J. Range Manage. 52:2Ð6 January 1999 Livestock-guarding in Norway: Part I. Interactions

INGER HANSEN AND MORTEN BAKKEN

The authors are with the Planteforsk Tj¿tta Rural Development Centre, N-8860 Tj¿tta, Norway and the Agricultural University of Norway, P.O. Box 5025, N-1432 Ås, Norway respectively.

Abstract Resumen

We documented behaviors of Great Pyrenees livestock- Documentamos el comportamiento de los perros guarding dogs toward people, livestock, dogs, horses, rein- guardianes de ganado de raza “Great Pyrenees” hacia la deer, and bear to determine if they might be suitable for gente, ganado, perros, caballos, renos y osos para determi- protecting livestock in Norway. None out of 13 dogs nar si ellos pudieran ser apropiados para proteger el gana- showed aggressive behavior towards unfamiliar people, do en Noruega. Ninguno de los 13 perros mostraron com- and aggressiveness towards dogs and livestock was also portamiento agresivo hacia la gente desconocida y la agre- low. However, 91% of the dogs tested chased reindeer. A sividad hacia perros y ganado también fue baja. Sin willingness to chase bears was apparent in all 3 dogs test- embargo, 91% de los perros bajo prueba persiguieron los ed. Although the Norwegian strains of the Great Pyrenees renos. La disponibilidad para perseguir osos fue aparente are bred mainly for exhibition, they obviously have en los 3 perros probados. Aunque la líneas noruegas de la retained some behavioral patterns important for the live- raza “Great Pyrenees” son criados para exhibición, ellos stock-guarding function. Their nonaggressive behavior obviamente han retenido algunos patrones de compor- towards people, dogs, and livestock, and their active reac- tamiento importantes para la función de proteger el gana- tion towards bears suggest that this breed could be suit- do. Su comportamiento no agresivo hacia la gente, perros y able for use as livestock-guardians in Norway. However, ganado y su reacción hacia los osos suguieren que esta raza the dogs’ tendency to chase reindeer is a trait that may pudiera ser apropiada para usarla en Noruega como per- cause conflicts in reindeer-herding areas. ros guardianes de ganado. Sin embargo, la tendencia de los perros a perseguir renos es una característica que pudiera causar conflicto en áreas donde hay manadas de renos. Key Words: Great Pyrenees, behavior, strangers, cattle, sheep, horses, reindeer, bear, predation

In Norway, 2.5 million sheep (Ovis aries L.) are grazed ficult in Norway. In addition, rangeland is by law publicly on mountain and forest summer range (SSB 1995), and accessible (Lov om friluftslivet 1957), and wildlife more than 100,000 of these disappear each year. In some (Viltloven 1981), free-ranging livestock (Lov om ansvar areas depredation on sheep may exceed 70% (Kvam et al. for skade på bufe ved hund m.v. 1926), and domestic rein- 1995b; Mysterud et al. 1994) of the total loss. deer (Reindriftsloven 1978) are legally protected from Scandinavian brown bears (Ursus arctos L.), lynx (Lynx stray dogs. Consequently, before the use of livestock- lynx L.) and wolverines (Gulo gulo L.)) are the main preda- guarding dogs can be considered in Norway, we need to tors of livestock in Norway. Only bears and wolves are know the behavioral traits of the dogs which could cause fully protected. Wolverines are hunted by licence in north- conflicts with people, wildlife, and other animals. ern Norway and a quota of lynx are hunted during a regular Significant differences in success between breeds of live- hunting season. stock-guarding dogs were not found (Green and Woodruff Livestock-guarding dogs may be an effective way of pro- 1983; Green and Woodruff 1988), but among different breeds tecting sheep from predators (Andelt 1992; Coppinger et evaluated, the Great Pyrenees was one of the least aggressive al. 1988; Green and Woodruff 1990; Lorenz et al. 1986), towards people, livestock, and other dogs. For this reason, we but widely dispersing sheep make using guarding dogs dif- started with the Great Pyrenees before examining other breeds. However, there is no tradition for using livestock- guarding dogs in Norway, so the Norwegian lines of the This project was funded by the Directorate of Nature Management and the Department of Agriculture. We would like to thank Ole Jakob S¿rensen Great Pyrenees are mainly bred as show dogs. (Norwegian Inst. for Nature Science) and Even Bj¿rnes (County Governor, Dept. This publication focuses on interactions between Great of Environmental Affairs) for valuable support during trial III. We also wish to Pyrenees and strangers, livestock, dogs, horses, reindeer, thank the handlers; Katrine Eldegaard, Hildebj¿rg Haugan, Torill J¿rgensen, Brynjulf Liland, Aud Rings¿, and Hanne-Brith Vold for indispensable assistance. and bear. Different ways of using guarding dogs under Thanks are also due to the private Pyrenees owners. Norwegian conditions will be reported in a Part II paper. Manuscipt accepted 18 Apr. 1997.

