Applied Animal Behaviour Science 119 (2009) 210–218
Contents lists available at ScienceDirect
Applied Animal Behaviour Science journal homepage: www.elsevier.com/locate/applanim
Ethiopian village dogs: Behavioural responses to a stranger’s approach Alessia Ortolani a,*, Hans Vernooij b, Raymond Coppinger c a
Department of Animal and Human Biology, University of Rome ‘‘La Sapienza’’, Viale dell’Universita’ 32, Roma 00185, Italy Department of Farm Animal Health, Veterinary Faculty, Utrecht University, Marburglaan 2, NL-3584CN Utrecht, The Netherlands c School of Cognitive Sciences, Hampshire College, Amherst, MA 01002, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Accepted 9 March 2009 Available online 17 May 2009
We studied the behavioural ecological characteristics of free-roaming dogs (Canis familiaris) in four Ethiopian villages via observational surveys. The Ethiopian village dogs surveyed in this study have similar characteristics to other free-roaming dog populations in the world: (1) they are almost entirely unrestrained, (2) there is a male-skewed sex ratio in the adult population, (3) the majority appear not to be owned (or at least not declared so) and (4) many of them are found around people’s dwellings. We also evaluated ‘village’ dogs’ attitudes towards people using a stranger-approach test. Eighty-five dogs were approached by an unfamiliar Caucasian woman in different contexts and the dogs’ vocal behaviour before the approach, reaction to the approach and approaching distance between dogs and observer were systematically recorded. We found variability in village dog behavioural responses to a stranger depending on dogs’ location (inside house, outside house, street), social setting (alone, pair, group) and village. Dogs inside homes and dogs that were alone were more likely to vocalize towards the observer than dogs in the street. Avoiding was the most common reaction exhibited by village dogs, especially when approached in the streets, suggesting that most of them are shy of people. However, 11% responded aggressively towards the observer; all of these dogs vocalized towards the observer before she approached them. One third of the dogs maintained a neutral attitude apparently ignoring the observer. Slightly less than one third of village dogs could be approached closely (0–2 m), one third could be approached at a medium distance (2–5 m), and over one third could not be approached at all (>5 m). Altogether, our findings showed that the dog populations in the villages we studied were not behaviourally homogeneous, within and between villages. Our study suggests that behavioural investigations are useful for characterizing free-roaming/village dogs’ attitudes, especially when assessing any risk dogs pose to the human population. Our observations suggest that people might have greater chances of being bitten by dogs that: vocalize towards people, are alone or in pairs, and are found inside households. Our ‘‘approach test’’ proved to be a simple, effective way of measuring dog attitudes, which could be easily implemented in different parts of the world by training local people. Quantifying behavioural responses of village dogs towards people would be particularly important in areas of the world where potentially fatal dog-transmitted diseases, such as rabies, are still prevalent. ß 2009 Elsevier B.V. All rights reserved.
Keywords: Dog Free-roaming Dog-human relationship Barking Avoiding Aggression
1. Introduction * Corresponding author. Present address: Egelinglaan 36, 3705TD Zeist, The Netherlands. Tel.: +31 30 6961055; fax: +31 30 6961055. E-mail address:
[email protected] (A. Ortolani). 0168-1591/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.applanim.2009.03.011
In 1988, the World Health Organization (WHO) classified dogs (Canis familiaris) in four categories according to their dependence (or lack of it) on humans
A. Ortolani et al. / Applied Animal Behaviour Science 119 (2009) 210–218
(WHO, 1988; Perry, 1993): (a) restricted dog: fully dependent and fully restricted by humans, (b) family dog: fully dependent and semi-restricted, (c) neighbourhood dog: semi-dependent and either semi-restricted or unrestricted, and (d) feral dog: independent and unrestricted. The term ‘‘stray dog’’ was removed from official vocabulary because it was considered misleading, and WHO recommends using it only for dogs ‘‘not in compliance with local regulations,’’ such as lost, abandoned or free-roaming dogs. The intent of this classification was to help fight the spread of canine transmitted rabies in humans by targeting dogs most at risk of carrying the disease. Classifying dogs in terms of their dependence on humans might be a uniquely western concept where it is generally assumed that all dogs are (or have been) owned by humans, and that they are either pets, working dogs, or strays that have escaped from human restrictions. In many parts of the world, however, such conditions do not exist or may not have yet developed. In fact, far from being uncommon, as many as 99% of dogs surveyed are free-roaming (Butler and Bingham, 2000). There are different kinds of free-roaming dogs in some aspect of their behaviour (see Boitani et al., 1995, 2005). Dogs that freely associate with humans in and/or around their dwellings have been called ‘‘village dogs’’ (Macdonald and Carr, 1995; Coppinger and Coppinger, 2001) and some of these would belong to WHO’s category of ‘‘neighbourhood dogs.’’ Village dogs around the world all appear to share similar characteristics: they live mostly unrestrained; they may or may not have owners; their adult sex ratio is usually skewed towards males; many are solitary (except during oestrous gatherings or around food sources) but pairs or trios are not uncommon; they are mainly active during the early hours of the morning and in the evening; they feed mainly on human refuse and garbage; and they use buildings, porches, vehicles or other human-built structures for shelter (Beck, 1975; Daniels, 1983; Berman and Dunbar, 1983; Boitani and Racana, 1984; WHO, 1988; Daniels and Beckoff, 1989; Brooks, 1990; De Balogh et al., 1993; Macdonald and Carr, 1995; Pal et al., 1998a,b, 1999; Butler and Bingham, 2000; Coppinger and Coppinger, 2001). Population densities of free-roaming/village dogs in urban areas, where recorded, range from 68 dogs/km2 in Zimbabwe (Brooks, 1990) to 936 dogs/km2 in Mexico (Daniels and Beckoff, 1989), but can also reach peaks of 1792 dogs/km2 and even 3700 dogs/km2, as found in Ecuador and Sri Lanka, respectively (WHO, 1988). Although data are limited, we believe that village dogs may be the most abundant dog category in underdeveloped areas of the world. We studied the behavioural characteristics of village dogs in Ethiopia, which has a large population of freeroaming dogs. Many Ethiopian villages still subsist on a predominantly rural economy where livestock ownership is the norm, and large predators, such as Ethiopian wolves (Canis simensis) and spotted hyenas (Crocuta crocuta), still threaten the livestock in many areas. The combination of these factors creates an ideal setting to study the interaction between dogs and people. Dogs may be of value to people in these areas if they help protect their livestock from wild predators and perhaps even human
