Explore the vast range of titles available.

In shelters it is usual to conduct standardised behaviour assessments on admitted dogs. The information gathered from the assessment is used to identify dogs that are suitable for adoption and assist in matching the dog with suitable adopters. These assessments are also used to guide behaviour modification programs for dogs that display some unwanted behaviours. For some dogs, the results may indicate that they are unsuitable either for re-training or for adoption. In these circumstances the dogs may be euthanised. We investigated the predictive value of a standardised behaviour assessment protocol currently used in an Australian shelter for dog behaviour post-adoption. A total of 123 dogs, aged 1–10 years and housed in an animal care shelter, were assessed before they were adopted. The new owners of the dogs took part in a post-adoption survey conducted 1 month after adoption, which explored the behaviour of their dog after adoption. Ordinal regression analyses identified that friendly/social, fear and anxiousness identified in the shelter assessment significantly predicted corresponding behaviours post-adoption. However, behaviour problems, such as aggression, food guarding and separation-related behaviours, were not reliably predicted by the standardised behaviour assessment. The results suggest that further research is required to improve the predictability of behaviour assessment protocols for more specific behaviour problems, including different categories of aggression and separation-related problems. We recommend that dog behaviour assessments in shelters are used only in conjunction with other monitoring tools to assess behaviour over the whole shelter stay, thus facilitating increased safety/welfare standards for dogs, shelters and the wider community.

Dogs exhibit behavioural heterogeneity as a result of their close proximity to people as pets, working animals, or research animals. This variability stems from their natural talents as well as contextual effects. This document examines the several types of dog behavioural tests, including those that are used to evaluate dogs and others that are used to categorize individual animals. This study revealed a lack of agreement on all of these testing procedures. Individual variations in behaviour, or personality differences, may now be quantified and described in the working dog literature. The predictive association between certain dog behavioural features (if any) and crucial working results is less well-known.

Guide dog organizations have strict criteria to breed, raise, and select dogs to assist people with visual impairments. In collaboration with Dr. James Serpell, several guide dog training organizations developed a scoring tool called the Behavior Checklist (BCL) to evaluate candidate guide dogs. The tool’s use has expanded to the entire assistance dog industry and is rapidly emerging as the standard behavior assessment. Since 2003, Guiding Eyes for the Blind (GEB) has used the BCL to measure individual dogs’ behaviors up to 8 times between puppyhood and final placement. Here, we evaluate the consistency of the BCL over multiple evaluations in a population of 3,969 Labrador Retrievers raised by Guiding Eyes. We grouped BCL evaluations by two methods, factor analysis, and trainer-defined groups, and summarized groupings of behavior in two ways, using mean and lowest scores. We then determined the agreement between pairs of evaluations using kappa statistics and the predictive capacity of early BCL scores to predict later scores using positive and negative predictive values. Evaluations that are similar in nature and those that are scored within 3 to 6 months of one another agree the most. When a dog scores well early in life, they are likely to consistently score well and the dog’s behavior is unlikely to regress over time. We also found that dogs who score poorly early in life either improve their scores on later evaluations with training intervention or are removed from training. One limitation of this data is that dogs who score poorly at early time points are often removed from training and the data from later BCL evaluations is biased toward higher-scoring dogs. Regardless, these data show that the BCL is an effective way to evaluate assistance dog behavior and has some predictive capacity.

Animal‐worn sensors have revolutionised the study of animal behaviour and ecology. Accelerometers, which measure changes in acceleration across planes of movement, are increasingly being used in conjunction with machine learning models to classify animal behaviours across taxa and research questions. However, the widespread adoption of these methods faces challenges from imbalanced training data, unquantified uncertainties in model outputs, shifts in model performance across contexts and noisy classifications in continuous data streams, where predicted behaviours change abruptly within a sequence. To address these challenges, we introduce an open‐source approach for classifying animal behaviour from raw acceleration data. Our approach integrates machine learning and statistical inference techniques to evaluate and mitigate class imbalances, changes in model performance across ecological settings and noisy classifications. Importantly, we extend predictions from single behaviour classifications to prediction sets: sets of behaviour labels guaranteed to contain the true behaviour with a pre‐specified probability, in a framework analogous to the use of prediction intervals in statistical analyses. We evaluate our approach via simulation and highlight its utility using data collected from a free‐ranging large carnivore, African wild dogs (Lycaon pictus), in the Okavango Delta, Botswana. We demonstrate significantly improved predictions along with associated uncertainty metrics in African wild dog behaviour classification, particularly for rare and ecologically important behaviours such as feeding, where correct classifications more than doubled following quality checks and data rebalancing introduced in our pipeline. Our approach is applicable across taxa and represents a key step towards advancing the burgeoning use of machine learning to remotely observe around‐the‐clock behaviours of free‐ranging animals. Future work could include the integration of multiple data streams, such as accelerometer, audio and GPS data, for model training and could be incorporated directly into our pipeline.

