(from Foxes, Wolves, Jackals, and Dogs,
the IUCN/SSC Canid Specialist Group's 1990 Action Plan)

Predation and Predator Control


The Carnivora, as the name implies, are generally predatory, and to a greater or lesser extent, all canids live by killing prey.This fact raises three related questions in the context of conserving and managing wild canids. First, to what extent are the populations of canids limited by their prey, and to what extent do they limit the numbers of their prey? Second, and with respect to valuable prey, is the impact of predation by canids disadvantageous to people? Third, where predation by canids throws them into conflict with people, how might such conflict best be resolved?

Problems are likely to arise with predation on three categories of prey: domestic stock, wild game, and endangered species. Clearly the delineation of these three categories is indistinct; for example, the management of incubator and pen-reared game birds such as pheasants has more in common with the domestic stock than it does with wild game such as woodcock or snipe.

Nearly every species of wild canid, from the abundant and successful coyote (Andelt 1987) to the endemic and highly endangered Simien jackal (Sillero-Zubiri and Gottelli pers. comm.) has been implicated in livestock damage. Of the 34 species discussed in the previous chapters, 21 have been reported to kill livestock or poultry at least occasionally. However, only a handful of canids are sufficiently numerous and find themselves in circumstances under which their predation is economically significant (e.g. wolves, coyotes, some foxes). Even for these, it is often a moot point as to whether they regulate the populations of their prey.

When a farmer sees a dead lamb being eaten by a jackal, or a deer stalker sees a pack of wolves gathered at the carcass of a moose, they commonly feel a wrath born of competition: both farmer and hunter had wanted the prey for themselves (albeit for different purposes). But underlying such anger is the assumption, rather than the proof, of competition. If the lamb or deer had, for example, been eaten as carrion, then it would already have been valueless. Equally, if the predators had singled out sickly individuals that were destined to perish then the measure of competition with people would be greatly devalued.

Predation does not necessarily affect long term measures of prey numbers. This paradox arises because, in prey populations which are limited by food, killing prey does not necessarily mean there are fewer of them, except in a very short-term sense! If some prey animals are destined to die from starvation, and a predator's feasting is confined to the proportion destined to die, then it acts merely as executioner. Only if predation eats into the breeding stock that could have been supported by the food supply (or some other limiting factor) could predation be said to be limiting a natural population. This point makes it vital to distinguish whether predators are taking only a "doomed surplus" or more.

Equally vital, people have to be clear whether they are competing with canids for the breeding stock or the doomed surplus. The quarry of hunters is often the doomed surplus of game, so whether or not foxes, for example, regulate the population of pheasants is displaced by the question of whether foxes eat any pheasants which the hunter might otherwise have shot: if they do, they are in direct competition with humans for pheasants.

These considerations direct attention to one salient point: evidence that predators eat a given prey is not evidence that by doing so they are a pest. To evaluate pest status much must be known of the circumstances, including both biological and economic information. Furthermore, having ascertained the magnitude of the problem, then the costs and benefits of proposed solutions also require careful assessment. Obviously, such an assessment hangs in the perspective of local economies: predation by maned wolves upon chickens may be trivial in the economics of Brazilian poultry farming, but a peasant losing his poultry flock may nonetheless be suffering a greater loss than his European counterpart losing the same number of chickens to a red fox.

Thus, in evaluating the damage done by canid through predation the following categories of information are helpful:

  1. A measure of competition: this involves measuring the frequency of relevant prey deaths. So, if the concern is over live prey, then the measurement must exclude those that were eaten as carrion. Similarly, if predators take sickly individuals, the cost of their predation must be devalued by what it would have cost to nurture the ailing prey. If concern is over the breeding stock of prey, then measurement of the competition would exclude predation upon that proportion of the population doomed to die for other reasons.
  2. A measure of the loss: this involves assessment of the costs of predation. Unfortunately, this may involve measuring costs in different currencies, for example, the deer stalker who is usurped by a wolf may lose both the commercial value of the carcass he might have sold, and he loses the enjoyment he anticipated in the hunt. Economic assessments must be made in appropriate contexts, particularly distinguishing the individual and the state. For example, it is relevant to know how the loss affects the micro-economic position of the individuals that bear it, and how it affects wider economics.
  3. The consequences of action: if action is taken to remedy the losses due to predation, then the benefits of that action should outweigh the costs. The costs have many and varied components. One might calculate the cost of time and effort expended by hill sheep farmers in the U.K. when they hunt red foxes with hounds and terriers, and ask whether that cost either exceeds any diminution in losses resulting from their actions (or even whether it exceeds the magnitude of the losses altogether) (Macdonald 1987). However, such an estimation of costs would, to be complete, have to account for the recreational value that the hunters put on fox control. Equally, the costs of an irate farmer in Montana killing a coyote are different to those of an irate Ethiopian killing a Simien jackal for just the same misdemeanour. Logically, it should be simple to evaluate the consequences of predator control: simply compare, in a scientific manner, places where control is, or is not, conducted. If the improvement is less than the cost, then the control is a waste. Even if the cost is taken in its simplest form, such comparisons are very rarely made. A more common approach is to assume that one less predator results in a proportionate improvement in the circumstances of prey; that assumption is seriously flawed.