2 JOURNAL OF RANGE MANAGEMENT 52(1), January 1999 Material and Methods to have provoked the dogs by entering Trial III. Interactions between their personal space. The test dog was Pyrenees and Bear Dogs tied to a long leash at a starting-point The Norwegian Institute of Nature A total of 13 Great Pyrenees from 7 located 150 m from both entrances, Research allowed us to arrange the litters and 2 breeders were tested. but was released by a dog handler at first controlled confrontation between They were all reared on sheep farms, the moment it became aware of the a radio-tracked bear and 3 radio- but due to the strong influence of the unfamiliar being. tracked dogs. The dogs tested were a breeders and Kennel Club, the dogs The following observations were 2-year-old female and two 3-year-old were not reared with sheep until the recorded: the dog’s behavior when the males (male A and B). The female and age of 12Ð16 weeks and were handled stranger entered the arena (behav.1); male B might have had earlier experi- a lot by people. Consequently, they the behavior when the stranger left the ence with bears during a field study were more socially bonded to people arena (behav.2); the minimum dis- (Hansen. 1996). The bear was a 150 than to sheep. The dogs were sired by tance between dog and stranger kg, 3-year-old male (Kvam et al. (min.dist); the latency time from when the most common stud males and 1995b; Kvam et al. 1996) that previ- the dog discovered the stranger until should therefore represent the major ously had killed sheep, and sheep the former sought physical contact genes of Great Pyrenees lines in were grazing in a mountain area 5 km (cont.time); and whether the dog Norway. All males were castrated, south of the bear’s location. The dogs barked () or urinated (uri). In were released at 5-minutes intervals at whereas none of the females were cases where the dog did not make con- spayed. a distance of approximately 100 m tact with the stranger, the cont.time from the bear, which was hidden in was recorded as a maximum time of dense vegetation. Trial 1. Test of Interactions between 300 seconds. The behavioral reper- Pyrenees and “Strangers” toire (behav.1 and behav.2) was cate- Of the 13 dogs tested, 11 were gorized into 13 behavior patterns Statistics males. Seven were 1 year old and 6 including defensive aggression; flee- Unless otherwise stated, data are were 2 years old. The test was per- ing; uninterested; staying/lying/sitting presented across breeding line, age, formed inside a 4-ha enclosure, which motionless, but attentive; ambivalent; sex, and rearing conditions. Standard was part of a pasture for 40 ewes and approaching, but keeps a distance and SAS procedures (SAS 1987) are used. lambs that were familiar with makes no contact; approaching and Categorized behavioral data were test- Pyrenees. The dogs were tied to a long making contact; non-aggressive greet- ed by non-parametric tests (Mann- chain inside the enclosure the night ing; playful; playfully jumping upon Whitney U-test, Spearmann’s ranked before being tested. the stranger; chasing without attack- correlation, and Chi-square), whereas The dogs were tested 3 times each at ing, the dog stops by itself; chasing variables following the normal-distrib- different days on 5 types of “stranger” without attacking, but the observer has ution were treated by analysis of vari- in the following sequence: an unfamil- to stop the dog; and aggressive attack. ance (GLM) and two-tailed t-tests. iar man; an unfamiliar dog guided by a Because these behaviors could be cat- Except for the descriptive presentation man; a herd of 6 sheep that were not egorized on a scale from strongly in Table 1, data regarding Trial I are used to dogs; an unfamiliar horse; and defensive to strongly offensive, they pooled across the 3 repetitions for an unfamiliar Hereford cow with calf. were evaluated statistically by both each type of stranger. All differences The livestock were followed by a shep- nonparametric and parametric tests. discussed are statistically significant at herd. The unfamiliar man wore 3 types the 5% level unless otherwise stated. of clothings: hunting clothes while Trial II. Interactions between bearing a shotgun, hiking clothes with Pyrenees and Reindeer Results a large backpack, and a rainsuit with a Ten male Pyrenees and 1 female bucket used for picking berries. The 3 were tested on a small flock of 20 unfamiliar dogs were males of different reindeer on an island. None of the Trial 1. Test of Interactions between breeds: a Kleiner Münsterländer (bird dogs had previous experience with Pyrenees and Strangers dog), a , and a medium- reindeer. A maximum of 3 dogs were None of the 13 dogs showed any sized mixed breed. The horse and the tested per day, and each dog was test- kind of aggressive behavior towards cow/calf pair were not the same in all ed only once. The dogs were released unfamiliar people (Table 1). Two presentations, however they were iden- one-at-a-time and, as long as no criti- Pyrenees displayed a highly domi- tical for all dogs within the same repe- cal situations occurred, they were nant/threatening posture toward the tition number. allowed to follow/chase the reindeer husky and the mixed breed, but with- The strangers entered the open field until the dogs stopped themselves. out fighting. One dog behaved aggres- from 2 different entrances, all follow- Distance to the reindeer at the moment sively towards the cattle, and 4 dogs ing a path that turned 90 degrees at a the dog became aware of them, mini- chased the flock of 6 sheep that were distance of 50 m from the test dog. In mum distance between dog and rein- unsocialized to dogs. One dog dis- this way the stranger was assumed not deer, and total time of the chase were played fear towards cattle. recorded.