211
thieves. On the other hand, village dogs still pose a threat to the health of the human population in Ethiopia, since canine rabies is widespread (Sillero-Zubiri et al., 1996; Laurenson et al., 1998) and dog bites can result in human fatalities (WHO, 2000). The goals of this study were to gain information about the behavioural ecological characteristics of Ethiopian village dogs via direct surveys, as well as to evaluate the dogs’ attitudes towards people via a stranger-approach test. We evaluated three aspects of village dogs’ behaviour: (1) their initial behaviour upon encountering a stranger, (2) their reaction to the stranger’s approach, and (3) their approachability (i.e. shortest reachable distance between stranger and dog(s)). 2. Material and methods 2.1. Study sites The study was conducted between April and July 2002 in the following Ethiopian villages: Dinsho (latitude 39.758N, longitude 6.758E): Oromia district; elevation: 3000 m; population 28,358; religion predominantly Muslim; predominant agriculture: barley; predominant livestock encountered in village: cattle (average 17/house, Flintan, 2000), goats, sheep, horses, poultry; wild predators encountered around village: Ethiopian wolf (C. simensis), common jackal (Canis aureus). Dinsho lies at the foot of the Bale Mountains National Park. The whole village stretches along one main unpaved road where the survey was conducted. Debre Zeyit (latitude 8.758N, longitude 38.998E): Oromia district; elevation 1900 m; population 95,600; predominant religion: orthodox Christian; predominant livestock encountered in village: horses, donkeys, cows, sheep, goats, chickens; wild predators encountered around village: spotted hyena (C. crocuta). Debre Zeyit is a small commercial town, which lies 50 km from Addis Ababa. Several commercial chicken farms are present in the area and the town’s major dump is regularly supplied with chicken remains, constituting a major attraction for the village’s dogs and hyenas. Since the main road of the village is also the main paved road to Addis Ababa, the survey was conducted on the northern parallel residential unpaved road leading to Lake Chelekleka. Harar (latitude 9.318N, longitude 42.138E): Harari district; elevation 1800 m; population 99,500; predominant religion: Muslim; predominant agriculture: coffee, grain, chat (Cata adulis), fruits, nuts; predominant livestock encountered in village: horses, cows, sheep, goats, chickens; wild predators encountered around village: spotted hyenas (C. crocuta). Harar is a 15th century walled town considered by some Muslims to be the fourth holiest town in the world. It is an important economic and trading centre especially for coffee and chat. Four surveys were conducted in this village: two outside along the town’s walls and two inside the walls along the main street crossing the town.
212
A. Ortolani et al. / Applied Animal Behaviour Science 119 (2009) 210–218
Gonder or Gondar (latitude 12.618N, longitude 34.468E): Amhara district; elevation 2225 m; population 146,300; predominant religion: orthodox Christian; predominant agriculture: teff, maize, barley, finger millet, nug, chick peas and coffee; predominant livestock encountered: horses, cows, sheep, goats, chickens; wild predators encountered: none. Gonder is an ancient medieval town situated at the foot of the Simien Mountains. Today it is a major regional trade and tourist centre. Most of the town main roads are asphalted.
approaching it (n.b. the flight distance from the observer was recorded), and (c) the dog attempted to attack the observer. Aggressive dogs were defined as individuals who bared their teeth at the observer for more than 3 s, moved towards the observer snapping their jaws, and/or attempted to bite her. Although some vocalizations, such as ‘‘bark-growling’’, might have also been displayed by aggressive dogs, they were not included in the definition of aggressive behaviour (see also Section 3). 2.3. Statistical analyses
2.2. Subjects and behavioural observations A total of 85 dogs (54 M, 26 F, 5?) were surveyed during the study. Surveys were conducted opportunistically at different times of the day and night along a 1 km transect along the main road of each village. The transect length represented approximately the span of the main village road (with the exception of Gondar’s which was approx. 1.5 km). The main road was chosen as it provided a crosssection of the village passing through its centre. Transects were defined by map (Briggs, 2000) and were traversed starting from the periphery of the village going towards its centre and ending at the opposite periphery. Whenever possible (e.g. depending on the observers’ safety conditions), repeated surveys were conducted in the same village (i.e. Debre Zeyit, N = 1; Gondar, N = 1; Dinsho, N = 3; and Harar, N = 4). All observations were made on foot by two observers (AO and an assistant). Any dog encountered within an area of 50 m of the observers (including dogs inside people’s homes or within fenced property) was identified by a combination of physical characteristics (e.g. coat colour pattern, body size) and photographed with a Pentax ME Super 35 mm automatic photo camera equipped with a 70–210 mm lens. This allowed us to avoid double counting individuals in repeated surveys (Beck, 1975). A focal animal was then chosen opportunistically and, from a distance of 10–15 m, the following behavioural ecological variables