Here, we focus on a few of the best studied species, with a view to illustrating the nature of the problems.

Canid Predation on Livestock

One might reasonably assume that canid predation on livestock must be a common event: the fox in the chicken coop and the wolf in sheep's clothing have been enshrined in English as metaphors for negligence and cunning. But the strength of these images may belie the frequency with which the actual events occur. Amidst the flying feathers, how many chickens or turkeys are actually eaten by canids?

Every British city-dweller 'knows', whether or not he has ever seen either fox or chicken, that red foxes are the scourge of the chicken-run; the power of ideas learnt on the parental knee may not take account of the fact that free-range poultry are an economic irrelevance to the British poultry business.

In a survey by Macdonald and Doncaster (1985) of red foxes killing urban pets, many of those questioned in one district responded with details of a child's guinea pig being killed-it turned out that all these accounts referred to the same guinea pig. In the same survey missing cats were often said to have been killed by foxes, despite the fact that this is a demonstrably rare event and that cats are very often killed in road traffic accidents. Similarly, a horrendous report of a surplus kill of lambs by a red fox, vividly related by an elderly hill farmer, turned out to have been the misfortune of his father many years before. These points are not to deny the importance of predation, but merely to highlight the difficulty in quantifying it.

Only in the last few decades have data been systematically collected to assess the extent, and costs, of livestock losses to carnivores in general, and to canids in particular (Andelt 1987).

It is exceedingly difficult to answer a relatively simple question: how does one measure the magnitude of livestock losses which are caused by canids? Most field studies of canid feeding ecology are made from the perspective of the predator, not the prey. They give some indication of the proportion of livestock in an animal's diet. From this, if one knows the population density of the carnivore being studied, an estimate can be made of the numbers of sheep, chickens, or cattle which might be lost to any particular species of canid. However, when measuring the impact of foxes, wolves, or jackals on livestock production, the question we need to answer is not "what percent of the diet is composed of livestock?" or even "how many sheep are eaten by canids?" but "what effect does canid predation have on total livestock production?"

To answer this question, two statistics are needed as a starting point: the number of viable domestic animals killed by the carnivore in question, and the total production of domestic animals in that region. In Table 1, we have summarized data from several studies on the impact of canid predation on livestock production.

Assessing the extent of Canid Induced Losses

The data in Table I show clearly that the percentage loss of domestic livestock to canid predation is, for the most part, small. Of the 15 studies cited, mean losses in excess of 2% occur in very few cases. In some cases, such as wolves in Norway (Naess and Mysterud 1987), the extremely small losses (0.02%) reflect extremely small wolf populations (4- 10 individuals). Nonetheless, even in areas where healthy canid populations persist (e.g. coyotes in the western United States), livestock losses as a percentage of total production are relatively small.

Even an estimate of 2% loss may exaggerate the losses which can be directly blamed on canids. Most of the known biases happen to inflate (but rarely deflate) estimates of canid predation on livestock. Pearson (1986 loc. cit. Andelt 1987), notes that many of the studies he reviewed in his manuscript have been conducted in areas where predation is known to be a problem. This is not surprising; there is little need, and little demand, to study the effects of predation in areas where predators are either uncommon or where predators do not disturb livestock. However, in many areas, the great majority of ranchers are rarely affected by predation by wild canids. In Minnesota, over 99% of all livestock producers were unaffected by wolves (Fritts 1982). In the western United States, most sheep ranchers suffer no loss or only minor losses to predators (Andelt 1987).

Table 1. Levels of Canid Predation on Livestock

Predator Prey % Crop Loss Region Source
Coyote Ewes 1.0-2.5%
Western U.S.A.
Western U.S.A.
Alberta, Canada
USFWS 1978 loc cit Andelt 1987
Pearson 1986, Andelt 1987
Dorrance and Roy 1976
Coyote Lambs 4-8%
Western U.S.A.
Western U.S.A.
Alberta, Canada
USFWS 1978 loc cit Andelt 1987
Pearson 1986 loc cit., Andelt 1987
Dorrance and Roy 1976
Coyote Beef calves 0-0.4%
Western U.S.A.
USFWS 1978 loc cit Andelt 1987
Gee 1978, Andelt 1987
Coyote Turkeys 0.8% Nebraska, U.S.A. Andelt and Gipson 1979
Coyote Goats (Adults)
Texas, U.S.A.
Texas, U.S.A.
Pearson 1986 loc cit., Andelt 1987
Pearson 1986 loc cit., Andelt 1987
Wolf Beef cattle 0.2%-3.1% Alberta, Canada Fritts 1982
Wolf Sheep 0.12%
Alberta, Canada
Fritts 1982
Naess and Mysterud 1987
Arctic fox Lambs 3-4% Iceland Hersteinsson unpublished data

In management terms, this suggests that the figures presented below may represent the worst case. If a study is undertaken in an area with perceived coyote problems and determines that livestock losses are approximately two percent, the average loss of livestock in all areas, those with and without perceived coyote problems, is apt to be much lower.

