Feralan: E. Font. English.

Applied Animal Behaviour Science, 17 (1987) 319-328

Elsevier Science Publishers B.V., Amsterdam Printed in The Netherlands






Faculty of Biological Sciences, University of Valencia (Spain) (Accepted for publication 24 September 1986)




Font, E., 1987. Spacing and social organization: urban stray dogs revisited. Appl. Anim. Behav. Sci., 17: 319-328.

Different aspects of the etho-ecology of the stray dog population in Valencia (Spain) were investigated. Densities of between 127 and 1304 stray dogs km -2 overlap with density estimates reported for other populations. A male:female ratio of 2:1 is also in agreement with earlier studies. Behavioral observations revealed that these dogs will occasionally form groups with dominance hierarchies and communal defense of a territory. From the stability of these groups, long-term affiliative bonds apparently exist among group members. This finding conflicts with the accepted notion that urban stray dogs are asocial and do not form stable social groups. Methodological problems may invalidate earlier claims that urban stray dogs are asocial animals. This is due to improper use of a Poisson model for assessing social organization. It is suggested that stray dogs possess, like most canids for which an adequate data base exists, remarkable behavioral plasticity allowing them to adjust their social system to prevailing ecological constraints.



In recent years, popular and scientific accounts indicate a growing awareness of problems posed by large populations of uncontrolled domestic dogs in urban and suburban environments. This concern has fostered a number of studies on the natural history of urban stray dogs in cities in the United States (Beck, 1971,1973; Fox et al., 1975; Bekoff, 1979; Westbrook and Allen, 1979; Rubin and Beck, 1982; Berman and Dunbar, 1983; Daniels, 1983a, b; Lehner et al., 1983; Reid et al., 1984) and elsewhere (Oppenheimer and Oppenheimer, 1975; Fox, 1978; Ghosh et al., 1984; Carr, 1985). Variations and inconsistencies in the collection and analysis of data make comparisons among studies difficult or impossible. Contradictory results have also been reported. In particular, urban stray dogs have been portrayed as being asocial animals unable to establish permanent social bonds (e.g., Beck, 1973; Berman and Dunbar, 1983; Daniels, 1983a). This idea, persistent in the literature for over a decade, is in conflict with other studies which show that urban stray dogs form stable social groups (Fox et al, 1975; Carr, 1985).

Between August 1981 and January 1982,1 conducted a study on the etho-ecology of urban stray dogs in Valencia, a city of about 800 000 inhabitants on the Mediterranean coast of Spain. Selected population parameters were meas­ured and analyzed in the context of other similar studies. In addition, behav­ioral observations were conducted on a number of groups of stray dogs in order to explore their social dynamics (Font, 1982). This report presents data on the abundance and sex ratio of the stray dog population of Valencia together with activity patterns, home range and territory size estimates calculated from observations of one group composed of 4 feral dogs. Finally, the question of whether or not stray dogs are truly social animals will be addressed.




Population surveys

Data on the number of free ranging (i.e. unsupervised) dogs, their location, sex and approximate age were collected during periodic censuses done along 7 fixed transects scattered throughout the city. Each transect was run at a con­stant speed of 15 km/h for 9 consecutive days using a small motorcycle. The transects encompassed a variety of urban environments from low income, eco­nomically depressed neighborhoods to middle class, mostly residential, areas. Abundance was estimated using the photographic mark-recapture method as described by Beck (1973). In this method a photograph replaces the actual capture of the animal.


Focal groups

Preliminary observations indicated that groups of stray dogs will occasion­ally remain stable for extended periods of time. These groups exhibit hierar­chical organization and communal defense of a territory, which is typical of many wild canids. The behavior of several such groups was analyzed in order to investigate their social dynamics. Of these one group, consisting of 2 adult males, 1 juvenile male, and 1 female, was studied intensively. All the dogs in this group were "feral" (i.e. unowned) mongrels. Their estimated weights ranged from 15 to 40 kg. All were in fair physical condition, and analysis of feces revealed no intestinal parasites. The female in this group was pregnant at the beginning of the study, and gave birth to 3 pups that were adopted by local residents.