were recorded: sex, age, body size, social condition, body condition, female reproductive condition, coat colour, presence of owner, location, activity and behaviour before the approach (see Table 1 for variable definitions). The behaviour before approach category recorded the dog’s behavioural state as soon as it was seen by the observer, usually from a distance of >20 m; although, some dogs may already have been reacting to the observer’s presence from a distance, no attempt was made to approach the dog at this time and the observer was standing still recording data. Photographed dogs that left the area before data could be collected were excluded from the study. After recording the above variables, one of us (AO) approached the dog, starting from a distance of 10–15 m, while a second observer recorded the dog’s reaction to the approach and the approach distance (Table 1). Approaches were made by walking slowly in a direct line towards the individual avoiding any eye contact with it and keeping a relaxed normal straight posture with the arms by the sides. The approach finished when one of the following conditions occurred: (a) the observer reached the dog by standing within 20 cm of it, (b) the dog fled while the observer was
All statistical analyses for Behaviour before the approach, Reaction to the approach and Approaching distance (Section 2.3.3) were performed using the statistical software SPSS1 version 15.0 (SPSS, Inc., Chicago, IL) with a significance level of 0.05. 2.3.1. Behaviour before the approach (hence Vocal behaviour) The categories barking, growling and bark-growling were combined to create the subcategory ‘‘vocal behaviour’’, and the categories neutral, tail-wagging and sleeping were combined to create the subcategory ‘‘non-vocal behaviour.’’ A logistic regression (LR) model was performed with the explanatory variables Location (in house, out house, street) and Social condition (alone, pair, group) as predictors of dogs’ vocal behaviour (vocal/non-vocal). The Hosmer and Lemeshow test was used to check the goodness of fit of the final model. 2.3.2. Reaction to the approach For statistical analyses, the neutral and approach categories were combined in neutral/approach since only a few dogs approached the observer. A multinomial logistic regression (MLR) model was performed with dog’s reaction to the approach (avoid, aggressive or neutral/approach) as the outcome. In the first MLR model, Vocal behaviour, Location, Village and Social condition were all entered as independent variables. However, Vocal behaviour invalidated the MLR model since all aggressive dogs vocalized towards the observer before she approached them, and thus was dropped from the model and tested separately with a Pearson’s x2 (Siegel and Castellan, 1988). Village was also dropped from the MLR model because the validity of the model fit was uncertain. In the final MLR model, Location (outside house/street, inside house) and Social condition (alone, pair, group) were entered as predictors of dog’s reaction to the approach. The best model was chosen based on the Likelihood Ratio Test and the Pearson’s Goodness of Fit Test was used to check the validity of the MLR model. 2.3.3. Approaching distance A multinomial logistic regression model was performed for Approaching distance with the distance (0–2, 2–5 and >5 m) as the outcome. In the first MLR model Vocal behaviour, Location, Village and Reaction to the approach were all entered as independent variables. However, the validity of the model fit was uncertain so Reaction to the approach and Location were dropped out of the final model. The Pearson’s Goodness of Fit Test was used to check the
A. Ortolani et al. / Applied Animal Behaviour Science 119 (2009) 210–218 Table 1 Ethiopian village dogs survey results, actual numbers and (%) of dogs shown unless otherwise stated (abbreviations: M = males, F = females, mo = months, yr = year, sh = shoulder height). Total number of dogs surveyed per village Debre Zeyit Dinsho Gondar Harar Total
24 46 6 9 85
Average number of dogs per survey Debre Zeyit Dinsho Gondar Harar
24.0 15.3 6.0 2.5
Average number of dogs per time of survey AM (10:30–12:00) PM (16:30–19:00) Evening (20:00–21:30)
10.0 9.6 8.0
Sex ratio Debre Zeyt Dinsho Gondar Harar
(M:F) 2:1 2.4:1 2:1 1.25:1
Overall
2.08:1
Age Pup (l yr)
2 (2.4%) 19 (22.9%) 62 (74.7%)
Body size Small (sh 65 cm)
6 (7.2%) 77 (92.8%) 0
Social condition Alone Pair Group (3)
41 (49.4%) 16 (19.3%) 26 (31.3%)
Body condition Fair Poor
72 (96%) 3 (4%)
Female condition Lactating Non-lactating
4 (15%) 22 (85%)
Coat color Tan Black and white Tan and white Black and tan Black, white and tan Black Yellow White
24 (31.6%) 19 (25%) 16 (21.1%) 8 (10.5%) 3 (3.9%) 3 (3.9%) 2 (2.6%) 1 (1.3%)
With owner Yes No
17 (20.5%) 66 (79.5%)
Location Inside house Outside house Street
19 (22.4%) 31 (36.5%) 35 (41.1%)
213
Table 1 (Continued ) Activity Laying Walking Restricted
43 (51.8%) 39 (47%) 1 (1.2%)
Behaviour before approach Neutral Barking Growling Bark-growling Sleeping Tail-wagging
49 (59) 10 (12%) 5 (6%) 3 (3.6%) 11 (13.3%) 5 (6%)
Reaction to approach Avoid Neutral Aggressive Approach
43 (51.8%) 28 (33.7%) 9 (10.8%) 3 (3.6%)
Approaching distance 0–2 m 2–5 m >5 m
25 (30.5%) 27 (33%) 30 (36.5%)
validity of the MLR model and the best model was chosen based on the Likelihood Ratio Test. In the final MLR model, Vocal behaviour (vocal, non-vocal) and Village (Dinsho, other villages) were entered as predictors of Approaching distance. Note: the GOF of the model of the approaching distance could not be calculated when Village was left out. This is because the remaining variable Vocal behaviour predicts exactly (perfect fit) the number for each approaching distance class. Finally, it is worth noting that the MLR tests the likelihood that Approaching distance is preceded by Vocal behaviour, not their casual relationship. 3. Results The results of the village dogs’ survey conducted in four Ethiopian towns are presented in Table 1. The overall sex ratio was 2:1 males to females. Three-quarters of the dogs sampled were adults estimated at being over 1 year of age. Most village dogs fell in an intermediate size category (30– 65 cm at the shoulder) with the exception of a few smaller individuals. Very few village dogs were found to be in poor body condition. Colour patterns were variable with tan being predominant, followed closely by black-and-white and tan-and-white combinations. All village dogs except one were unrestrained, and the majority did not appear to have an owner. We asked villagers if they owned the dogs that were inside a house or lying just outside it. 35.3% of the dogs lying inside the house were claimed to be owned and 29% of the dogs outside the house were claimed to be owned. For dogs encountered in the streets, it was not always possible to ask people if they owned them because dogs were often alone. About half of the dogs were alone, while groups (3) were encountered more often than pairs. Dogs were mainly seen in the streets or just outside households, while only a minority was observed inside people’s homes. When first detected, village dogs were resting, lying down either sleeping or alert, or moving