JOURNAL OF RANGE MANAGEMENT 52(1), January 1999 3 Table 1. The most common behavior pattern displayed by the dogs (N=13) when the different types of “stranger” entered the arena (behav. 1) and when they left (behav. 2), as shown by the number of dogs performing these behaviors at least twice1 (N=13), and by the percentage of total trials per stranger type (N=39).

Stranger Variable Most Common Behavior Pattern No. of Dogs % of Total Trials Man Behav. 1 Standing/lying/sitting motionless, but attentive 5 39 Behav. 1 Uninterested 2 21 Behav. 2 Uninterested 7 44 Behav. 2 Non-aggressive greeting 5 33 Dog Behav. 1 Approaching to make contact 9 72 Behav. 2 Non-aggressive greeting 9 59 Sheep Behav. 1. Approaching, but keeps a distance and makes no contact 3 28 Behav. 1 Standing/lying/sitting motionless, but attentive 3 26 Behav. 2 Uninterested 5 41 Behav. 2 Chasing 3 18 Horse Behav. 1 Approaching, but keeps a distance and makes no contact 3 36 Behav. 1 Standing/lying/sitting motionless, but attentive 2 36 Behav. 2 Uninterested 4 31 Behav. 2 Approaching, but keeps a distance and makes no contact 3 26 Cattle Behav. 1 Standing/lying/sitting motionless, but attentive 5 39 Behav. 1 Approaching, but keeps a distance and makes no contact 5 39 Behav. 2 Uninterested 4 26 Behav. 2 Standing/lying/sitting motionless, but attentive 3 28 1These 2 measures will not correspond, because each dog may show different behavioral patterns during the 3 trials.

