1 INTRODUCTION

Domestic dogs (Canis lupus familiaris) have worked alongside humans for thousands of years, primarily used for hunting, guarding and even forensic work by the ancient Greeks (Bergström et al., 2020; Helton, 2009; MacKay et al., 2008; Otto et al., 2019; Shields & Austin, 2018; Whitehouse-Tedd et al., 2021). Even now, dogs support humans by assisting those with disabilities, herding livestock in agriculture, providing protection in law enforcement and military and utilising their sense of smell to find a vast range of substances (Otto et al., 2019; Whitehouse-Tedd et al., 2021; Woollett et al., 2013). Dogs have searched for numerous targets including accelerants, hazardous chemicals, explosives, illegal drugs, human diseases such as cancer, diabetes and epilepsy, live humans, cadavers and more (Browne et al., 2006; Kokocińska-Kusiak et al., 2021; Rooney et al., 2019; Whitehouse-Tedd et al., 2021; Woollett et al., 2013). Within canine scent work, one of the most up and coming areas is that of conservation.

Dogs began working in conservation in the 1890s in New Zealand when they supported efforts to translocate kiwis and kakapos away from areas inhabited by invasive predators (Hill & Hill, 1987). Since then, there has been an almost unlimited scope of their application. Conservation detection dogs (CDD) can perform a variety of tasks like searching for live or dead specimens, nests or burrows and residual scent from hair or urine (Bennett et al., 2022; Helton, 2009; Kokocińska-Kusiak et al., 2021; Woollett et al., 2013). Additionally, scat surveys have been used for indicating animal presence particularly by using DNA analytical techniques like barcoding (i.e. species identification (Arnot et al., 1993)) and profiling (i.e. identification of an individual organism (Giardina, 2013)) located scats, especially when the scat of different species is visually indistinguishable (Bennett et al., 2022; Helton, 2009; MacKay et al., 2008). CDD use has been documented in 62 countries across over 480 biological species including terrestrial, avian and aquatic mammals, birds, reptiles, amphibians, insects, molluscs, fungi, bacteria and invasive plants (Grimm-Seyfarth et al., 2021). Seemingly, scent detection dogs have ‘limitless potential’ (Woollett et al., 2013, p. 261) and their application is restricted only by the ‘human imagination’ (Browne et al., 2006, p. 101). They are invaluable, especially during a time when biodiversity is deeply threatened and the risk of extinction faces many species (Ceballos et al., 2017; Díaz et al., 2019).

Given that most animals have olfactory capabilities for navigation and communication (Kokocińska-Kusiak et al., 2021), why are dogs used most frequently for conservation detection work rather than other species? A key factor is the sheer capacity of canine olfaction. Dogs have up to 250 million olfactory receptors, depending on breed, in comparison to five million in humans (MacKay et al., 2008; Woollett et al., 2013) and can detect odours at concentrations as low as one part per trillion whereas analytical instruments are restricted to parts per billion (Otto et al., 2019). This is due to the unique anatomy of the canine nasal organ and brain (Jenkins et al., 2018; Jezierski et al., 2016; Kokocińska-Kusiak et al., 2021; MacKay et al., 2008). However, rats, insects and pigs also have the ability to be trained to perform scent discrimination like CDD (Bijland et al., 2013; Cambau & Poljak, 2020; Oh et al., 2015; Teodoro-Morrison et al., 2014), so why are these species used less frequently?

For conservation work, trainability and capability in the field are required in addition to olfactory acuity (DeMatteo et al., 2019). Canine domestication means that the species has been selected for sociability, motivation and flexibility of learning (Beebe et al., 2016; Helton, 2009; Otto et al., 2019); psychological traits necessary for conducting complex scent work alongside humans. Furthermore, most conservation work takes place outdoors for several hours in varied weather, topographical and vegetation conditions, meaning CDD must be able to traverse great distances, over extended periods of time, while manoeuvring through obstacles. As such, specific physical features are sought when selecting a dog: stamina, agility and resilience to temperature to name a few (Beebe et al., 2016; DeMatteo et al., 2019; Helton, 2009; Otto et al., 2019). These are characteristics seen in many dogs that are rarely found in smaller or less domesticated species.