214
A. Ortolani et al. / Applied Animal Behaviour Science 119 (2009) 210–218
around the village, either meandering or walking in a precise direction. More than 70% of the dogs initially ignored us and this appeared to be their main attitude towards villagers as well. About one fourth of the dogs reacted to our presence by vocalizing, while only 6% were friendly towards us. The most common reaction to the observer’s approach was to avoid her by fleeing, while 11% showed an aggressive response towards the observer. Dogs’ approaching distances were distributed about equally: slightly less than one third could be approached closely (0–2 m), one third could be approached at a medium distance (2–5 m), and over one third could not be approached at all (>5 m). The dogs’ behaviour before the approach (Section 3.1), their reaction to the observer’s approach (Section 3.2) and their approaching distance (Section 3.3) are analyzed in more detail below. 3.1. Behaviour before the approach: vocal versus non-vocal Location was found to be a significant predictor of dogs’ Vocal behaviour (i.e. barking, growling and bark-growling) displayed towards the observer before the approach (Fig. 1). Social condition was only a marginally significant predictor of Vocal behaviour (LRT: x2 = 5.4, d.f. = 2, P = 0.068). Location: dogs inside a house were 18 times more likely to vocalize towards the observer than dogs in the street, while dogs outside a house were 5.5 times more likely to vocalize towards the observer than dogs in the street (Table 2). Social condition: dogs alone were five times more likely to vocalize towards the observer than dogs in groups (3), while dogs in pairs had also higher odds of vocalizing with respect to dogs in groups but the effect was not significant (Table 2).
Table 2 The table below shows the results from a logistic regression of Behaviour before the approach (i.e. vocal vs. non-vocal) explained by dogs’ Location and Social condition (LRT: X2 = 15.08, d.f. = 4, N = 83, P = 0.005). The reference category is: non-vocal. Predictor
OR
Behaviour: vocal Social = alone Social = pair Social = group Location = in house Location = out house Location = street
95.0% C.I. for OR Lower
Upper
5.0 3.4 1.0
1.2 0.5
21.0 21.6
17.8 5.5 1.0
3.3 1.2
94.7 25.5
included in the multivariate logistic regression because it invalidated the model (see Section 2.3.2). The numbers of dogs with different vocal behaviour scores were very unbalanced in the multidimensional table since all aggressive dogs vocalized towards the observer before she approached them. 3.2.1. Avoiding reaction Social condition: dogs alone were seven times more likely to avoid the observer’s approach than dogs in groups (i.e. 3); dogs in pairs were 9.5 times more likely to avoid the observer’s approach than dogs in groups (Table 3).
3.2. Reaction to the approach: avoid, aggressive, neutral Social condition and location (Fig. 2) were both found to be significant predictors of dogs’ reaction to the observer’s approach. Vocal behaviour was also significantly associated with Reaction to the approach (Pearson’s x2 = 38.96, d.f. = 2, P < 0.0001; Fig. 1); however, this variable could not be
Fig. 2. Dogs’ reaction to the approach by dogs’ location (N = 83, LRT: x2 = 10.993, d.f. = 2, P = 0.004) and by social condition (N = 83, LRT: x2 = 15.269, d.f. = 4, P = 0.004). Table 3 The table below shows the results from a logistic regression of Reaction to the approach explained by dogs’ Location and Social condition (LRT: 2 X = 27,955, d.f. = 6, N = 85, P = 0.000). The reference category is: neutral/ approach. Predictor
Reaction: avoid Social = alone Social = pair Social = group Location = out/street Location = in house
Fig. 1. Dogs’ vocal behaviour before the approach by location where dog was encountered (N = 85, LRT: x2 = 13.7, d.f. = 2, P = 0.001) and by reaction to the observer’s approach (N = 85, Pearson’s x2 = 38.96, d.f. = 2, P < 0.0001); vocal = barking, growling, bark-growling.
Reaction: aggressive Social = alone Social = pair Social = group Location = out/street Location = in house
OR
95% C.I. for OR Lower bound
Upper bound
7.1 9.5 1.0
2.1 1.9
23.7 47.2
4.2 1.0
0.9
19.4
5.2 7.8 1.0
0.8 0.7
34.3 87.3
0.3 1.0
0.1
1.3
A. Ortolani et al. / Applied Animal Behaviour Science 119 (2009) 210–218
Location: dogs in the street or outside homes were four times more likely to avoid the observer’s approach than dogs inside a house, although this result was only marginally significant (Table 3). 3.2.2. Aggressive reaction The overall results for Location and Social condition as predictors of dogs’ aggressive reactions were only marginally significant, probably due to the fact that, in general, very few aggressive dogs were encountered in each category. Social condition: dogs alone or in pairs were more likely to display an aggressive reaction towards the observer than dogs in groups, but these results were only marginally significant (Table 3). Location: there is a trend showing that dogs in the street or outside homes were less likely to be aggressive than dogs inside homes (Table 3). Vocal behaviour: 19% of avoiding dogs vocalized towards the observer versus 3% of neutral dogs. All aggressive dogs (N = 9) vocalized towards the observer before she approached them (Fig. 1). However, only 50% of all vocalizing dogs displayed an aggressive reaction. Among ‘barking dogs’: 7/10 were aggressive, 3/10 avoiding; among ‘bark-growling’ dogs: 2/3 were aggressive, 1/3 avoiding; and among ‘growling dogs’: 4/5 were avoiding, 1/ 5 neutral.