Each dog showed consistency in The dogs detected the entering Trial II. Interactions between behavior pattern between the 1st, 2nd stranger at ≤ 50 m in 8% of the trials, Pyrenees and Reindeer and 3rd meeting with same stranger at 51Ð100 m 67% of the trials, and at Ten of 11 dogs chased reindeer type (P>0.05). Therefore, the 3 repe- ≥ 100m (200m max.) in 25% of the (Table 2), and all were interested in titions for each dog were pooled (df occasions. In 62% of the trials, the the unfamiliar scent. However, none = 64). dogs did not seek physical contact of the dogs achieved physical contact Of the 5 stranger types, unfamiliar with the stranger at all (cont.time = with the reindeer. One dog was afraid dogs elicited the most offensive max = 300sec.). In 15% of the times of the flock. This dog also displayed behavior pattern regarding behav.1 the contact time was <30 seconds. fear towards the cattle in Trial I. and behav.2 (Table 1). They also trig- Dogs that showed the most offen- Our observer had to intervene and gered the fastest contact time (Fig. 1) sive behavior when the stranger stop 2 dogs during the testing. The and the least minimum distance (Fig. entered the arena (behav.1) also were first had chased a reindeer for more 1). In addition, the strange dogs the most offensive when the stranger than 1 km, and the second chased a induced the Pyrenees to urinate more left (behav.2, rs=0.78). The faster the female out into the sea. Most often, often and to follow them out of the Pyrenees approached to investigate singles or small family groups were pasture more often. The contact time the stranger (cont.time), the closer chased. The reindeer had no problems and minimum distance between the contact they established (min.dist., rs outrunning the dogs, and the reindeer other stranger types were not signifi- = 0.86). No significant differences in were often found in the same area cantly different (Fig. 1). The Pyrenees behav.1, behav.2, cont.time or immediately after being chased. Panic barked most frequently at cattle and min.dist between 1-year-old and 2- among reindeer was observed only least often at sheep (15% versus 3%). year-old dogs were found. during the 2 occasions described above.

Table 2. Distance to the reindeer at the moment the dog discovered it (discovery distance), the Trial III. Interactions between closest distance between dog and reindeer (minimum distance) and chasing time. Pyrenees and Bear When first released and after Mean Minimum Maximum smelling the bear, the dogs ran about Discovery distance(m) 86 10 300 rather unsystematically, trying to locate Minimum distance(m) 59 5 100 Chasing time(sec) 176 60 400 the bear. The bear started moving in a circle around the dogs and observers at 60Ð100 m away, but out of sight. Ten minutes after the last dog was released,

4 JOURNAL OF RANGE MANAGEMENT 52(1), January 1999 they finally began chasing the bear. The chase lasted approximately 25 minutes, by which time the bear moved about 1 km away. However, radio-tracking revealed that the bear was already on its way back to the starting-point one hour later. Of the 3 dogs, the female was consistently the most offensive and remained closest to the bear, whereas male B returned to the observers twice. The dogs’ response towards the bear was modest compared with other breeds bred for bear hunting as the , Karelsk bjørnehund or Jämthund (S¿rensen, pers. comm.). None of the dogs were injured during the test.