Of course, there is a corollary to this point: if most ranchers and farmers are unaffected by canid predation, then the few that are will suffer heavier losses. This may, in turn, result in those producers most affected taking drastic measures in an attempt to reduce loss. For example, in South Africa, one of the last surviving populations of approximately 300 African wild dogs is found in Kruger National Park. In a single year (1987), one farmer bordering the park is known to have shot 20 dogs (M.G. Mills, pers. comm.). Although livestock losses to African wild dogs in South Africa must be measured in the hundredths of a percent, the farmer bordering the park obviously found his personal losses intolerable. A second problem common to many studies is that losses reported are frequently based on reports made by ranchers. Fritts (1982) notes that this type of data collection may introduce many biases. The first such problem is verification that animals reported as being killed by canids were actually killed by canids (Macdonald and Doncaster 1985, Macdonald 1987).

In his study on wolf predation in Minnesota, Fritts extracted data from reports made to the state by ranchers seeking compensation. Many of the reports of wolf predation were completely unverified: the carcasses of 76% of the cattle, and 73% of the calves reported missing were never found, and wolf involvement in the death of these animals could not be verified. In an area of northwest Minnesota where wolves were recently protected, there was only one confirmed report of wolf predation in 5 years; only 1% of scats examined had remains of cattle suspected to have been killed by wolves (Fritts and Mech 1981).

In surveys where a large proportion of "kills" are unverified, a great majority of animals "killed" by predators may have died from other causes. The magnitude of the error introduced by unverified reporting can be seen in the results of a study on wolves and cattle in Alberta, Canada (cited by Fritts 1982). In 121 cases where the cause of death of the animal could be determined, only 19 deaths (16%) were caused by predators. The great majority of deaths (67 or 55%) were ascribed to natural causes such as pneumonia and the ingestion of poisonous plants. Although it concerns an avian predator, and not a canid, Houston and Maddox's studies (1974) of predation by carrion crows on lambs show elegantly the grave inequality between the farmer's suspicions and the biologist's data. Similarly, Hewson (1984) shows how, despite a fearful representation, red foxes in west Scotland were responsible for killing only a small percentage of lambs, which were actually dying.

Even livestock that is seen being consumed by a predator may have died of natural causes and, subsequently, been scavenged by the predator. For instance, although the South American grey zorro is believed by some to kill livestock, evidence suggests that the great majority of all livestock consumed is scavenged (Jaksic and Yanez 1983). On the other hand, it appears that crab-eating zorros in Brazil are a persistent minor nuisance through their depredations on chickens (D. Macdonald, pers. obs.).

Oddly enough, not all animals reported as dying are known to have lived. Calf losses are frequently assumed to have occurred if a cow thought to be pregnant is put out to pasture and, sometime later, is then sighted without a calf. However, when cows thought to be pregnant by ranchers were tested, 27% (40 of 150) were found to be "false" pregnancies (Fritts 1982). This overestimate of pregnancy leads to an overestimate of the number of calves in a herd; this then leads to an overestimate of losses. The wolves, in the end, are blamed.

Deaths that are verified as canid kills may not have been caused by the species of canid suspected of doing the damage. In Minnesota, Fritts (1982) reported that coyotes were responsible for approximately 10% of the deaths thought to have been caused by wolves. In Leon, Spain, 47 sheep and 11 goats were "killed" by wolves in a two year period. In fact, nearly 50% of the sheep thought to have been killed by wolves were actually killed by feral dogs (Salvador and Abad 1987). In Italy, 50% to 80% of the sheep thought to have been killed by wolves were probably killed by feral domestic dogs (Boitani 1982).

Why are so many losses that appear to be due to natural causes, or other predators, blamed on a particular species such as wolves? In many cases, the blame for such mis-reporting is, in part, due to abuses of programmes developed to protect wolves. In many parts of the world, where wolves are rare or endangered, ranchers are compensated for losses due to wolves, but not for other causes of livestock mortality. Hence, kills known to have been made by coyotes, foxes, or feral domestic dogs are called wolf kills so that a rancher can collect compensation (Italy-Zimen and Boitani (1979), Macdonald and Boitani (1979); Minnesota, U.S.A.-Fritts (1982); Leon, Spain, Salvador and Abad (1987).

The data in Table 1 also clearly show that juvenile animals are at a greater risk than adults. Calves are eaten at higher rates than cattle, lambs more frequently than sheep. This suggests that canid predation will be the greatest problem at times when livestock are bearing and raising their young. Of course, at other times of the year, canids are surviving on other types of food. Hence, if young animals can be protected, canids may well stop eating livestock, and switch to more easily acquired foods.