A total of 150 h was devoted to observing behavior within this social group. Observations were scheduled in 3-h periods distributed so as to cover a daily cycle. During each observation period a combination of ad libitum and focal-animal sampling (Altmann, 1974a) yielded data on the dogs' daily activity patterns, home range and territory sizes, shelter types, food and water pro­curement and social behavior. Daily activity patterns were constructed by averaging the time each individual animal spent sleeping, lying or sitting, over all observation periods in which the animal was focal. Home range and move­ment patterns were examined by plotting the location of the animals at 10-min intervals on scale maps of the area. Home range size was estimated by drawing a convex polygon connecting the outermost sightings and computing its area exclusive of inner buildings and fenced areas to which stray dogs were not permitted access. A territory and a core area (Jewell, 1966) were also measured.




Population surveys

A sex ratio of 2 males to 1 female observed in all study areas (N 94 males, 44 females) is in agreement with previous studies on urban stray dogs (Beck, 1973; Westbrook and Allen, 1979; Daniels, 1983a). Since most stray dogs were once pets, Daniels (1983a) suggested that the disproportionate numbers of males reflect a preference for male dogs as pets. However, sex ratios heavily biased towards males have also been found in other canid species (Kleiman and Brady, 1978).

Population densities in the 7 study areas ranged from 127 to 1304 stray dogs/km² . Higher densities tended to occur in the most economically depressed neighborhoods, particularly those with abundant vacant lots and permanent litter accumulation. Approximately one third of the ground surface in an average urban area is occupied by buildings and fenced areas that are not available for use by the stray dogs. This space was excluded from density and home range size calculations. These features of the urban topography were not taken into account in the Baltimore and Newark studies, which reported densities of 232 and 138-204 (range, 3 study areas) dogs/ km² , respectively (Beck, 1973,1975; Daniels, 1980,1983a).


Focal groups: home range, territory and activity patterns



Home range size of 2 groups of stray dogs studied in Valencia together with home range size estimates for other groups collected from the literature



Group size

Home range size (ha)




7.8 1

This study



28.0 1

This study




Beck, 1973,1975




Beck, 1973,1975




Beck, 1973,1975

St. Louis


61.0 1

Fox et al., 1975


1 ( N=9, range summer)

0.2-11.1 1

Daniels, 1980, 1983a


1 (N=13, range winter)

0.1-5.7 1

Daniels, 1980, 1983a

Queens (NYC)



Rubin and Beck, 1982




Rubin and Beck, 1982


1 ( N=15, average)


Rubin and Beck, 1982


1 (N=8, range)

0.01-4.18 1

Berman and Dunbar, 1983

'Corridor home range, exclusive of inner buildings and fenced areas.


Table I shows home range size estimates for the group of 4 strays as well as another group consisting of 2 adult males that was also studied in Valencia (Font, 1982). Reported data for a number of different group sizes and solitary dogs are presented for comparison. In some cases, however, home range size is calculated as the total area enclosed by the convex polygon connecting the outlying sightings. In others the presence of buildings and fenced areas is accounted for and excluded from the calculations ("corridor home range"; Berman and Dunbar, 1983). Home range size estimates for groups of stray dogs in rural environments are one or two orders of magnitude larger than those reported here (Scott and Causey, 1973; Nesbitt, 1975; Carr, 1985), prob­ably due to a wider dispersion of the trophic resources (Fox et al., 1975 ).

Patterns of land use in this study were not exactly the same for all members of a group, although there was a considerable overlap. Therefore group home ranges were calculated by superimposing the home ranges of individual group members on a map and measuring the area enclosed by a convex polygon con­necting the outermost sightings for every animal.