Discussion

The Pyrenees tested showed no Fig. 1. Time from when the dog discovered the “stranger” until it sought physical contact aggressive behavior towards unfamil- with the unfamiliar being (contact time) and minimum distance between dog and iar people and only slight aggressive- “stranger” after discovery (mean ± SE). ness towards dogs and livestock. These findings agree partly with char- In Norway, sheep and reindeer graz- returned to their flock (the observers) acteristics of the breed documented by ing areas often overlap. Reindeer relatively soon after the encounter, Green and Woodruff (1983), who farmers are, in general, opposed to the which is typical for canine guardians found that 4% of 437 Pyrenees use of dogs since stray dogs often (Green and Woodruff 1990). injured people, 7% injured sheep and chase their reindeer. More than 90% However, the bear returned soon 67% were aggressive to dogs. These (10 dogs) of the Pyrenees that partici- thereafter. This observation corre- results are of course dependent on the pated in this trial chased reindeer; sponds to a very similar trial in Pasvik quality and the strength of the social- thus the use of guarding dogs in rein- (Norway) during the summer of 1994, ization to other species during rearing. deer grazing areas could result in con- where 3 other Pyrenees were allowed Further, the dogs in this study did not flicts. Indeed, the mere presence of a to chase bears (Wikan 1996). A bear show any offensive protective behav- dog can be enough to cause reindeer needed at least 4Ð5 encounters with ior, i.e. they did not seem to be highly to avoid an area. This was observed dogs before it finally left the area. protective of the sheep. A stronger during another field trial, where dogs This is also in accordance with the bonding to the sheep and a longer barked frequently at night (Hansen experiences of encounters between habituation period inside the test area 1996). The problem should be dis- livestock-guarding dogs and grizzlies would probably have resulted in more cussed with the Reindeer Farmers’ (Green and Woodruff 1989). The 2 aggressive behavior towards the Association. Perhaps, through cooper- Norwegian trials show that the will- intruders. ative efforts, methods for preventing ingness to chase bears is latent in all 6 When the sheep that were unfamil- dogs from chasing reindeer could be Pyrenees tested thus far. iar with dogs discovered the Pyrenees, found. One solution to this problem The behavior displayed towards they immediately turned and ran out could be aversive conditioning of the bears is also applicable toward lynx of the arena. This flight behavior trig- dogs using electric shock collars. A and wolverines. In another study gered a chase by some of the youngest better solution might be to expose (Staaland et al. 1998) we documented dogs. Nevertheless, none of the 40 pups regularly to reindeer. Because that the female from the bear-dog grazing sheep that were familiar with reindeer are also subject to severe pre- interaction chased a wolverine in dogs were ever chased. These dation (20% of the flock can be much the same way that she chased episodes indicate that there could be preyed upon annually (Kvam et al. the bear. Furthermore, losses of lambs difficulties involved in using guarding 1995a)), one could even try using due to lynx were reduced in the pres- dogs on open rangeland where differ- dogs to guard reindeer herds. This ence of guarding dogs (Hansen et al. ent herds graze together, as seen else- might help change the attitude of rein- 1997). Due to the strict legislations where (Green and Woodruff 1990; deer farmers towards guarding dogs. for dog-keeping, predation by stray Lorenz and Coppinger 1986). This The dogs proved their bear-chasing dogs is not a large problem in problem will be focused further in capacity. They chased the bear away Norway. This study shows, however, Part II. from the close surroundings and that the Pyrenees may be less effec-