Clearly, the influence of wolves on domestic livestock will depend on the interactions between wolves and their wild prey. The extent of predation on livestock, for instance, may be directly related to the quality or quantity of other prey species.

An example of such a interaction has recently been elucidated by Mech et al. (1988). In the summer in Minnesota, white-tailed deer fawns constitute a large part of a wolf's diet. The vulnerability of fawns to predation appears to be a function of the previous winter: a bad winter results in more vulnerable fawns. Furthermore, there is less wolf predation on domestic livestock after a bad winter. The increased vulnerability of fawns appears to result in a decline in wolf predation on domestic livestock.

In a study of a small pack of wolves in the availability of a third type of "prey," human refuse, might also influence patterns of predation on livestock. In Spain, wolves appeared to compensate for reduced prey (roe deer) numbers by eating more garbage (Salvador and Abad 1987). If garbage had not been an easily acquired resource, perhaps wolves might have switched to domestic livestock.

Fritts (1982) notes that in Alberta, cattle are much more heavily preyed upon in closed, brushy habitats (3.3%) than in open habitats (1.3%). The increase in predation in closed habitat may be due to increased risk of predation due to limited visibility, or to wolf habitat preference: in northwest Minnesota, wolves were rarely observed in pasture areas and spent most of their time in the woods (Fritts and Mech 198 1).

Canid Predation on Wildlife

It is often assumed that canid predation on wildlife reduces the amount of wildlife available for human consumption and sport. Before such an assumption can be made, data must be collected which address the following questions: 1) Is the harvest of the prey species by humans on a scale with that of canids? If, for the most part, humans are the major predator in a system, removing other causes of predation may result in only a marginal increase in human harvests. 2) If canids are removed from an ecosystem, or reduced in number, does the prey they eat become available to man or do these animals die from other natural causes? For example, canids often specialize on young animals. If removal of canids results in a greater rate of predation by other predators (e.g. bears, birds of prey) or increased natal mortality from starvation or disease, canid reduction alone is unlikely to result in greater human harvests of adult animals.

In this section, we review the available literature and attempt to answer the following questions: 1) Do canids control prey populations? 2) In those circumstances in which canids do control prey populations, will killing predators ("lethal control measures") significantly reduce the damage done by canids? 3) If killing predators does not work, are other non-lethal control measures possible?

Do Canids Regulate Prey Populations?

Assessing the impact that canid populations have on prey populations can be extremely complex. The complexity of the ecosystem (e.g. the number of prey and predator species), variations in weather patterns, and the ability of both prey and predator to emigrate and immigrate will influence a particular predator's ability to regulate the population of a particular prey.

Despite these complications, carnivores in general, and canids in particular, are often thought to regulate game animal populations. If this is the case, then removal of canids will increase the amount of game available to hunters. How often, and in what circumstances, do canids regulate prey populations? We will answer this question by looking at a few well documented case studies.

Wolves, Caribou, and Moose

Perhaps the most frequently studied canid/prey systems are those including wolves, caribou, and moose. In some studies, wolves are found to regulate prey numbers, while in others wolves are thought not to be important in these processes. What is critical to note is that two studies of a single ecosystem have come up with contradictory results.

The best such example is the relationship between wolves and the caribou of the Nelchina caribou herd in south-central Alaska. In the period 1950 to 1961, the caribou population increased dramatically. In 1961-1962, the population crashed and fluctuated at a low level until 1972. Caribou population levels were stable from 1972 to 1976, and increased thereafter (Van Ballenberghe 1985).

Caribou calf mortality during this period was correlated with winter severity, not wolf predation (Van Ballenberghe 1985). The increase in herd size preceded wolf control measures which were instituted in the early 1950s. This suggested that wolf control measures were not responsible for the large increase in caribou numbers. This appears to be conclusive evidence that wolves do not control caribou populations.

A second study, using the same data, came to the opposite conclusion. Re-analysis of the data suggests that predation by wolves on young animals was the most consistent natural limiting factor in the dynamics of the Nelchina herd (Bergerud and Ballard 1988).

The population dynamics of the Nelchina caribou herd are obviously complex. Usually, female caribou in the Nelchina herd reduce the level of predation to which they are exposed by calving on a high plateau where wolf population densities are lower. Bergerud and Ballard (1988) note that in 1964-1966 late snowfall prevented caribou females from reaching the calving grounds. In these years, there was large calf mortality due to predation. Obviously, without wolves, there would have been no predation: but is the increased predation due to high wolf population density or random weather patterns? Would a large majority of the calves have died as they were born in late snow outside the usual calving grounds? In other words, did wolves kill animals that were already doomed? Perhaps the focus on wolves is also misguided. Wolves, alone, certainly are not the sole cause for the subsequent decline of the Nelchina herd. Both authors (Bergerud and Ballard 1988; Van Ballenberghe 1985) agree that whether wolves or snow were to blame for the initial reduction in the recruitment of young, the major cause of population decline after 1966 was over-hunting by humans: at its worst, in 1971/1972, hunters killed 44% of the herd.