CDD have been highly beneficial to conservation outcomes. Their use is non-invasive which protects environmental and wildlife welfare and is preferable to capture-recapture methods (Browne et al., 2006; Grimm-Seyfarth & Klenke, 2018; Kerley, 2010; Richards, 2018). Across many circumstances, CDD are faster, can find more samples and cover greater distances during a survey than other methods (Browne et al., 2006; Grimm-Seyfarth & Klenke, 2018; Kerley, 2010; MacKay et al., 2008; Stanhope & Sloan, 2019). For example, Mathews et al. (2013) found that when comparing humans and CDD during searches for bat carcasses at wind turbine sites, CDD took on average 40 min to conduct a search versus humans taking 2 h and 46 min and CDD found 75% of targets versus humans finding 20%. Furthermore, using CDD can reduce sampling bias as they do not rely on visual information to find targets the way methods like human surveys and camera-trapping does. Therefore, CDD are more capable of finding obscured samples and those in visually less obvious places (Kerley, 2010; MacKay et al., 2008). Additionally, CDD can play the role of ambassador for conservation work through people's affinity towards dogs (Richards, 2018; Witherington et al., 2017).

However, like any detection tool, disadvantages must also be considered. CDD teams are expensive both in terms of time and money. It takes months, if not years, to train a CDD and its handler along with the monetary cost of training and maintaining the dog through transport, housing, food, etc. (Kerley, 2010; MacKay et al., 2008). Acquiring samples for training can be difficult both practically and legally depending on whether the target species is elusive, endangered, or invasive (Kerley, 2010; MacKay et al., 2008). Moreover, despite generally high efficacy rates, substantial variation occurs (MacKay et al., 2008) which brings CDD reliability into question. Indeed, modern guidelines for conservation methods, such as ‘What Works in Conservation 2021’ by Sutherland et al. (2021) along with governmental protocols for target species searches (Thompson et al., 2020), do not include CDD despite their widespread use, which may be indicative of the concerns around their efficacy. Given that conservation suffers from underfunding (Cozzi et al., 2021; Cristescu et al., 2020), the tool used for a project must be worth the cost.

Hence, this review aims to answer the questions of how, why and to what extent does efficacy varies, as these must be understood to achieve the best results possible when using CDD. To do this, available CDD studies were searched for (n = 67) and analysed in light of these questions. A major difficulty facing CDD work is a lack of standardisation across the field (Bennett et al., 2020; Hayes et al., 2018; Johnen et al., 2017; Otto et al., 2019). This is despite efforts made to standardise procedures for the use of scent detection dogs in general (Furton et al., 2010), and for the testing and reporting of CDD work (Bennett et al., 2020; Johnen et al., 2017). At present in CDD literature, terminology for analytical measures is inconsistent (Hayes et al., 2018; Johnen et al., 2017), sample sizes are small leading to low statistical power (Lazarowski et al., 2020; Whitehouse-Tedd et al., 2021) and according to a systematic review by Johnen et al. (2017), up to 70% of CDD studies report limited training details and almost 25% were considered to be poor quality. All these factors together greatly harm the field's reliability and replicability, which is key to verifying results and improving future research. By assessing efficacy and methodology, issues in the literature can be highlighted, thereby increasing understanding of best practices. In this review, the efficacy of CDD will be investigated across training, testing and operational searches and when searching for different target species. Once efficacy rates have been established, the factors affecting efficacy will be discussed along with how methodological problems may be contributing.