215
Table 4 The table below shows the results from a logistic regression of Approaching distance explained by Behaviour before approach and Village (LRT: X2 = 45.76, d.f. = 4, N = 85, P = 0.000). The reference category is: 0– 2 m; beh = behaviour before the approach, HDZG = Harar, Debre Zeyit, Gondar. Predictor
OR
95% C.I. for OR Lower bound
Upper bound
Distance: >5 m beh = vocal beh = non-vocal
12.5 1.0
1.7
93.8
Village = Dinsho Village = HDZG
80.3 1.0
12.1
530.6
Distance: 2–5 m beh = vocal beh = non-vocal
1.3 1.0
0.2
7.5
Village = Dinsho Village = HDZG
5.2 1.0
1.4
19.5
Note: The goodness of fit for this LR model was borderline (GOF: x2 = 6.1, P = 0.048) due to the low numbers in some cells of the multidimensional table. When ‘‘village’’ was left out of the model the GOF could not be calculated and the OR for vocal behaviour in the distance category ‘‘>5 m’’ were diminished by more than half (i.e. OR = 5.1, P = 0.023).
were also less approachable in the intermediate approaching distance (i.e. 2–5 m; OR = 5.2) although it was not significant (Table 4).
3.3. Approachability of dogs: close, medium, far
4. Discussion
Dogs’ Vocal behaviour before the approach (Fig. 3) and Village (LRT: X2 = 36.77, d.f. = 2, P = 0.000) were both found to be significant predictors of how close the observer could approach the dog(s). Reaction to the approach was also significantly associated with Approaching distance (Pearson’s x2 = 26.77, d.f. = 4, P < 0.0001) but it could not be included in the multivariate logistic regression because it invalidated the model (see Section 2.3.3). Vocal behaviour: dogs that vocalized towards the observer were 12.5 times less approachable (i.e. >5 m) than non-vocalizing dogs (Table 4), even though vocal behaviour alone never caused the end of an approach (see Section 4). Village: dogs in the village of Dinsho were 80 times less approachable (i.e. >5 m) than dogs in other villages. Dogs
The Ethiopian village dogs surveyed in this study have similar characteristics to other free-roaming dog populations studied in different parts of the world: (1) they are almost entirely unrestrained, (2) there is a male-skewed sex ratio in the adult population, (3) the majority appear not to be owned (or at least not declared so), and (4) many of them are seen around people’s dwellings. Moreover, dogs were often observed feeding on garbage in the villages. These findings suggest that village dog populations around the world are fairly homogenous in their behavioural ecology. However, when we look at their behavioural attitudes towards strangers, or an unfamiliar passer-by, our findings suggest that different types of village dogs exist, and that the types often transcend those categories outlined by the WHO (1988) study. Ethiopian village dogs showed variability in their initial vocal behaviour, their reaction to the approach, and the approaching distance reached by an unfamiliar Caucasian female. All three behavioural variables were correlated with each other and had predictive value over each other: vocal behaviour predicted reaction to the approach by about 50% (lambda = 0.5), and, reaction to the approach predicted approaching distance by about 30% (lambda = 0.325). Approaching distance was also predicted by vocal behaviour. While the majority of dogs initially ignored the observer (i.e. AO), and indeed this appeared to be their general attitude towards villagers that were casually walking by (personal observation), over one quarter of the dogs immediately responded to the observer’s presence by vocalizing (i.e. barking, growling or bark-growling). When the observer attempted to
Fig. 3. Dogs’ vocal behaviour before the approach by distance to the dog reached by the observer after the approach (N = 83, LRT: X2 = 10.221, d.f. = 2, P = 0.006).
