male dogs, respectively) in Qaanaaq (77�270 N 69�150 W) inthe summer of 2007 and the winter of 2008. We used aportable respirometry unit and the doubly labeled water(DWL) method to measure energy metabolism in resting andworking dogs under different temperatures. The key meth-odologic problem was designing a breathing mask for dogsthat is airtight, protects electronics against damage, andavoids freezing of tubing during winter. We obtained highlyrepeatable respirometry measurements down to temperaturesof 215�C. At lower temperatures, material properties didnow allow to continue the measurements. For the DLWtechnique we followed standard protocols: injection of DLW,repeated blood sampling during 10 days postinjection, andmass spectrometry for isotope ratio analysis from the bloodsamples. To measure the dogs’ activity level we used activitymonitors (ActiTrac [IM Systems Inc., Baltimore, MD]) andheart rate monitors (Polar [Polar S610i, Polar Electro GmbH,Germany]). ActiTracs and heart rate monitors were fixed tothe each dog’s collar, elastic waist belts with heart ratetransmitters were fixed to the dog’s chest and electrodecontact was established with contact gel. For the determina-tion of the energy balance and digestive efficiency formacronutrients we collected and quantitatively measuredfeces for 96 hours after feeding. Because we could not usemetabolic cages we developed fecal collection devices withexchangeable collection units that were carried by the dogslike a harness during the feeding trials. Transcutaneousultrasonography was conducted using a portable Sonosite,Titan (SonoSite, Bothell, WA) with Doppler-function torecord the thickness of selected muscles of the locomotorapparatus, blood flow to the leg of locomotor muscles,intestine size, heart size, and cardiac output. The mostchallenging aspects of using ultrasonography in the higharctic region were to overcome logistic problems with thetransport of huge amounts of ultrasonography gel and to useultrasonography in the open field where strong ultravioletlight and temperatures below 215�C disrupt the functioningof most sophisticated electronics. Pushing the equipment totheir limits, we were able to record with high repeatability allparameters of interest during all conditions down to 230�C.To explore structural and ultra-structural changes of muscletissue we took microbiopsies from leg muscles under localanesthesia. This technique allows evaluation of muscletissue by light and transmission electron microscopy with-out impairment of the performance of the dog.Problems and limitations of these methods under arcticfield conditions are discussed and solutions are presented.

Key words: non-invasive methods; respirometry; portableultrasonography; activity monitoring; heart rate monitoring

Shelter and working dogs 77

55

RESPONSE TO A STANDARDIZED PSYCHOGENIC STRESSORAS AN INDICATOR OF WELFARE STATUSNicola J. Rooneya,*, Samantha A. Gainesa,b, HamishD.C. Denhamc, John W.S. Bradshawa

aAnthrozoology Institute, Department of ClinicalVeterinary Science, University of Bristol, United KingdombDefence Science and Technology Laboratory, FortHalstead, United KingdomcDefence Animal Centre, Asfordby Road, MeltonMowbray, Leicestershire, United Kingdom*Corresponding author: Nicola.Rooney@bristol.ac.uk

Urinary cortisol/creatinine (C/C) ratios have been used tomeasure stress and hence the welfare of dogs. However,resting or basal levels of cortisol may not be reliable indicatorsof chronic stress, since HPA axis functioning may be alteredby long-term exposure to stressors (Jarvis et al., 2006).We carried out 2 studies to explore the use of a validatedpsychogenic challenge (van Vonderen et al., 1998) to mea-sure C/C responsiveness as an indicator of welfare. Wecompared the cortisol responses of German shepherd dogs ofdiffering histories when exposed to a mild acute stressor.Urine samples were collected for cortisol and creatinineanalysis before and at several time periods after a standard-ized veterinary examination. In the first study, we compared 3populations: home-kept, pet dogs (n 5 11); kenneled policedogs (n 5 10); and kenneled military dogs (n 5 13). In thesecond study, we compared the cortisol responses ofkenneled military dogs within a single kennel establishment(n 5 30) but divided according to their propensity to exhibitrepetitive behaviors. To classify each according to its repet-itive behavior profile, we used each dog’s responses to 9different stimuli and when unstimulated.In the first study, we saw differences in baseline C/C betweenthe 3 populations of dogs (F(2,31) 5 4.9, P 5 0.013). However,the response to the stressor varied greatly between individualdogs, and only those in the police dog population showed a sig-nificant change after the stressor (F(3,7) 5 13.7, P , 0.001). Inthe second study, urinary C/C response curves differed signif-icantly between groups of dogs showing different behavioralprofiles (F(9,78) 5 2.79, P 5 0.007). Dogs that showed repet-itive behaviors at times of low arousal differed from dogs thatshowed repetitive behaviors only in response to high arousalactivities and those that stereotyped when alone.We concluded that basal C/C ratios are effective atdistinguishing between different populations of dogs sub-jected to differing levels of background stress. However,within a population, the use of a psychogenic stressor andmeasurement of consequent C/C response was proven to bea useful method to examine interdog variation. The varyingC/C responses of dogs that exhibited repetitive behaviorsunder differing conditions implies that HPA axis respon-siveness varies between animals with different behaviorprofiles. The motivational basis of repetitive behaviors andthe relationship between repetitive behaviors and the HPAaxis require further research to aid the understanding ofthese widely used behavioral and physiological measures inwelfare assessment.