2 METHODOLOGY

3 EFFICACY RATES ACROSS CONTEXTS

When assessing the efficacy of CDD, one must be consistent in which measures are considered to ensure as little bias as possible. However, it can be unclear what a study is measuring and terms like ‘detection rates’ may be used without stating what they quantify regarding the search and dog performance (Hayes et al., 2018; Johnen et al., 2017). Bennett et al. (2020) recommend sensitivity (i.e. proportion of target samples found out of total available), and precision (i.e. proportion of alerts that are true positives), also known as ‘reliability’ or ‘predictive positive value’, as measures to be used for evaluating CDD performance. Sensitivity can investigate performance during training and testing which can then help predict the probability of detection during operational searches, as sensitivity in the field is difficult to ascertain without estimating the total number of targets in an area often using techniques with high margins of error like playback (Bolton et al., 2021). Precision aids in determining the ability of the CDD to distinguish and discriminate the target scent from other odours. Lazarowski et al. (2020) propose measuring sensitivity and ‘specificity’ (i.e. proportion of non-alerts that are true negatives) in tandem as key to scent detection work. However, they also acknowledge that specificity is often challenging to accurately measure due to the limitless number of distractor scents that may be available during field trials or operational searches, as well as the difficulty of ascertaining that the target scent is completely absent in a natural environment. As such for this review, sensitivity and precision will be the measures of focus (see Table 1). Of the studies reviewed, 46% stated sensitivity rates, though this rate rises to 79% if counting only studies where sensitivity could have been assessed i.e. because the number of potential targets is known (n = 39), and 55% provide precision rates.

However, it must be acknowledged that in the field of CDD research and applications, one-to-one comparisons of efficacy rates between studies are made difficult by small sample sizes (Lazarowski et al., 2020; Whitehouse-Tedd et al., 2021) and lack of detail about the studies (Johnen et al., 2017). Indeed, of the 67 studies identified for this review, the sample sizes of CDDs ranged from 1 to 20, and 6% (n = 4) did not report this clearly (see Table 2). For the purposes of detecting statistically significant differences and effect sizes, these sample sizes are far too low. But in many ways, this is the nature of the field given the costs and practicalities associated with training and maintaining a CDD (Kerley, 2010; MacKay et al., 2008) and it is unlikely that future studies with vastly larger sample sizes will be produced any time soon. Therefore, there is a need to assess the current state of research in the field, comparing specific comparable measures where available (Table 1) while also acknowledging limitations.

In controlled training and testing trials, the ability to find present targets accurately appears to vary greatly. For insects like beetles, bumble bees and stoneflies, reported sensitivity ranged from 55% to 100% and precision from 75% to 100% with the use of targets like nest material, infested wood, or larvae (Arnesen & Rosell, 2021; Hoyer-Tomiczek et al., 2016; Mosconi et al., 2017; O'Connor et al., 2012; Rutter, Mynott, et al., 2021; Waters et al., 2011). For plant species, rates were high with 99% to 100% sensitivity and 85% to 100% precision (Goodwin et al., 2010; Needs et al., 2021; Vesely, 2008). Work with reptiles and amphibians reported rates of between 61% to 98% for sensitivity and 27% to 100% for precision (Browne et al., 2015; Cablk & Heaton, 2006; Engeman et al., 2002; Matthew et al., 2021; Stevenson et al., 2010; Witherington et al., 2017). CDDs detecting carcasses of birds and bats on windfarms were reported to show sensitivities between 71% and 96% (Arnett, 2006; Domínguez del Valle et al., 2020; Mathews et al., 2013; Smallwood et al., 2020). Searching for bird species through scat, carcasses, or eggs has resulted in sensitivity rates between 62.6% and 100% with precision reported between 50% to 100% (Arnesen et al., 2020; Bolton et al., 2021; Fukuhara et al., 2022; Reynolds et al., 2021). However, the study by Arnesen et al. (2020), where dogs searched for rock ptarmigan (Lagopus muta) scat in lab conditions, had three dogs out of four perform no better than chance and none of the dogs or handlers had any previous experience of training for CDD work.