216
A. Ortolani et al. / Applied Animal Behaviour Science 119 (2009) 210–218
approach them about half of the vocalizing dogs fled, while the other half displayed an aggressive reaction (see aggression discussion below). Only one dog had a neutral reaction to the approach after initially vocalizing towards the observer. The function of vocalizations such as ‘barking’ in dogs has recently gained the interest of behavioural scientists (see Coppinger and Feinstein, 1991), with some authors finding that dogs’ barks vary in structure depending on context (Yin and McCowan, 2004), and others reporting that humans can assign different emotional values (i.e. aggressive or non-aggressive) to different dog barks (Pongra´cz et al., 2006). Our results show that village dogs were much more likely to bark if they were inside or just outside people’s households. Dogs were also much more likely to vocalize towards the observer when alone. Although we do not know whether dogs would have exhibited the same response towards local villagers—even though Caucasian females were not uncommon in any of the villages—it is nonetheless interesting that they manifested a clearly distinct response depending on their location. Village dogs might get food benefits by associating with particular households, as has been reported in other studies (Brooks, 1990; Macdonald and Carr, 1995; Butler and Bingham, 2000). In Ethiopia, one of the authors witnessed some household owners throwing the leftovers of a meal or butchering scraps to the dogs outside their homes. Barking in canids has been associated with territoriality (Lehner, 1978; Cohen and Fox, 1976) and if village dogs gain food from people, they might defend their homes from strangers. Although we did not record the village dog barks, since vocalizations per se were not a specific aim of this study, we observed that many dogs included growls in their barking (which were classified as ‘‘bark-growling’’). Interestingly, Yin and McCowan (2004), in their study of urban pet dogs in the USA, found that harsh low-frequency sounds, such as growls, are the main acoustic characteristic of barks elicited by a stranger ringing the doorbell. Altogether our findings in village dogs are not incompatible with the hypothesis that barking may have a ‘‘mobbing’’ function towards intruders, as recently proposed by Lord et al. (2009). In different areas of the world, two main reasons, other than companionship, have been consistently given by people for keeping dogs inside their homes: (1) to guard personal property and livestock from human burglars, and/ or (2) to guard livestock from wild predators (Beck, 1975; Daniels, 1983; Boitani and Racana, 1984; De Balogh et al., 1993; Brooks, 1990; Butler and Bingham, 2000; Wandeler et al., 1993; WHO, 1988). In Ethiopia, we observed many dogs standing, lying or sleeping just outside the main entrance of people’s homes, but, curiously, when we asked people in the house if they owned the dog(s), the majority of them (65%) denied it. In general, the villagers’ main attitude towards dogs was not positive, probably due to cultural and/or religious beliefs. Therefore, it was not surprising to find fewer dogs inside people’s dwellings than in the streets, but it was unexpected that they would be inside homes at all. Many households we visited also kept livestock such as poultry, goats, sheep and even cows inside the area where the people also slept and ate
(personal observation, but see also Flintan, 2000). We do not know whether Ethiopian village dogs are effective in providing protection against livestock theft or predation, but we believe it is possible that villagers may tolerate, or even encourage, the presence of dogs around their homes because of a ‘perceived’ protection of their livestock and/or other property. Indeed, one cattle owner near Dinsho with five large dogs roaming around his dwelling told us he kept them to protect the herd from Ethiopian wolves. In contrast to dogs associated with households, the majority of dogs that were encountered in the streets did not vocalize towards the observer and avoided her approach by fleeing. In general, avoiding was the most common reaction exhibited by village dogs, suggesting that most of these animals are shy of people. Moreover, dogs tended to avoid the approach especially when they were alone, suggesting that fear of humans might have played a role in explaining their reaction. Although dogs in the streets or outside households were more avoiding compared to dogs inside homes, the approaching distances among these groups were not significantly different. In fact, approaching distances were variable within each location and among different dog reactions (although avoiding and aggressive dogs were least approachable). Vocalizing dogs were least approachable, even though one third of vocalizing dogs could be approached within short distances (0–5 m). In their theory of dog domestication, Coppinger and Coppinger (2001) proposed that the earliest ‘‘dogs’’ exhibited variation in flight distances towards humans in the first Mesolithic villages. Dogs with the shortest flight distances, i.e. greater tolerance of humans, would have been able to fully exploit a new food source – the village’s garbage dump. In our study, the fact that individual dogs displayed different degrees of tolerance towards people, despite their predominantly shy response, may indicate that selection for tameness might be still at work in Ethiopian village dogs. On the other hand, dogs’ differential responses may also reflect their previous experiences with people. We encountered the highest proportion of avoiding dogs in the village of Dinsho, where dogs were also least approachable. Dinsho lies at the base of the Bale Mountains National Park, home to one of the last remaining populations of Ethiopian wolves. In an effort to protect wolves from rabies and from interbreeding with dogs, a team of researchers from the Ethiopian Wolf Conservation Programme made efforts to vaccinate and sterilize dogs in the Dinsho area some time prior to our visit (Laurenson et al., 1998; Williams, 2003). It is possible that these prior trapping attempts somehow conditioned the dog population to avoid (European) people. However, all of the males we encountered were intact and we do not know if any of the encountered dogs would have been alive at the time of the previous vaccination attempts. Although we have no evidence of this, it is also possible that the villagers’ cultural and religious beliefs, in this predominantly Muslim village (Flintan, 2000), could have influenced dogs’ behavioural responses towards humans. Aggressive reactions in Ethiopian village dogs were relatively uncommon—contrary to the belief that stray and free-roaming dogs must be aggressive to survive
A. Ortolani et al. / Applied Animal Behaviour Science 119 (2009) 210–218
(Cameron, 1997)—but when they did occur they were always preceded by vocal behaviour. Animal conflicts are often settled by elaborate displays, which may reveal the opponents’ true fighting ability thus avoiding lethal, aggressive encounters (Maynard Smith, 1974). Thus, village dogs may use vocalizations, such as barking, to warn their opponents before escalating to aggression. It is worth noting that barking did not prevent the observer from initiating an approach towards the dog(s), nor did it cause its termination. The aggressive behaviour observed in some dogs might, in fact, have resulted from the observer’s persistence despite the warning. The majority of aggressive reactions occurred inside households, although this result was not statistically significant. During the approaches, dogs were never cornered in any way; in fact when dogs reacted aggressively they did so before the observer even got a chance to enter the household. None of the aggressive reactions resulted in biting, even though some dogs made attempts. Several factors influence aggressive behaviour in (pet) dogs (see Houpt, 1983; Polsky, 1984). The most commonly diagnosed form of (pet) dog aggression towards humans is dominance aggression (Cameron, 1997), but other causes, such as play, fear, health related, protective, and redirected aggression, have