Factors Affecting Efficacy and Methodological
Issues
Training
Training is the foundation of CDD performance with several stages
including imprinting, indication, search tasks, and discrimination
trials. Each has the potential to affect efficacy. In the context of
scent detection dogs, imprinting is the process of familiarising the CDD
with the target odour and is therefore the basis for conducting CDD work
. Given the sensitivity of the canine nose, sample handling during
training must be conducted with care . Subtle aspects of sample
preparation can lead to the dog learning that another odour is paired
with the reward rather than the target itself . Papers often provide
only limited information on sample storage and handling so no inference
can be made on whether this affected efficacy. Indeed, issues identified
regarding sample use include sample contamination with human scent or
other non-target scents , poor decontamination procedures like running
under hot water rather than sterilisation, dog saliva touching sample
containers , and urination and/or defecation by dog during searches
which poses a threat to samples and ecosystems . Furthermore, a review
of bias in scent detection dog work suggests that over 20% of studies
may have used the same samples across training and testing which means
the dog may have learnt the specific samples rather than the target
odour profile .
Given that CDD are biological systems, their olfactory function is
subject to many influences.
Factors linked to reductions in
olfaction capability include older age, use of certain pharmaceuticals,
diseases, dehydration, diet and nutrition, activity levels, temperature,
and humidity and precipitation . There is simply no way to know if any
internal variables have played a role in CDD efficacy, if details are
missing about the dogs used and their care. Furthermore, the target
odour which a CDD has been trained to find, can also affect operational
search efficacy. It is unclear whether CDD search for complete odour
signatures or simply components of the target odour that are present
across samples and conditions . Indeed, CDD are very capable of
generalising from low scent profiles during training to full specimens
in the field and vice versa . However, depending on the samples used to
train the dog, different errors may be made in the field. For example,
if trained on extremely low concentrations of odour then CDD may alert
where no visual sample can be found due to residual scent from past
specimen presence. Alternatively, smaller samples may be missed if
training involved only high odour concentration samples or failed to
simulate any aspect of search environments through field tests and
discrimination training, meaning the sample can be masked by non-target
scents from wildlife or the environment.
Indication or alerting is how a CDD informs a handler that they have
found a target through a distinct and consistent change in behaviour.
Indications can be passive (i.e., no interaction with target) or active
(i.e., body contact with target) depending on the needs of a project.
Passive indication is recommended for CDD work to protect sample
integrity and the safety of both the dog and wildlife . However, details
and definitions of CDD indications are regularly omitted in the
literature. Furthermore, some authors report changes of behaviour (COB)
(i.e., notable shifts in CDD behaviour that suggest the dog has found or
is tracking a scent) or partial indications as a suitable criteria for a
true positive which is far more subjective and open to interpretation
and unable to be standardised, thus altering efficacy rates .
Several types of search tasks can be used when training and testing CDD
efficacy. Multiple-choice tasks are where the CDD has the option to
investigate multiple containers and is rewarded if they alert on the
correct one. These can simulate exposure to different scents available
in the field and also facilitate discrimination training which is key to
ensuring CDD are exposed to commonly encountered scents that should be
ignored in favour of the target odour . However, they also provide more
sensory interference for the dog and can cause preferences for specific
container positions which makes assessing true odour discrimination and
indication performance more difficult . Alternatively, yes/no or
go/no-go tasks involve presenting the dog with a singular sample and
rewarding if they make the correct choice in alerting or ignoring. These
allow for a clear examination of where the dog may be making mistakes
and whether they are making choices more liberally (i.e., more false
positives) or conservatively (i.e., more false negatives . However,
requiring the dog to have greater response inhibition during these tasks
can make them needlessly challenging . Yes/no tasks have been
recommended for CDD , but multiple-choice tasks are commonly seen in the
literature. Although this method has benefits, it lacks details on dog
performance which can help estimate and explain field efficacy rates.
A vital factor for ensuring efficacy results are reliable is blinding.
Single blinding is done to ensure the dog is using olfaction rather than
memory to find the target. But double blinding is preferred where both
the handler and tester also do not know where the target is . This
avoids the ‘Clever Hans effect’ which is an example of a horse seemingly
being able to count but instead was reading human behaviour to determine
when the correct response was given to receive a reward . Domesticated
animals like dogs are highly skilled at reading human behaviour so even
in cases where the handler or tester knows the target location and
believes that efficacy will be unbiased due to the dog ignoring them for
the most part , they may still unconsciously and unintentionally signal
the location of the target to the CDD. Indeed, found that within
ecological, evolutionary, and behavioural research, only 13.3% of
studies susceptible to observer bias, reported use of blinding. IN our
own analysis we found that 43% of the studies described in Table 1 used
blinding, with 90% of these being double-blinding and 10%
single-blinding. In all other cases, it is either unreported or more
worryingly not being conducted at all.