The territory, or that part of the home range actively defended against intruders (Jewell, 1966), was measured for the group of 4 feral dogs by plotting locations where aggressive encounters took place between group members and potential intruders. Territorial boundaries were assumed to be the same for all group members since they all participated in these acts of territorial defense. In all encounters observed, intruders entering this exclusive area were repelled. However, pet dogs of local residents with which the 4 feral dogs were familiar were not treated as intruders. Some of these local residents were occasionally seen feeding handouts to the dogs. The group's territory consisted of a 2-ha vacant lot containing abundant shrubs and litter deposits. A smaller core area of 0.65 ha enclosed 60% of all sightings and included individual "favorite" resting sites. Although no quantitative data on scent marking were collected, urine and fecal deposition by group members appeared to be most pronounced along the territorial boundaries.

The activity of the 4 animals in the group was greatest during early morning (7.00-10.00 h) and early evening (17.00-21.00 h). A third activity peak cor­responding with long solitary excursions was observed after midnight. Activity measurements in other studies were based on head counts of the number of dogs observed at different time-intervals during the day. Despite the different techniques used to record activity patterns, most studies report a bimodal dis­tribution, with activity peaks in the morning and late afternoon or early eve­ning. On a typical day the dogs would spend up to 18 h resting (i.e. sleeping, lying or sitting) inside the territorial boundaries. When the animals first became active in the morning, they were often seen to engage in a "morning greeting" ceremony consisting of long bouts of reciprocal social investigation (Fox et al., 1975).

Agonistic interactions among group members were very scarce, and ritual­ized displays of dominance or submission were observed only on a few occa­sions. This lends support to the contention that most communication among group members is of a very subtle nature and is based on mutual recognition (Fox et al., 1975 ). A dominance hierarchy was constructed based on those rare aggressive incidents. The largest male in the group was dominant to the other 2 males. Of these, the juvenile was lowest in the social hierarchy, subordinate to all other members in the group. No aggressive interactions were observed between the female and the adult males, although they were seen to withdraw when the female made claims concerning food or a resting site.




Are urban stray dogs social?

The results of this study are consistent in many respects with previous stud­ies on urban stray dog populations. One issue in conflict is whether or not urban stray dogs form stable social groups. The findings of this study contra­dict the widely accepted notion that urban stray dogs are "asocial" animals (e.g. Scott and Fuller, 1965; Beck, 1973; Kleiman and Brady, 1978; Berman and Dunbar, 1983; Daniels, 1983a) and agree with other studies suggesting that long-term affiliative bonds exist within urban stray dog groups (e.g. Fox et al., 1975; Carr, 1985). Members of the social groups examined in Valencia were always seen to travel through their home ranges alone or, occasionally, as pairs. However, they defended a common territory, had dominance hierar­chies, shared food, and group membership remained unchanged well beyond the termination of the study (unpublished observations). This strongly sug­gests that long-term affiliative bonds exist among the members of these groups. Unfortunately, a population in which individuals forage alone resembles one in which social groups are non-existent. Waser and Jones (1983) point out that an individual that forages alone need not decrease the number of individ­uals recognized and may still live within a complex social network. An alter­native explanation is that stable social groups are a peculiarity of some populations but not others. However, a re-evaluation of techniques used in earlier studies to investigate the stray dogs' social system raises the possibility that claims of asociality result from a methodological artifact.

The studies in question involved recording the frequencies of different group sizes observed during a population survey and comparing these to a theoretical frequency distribution (Beck, 1973 ). The procedure requires selection of some criterion to measure the amount of "grouping" occurring in the population. This criterion may be based on spatial proximity of the animals (e.g. animals within 10 m of each other are assigned to the same group; Berman and Dunbar, 1983 ), or a combination of spatial and temporal considerations (e.g. animals that "remain together for at least 1 min" constitute a group; Daniels, 1983a, p. 343 ). A census of the population is then conducted that yields data on the abundance of different size groups as defined by the above criteria. The result­ing frequency distribution of observed group sizes is subsequently compared (X² test) to a set of expected frequencies generated by a Poisson process with the zero value truncated. The zero truncation is necessary since groups of size zero are impossible. Underlying this argument is the assumption that a random distribution of animals in space will yield a zero truncated Poisson distribution of group sizes. Since animals that are distributed randomly in space would not be expected to group more or less than is determined by chance alone, an observed distribution of group sizes that fits the Poisson model is taken as evidence that the observed groups of stray dogs are produced by random or casual encounters among animals wandering freely about their home ranges, i.e. they do not represent "true" social groups.

