single and double blind conditions. Overall, when handlers knew the number of hides, it led to significant changes in search behavior of the detection team but did not influence the overall false alert rates.
Article
Full-text available
Alpine and arctic bird populations have shown an unmistakable decrease over the last three decades, and the need for conservation is highly necessary. We investigated the use of five privately-owned dogs (Canis lupus familiaris) as a non-invasive tool to determine the presence of rock ptarmigan (Lagopus muta), through sniffing out faecal pellets, using a yes/no training regime. We carried out 36 double-blind experimental trials per dog and hypothesised that dogs could discriminate the rock ptarmigan from similar species, such as black grouse (Tetrao tetrix), western capercaillie (T. urogallus) and willow grouse (L. lagopus). Our dogs detected differences between the avian species with an average accuracy of 65.9%, sensitivity of 66.7% and specificity of 65.3%. We showed that privately-owned dogs have the potential to be used as biodetectors for conservational work within controlled laboratory conditions for declining species, but overall, only one dog was considered proficient enough. We concluded that dogs could be used as a non-invasive tool to detect the rock ptarmigan, and with further field training and testing, operate in the field for detection surveys.
Article
Full-text available
Detector dogs could be trained to find invasive insect pests at borders before they establish in new areas. However, without access to the live insects themselves, odour training aids are needed to condition dogs to their scent. This proof-of-concept study assessed two potential training aids for insect detection: a scent extract and dead specimens of the target species. Using Musgraveiasulciventris (Hemiptera: Tessaratomidae) as an experimental model, gas chromatography-mass spectrometry (GC-MS) analyses were carried out to compare the chemical headspaces that make up the odours of live specimens and these two training aids. This was then followed by canine scent-detection testing to investigate biosecurity detector dogs' (n = 4) responses to training in an ecologically valid context. Both the scent extract and the dead specimens shared the majority of their volatile organic compounds (VOCs) with live insects. Of the dogs trained with scent extract (n = 2), both were able to detect the live insects accurately, and of those trained with dead specimens (n = 2), one detected the live insects accurately. These findings lend support for these training aids as odour-proxies for live insects - particularly scent extract, which is a relatively novel product with the potential for broad application to facilitate and improve insect-detection training.
Article
Full-text available
The invasion of a species can cause population reduction or extinction of a similar native species due to replacement competition. There is a potential risk that the native Eurasian beaver (Castor fiber) may eventually be competitively excluded by the invasive North American beaver (C. canadensis) from areas where they overlap in Eurasia. Yet currently available methods of census and population estimates are costly and time-consuming. In a laboratory environment, we investigated the potential of using dogs (Canis lupus familiaris) as a conservation tool to determine whether the Eurasian or the North American beaver is present in a specific beaver colony. We hypothesized that dogs can discriminate between the two beaver species, via the odorant signal of castoreum from males and females, in two floor platform experiments. We show that dogs detect scent differences between the two species, both from dead beaver samples and from scent marks collected in the field. Our results suggest that dogs can be used as an "animal biosensor" to discriminate olfactory signals of beaver species, however more tests are needed. Next step should be to test if dogs discern between beaver species in the field under a range of weather conditions and habitat types and use beaver samples collected from areas where the two species share the same habitat. So far, our results show that dogs can be used as a promising tool in the future to promote conservation of the native beaver species and eradication of the invasive one. We therefore conclude that dogs may be an efficient non-invasive tool to help conservationist to manage invasive species in Europe, and advocate for European wildlife agencies to invest in this new tool.
Article
Full-text available
Key message The dog detection allows timely removal by sanitation logging of first beetle-attacked trees before offspring emergence, preventing local beetle increases. Detection dogs rapidly learned responding to synthetic bark beetle pheromone components, with known chemical titres, allowing search training during winter in laboratory and field. Dogs trained on synthetics detected naturally attacked trees in summer at a distance of > 100 m. Context An early detection of first beetle-attacked trees would allow timely sanitation felling before offspring emergence, curbing local beetle increase. Aims We tested if detection dogs, trained off-season on synthetic pheromone components from Ips typographus, could locate naturally bark beetle–infested spruce trees. Methods Indoor training allowed dogs to discriminate between the infestation odours (target) and natural odours (non-target) from the forest. Odour stimuli were shown by chemical analysis to be bioactive at extremely low-levels released (< 10⁻⁴ ng/15 min) in the laboratory. Results Detection dogs, trained to recognise four different synthetic pheromone compounds in the wintertime, were able to detect naturally infested spruce trees unknown to humans the following summer. The dog-handler pairs were able to detect an infested spruce tree from the first hours of beetle attack until several weeks after first attack, long before discolouration of the crown. Trained sniffer dogs detected infested spruce trees out to ≥ 100 m, as measured by GPS-collar tracks. Conclusion Dog-handler pairs appear to be more efficient than humans alone in timely detecting bark beetle infestations due to the canine’s ability to cover a greater area and detect by olfaction infestations from a far longer distance than can humans.