Despite the apparent differences in interpretation of the Nelchina data, and differences in conclusions drawn in so many studies, several recent reviews of wolf-caribou-moose systems agree with Mech (1970:268-277) that in certain circumstances, at certain times, wolves can control ungulate populations.

Most prey populations appear to cycle, the length of the cycle being related to the species' body size (Peterson et al. 1984). Once ungulate numbers are at or near their peak, most authors are in agreement that wolves do not produce a decline in ungulate numbers (but see Bergerud 1988). Declines are caused by combination of weather, over-hunting, and food limitation. (Peterson et al. 1984, Gauthier and Theberge 1987; Mech et al. 1987).

When ungulate numbers are declining, predation alone may be sufficient to depress herbivore density during the final period of a herbivore population decline (Keith et al. 1977; Peterson et al. 1984; Gauthier and Theberge 1987; Mech and Karns 1977). If prey densities are low, then predators may exert a controlling effect upon their numbers (Peterson 1977). The ready availability of other prey (beaver, moose) may allow wolves to survive at high numbers even when their primary ungulate prey has declined (Gauthier and Theberge 1987, Bergerud 1988).

In situations where the initial decline is caused by the ungulate population exceeding its food supply, wolf predation may further reduce ungulate numbers, extending the period over which prey populations remain at low levels. The decline in herbivore density may allow forage to recover. If wolf numbers decline due to decreased prey numbers, disease, or active control, forage conditions are sufficiently improved to allow herbivore population numbers to increase rapidly (Peterson et al. 1984). Wolves do not usually cause the initial decline in prey numbers. They may, in some circumstances, prolong prey declines.

Zorros and Rabbits

In Chilean Tierra del Fuego, 24 South American grey zorro, Dusicyon griseus, were introduced in 1951 in an attempt to control a burgeoning population of 30 million introduced European rabbits (Oryctolagus cuniculus) (Jacksic and Yanez 1983). The culpeo, Dusicyon culpaeus, was already present on the island but at low densities due to persecution by humans.

Before the effect of introductions could be assessed, myxomatosis was introduced and the rabbit population crashed and has remained low since. Jaksic and Yanez (I 983) suggest that both zorro species may play a part in controlling rabbits but that zorros alone probably could not control an infestation: rabbits account for 18.4% of the diet of the culpeo and only 3.3% of the diet of the grey zorro. Foxes might show a greater preference for rabbits if their numbers greatly increased; however, because rabbit populations increase geometrically faster than those of foxes, they believe it is unlikely that foxes would be able to regulate a rapidly expanding population. Similarly, red foxes introduced to Australia have not controlled rabbit populations there (Macdonald 1987).

Foxes and Gamebirds

In Europe canids have long been viewed as important in reducing bags of gamebirds, and have been persecuted as a result. In 1912 there were 22,000 gamekeepers in Britain alone (Potts 1986). This view has been rigorously tested only recently, however, when a number of predator-removal experiments have sought to determine exactly what effect canid predation has upon the size and available harvest of gamebird populations.

One study on partridges (Perdix perdix) was carried out in southeast England. Of two experimental areas, of similar size, partridge density and history of keepering, predators (foxes, stoats, and corvids) were controlled intensively on one and protected on the other (Potts 1986). After just one year, partridge density on the predator control area has risen from 223 birds to 338, while that on the predator protection area had fallen from 230 to 196. This rise seemed to occur through nest losses, which remained constant in the predator protection area, but fell where predators were controlled. Under predator control, the numbers of broods seen at the age of 6 weeks rose from 21 to 34, while under predator protection this rise was from 16 to just 17. However, these preliminary results cannot be evaluated until the experiment is complete. This will involve reversing the procedures so that the predator control area becomes a protected area, and vice versa. A similar study near Bonn in West Germany (cited in Potts 1986) showed a mean annual bag of 5.7 ±0.8 under predator protection, and 10.9 ±1.9 under predator control. Potts was able to mimic this difference by a computer simulation using population parameters "stolen" from his own study, with nest predation as the only factor differing between areas with and without predator control.

Potts tried to model his study population over 20 years and got a very good approximation to the real situation. Then he put chick mortality (inflated by modem herbicides) and nest predation (inflated by the cessation of predator control) back to their 1976 levels from the beginning of the study. By doing this, he was able to "prevent' (retrospectively) the decline in the partridge population that he had observed. By the end of the run, an annual bag of 40 birds per square Ian would have been possible. However, when he considered only predator control, (i.e. left in the herbicides), he found that restoring it "would have increased breeding stocks in recent year but it would not have prevented a steady decline" (his italics). The bag would have been just 5 birds per square kilometre, and the cost of keepering to control the predators would have amounted to £300 per bird shot. He concluded that predation pressure had simply accelerated a decline that had been ultimately caused by modem herbicides.