JOURNAL OF RANGE MANAGEMENT 52(1), January 1999 5 tive against stray dogs. Under Green, J.S. and R.A. Woodruff. 1983. The Statistisk sentralbyrå (SSB). 1995. NOS Norwegian conditions this could be use of three breeds of dog to protect range- Jordbruksstatistikk 1994. best, since hikers might walk their land sheep from predators. Appl. Anim. Viltloven. 1981. Lov av 29. mai, nr. 38, Kap. Ethol. 11:141Ð161. XI, ¤ 52. dogs through the sheep herding area. Green, J.S. and R.A. Woodruff. 1988. Livestock-guarding dogs have behav- Wikan, S. 1996. Bruk av pyreneerhunder mot Breed comparisons and characteristics of bj¿rn. Erfaringer fra Pasvik 1994. Svanhovd ioral characteristics that distinguish use of livestock guarding dogs. J. Range milj¿senter. Rapport No. 23. 25 p. them from other breeds. They are sub- Manage. 41:249Ð250. missive and show no predatory behav- Green, J.S. and R.A. Woodruff. 1989. ior towards livestock; they are strongly Livestock-guarding dogs reduce depreda- bonded to their flock-mates whom they tion by bears. Bear-People Conflicts. Proc. of Symp. on Management Strategies. will protect; and they show ambivalent North-West Territories, Dept. of Renew. behavior (barking and approach/with- Res. drawal) rather than aggression when Green, J.S. and R.A. Woodruff. 1990. something novel approaches (Lorenz Livestock guarding dogs: Protecting sheep and Coppinger 1986). The result is a from predators. USDA Agr. Info. Bull. “preventive” defense, usually without No. 588. 32 p. physical contact between dog and Hansen, I. 1996. Bruk av vokterhund som vern mot rovdyr i beiteområder for sau. Planteforsk predator. Because the livestock-guard- Tj¿tta fagsenter. Sluttrapport. 26 p. ing dogs’ way of working is more sub- Hansen, I., A. Rings¿ and T. Staaland. tle than that of the bear-hunting dogs, 1997. Bruk av vokterhund som vern mot which are generally bred to be rovdyr i beiteområder for sau. Erfaringer extremely eager hunters, the guarding fra feltfors¿k i Hattfjelldal. Planteforsk dogs’ working capacity is often misin- Tj¿tta fagsenter, Rapport No. 7/97. 21 p. Kvam, T., K. Nybakk, K. Overskaug, O.J. terpreted. S¿rensen, and K. Br¿ndbo. 1995a. Gaupa tar mye mer rein enn antatt. Reindriftsnytt (4), 40Ð43. Conclusions Kvam, T., O.J. S¿rensen, T. Eggen, K. Knutsen, K. Overskaug, F. Berntsen, Although the Great Pyrenees in and J.E. Swenson. 1995b. Årsrapport fra Norway are bred mainly for exhibition, Rovdyrprosjektene i Nord-Tr¿ndelag 1994. NINA Oppdragsmelding No. 364. they have obviously maintained some 37 p. of the behavioral patterns important for Kvam, T., O.J. S¿rensen, K. Overskaug, their use as livestock-guarding dogs. T. Eggen, F. Berntsen and J.E. Based on their calm and nonaggressive Swenson. 1996. Årsrapport fra behavior towards people, dogs and Rovdyrprosjektene i Nord-Tr¿ndelag livestock, and their conservative (i.e., 1995. NINA Oppdragsmelding No. 424. 40 p. minimal physical contact) way of chas- Lorenz, J. and L. Coppinger. 1986. ing predators, we believe the Great Raising and training a livestock-guarding Pyrenees breed has a good potential for dog. Ore. St. Univ. Exten. Circ. 1238. use as livestock guardians in Norway. Lorenz, J., R. Coppinger, and M.R. Still, there may be other breeds, yet to Sutherland. 1986. Causes and economic be tested, that are as good or better. The effects of mortality in livestock guarding problems with dogs chasing reindeer dogs. J. Range Manage. 39:293Ð295. Lov om friluftslivet. 1957. Lov av 28. juni, must be solved before guarding dogs nr. 16, ¤ 2. are used on rangeland used by both Lov om ansvar for skade på bufe ved sheep and reindeer. hund m.v. 1926. Lov av 9. juni, nr. 4, ¤ 3. Mysterud, I., J.T. Warren, T. Lynnebakken. 1994. Tap av sau i Målselv Literature Cited 1993. Sau og geit 1/94, 66Ð70. Reindriftsloven. 1978. Lov av 9. juni, nr. Andelt, W.F. 1992. Effectiveness of live- 49, Kap. VII, ¤ 29. stock guarding dogs for reducing predation SAS. 1987. SAS/STAT Guide for Personal on domestic sheep. Wildl. Soc. Bull. Computers, 6 Ed. Stat. Anal. Sys. Inst., 20:55Ð62. Inc. Cary, N.C. Coppinger, R., L. Coppinger, G. Staaland, T., A.J. Rings¿, and I. Hansen. Langeloh, L. Gettler, and J. Lorenz. 1998. Vokterhund som forebyggende tiltak 1988. A decade of use of livestock guard- mot rovdyrskader. Proc. Husdyrfor- ing dogs. Univ. of Calif., Davis. Proc. søksmøtet, Norges landbruksøgskole, Ås, Vertebr. Pest. Conf. 3:209Ð214. Febr. 10.Ð11. 1998. 453Ð457.

6 JOURNAL OF RANGE MANAGEMENT 52(1), January 1999