been reported as well (Guy et al., 2001). In a recent study, Duffy et al. (2008) found that the most common types of aggression in pet dogs were inter-dog and stranger-directed aggression. Vas et al. (2005) compared the reactions of adult pet dogs in Hungary towards an unfamiliar woman approaching them showing either signs of friendliness or threat. Their results suggest that dogs’ reactions can be affected by human intentions. In our study, however, observer’s intentions could not have been a (important) factor in explaining the differences found in Ethiopian village dogs’ reactions to the approaches. In Vas et al.’s (2005) study, all approaches were started at a distance of 5 m from the dogs. During the friendly approach, the person used eye contact with the dog while talking to the dog; during the threatening approach, the person moved slowly and haltingly, looking steadily into the dog’s eyes but without verbally communicating. In this study: (1) the starting distance at which the observer began approaching the Ethiopian village dogs was 2–3 times greater, and more than one third of the approaches were already terminated by 5 m, (2) eye contact with the dog(s) was avoided and the observer was always wearing a baseball-type hat which covered the eyes, (3) the observer never talked during the approach, and (4) the observer strived to maintain a totally neutral attitude, movements and body position, while she approached the dog(s), as if she was just casually walking by. 5. Conclusions It is important to understand that even though the vast majority of (the world’s) dogs have an obligatory symbiotic relationship with humans, the structure of that symbiosis varies greatly. Altogether, our findings showed that the dog populations in the villages we studied were not behaviourally homogeneous, within and between villages. As a
217
result, classifying free-roaming/village dogs in terms of their dependency on and restriction by humans, as suggested by WHO (1988), may not be possible, nor useful, because ‘‘village dogs’’ might conceivably belong to all of WHO’s categories at any point in time. Moreover, the fact that just over 10% of village dogs showed an aggressive reaction towards a casual passer-by has important implications in a country such as Ethiopia where dogtransmitted rabies still causes human fatalities and where the vast majority of dogs is not vaccinated (WHO, 2000). Our study suggests that behavioural investigations are useful for characterizing free-roaming/village dogs’ attitudes, an important component when assessing any risk dogs pose to the human population. Our research in Ethiopian villages showed that most dogs are shy and prefer to avoid human contact. However, our observations also suggest that people might have greater chances of being bitten by dogs that: vocalize towards people, are alone or in pairs, and are found inside households. Because dog behaviour varied among villages, we cannot be sure that these findings would apply to other areas, but this further stresses the importance of behaviourally evaluating single dog populations before measures are applied. Our ‘‘approach test’’ proved to be a simple way of measuring dog attitudes, which could be easily implemented in different parts of the world by training local people to control for potential differences in dogs’ reactions (e.g. local vs. foreign, men vs. women). Moreover, all three behavioural variables used in our study (Vocal behaviour before the approach, Reaction to the approach and Approaching distance) were shown to have predictive value over each other (i.e. Vocal behaviour over Reaction and over Distance, and Reaction over Distance) and thus can provide useful measures for field researchers assessing village dog behaviour in different areas of the world. Although it may be possible to predict which dogs would be more likely to display aggressive reactions using this methodology, we cannot speculate on what causes the aggression. We believe the causes must be carefully researched in further ad hoc studies. While village dogs appear to thrive and are ubiquitous in many areas of the world, their symbiotic relationship with humans still remains poorly understood, especially in terms of the benefits, if any, that people might gain from sharing their niche with these canines. Further studies quantifying behavioural responses of village dogs towards people in different areas of the world are needed, as well as more refined measures to quantify people’s attitudes towards dogs in different cultures. Such studies would be particularly important in areas where potentially fatal dogtransmitted diseases, such as rabies, are still prevalent. We believe that free-roaming village dogs can be ecologically important populations and may hold a key to understanding how a predator/scavenger evolved into domestic dog. Acknowledgements Financial support for part of the field studies was provided by Hampshire College. We would like to thank: Jane Brackman for financial support, Roland Bronneberg
218
A. Ortolani et al. / Applied Animal Behaviour Science 119 (2009) 210–218
for invaluable assistance during the field study, Lorna Coppinger for editing the manuscript, The Ethiopian Wildlife Conservation Department for allowing the study, Gert van Putten for hospitality in Debre Zeyit, Stuart Williams and Claudio Sillero-Zubiri from the Wildlife Conservation Research Unit of the University of Oxford for facilitating the study in Dinsho, and all the Ethiopians for their hospitality, generosity and warm heartedness. References Beck, A.M., 1975. The ecology of ‘‘feral’’ and free roving dogs in Baltimore. In: Fox, M.W. (Ed.), The Wild Canids. Van Nostrand Reinhold Co., New York, pp. 380–390. Berman, M., Dunbar, I., 1983. The social behaviour of free-ranging suburban dogs. Appl. Anim. Ethol. 10, 5–17. Boitani, L., Racana, A., 1984. Indagine eco-etologica sulla popolazione di cani domestici e randagi di due comuni della Basilicata. Silva Lucana (Bari, Italy) 3, 1–86. Boitani, L., Francisci, F., Ciucci, P., Andreoli, G., 1995. Population biology and ecology of feral dogs in central Italy. In: Serpell, J. (Ed.), The Domestic Dog, Its Evolution, Behaviour and Interactions with People. Cambridge University Press, Cambridge, pp. 217–244. Boitani, L., Ciucci, P., Ortolani, A., 2005. Behaviour and social ecology of free-ranging dogs. In: Jensen, P. (Ed.), The Behavioural Biology of Dogs. CABI, Oxfordshire, pp. 147–165. Briggs, P., 2000. Guide to Ethiopia. Bradt Travel Guides, 2nd ed. (June 1, 1998), pp.354. Brooks, R., 1990. Survey of the dog population of Zimbabwe and its level of rabies vaccination. Vet. Rec. 127, 592–596. Butler, J.R.A., Bingham, J., 2000. Demography and dog-human relationships of the dog population in Zimbabwean communal lands. Vet. Rec. 147, 442–446. Cameron, D.B., 1997. Canine dominance-associated aggression: concepts, incidence, and treatment in a private behavior practice. Appl. Anim. Behav. Sci. 52 (3), 265–274. Cohen, J., Fox, M., 1976. Vocalizations in wild canids and possible effects of domestication. Behav. Process. 1, 77–92. Coppinger, R., Feinstein, M., 1991. Hark! Hark! The dogs do bark. . .’ and bark and bark. Smithsonian, January 1991, pp. 119–129. Coppinger, R., Coppinger, L., 2001. Dogs: A Startling New Understanding of Canine Origin, Behavior and Evolution. Scribner, New York, pp. 1–352. Daniels, T.J., 1983. The social organization of free-ranging urban dogs. I. Non-estrous social behaviour. Appl. Anim. Ethol. 10, 341–363. Daniels, T.J., Beckoff, M., 1989. Populations and social biology of freeranging dogs, Canis familiaris. J. Mammal. 70, 754–762. De Balogh, K.K.I.M., Wandeler, A.I., Meslin, F.X., 1993. A dog ecology study in an urban and semi-rural area of Zambia. Onderstepoort J. Vet. Res. 60, 437–443. Duffy, D.L., Yuying, H., Serpell, J.A., 2008. Breed differences in canine aggression. Appl. Anim. Behav. Sci. 114 (3–4), 441–460. Flintan, F., 2000. A gender-sensitive study of perceptions and practices in and around Bale Mountains National Park, Ethiopia. WWF Technical Progress Report–Ethiopia, International Famine Centre and WWF International, http://www.ucc.ie/famine/Staff/fiona_flintan.htm.