Another study was carried out on two small islands in the Gulf of Bothnia-predators (red foxes and martens) were trapped and shot each winter on one of the islands, while no predators were removed from the other (Marcstrom et al. 1988). After five years, predators were no longer controlled on the one island, and were allowed to recolonize across the sea ice. Predators were then removed from the other island, and the study continued for four more years. Populations of capercaillie (Tetrao urogallus), black grouse (T. tetrix), hazel grouse (Bonasa bonasia), and willow grouse (Lagopus lagopus) were monitored on both islands throughout the study, as were those of small mammals.

Predator removal on each island caused increased chick production-on average, broods from predator-control areas contained 68% more young than those in areas where predators were not removed. Furthermore, on islands with predators, 59% of females produced broods, as compared with 77% on islands without. These figures correspond to a 2.2-fold increase in chick productivity following predator removal, from 1.94 young/hen to 4.25. Adult populations were, however, less affected by the removal of predators-a 2.2-fold increase in productivity was predicted to lead to a 2.5-fold increase in adult population, but the counts that revealed the productivity change showed just a 1.6-1.8-fold increase. No change was detectable in the small mammal populations, which continued to cycle on both islands throughout the study. However, when predators remained on each island, the gamebird populations fluctuated in synchrony with the small mammals, while this relationship

was destroyed in the absence of strong predation. It has been suggested that predators "transmit' population cycling to gamebirds by preying more heavily upon them when their principal prey, small mammals, are at low abundance.

It seems, then, that fox and marten predation influences the gamebird populations through the medium of chick survival, and that the predation pressure varies through time according to the abundance of alternative prey. Thus, when small mammals are rare, the predators suppress the gamebird populations more effectively--especially just after a rodent "peak" when large numbers of predators, experiencing a rapid decline of their mammalian prey, take many gamebird chicks.

Studies of predators other than canids suggest that factors exogenous to predator-prey interactions may be responsible for regulating ungulate populations. In the Serengeti, Tanzania, wildebeest populations have increased dramatically since the elimination of rinderpest despite concurrent increases in the lion population (Norton-Griffiths and Sinclair 1979). In Idaho, both mule deer and elk increased in numbers despite heavy predation by both mountain lions and humans (Hornocker 1970). That predators may, in some circumstances, exert control over prey populations is highly probable. Predators are not, however, uniquely responsible for fluctuations in prey numbers.

Reducing Losses-Lethal Control

In the great majority of cases the level of livestock losses attributed to canids appears to be exaggerated. Similarly, the data suggesting that canids and man are in direct competition for game are ambiguous. Even if every death ascribed to canid predation is verifiers in many cases such small percentage losses suggest that it should be possible to reduce losses to a level which will allow wild canids, livestock, and game to coexist.

In principle, one can seek to cut losses to predators either by reducing the numbers of predators (lethal control) or by reducing their access to, or the availability of, the prey (non-lethal control). In practice, predator-reduction is much the most common approach and, on first principles, much the least promising. Canids can be killed in a number of ways, by shooting, poisoning, or trapping. Each of of these methods has its relative advantages and disadvantages.

From the point of view of conservation, lethal methods that are not species specific (trapping, poisoning) frequently result in the inadvertent killing of "non-target' animals. In areas where common predators coexist with rarer animals, nonspecific lethal controls result in what can only be called reckless endangerment of the rare or vulnerable species. In the Soviet Union, Ovsyanikov (pers. comm.) suggests that poison bait programmes aimed at elimination of wolves may have inadvertently eliminated dholes in areas where the two species overlap. In Italy, poison baits laid to kill red fox are known to be dangerous for dogs, children, and wolves (Boitani 1982).

No matter what method of lethal control is employed, however, reducing predator numbers is an expensive process that requires a long-term commitment on the part of a government or private producers. Although data that elucidate this are few, we have chosen two examples which illustrate this point. The case of the arctic fox in Iceland will serve to demonstrate this point in relation to livestock (P. Hersteinsson has kindly provided the following unpublished data); the second, that of wolves, can be used to examine the problem in relation to game species.

Loss of livestock, particularly lambs, is the major reason for attempts at controlling arctic fox in Iceland. Each year, a maximum of 2-4% of Iceland's lambs are lost to predation by foxes (the true figure, for reasons cited previously, is probably much smaller). The annual crop of lambs is approximately 900,000 individuals, and the value of each lamb is U.S. $50, resulting in a mean loss to ranchers of U.S. $0.9 to U.S. $3.6 million per annum from predation.

Hunting to control foxes has been undertaken for centuries. The earliest laws promoting hunting of arctic fox in Iceland date from 1295 A.D., while the legislation governing the present hunting was enacted somewhat more recently (1958). Hunting of the arctic fox in Iceland is jointly sponsored by the Department of Agriculture (2/3 of costs) and local authorities (1/3 of costs). The hunting is undertaken by professional hunters hired by local authorities. The cost of this program is approximately U.S.$200,000 per annum. The hunt is a year-round activity and kills, annually, approximately 900 adult foxes and 1,300 cubs in a population of 2,000 adults.