This approach has also been applied to other species (e.g. kangaroos; Caugh-ley, 1964) and presents a number of difficulties. In the first place, it should be clear that the use of criteria of spatial proximity renders the argument invalid if the members of social groups are solitary foragers. It should also be noted that casual group formation may have little to do with sociality. There is no doubt that by some arbitrarily chosen criterion the groups of humans that form at a cocktail party will closely approximate a zero truncated Poisson distribu­tion or some other related theoretical model (Cohen, 1971 ). We may even feel justified in qualifying these groups as casual. However, to cite this as evidence that humans are asocial would clearly be erroneous. Moreover, use of a Poisson model to test for sociality in stray dogs is incorrect for the following reasons. First, the procedure described above compares expected vs. observed distri­butions of group sizes. It does not take into account the animals' spacing pat­terns. One may argue that the spatial distribution of animals is not the question at issue here, but the distribution of group sizes. However, so long as groups are defined on the bases of criteria of spatial proximity among individuals, comparison to a Poisson model is simply not correct. Second, regardless of the actual distribution of animals in space, a distance criterion could always be found such that the animals will fall into groups fitting a zero truncated Pois­son distribution. For instance, if a very small distance criterion were used, one would expect the size of most groups to be one. We would then conclude that our animals were over-dispersed, spacing themselves out regularly by some sort of territorial behavior. Conversely, if a very large distance criterion were to be used, most animals would be found to fall into groups most of which consisted of many animals. Between these two extreme situations there will necessarily be a distance criterion that yields a good fit to the zero truncated Poisson distribution. If this criterion were used, we would conclude that the animals form random groups. Thus, how closely our distribution of group sizes resem­bles a Poisson model is determined by two factors: (1) our initial choice of a distance criterion; (2) the distribution of animals in space. This is analogous to the problem, familiar to ecologists, of choosing the "right" size of grid in order to determine patterns of distribution of plants or animals (i.e. clumped, random or uniform). For the researcher trying to analyze the stray dogs' social system, two parameters are unknown: (1) the "true" largest distance below which two animals are social; (2 ) the distribution of animals in space. No mat­ter what distance criterion is used, not knowing the distribution of animals in space (i.e. the distances between individuals within a group as well as the dis­tances among groups) precludes comparison to a Poisson model. Third, even though casual groups may under certain circumstances be adequately described by a Poisson distribution (Cohen, 1975), the inference should not be made that all distributions of groups fitting a Poisson distribution consist of casual groups only.

Given the above arguments, reports of asociality in urban stray dogs seem questionable. Social groups may have existed in all the populations studied, but were overlooked by the techniques used and by a tendency to equate gre-gariousness to sociality. To the best of my knowledge no statistical manipula­tion can replace careful and prolonged observation of individual animals and groups of animals in the study of social organization.

Even though members of social groups such as the ones reported here travel independently when wandering through their home ranges, stray dogs in rural areas show a definitive tendency towards pack formation (Scott and Causey, 1973; Nesbitt, 1975; Carr, 1985). One may then ask, is solitary foraging an adaptation to life in the city or a product of the dogs' social "misbehavior"?