This study investigates the efficacy of various genomic prediction models—Genomic Best Linear Unbiased Prediction (GBLUP), Random Forest (RF), Support Vector Machine (SVM), Extreme Gradient Boosting (XGB), and Multilayer Perceptron (MLP)—in predicting genomic breeding values (gEBVs). The phenotypic data include three binary health traits (anodontia, distichiasis, oral papillomatosis) and one behavioral trait (distraction) in a population of guide dogs. These traits impact the potential for success in guide dogs and are therefore routinely characterized but were chosen based on differences in heritability and case counts specifically to assess gEBV model performance. Utilizing a dataset from The Seeing Eye organization, which includes German Shepherds (n = 482), Golden Retrievers (n = 239), Labrador Retrievers (n = 1188), and Labrador and Golden Retriever crosses (n = 111), we assessed model performance within and across different breeds, trait heritability, case counts, and SNP marker densities. Our results indicate that no significant differences were found in model performance across varying heritabilities, case counts, or SNP densities, with all models performing similarly. Given its lack of need for parameter optimization, GBLUP was the most efficient model. Distichiasis showed the highest overall predictive performance, likely due to its higher heritability, while anodontia and distraction exhibited moderate accuracy, and oral papillomatosis had the lowest accuracy, correlating with its low heritability. These findings underscore that lower density SNP datasets can effectively construct gEBVs, suggesting that high-cost, high-density genotyping may not always be necessary. Additionally, the similar performance of all models indicates that simpler models like GBLUP, which requires less fine tuning, may be sufficient for genomic prediction in canine breeding programs. The research highlights the importance of standardized phenotypic assessments and carefully constructed reference populations to optimize the utility of genomic selection in canine breeding programs.

Working dogs play a variety of important roles, ranging from assisting individuals with disabilities, to explosive and medical detection work. Despite widespread demand, only a subset of dogs bred and trained for these roles ultimately succeed, creating a need for objective measures that can predict working dog aptitude. Most previous research has focused on temperamental characteristics of successful dogs. However, working dogs also face diverse cognitive challenges both in training, and throughout their working lives. We conducted a series of studies investigating the relationships between individual differences in dog cognition, and success as an assistance or detection dog. Assistance dogs ( = 164) and detection dogs ( = 222) were tested in the Dog Cognition Test Battery, a 25-item instrument probing diverse aspects of dog cognition. Through exploratory analyses we identified a subset of tasks associated with success in each training program, and developed shorter test batteries including only these measures. We then used predictive modeling in a prospective study with an independent sample of assistance dogs ( = 180), and conducted a replication study with an independent sample of detection dogs ( = 90). In assistance dogs, models using data on individual differences in cognition predicted higher probabilities of success for dogs that ultimately succeeded in the program, than for those who did not. For the subset of dogs with predicted probabilities of success in the 4th quartile (highest predicted probability of success), model predictions were 86% accurate, on average. In both the exploratory and prospective studies, successful dogs were more likely to engage in eye contact with a human experimenter when faced with an unsolvable task, or when a joint social activity was disrupted. In detection dogs, we replicated our exploratory findings that the most successful dogs scored higher on measures of sensitivity to human communicative intentions, and two measures of short term memory. These findings suggest that that (1) individual differences in cognition contribute to variance in working dog success, and (2) that objective measures of dog cognition can be used to improve the processes through which working dogs are evaluated and selected.

Background This study aimed to evaluate the accuracy of prediction of breeding values in a genomic selection program for behavior traits in a population of Labrador Retrievers used as guide dogs. Implementing genomic selection as a new tool in service dogs has the potential to increase genetic gain, improving the performance of populations. Additionally, genomic predictions may help service dog organizations in identifying training candidates with higher accuracy. Results Phenotypes for 17 traits on 4,841 Labrador Retrievers collected from 2008 to 2019 from the International Working Dog Registry’s (IWDR) behavior checklist were analyzed. The Behavior Checklist (BCL) standardizes a scoring system for a dog’s reaction to a variety of environmental stimuli. Data are used to assess a dog’s behavior and suitability for training as well as genetic selection using a selection index of prioritized traits with estimated breeding values. Genomic data were available for 1076 individuals from whole genome sequences and reduced to 94 K SNPs. Variance components were estimated using AIREML. Genomic information was included under a single-step GBLUP approach. Accuracies were evaluated among a sample of the higher accuracy animals using the linear regression method. Genomic estimates of heritability ranged from 0.08 to 0.21. Accuracies were calculated with the LR method and ranged from 0.30 to 0.58 for pedigree information, with an average of 0.46. Accuracies of genomic predictions ranged from 0.32 to 0.63, with an average of 0.50, and were higher than pedigree predictions for all traits. Conclusions The gains in accuracy from inclusion of SNP genotype data show that genomic prediction using single-step GBLUP can improve selection by identifying the cohort of young dogs that have the highest genetic merit for the desired traits. Gains in validation accuracy were limited by the small number of genotyped animals and are expected to increase as more animals are genotyped.