A hunting programme which costs the government approximately 10% of the value of the annual loss in lambs might be viewed to be a success. However, whether the hunting has any impact on fox populations, or livestock losses for that matter, is debatable. Despite intensive hunting, with the annual harvest of nearly 50% of the adult fox population, arctic fox population levels in many parts of Iceland have been increasing since they reached their low in 1974. Fox populations appear to be cyclical and are probably not regulated by hunting. The pattern of population and sub-population growth is unclear, however it appears that a protozoan parasite, Encephatitozoon cuniculi, may contribute to regulation of the long-term cyclical patterns of population size. Of course, control of predators on an island is easier than it is on the mainland where immigration and emigration further complicate control efforts.

In those situations in which wolves have been shown to depress ungulate population levels, the most frequent management response is to kill wolves. But wolves, like many canids, can be very productive. A wolf pack usually produces one relatively large litter of five to six pups. When populations are not saturated, 2241% of all wolf packs produce multiple litters (Harrington et al. 1982). In other areas, only 10% of the packs may produce multiple litters (Ballard et al. 1987). Hence, reducing wolf numbers may result directly in a larger number of young wolves.

What this means is that even when suffering an annual mortality of 50%, a wolf pack can remain stable in its numbers. Natural mortality appears to vary between 10% (Mech 1970) and 20% (Ballard et al. 1987). Removing 20% to 30% of the population appears to result in stable population numbers (Pimlott 1967, Gassaway et al. 1983). To effect a significant reduction in wolf population levels, a control programme must kill, annually, between 40% to 50% of the wolf population, although in some areas, removal of greater than 35% may be "all" that is necessary to reduce wolf populations (Keith 1983). These estimates were confirmed in a recent study (Ballard et al. 1987). An experimental reduction of wolves of 42%-58%, although resulting in reduced wolf numbers, also led to several new waves of immigration and an increase in births.

Nature abhors a vacuum, and carnivores appear to be no exception to natural laws. Canids are frequently killed as a control measure in areas where ecological factors such as abundant food have made them a pest. But for every predator killed, there may be another just waiting to move into such prime habitat. A local lethal control programme for wolves in Canada did not decrease wolf number but led to an increase in immigration with new wolves moving into the area to compensate for artificially low population densities (Ballard et al. 1987). This clearly shows that even total removal will only result in a temporary respite from canid predators unless an area is completely isolated. In those cases where a short respite from predation allows a prey species to "escape" the predator's control, a temporary reduction may be all that is needed.

The role of humans in altering the ecological landscape cannot be ignored. Development and human activities, even in the most remote areas, can have a profound impact on predator-prey interactions. For example, during calving, caribou "escape" predation from wolves either by spacing themselves out in areas where wolves are rare (woodland caribou) or by giving birth in calving grounds outside the range of wolves (tundra caribou). If humans disturb the mobility of the caribou, the impact of predation can greatly increase (Bergerud 1988).

Even in areas where predator numbers have been drastically reduced, such as wolves in Spain (Salvador and Abad 1987), Italy (Boitani 1982), and Norway (Naess and Mysterud 1987), or African wild dogs in Zimbabwe (Townsend 1988) or South Africa (Mills pers. comm.), the conflict between producer and predator continues. Several authors have noted that a few wolves have evoked spirited and vociferous debate throughout Norway (Naess and Mysterud 1987; Anon 1989).

Reducing Losses-Non-Lethal Controls

While humans may be limited in the means available for killing predators, numerous methods of non-lethal control have been pursued. A variety of non-lethal controls have been tested. The effectiveness of various control measures for the prevention of predation by coyotes has recently been reviewed by Andelt (I 987) and in a volume edited by Green (1987). Their findings, augmented by results from studies on other species of predators, are summarized in Table 2.

The path of least resistance to a state resembling harmonious coexistence of predators and livestock producers involves limiting the opportunity for conflict. Improvement in animal husbandry may not be costly and may have significant results.

Table 2. Efificacy of Non-lethal Control

Method Canid Prey Region Effectiveness and Comments Source
Herders Coyotes Sheep U.S.A. Presence of herders reduces losses. Andelt 1987
Guard Dogs Wolves Sheep Portugal Greatest deterrent is sheep dog and herder. Flower 1971
Coyotes Sheep U.S.A. Dogs can greatly reduce predation problems. Andelt 1987
Wolves Sheep Italy Losses greatest in areas not using traditional herding/sheep dog methods. Boitani pers. comm.
Coyotes Sheep Western U.S.A. 80% success rate. Dogs cost effective. Green and Woodruff 1987
Disposal of Prey Carcasses Coyotes Sheep Various studies Carrion may attract coyotes, introduce sheep as food. Burial or removal reduced predation loses. Andelt 1987
Wolves Cattle Minnesota Carrion may attract wolves and introduce them to cattle as potential food. Fritts 1982
Confinement Coyotes Sheep Kansas U.S.A .Predation largely nocturnal; night-time confinement greatly reduced losses. Andelt 1987

Proper disposal of livestock carcasses, either those killed by predators or those resulting from natural mortality, appears to reduce subsequent predation. Confining animals at night, or during their infancy when they are most vulnerable, also reduces losses to predators.