CDD Selection and the
Handler
Although CDD are used as a tool for detection, unlike analytical devices
each individual dog will differ which means the selection criteria of
CDD for efficacy is vital. There is little doubt that all dogs with a
functioning sense of smell can detect any target that emits odour . This
has been demonstrated with pet dogs and their owners that have been
trained to perform scent discrimination and search tasks for novel
odours similar to CDD teams . However, the breed of CDD is often
considered influential in achieving the biological and psychosocial
traits necessary for field work. Breeds that have been historically
selected for their scent abilities are frequently used under the belief
that they will inherently perform well . However individual differences
can affect efficacy . Across CDD literature 128 breeds of dogs have been
used and minimal differences found in suitability . Furthermore, the
assumption that brachycephalic breeds will perform worse is unverified
with pugs outperforming German Shepherds in scent discrimination tests ,
although their ability to physically endure under field conditions is
untested.
More important than breed specific differences is individual
personality. No standardised measures for conducting personality testing
exist and it is unknown when in the dog’s life cycle their ability to
work can be determined . Indeed, wastage (i.e., failing training) is a
major problem in breeding for CDD as the dog may be unsuited to
conservation work . The essential characteristics for CDD are high play
and/or food drive, high hunt drive, and low prey drive . However, the
lack of quantitative measures means that most assessments of these
traits rely on the subjective view of whoever chooses the dog .
Moreover, dogs are biological systems and there will always be an amount
of variability in performance based on countless internal and external
factors throughout their development .
CDD must work as a team alongside a human handler who oversees searches,
verifies finds, and reinforces training. As such, the handler also plays
a crucial role in CDD outcomes. Similar to dogs, specific skills and
traits must be demonstrated to be a handler: ability to direct a search
by assessing where the dog has yet to investigate, understanding of
animal behaviour, learning, and scent theory, attention to detail,
consistency, and endurance for working in field conditions . Handlers
can both positively and negatively influence dog performance. The
handler’s beliefs about how a search will go or the dog itself , the
handler’s behaviour during a search regarding possible finds, the
handler’s level of experience , and the handler’s personality can all
result in changes to the dog’s behaviour . Furthermore, the bond between
a CDD and handler matters for search performance . Dogs working with an
unfamiliar handler, display more stress-related behaviours and have
reduced search efficacy, if they will even search at all .
Search Environment and
Method
Various elements of a search including the area and methods used, also
play a role in efficacy. The environment is cited as a part of efficacy
variation , but the results for how it can alter CDD performance are
mixed . In some cases, detection rates have been seen to have a positive
relationship with wind speed and a negative relationship with vegetation
density . Precipitation can be a concern as it can wash away or degrade
samples . In other cases, no effects for temperature, wind, humidity, or
vegetation were found across studies looking for a range of targets
including mammalian carnivore scats, bat and bird carcasses at
windfarms, scat from different species of quoll, Hermann tortoises,
cheetah scat, and bird carcasses infected with avian botulism , and it
can be difficult to determine why results differ considering the wide
range of climates and locations that these studies took place in.
Regarding search methods, elements that differ include searching on or
off leash, operational time, and effective search distance. In terms of
how dogs search alongside handlers, it is recommended that CDD perform
off-leash searches to avoid handler bias and allow the dog to move
freely and make independent decisions regarding following scent trails .
This would mean that those who opt for line search where the dog is
leashed may be inadvertently altering efficacy. However, line search
must be conducted in some circumstances due to safety concerns for the
dog regarding the environment or predators, dense vegetation, or safety
for wildlife . Traditionally, operational searches occur in 30-minute
intervals , but evidence suggests dogs may be able to work continuously
for up to two hours if so trained . As such, if the dog has been
conditioned poorly for operational searches, they may become demotivated
or fatigued too soon into a search which could cause their efficacy to
drop. Lastly, CDD have an effective operational search distance from the
handler or transect lines. Despite maximum recorded search distances of
up to 62.8m , handlers should have continuous visuals of the dog for
safety and noticing alerts promptly. In addition, efficacy does appear
to be negatively related to search distance . Therefore, the
recommendation is usually less than 10m - 15m for the most efficient and
productive search , although even this can vary if wind directions and
speed are more optimal for the dog which can increase olfaction
abilities .
Future Progress
Although there are clearly issues that need to be addressed regarding
CDD use, research, and efficacy, the benefits CDD can offer to
conservation in this time of worldwide ecological crisis demonstrates
the necessity to improve their utilisation for the future. Conservation
in general requires more funding to achieve its goals and slow down
species decline . If CDD teams had more financial resources, then the
budget constraints which prevent some studies from conducting efficacy
assessments or deploying CDD on larger scales would be less of a
problem. Furthermore, greater communication between CDD practitioners
and researchers across institutions could lead to the development of
empirical standards of practice with the subsequent following of
standards by authors, researchers, and CDD teams globally.