Guy, N.C., Luescher, U.A., Dohoo, S.E., Spangler, E., Miller, J.B., Dohoo, I.R., Bate, L.A., 2001. A case series of biting dogs: characteristics of the dogs, their behaviour, and their victims. Appl. Anim. Behav. Sci. 74 (1), 43–57. Houpt, A.H., 1983. Disruption of the human-companion animal bond: aggressive behavior in dogs. In: Katcher, A.H., Beck, A.M. (Eds.), New Perspectives on Our Lives with Companion Animals. University of Pennsylvania Press, Philadelphia, pp. 181–186. Laurenson, K., Sillero-Zubiri, C., Thompson, H., Shiferaw, F., Thirgood, S., Malcom, J., 1998. Disease as a threat to endangered species: Ethiopian wolves, domestic dogs and canine pathogens. Anim. Conserv. 1, 273– 280. Lehner, P.N., 1978. Coyote vocalizations: a lexicon and comparisons with other canids. Anim. Behav. 26, 712–722. Lord, K., Feinstein, M., Coppinger, R., 2009. Barking and mobbing. Behav. Process. (under review). Macdonald, D., Carr, G., 1995. Variation in dog society: between resource dispersion and social flux. In: Serpell, J. (Ed.), The Domestic Dog: Its Evolution, Behaviour and Interactions with People. Cambridge University Press, Cambridge, pp. 199–216. Maynard Smith, J., 1974. The theory of games and the evolution of animal conflict. J. Theor. Biol. 47, 209–221. Pal, S.K., Gosh, B., Roy, S., 1998a. Agonistic behaviour of free-ranging dogs (Canis familiaris) in relation to season, sex and age. Appl. Anim. Behav. Sci. 59, 331–348. Pal, S.K., Gosh, B., Roy, S., 1998b. Dispersal behaviour of free-ranging dogs (Canis familiaris) in relation to age, sex, season and dispersal distance. Appl. Anim. Behav. Sci. 61, 123–132. Pal, S.K., Gosh, B., Roy, S., 1999. Inter- and intra-sexual behaviour of free-ranging dogs (Canis familiaris). Appl. Anim. Behav. Sci. 62, 267– 278. Perry, B.D., 1993. Dog ecology in eastern and southern Africa: implications for rabies control. Onderstepoort J. Vet. Res. 60, 429–436. Polsky, R.H., 1984. Factors influencing aggressive behaviour in dogs. Aust. Vet. Pract. 14 (2), 66–69. Pongra´cz, P., Molna´r, C., Miklo´si, A., 2006. Acoustic parameters of dog barks carry emotional information for humans. Appl. Anim. Behav. Sci. 100, 228–240. Siegel, S., Castellan, N.J., 1988. Nonparametric Statistics for the Behavioural Sciences, 2nd ed. McGraw-Hill, New York, pp. 1–399. Sillero-Zubiri, C., King, A.A., Macdonald, D.W., 1996. Rabies and mortality in Ethiopian wolves (Canis simensis). J. Wildlife Dis. 32, 80–86. Vas, J., Topal, J., Ga´csi, M., Miklosi, A., Csa´nyi, V., 2005. A friend or an enemy? Dogs’ reaction to an unfamiliar person showing behavioural cues of threat and friendliness at different times. Appl. Anim. Behav. Sci. 94, 99–115. Wandeler, A.I., Matter, H.C., Kappeler, A., Budde, A., 1993. The ecology of dogs and canine rabies: a selective review. Rev. Sci. Tech. Off. Int. Epizoot. 12, 51–71. World Health Organization, 1988. Report of WHO Consultation on Dog Ecology Studies related to Rabies Control, Geneva, Ref. WHO/Rab.Res/ 88.25. World Health Organization, 2000. RabNet: Human and Animal Rabies. Country: Ethiopia. http://www.who.int/rabies/rabnet/en. Williams, S. D., 2003. Ethiopian Wolf Conservation Programme: annual report 2002-03. http://www.ethiopianwolf.org. Yin, S., McCowan, B., 2004. Barking in domestic dogs: context specificity and individual identification. Anim. Behav. 68, 343–355.