Perhaps the most cost effective method of non-lethal predator control is the one we have used historically: guard dogs. Guard dogs, in conjunction with shepherds, have been used for millennia throughout the world. Of course, dogs are most effective in certain situations. Their efficacy is increased in small herds, and in the presence of a shepherd. An international trend to increasing the scale of production in all aspects of agriculture may limit the traditional use of guard dogs.

Several studies note, however, that guard dogs in conjunction with fencing can greatly reduce livestock depredation by carnivores (see Green and Woodruff 1987). As 80% of all of sheep producers, and 50% of all sheep produced in the western United States are raised within fenced pastures (Green and Woodruff 1987), the increased use of guard dogs may be particularly effective in these areas. Boitani (pers. comm.) argues strongly that traditional use of guard dogs by Italian shepherds was pivotal to the historical coexistence of wolves and sheep.

Most lethal control programmes, such as the one described in Iceland, attempt to limit the growth of the predator population. A simpler, and possibly more cost effective way to achieve the same goal is to reduce predator fertility. Behavioural techniques seem destined to failure (Barnum et al. 1979). Recent studies suggest that anti-fertility drugs, administered through improved baits, may be an economical means of predator control (Stellflug et al. 1978). A second method of fertility reduction, hormone implants, has been used successfully in field trials to limit reproduction in the African lion, Panthera leo (Orford et al. 1988).

Methods of repelling canid predators have not been extensively explored. Those trials which have been made have found that most repellents appear to be either useless, or not cost-effective. Behavioural modification of reproduction, whereby coyote calls are played at a high frequency to simulate dense coyote populations, had no effect on reducing fertility. The use of emetics and repellents, although occasionally providing short-term deterrence, appeared not to be cost effective. However, this approach merits more attention. Frightening devices and live trapping were of some use in particular circumstances.

One option in predator control that is rarely discussed involves the choice of "prey." The data in Table I suggest that rates of predation by canids vary with the species of livestock being raised and the predators which are most abundant Goats appear particularly vulnerable to predation from coyotes, while cattle, for the most part, are relatively immune. The choice of the type of livestock to be raised in a particular area may be influenced by various factors: market demand, the type of forage available, cultural traditions, or historical accident. However, in areas where a particular species of predator is causing problems with a particular species of livestock, one answer is to re move the predator. Where culturally, ecologically, and economically possible, a simpler solution might be to raise a different type of animal.

Reducing Losses-Public Opinion

Public attitudes to the various methods of lethal and non-lethal control of canid predators, and regional differences in opinion, have to be taken into account by the controlling authorities. In a survey conducted in the United States, attitudes to various coyote control methods varied widely among methods (Fig. 1). Clearly, non-lethal control measures are far more acceptable than methods that involve killing coyotes. The only exceptions to this are forms of direct compensation to farmers and ranchers. However, compensation is not a form of control, but a form of governmental recognition that the costs of predation should be borne by the public at large. As noted above, compensation without confirmation of canid depredation can lead to abuse; perhaps people are reacting to abuse of the system, rather than justified claims for compensation?

High Public Acceptability
Non-Lethal Controls Lethal Controls
Guard Dogs (7.1)
Repellent Chemicals (7.0)
Antifertility Drugs (5.8)
Shooting from ground or fast acting poisons (4.3)
Compensate Ranchers (3.2)
Aerial Gunning (2.5)
Killing pups at den (2.3)
Pay ranchers not to raise sheep (1.8)
Steel leg-hold traps (1.6)
Slow-acting poisons (1.3)
Low Public Acceptability

Figure 1. Public acceptance of methods of coyote control (after Andelt 1987).

In the above analysis and discussion, we have assumed that the important question to ask is how to mitigate the costs of depredation of livestock by canids. However, in many circumstances, the costs of predation may be irrelevant. When wild predators kill domestic animals, people often react instinctively rather than logically. For instance, in Norway, five or ten wolves eat approximately 0.02 percent of the county's annual sheep production. Despite the infinitesimal losses incurred, these few wolves have invoked a spirited public controversy (Naess and Mysterud 1987). Such a controversy would not have occurred had domestic dogs been responsible for the losses, or had a few sheep died in a spring snowfall. Canid predation, or more generally predation by wild carnivores, is seen as a loss which can and should be controlled: not an act of God, but the result of negligence on the part of the producer. Altering this pre-conception sufficiently that a low level of canid depredation is acceptable may do more for the conservation of canids than anything else.

Canid Action Plan Table of Contents CSG homepage References

© 1990 International Union for the Conservation of Nature and Natural Resources