Mammals are by far the most common animals searched for in published studies. For small mammals, sensitivity during training and testing contexts ranged from as low as 29% to as high as 100% with 44% to 100% precision (Chambers et al., 2015; Cristescu et al., 2015, 2019; Duggan et al., 2011; Fukuhara et al., 2010; Grimm-Seyfarth et al., 2019; Reindl-Thompson et al., 2006; Thompson et al., 2020). Regarding the 29% sensitivity in Chambers et al. (2015) for finding natural bat roosts, this was during a search for both natural bat roosts and suspended bags of guano where guano was the original trained target. This could have caused the CDD to have a preference for the guano samples (i.e. on which they had been imprinted and trained) over the bat roosts which were novel. Indeed, sensitivity was 79% on guano bags alone, and increased to 77% for finding bat roosts, when only searching for bat roosts not in the presence of guano bags. The concepts of using different samples in training versus testing, generalisation of CDD to non-trained targets and the effects of odour concentration in search performance are elaborated on further in the Training section.

For larger mammals, sensitivity rates during training and testing of between 23.8% for sheep remains and 93.3% for cheetah scat are reported (de Oliveira et al., 2012; Hansen & Winje, 2021; Hofmann et al., 2021; Reed et al., 2011) with Hofmann et al. (2021) demonstrating 100% precision on cheetah scat. Although 23.8% sensitivity for CDD seems low, this was compared to the 2.5% sensitivity of human searchers looking for the same carcasses (Hansen & Winje, 2021). Improvements in detection by even small proportions can be hugely beneficial as conservation projects often rely on methods with overall low detection rates (Mathews et al., 2013). These examples demonstrate little pattern regarding the target species when it comes to success during training and testing except for greater variation with mammal targets which could be due to a few things: an inherent issue with the target odours, variation in the quality of the studies, or the simply greater number of papers in that area (i.e. of 67 studies reviewed: 44 on mammals, eight on reptiles, seven on birds with three of these overlapping with mammal studies, seven on invertebrates, three on plants, one on amphibian) (see Table 2).

CDD efficacy should be evaluated during training and testing rather than waiting until operational searches to assess performance, however, many published studies simply investigate whether CDD can discriminate the target odour in a simple controlled trial and do not progress to testing the CDD in a field environment under operational conditions. Indeed, of the 67 studies examined in this review, 37% focus only on training and testing (n = 25), 42% assess solely field performance (n = 28) and 21% look at both (n = 14). Of those studies that measure training and testing performance, 33% conduct their experiments in purely lab-based or controlled field conditions. Moreover, seemingly obvious statistics are sometimes stated such as strong positive correlations between CDD alerts and true positives (Bolton et al., 2021; Oldenburg et al., 2016) which simply means that the dog is doing what it has been trained to do; an unsurprising result given the decades of effective scent detection work performed by canines. This breakdown shows a skew towards laboratory-based and controlled trials that do not translate into assessing fieldwork capabilities or improving methodological practices.

Sensitivity and precision rates within fieldwork vary similarly to those of training and testing. Although most operational windfarm mortality searches did not report precision, Paula et al. (2011) achieved rates of 100% meaning all indications were true positives. Of studies assessing performance in the field, scat surveys of mammals are the most prevalent with precision rates of between 30.8% to 100% (Beckmann, 2006; Clare et al., 2015; Cozzi et al., 2021; Davidson et al., 2014; DeMatteo et al., 2014; Furtado et al., 2008; Harrison, 2006; Hatlauf et al., 2021; Hollerbach et al., 2018; Kretser et al., 2016; Orkin et al., 2016; Sentilles et al., 2021; Smith et al., 2003; Statham et al., 2020; Thompson et al., 2012; Vynne et al., 2011). Low rates of precision may occur as it can be difficult for the handler to accurately identify scats visually which can lead to them accidentally rewarding indications on non-target scats (i.e. false positives) hence reinforcing and leading to a subsequent increase in their frequency. Additionally, CDD may be correctly alerting and DNA barcoding and profiling of the scat can be wrong due to contamination from non-target species resulting from coprophagy, urination, and contact with saliva (DeMatteo et al., 2018). Furthermore, both Hollerbach et al. (2018) and Kretser et al. (2016) used CDDs which had also been trained to indicate on other targets as part of previous work. Training CDD to detect multiple species with overlapping habitats can lead to indications on all targets. As such, most of the false positives in these studies were for the previously trained targets which although classified as a false positive in the context of the study, is not a false positive in the context of the dog's training.