Article
Full-text available
Context Dogs are often used to find rare or cryptic species, but search methods are not standardised, making it difficult to interpret and compare results. Standardised approaches are needed to optimise search effectiveness and/or efficiency. Designing an optimal search strategy requires knowledge of the effective sweep width, which is related to the probability of detection (POD) at various distances between the searcher and the search object. Aims Our primary aim was to estimate effective sweep width for wildlife-detector dogs searching for rodents. We also tested whether dogs differed in their reaction on encountering a laboratory rat (Rattus norvegicus) or a wild-caught Norway rat (wild-type R. norvegicus). Methods We conducted field trials using trained rodent-detector dogs to locate dead laboratory rats. We used the numbers of detections and non-detections at distances of 0-100 m to estimate detection probability and effective sweep width. Key results Dog teams located 100% of rats (regardless of strain) placed directly in their search path. POD declined rapidly with an increasing distance, yielding an observed detection rate of 33% at 10 m, and close to zero at ≥20 m. The data were best described by an exponential decay function. Effective sweep width was estimated to be 16.8 m (95% confidence interval 12.3-21.4 m), corresponding to a strip extending 8.4 m on either side of a walked track. Handlers could not consistently judge whether a dog had encountered a laboratory rat or a wild rat. Conclusions Our results suggest that when dogs are >10 m from a source of rat odour, POD declines sharply. We estimate that the effective distance explored when searching for a stationary rodent is 8.4 m either side of the search path. Implications This information will allow users to optimise the search pattern that dog teams should follow for a given search scenario.
Article
Early detection of infestation by the emerald ash borer (EAB), Agrilus planipennis is extremely difficult; hence developing additional methods is desirable. We built on the successful use of canine scent detection for the invasive long-horned beetles Anoplophora glabripennis and Anoplophora chinensis and trained six dogs in detection of EAB. A first test series was performed to evaluate detection accuracy of five of these dogs. Seven different experimental settings were tested under single blind conditions: (1) forest nursery, (2) piles of firewood, (3) firewood on the ground, (4) ash logs on the ground, (5) old urban ash trees, (6) urban forest with ash trees and (7) natural forest with ash trees. In total, 214 positive samples were presented to the dogs, out of which 20 remained undetected. The experiments ascertained sensitivity (correct positives of all positives) ranging from 73.3 to 100 percent and specificity (correct negatives of all negatives) from 88.9 to 99.8 percent in the tested settings. This initial study demonstrates that trained dogs are able to detect EAB scent from sources such as larval galleries in bark/wood, frass, living or dead larvae or dead dry beetles. The numbers of tested dogs and test series were limited, and further studies are needed to confirm the initial results. However, the preliminary findings demonstrate the potential of the method particularly for inspection of wood or plants at entry points.
Chapter
The value of the canine nose is well-documented, and working dogs are being utilized for their olfactory skills in an increasing number of fields. Not only are dogs used by police, security, and the military, but they are also now used in forensic science, in medical detection of disease, in calculating population trends of endangered species and eradicating invasive species in protected environments, and in identifying infestations and chemical contaminants. Edited and contributed to by eminent scholars, Canine Olfaction Science and Law: Advances in Forensic Science, Medicine, Conservation, and Environmental Remediation takes a systematic scientific approach to canine olfaction. It includes work from scientists working in pure and applied disciplines, trainers and handlers who have trained and deployed detection dogs, and lawyers who have evaluated evidence produced with the aid of detection and scent identification dogs. The book is divided into six sections covering The anatomy, genetics, neurology, and evolution of canine olfaction as well as diseases affecting it The chemistry and aerodynamics of odors Behavior, learning, and training Uses of canine olfaction in forensics and law Uses in conservation and remediation Uses in detection of diseases and medical conditions The various contributors describe cutting edge research, some conclusions of which are the subject of vigorous debates between various laboratories and researchers. The editors have added cross-references so that readers can consider the different perspectives that are currently being advanced and understand where consensus is being built and where more research needs to be done. A useful practical reference, Canine Olfaction Science and Law provides a wealth of information beneficial to a wide range of disciplines. It aids trainers and handlers of detection dogs as well as various professionals in healthcare, law enforcement, forensic science, and environmental conservation to gain a better understanding of the remarkable power of the canine nose while encouraging further advances in applications.
Article
The use of dogs in biomedical diagnosis, detection and alert as well as for the search and monitoring of species-at-risk is an emerging field of research. Standard practices are converging towards models that are not necessarily consistent with the well established field of (animal) psychophysics. We briefly discuss the different challenges of applied canine olfactory processing and discuss the adoption of more valid and reliable methods. For mostly historical reasons it seems, scent processing dogs are trained and tested using multiple alternative stimuli in choice tasks (e.g., line-ups including 6 alternative choices, or 6AFC). Data from psychophysics suggest that those methods will reduce or at the very least misrepresent the accuracy of canines. Unless canines are an exception to the rule, sensory, perceptual and cognitive arguments (e.g., Gadbois & Reeve, 2014) can be made against most multiple alternative forced choice tasks (mAFC's) in favor of detection tasks (yes/no and go/no-go procedures) or, as a compromise, simpler discrimination tasks (2AFC or 3AFC at most). We encourage the use of Signal Detection Theory as it focusses on two important factors in defining the validity and reliability of scent processing dogs: 1) It is a robust measure of sensitivity, an important factor in both diagnosis and sensory detection, and, 2) It describes the type of errors (false alarms vs. misses) that a given dog is most likely to commit, allowing for a solid assessment of performance and potentially a readjustment in training. We give an example with Diabetes Alert Dogs (DAD's) specialized in Hypoglycemia Detection in vitro and discuss the potential advantages of keeping a low number of alternatives during training and testing, the importance of low saliency training (LST), as well as adopting pure detection tasks requiring a response commitment from the dogs for both "yes" and "no" responses. The value of d' (a detectability or discriminability measure) and bias measures (criterion) are discussed in the context of canine selection, performance assessment and diagnostic accuracy across applications.