Key words: cortisol; HPA axis; domestic dog; welfare;repetitive behavior

57

SOCIAL LEARNING IN SHELTER DOGSJennifer Templeton*, J. ThornDepartment of Biology, Knox College, Galesburg, IL*Corresponding author: jtemplet@knox.edu

Socialization and training can make shelter dogs moreadoptable. However, time constraints make it difficult for

78 Journal of Veterinary Behavior, Vol 4, No 2, March/April 2009

References:

Jarvis, S., D’Eath, R.B., Robson, S.K., Lawrence, A.B., 2006. The effect of

confinement during lactation on the hypothalamic-pituitary-adrenal

axis and behavior of primiparous sows. Physiol. Behav. 87, 345–352.

van Vonderen, I.K., Kooistra, H.S., Rijnberk, A., 1998. Influence of veter-

inary care on the urinary corticoid:creatinine ratio in dogs. J. Vet.

Intern. Med. 12, 431–435.

56

HUMAN-DIRECTED AGRESSION IN SHELTER DOGS: HOWTO TEST FOR BETTER PREDICTION OF OUTCOMESBarbara Klausz*, Anna Kis, Eszter Persa, Marta GacsiDepartment of Ethology, Eotvos University, Hungary*Corresponding author: klauszbarbara@yahoo.com

In the last few decades many test series were developed andused to measure behavior in shelter dogs to identify individ-uals with aggressive tendencies who may not suitable for re-homing. However, relatively few studies evaluated theeffectiveness of these test procedures. According to a surveybased on questionnaire data, 40.9% of the dogs that hadpassed a temperament test showed aggressive behavior intheir new home within 1 year of the adoption. The aim of thepresent study was to explain this high error-percentage byfinding the possible shortcomings of the procedure. Accord-ing to our hypothesis, dogs’ behavior is suppressed in manyrespects when kept in a shelter, thus tests carried out duringthe first few days might have less predictive value.Human-directed aggression was investigated in 25 shelterdogs. We first tested the dogs 20–40 hours after intake and asecond time 2 weeks later. The same test series was appliedon the 2 occasions and we analyzed the difference betweenthe behavior shown in the 2 tests. Based on our experiencesfrom a previous study on pet dogs, the following subtestswere used: friendly greeting, taking away a bone, threateningapproach, and dominant approach. The dogs were tethered to2 trees in a V-shape with two 3-meter long chains, so that theycould not do semicircular movements. The test procedurewas carried out by 2 women experimenters. We coded thelevel of aggression in all the 4 subtests.We found significant differences in the dogs’ aggressivebehavior between the first and the second test. The number ofdogs showing aggression in the taking away a bone test andalso the level of aggression observed increased remarkablyfor the second test date. In contrast to results in pet dogs testedwith their owners, the threatening or dominant approach didnot evoke aggressive behaviors from shelter dogs.We suggest that timing is a crucial factor of the testingprocedure to gain more predictive results in tests trying topredict human-directed aggression in shelter dogs. More-over, we propose that threatening approaches from humansin a shelter setting tend to mainly evoke avoidance behaviorin shelter dogs (partly because of the owner’s absence),whereas the same stimuli might facilitate (fear-related)aggression in pet dogs (in the presence of the owner).

Key words: shelter dog; aggression; behavioral test; timing

shelter staff to complete tasks beyond caring for theanimals’ physical well-being. To examine the feasibilityof training in a shelter environment, we first carried out aseries of experiments (Thorn et al., 2006), which showedthat (1) dogs could be shaped to sit in as few as 10 trials;(2) a verbal secondary reinforcer (‘‘Good Dog!’’) wasmore effective than a clicker when used by untrained per-sonnel; and (3) dogs retained and generalized their newskill to new people and locations. We have now begun a se-ries of experiments examining whether social learning en-hances the speed of training in shelter dogs, and if so, how.Fourteen dogs were randomly assigned to No Observe(NO) and Observe (O) treatment groups (n 5 7 each).Trials took place with dogs in individual kennels. NO dogsand O dogs were in adjacent kennels, separated by chain-link fencing. A NO trial consisted of a ‘‘stranger’’ walkingup to a kennel and standing in front of it; no verbalcommands or gestures were made. If the dog sat, thestranger said, ‘‘Good Dog!’’ and gave a small piece of hotdog, recorded time to sit, and left the room. A trial lasted amaximum of 60 seconds, and there was a 30-secondintertrial interval. NO dogs received a total of 10 trials,while the O dog looked on. To ensure that O dogs had theopportunity to observe 10 successful, rewarded sits, addi-tional NO dog trials continued until this number had beenmet. Only the first 10 NO trials were included in the dataanalysis. Two minutes after NO trials had been completed,a new stranger began O trials. These were identical to NOtrials, but there were no extra trials.A 2-way repeated analysis of variance (ANOVAR) found asignificant decline in latency to sit over trials (F 5 4.75, df5 9,108, P , 0.001); a nonsignificant trial x treatment ef-fect indicated that latency declined at a similar rate for the 2treatment groups. O dogs sat significantly earlier than NOdogs (F 5 3.43, df 5 1,12, P , 0.05). The difference inperformance between the 2 treatment groups was most no-ticeable in the first trial; a 2-sample t test (1-tailed) showedthat O dogs sat significantly more quickly in their first trialthan NO dogs (t 5 -1.98, df 5 12, P , 0.05).O dogs learned to sit in response to a discriminative stimulus(arrival of a stranger) faster than NO dogs. This preliminaryfinding suggests that training of shelter dogs may proceedeven more rapidly than expected, thus saving shelter stafftime and effort. Next, we plan to compare physiologicalresponses of O and NO dogs during observation and trainingtrials, as well as the potential for multiple dogs to learn fromthe same dog at the same time.

Key words: shelter dogs; training; social learning