Unfortunately, even while assessing the ability of CDD using these set measures, not every study reports results clearly enough to make inferences. Sometimes, the number of targets found is the only measure reported due to budget constraints, being unable to verify true positives in the field (e.g. small mammals hiding or denning in inaccessible places (Thomas et al., 2020)), or simply a lack of information given within the study itself (Arandjelovic et al., 2015; Bearman-Brown et al., 2020; Becker et al., 2017; Brook et al., 2012; Dematteo et al., 2009; Glen et al., 2016; Jean-Marie et al., 2019; Kapfer et al., 2012; Liczner et al., 2021; Long et al., 2007; McGregor et al., 2016; Petroelje et al., 2021; Rolland et al., 2007; Thomas et al., 2020; Wasser et al., 2004, 2012). Although these results are still valuable for comparisons with other methods and establishing species presence, without any information on error rates, it cannot be determined whether the CDD is performing efficiently or if the authors are merely reporting successes and ignoring mistakes (i.e. are the dogs actually indicating on a number of species including species unwanted by the study (false positives) or are the finds even being verified in the field and the dog's indication taken as writ?).

Despite this, it is clear that across training, testing and operational tasks, CDDs perform generally well and much better than other methods with CDDs outperforming humans and other analytical tools in 91% of the 34 studies analysed where comparisons were made (see Table 2, Columns 3, 4 and 7), excluding select cases: bumble bee nest detection where performance was equivalent to humans (O'Connor et al., 2012) and rhinoceros scat searches where the size of the scat means CDDs do not provide a distinct advantage over the standard method (Brook et al., 2012). However, this review has established that sensitivity and precision rates still vary by a large margin across the studies regardless of target species and search context. So, the question remains, what drives the variation in CDD efficacy?

4 FACTORS AFFECTING EFFICACY AND METHODOLOGICAL ISSUES

5 CONCLUSION

In conclusion, there can be little doubt based on efficacy rates and comparison with other techniques, that using CDD is an effective and beneficial method for conducting a wide range of conservation work. However, the variation in CDD efficacy reported across studies signifies that substantial longstanding issues with standardisation and methodology within the field that are interfering with the understanding of and use of dogs in conservation. CDD are biological systems, meaning their performance is affected by factors including traits of the dog, training methods, experience of both the dog and handler, variables altering olfactory capabilities and the techniques used during a search as well as the search environment itself. This review has critiqued and described ongoing difficulties facing CDD methodology, namely a lack of detail on dogs, handlers, training, experience and study results and contamination of samples during training and searches. The performance of CDD may vary for numerous reasons and as such a cause cannot be determined in any one case without the relevant information. The question is no longer can CDD work in conservation, but rather what can be done to achieve the highest quality performance, while mitigating error and bias. Highlighting these outstanding problems within the literature can enhance future efforts to standardise and improve the CDD research quality, as until then these issues will overshadow the outstanding abilities of CDD.

AUTHOR CONTRIBUTIONS

Beth McKeague: Conceptualization (equal); writing – original draft (lead); writing – review and editing (equal). Caroline Finlay: Conceptualization (equal); supervision (equal); writing – review and editing (equal). Nicola Rooney: Supervision (equal); writing – review and editing (equal).

CONFLICT OF INTEREST STATEMENT

There was no conflict of interest for the authors in this study.