Assistance dogs can greatly improve the lives of people with disabilities. However, a large proportion of dogs bred and trained for this purpose are deemed unable to successfully fulfill the behavioral demands of this role. Often, this determination is not finalized until weeks or even months into training, when the dog is close to 2 years old. Thus, there is an urgent need to develop objective selection protocols that can identify dogs most and least likely to succeed, from early in the training process. We assessed the predictive validity of two candidate measures employed by Canine Companions for Independence (CCI), a national assistance dog organization headquartered in Santa Rosa, CA. For more than a decade, CCI has collected data on their population using the Canine Behavioral Assessment and Research Questionnaire (C-BARQ) and a standardized temperament assessment known internally as the In-For-Training (IFT) test, which is conducted at the beginning of professional training. Data from both measures were divided into independent training and test datasets, with the training data used for variable selection and cross-validation. We developed three predictive models in which we predicted success or release from the training program using C-BARQ scores ( = 3,569), IFT scores ( = 5,967), and a combination of scores from both instruments ( = 2,990). All three final models performed significantly better than the null expectation when applied to the test data, with overall accuracies ranging from 64 to 68%. Model predictions were most accurate for dogs predicted to have the lowest probability of success (ranging from 85 to 92% accurate for dogs in the lowest 10% of predicted probabilities), and moderately accurate for identifying the dogs most likely to succeed (ranging from 62 to 72% for dogs in the top 10% of predicted probabilities). Combining C-BARQ and IFT predictors into a single model did not improve overall accuracy, although it did improve accuracy for dogs in the lowest 20% of predicted probabilities. Our results suggest that both types of assessments have the potential to be used as powerful screening tools, thereby allowing more efficient allocation of resources in assistance dog selection and training.

Studies on behavioural development in domestic dogs are of relevance for matching puppies with the right families, identifying predispositions for behavioural problems at an early stage, and predicting suitability for service dog work, police or military service. The literature is, however, inconsistent regarding the predictive value of tests performed during the socialisation period. Additionally, some practitioners use tests with neonates to complement later assessments for selecting puppies as working dogs, but these have not been validated. We here present longitudinal data on a cohort of Border collies, followed up from neonate age until adulthood. A neonate test was conducted with 99 Border collie puppies aged 2-10 days to assess activity, vocalisations when isolated and sucking force. At the age of 40-50 days, 134 puppies (including 93 tested as neonates) were tested in a puppy test at their breeders' homes. All dogs were adopted as pet dogs and 50 of them participated in a behavioural test at the age of 1.5 to 2 years with their owners. Linear mixed models found little correspondence between individuals' behaviour in the neonate, puppy and adult test. Exploratory activity was the only behaviour that was significantly correlated between the puppy and the adult test. We conclude that the predictive validity of early tests for predicting specific behavioural traits in adult pet dogs is limited.

There are several hypotheses that could explain territory size in mammals, including the resource dispersion hypothesis (RDH), the intruder pressure hypothesis (IPH), and the intraguild predation hypothesis (IGPH). In this study, we tested predictions of these three hypotheses regarding territories of 19 packs of endangered African wild dogs (Lycaon pictus) over 2 years in the Kruger National Park, South Africa. If territory size was supported by the RDH, then we would observe (1) wild dog territories would be larger when resource patches were more dispersed, (2) pack sizes would be larger when resource patches were rich, and (3) pack size would not affect territory size. If supported by the IPH, then we would observe (4) larger territories would experience less intrusions, and (5) there would be an increase in territory overlap in areas of low resource dispersion. Finally, if supported by the IGPH, we would observe (6) territories would be larger in areas of higher lion (Panthera leo) density, as evidence of a spatial avoidance strategy. We found that the IGPH was fully supported (6), the IPH half supported (5), and the RDH partially supported (1 and 3), where we found spatial partitioning of wild dogs with lions, potentially mediated by resources and territory overlap with conspecifics. Ultimately, our results show that subordinate carnivores must balance a trade-off between dominant interspecific competitors and conspecifics to successfully coexist in areas with dominant carnivores.