Wolves, which share a common gene pool with dogs, live in packs of up to 36 individuals (Mech, 1970). They are able to adjust to changes in the abundance and distribution of their food resources by shifting from a solitary to a pack existence (Fox, 1975). The same is true of most wild canids for which an ade­quate data base exists (Kleiman and Brady, 1978 ). Urban stray dogs feed chiefly on garbage refuse and occasional handouts from local area residents (Beck, 1973; Font, 1982; Daniels, 1983a). Garbage is a patchily distributed food resource. It is available to the stray dogs in the form of small packages widely scattered throughout their home ranges. This distribution of food resources promotes foraging on an individual basis or in very small groups (Altmann, 1974b; Fox et al., 1975 ). The highly integrated packs of wolves and other canids appear to have evolved as an adaptation to hunting large herbivorous prey (Kleiman and Eisenberg, 1973; Fox, 1975 ). However, urban stray dogs no longer depend on cooperative hunting to procure their food supply. Further, pack formation in an urban environment may even be maladaptive, since large groups of dogs are rarely tolerated by people and are subject to intense predation from dog catchers (Beck, 1973; Fox et al., 1975; Daniels, 1983a). In rural areas, on the other hand, the cooperation of all group members may be required in order to bring down large prey items. Observations of pariah dogs in India revealed an added feature of behavioral plasticity. Here dogs with non-overlapping ter­ritories would occasionally band together to hunt deer and to drive strange dogs off shared foraging/hunting ranges (Fox, 1978).



Four main conclusions may be drawn from this study.

•  In order to be able to make comparisons among studies, a consensus must be reached with respect to techniques that are used to collect and analyze stray dog data (e.g. density and home range size calculations, daily activity patterns).

•  Comparing frequencies of observed group sizes to a zero truncated Pois­son distribution as a means to analyze the stray dogs' social tendencies is inap­propriate. Therefore, claims that urban stray dogs are asocial are probably unwarranted.

•  The results of this study show that urban stray dogs live in complex interactive networks with all the attributes of permanent social groups.

•  The differences between groups of urban and rural stray dogs are explainable on the basis of food distribution. The flexibility exhibited by the stray dogs' social system parallels that observed in other wild canids.



I am deeply indebted to my thesis director, Dr. J. Sabater-Pi, and the mem­bers of my graduate committee, M. R. Miracle, J. D. Acuna and E. Andreu, for their generous support during the early stages of this research. Matt Kramer spent many hours helping me understand the statistical end of things. The little I know about Poisson distributions I owe entirely to him. G. McCracken, M. Okoniewski and an anonymous reviewer read various drafts of this paper and improved its quality substantially.

A shorter version of this paper was presented at the 19th International Eth-ological Conference, Toulouse (France).





Altmann, J., 1974a. Observational study of behavior: Sampling methods. Behaviour, 49:227-267.

Altmann, S. A., 1974b. Baboons, space, time, and energy. Am. Zool., 14:221-248.

Beck, A. M., 1971. The life and times of Shag, a feral dog in Baltimore . Nat. Hist., 80:58-65.

Beck, A. M., 1973. The Ecology of Stray Dogs: A Study of Free-ranging Urban Animals. York Press, Baltimore , MD , 98 pp.

Beck, A. M., 1975. The ecology of "feral" and free-roving dogs in Baltimore . In: M. W. Fox (Edi­tor) , The Wild Canids: Their Systematics, Behavioral Ecology and Evolution. Van Nostrand Reinhold , New York , pp. 380-390.

Bekoff, M., 1979. Scent-marking by free-ranging domestic dogs. Biol. Behav., 4:123-139.

Berman, M. and Dunbar , I. , 1983. The social behaviour of free-ranging suburban dogs. Appl. Anim. Ethol., 10: 5-17.

Carr, G. M., 1985. Behavioural ecology of feral domestic dogs ( Canis familiaris) in Central Italy . XIX Int. Ethological Conference, Toulouse , France , abstract 267.

Caughley, G., 1964. Social organization and daily activity of the red kangaroo and the grey kan­garoo. J. Mammal., 45:429-436.

Cohen, J. E., 1971. Casual Groups of Monkeys and Men. Harvard University Press, Cambridge , 175 pp.

Cohen, J. E., 1975. The size and demographic composition of social groups of wild orang-utans. Anim. Behav., 23:543-550. Daniels, T. J., 1980. The social behavior of free-ranging urban dogs. Master's Thesis, Ohio State University.

Daniels, T. J., 1983a. The social organization of free-ranging urban dogs. I. Non-estrous social behavior. Appl. Anim. Ethol., 10: 341-363. Daniels, T. J., 1983b. The social organization of free-ranging urban dogs. II. Estrous groups and the mating system. Appl. Anim. Ethol., 10: 365-373. Font, E., 1982. Eto-ecologia del perro vagabundo. Master's Thesis, University of Valencia ( Spain ). Fox, M. W., 1975. Evolution of social behavior in canids. In: M. W. Fox (Editor), The Wild Canids: Their Systematics, Behavioral Ecology and Evolution. Van Nostrand Reinhold, New York , pp. 429-460.

Fox, M. W., 1978. The Dog: Its Domestication and Behavior. Garland STPM Press, New York , 296 pp.

Fox, M. W., Beck, A. M. and Blackman, E., 1975. Behavior and ecology of a small group of urban dogs (Canis familiaris). Appl. Anim. Ethol., 1:119-137. Ghosh, ., Choudhuri, D. K. and Pal, ., 1984. Some aspects of the sexual behaviour of stray dogs, Canis familiaris. Appl. Anim. Behav. Sci., 13:113-127. Jewell, P. A., 1966. The concept of home range in mammals. Symp. Zool. Soc., London , 18:85-109. Kleiman, D. G. and Brady, C. A., 1978. Coyote behavior in the context of recent canid research: Problems and Perspectives. In: M. Bekoff (Editor), Coyotes: Biology, Behavior and Manage­ment. Academic Press, New York , pp. 163-188.

Kleiman, D. G. and Eisenberg, J. F., 1973. Comparisons of canid and felid social systems from an evolutionary perspective. Anim. Behav., 21:637-659.

Lehner, P. N., McCluggage, C, Mitchell, D. R. and Neil, D. H., 1983. Selected parameters of the Fort Collins , Colorado , dog population, 1979-80. Appl. Anim. Ethol., 10:19-25.

Mech, L. D., 1970. The Wolf: The Ecology and Behavior of an Endangered Species. Natural His­tory Press, Garden City, New York , 384 pp.

Nesbitt, W. H., 1975. Ecology of a feral dog pack on a wildlife refuge. In: M. W. Fox (Editor), The Wild Canids: Their Systematics, Ecology and Evolution. Van Nostrand Reinhold , New York , pp. 391-395.

Oppenheimer, E. C. and Oppenheimer, J. R., 1975. Certain behavioral features in the pariah dog (Canis familiaris) in West Bengal . Appl. Anim. Ethol., 2: 81-92. Reid, J. ., Chantrey, D. F. and Davie , C, 1984. Eliminatory behaviour of domestic dogs in an urban environment. Appl. Anim. Behav. Sci., 12:279-287. Rubin, H. D. and Beck, A. M., 1982. Ecological behavior of free-ranging urban pet dogs. Appl.

Anim. Ethol, 8:161-168. Scott, J. P. and Fuller, J. L., 1965. Genetics and the Social Behavior of the Dog. University of Chicago Press, Chicago, 468 pp. Scott, M. D. and Causey, K., 1973. Ecology of feral dogs in Alabama . J. Wildl. Manage., 37:253-265. Waser, P. M. and Jones, W. ., 1983. Natal philopatry among solitary mammals. Q. Rev. Biol., 58: 355-390.

Westbrook, W. H. and Allen, R. D., 1979. Animal field research. In: R. D. Allen and W. H. West-brook (Editors), The Handbook of Animal Welfare. Garland STPM Press, New York , pp. 148-167.