Review of the Victorian Wild Dog Management Program and Recommendations for Future Approaches

Arrangements for the control of pest coyotes vary across United States of America jurisdictions. Coyotes are a native animal and come under the jurisdiction of agriculture and wildlife services both federally and at a state level. In most states, coyotes are not declared game but a licence is required to hunt or trap.

Interventions can include non-lethal and lethal interventions (Table ). In some states there is a preference for using lethal means where predation has occurred. Moreover, most states adopt a hierarchy system of intervention such that there is a preference for more intervention as the type of incident increases in seriousness from sighting through to predation and threatening behaviour towards humans.

Table : Types of coyote control measures

The provision of bounties varies: some states operate bounty systems, for example, North Dakota, Utah ($50 per head) and Missouri, some local jurisdictions of certain Canadian provinces such as Alberta and Saskatchewan; in other states such as Kentucky there are no bounties. There is debate over the efficacy of bounties and jurisdictional legislatures have debated their introduction as a means of encouraging local participation and control. However, the United States Department of Agriculture (USDA) Wildlife Service note that bounties are not a successful tool because they do not reduce coyote populations and redirect resources from more effective control activities by the agency. There are also concerns that bounties can encourage inhumane take, with local less skilled hunters using predominately shooting as their control methodology as opposed to highly skilled professional coyote control services.

There are restrictions on the use of control measures. In some jurisdictions a hunting licence is required to undertake shooting, trapping and snaring. Use of lethal toxicants is regulated federally through the Environment Protection Authority (EPA), and can only be used by authorised USDA Wildlife Service controllers, but their use is restricted by state laws – some allow the use of lethal toxicants whereas other do not. For example, CPEs are allowed in some states such as Virginia but not in others such as Kentucky.

In some states these restrictions on use reflect differences in management strategies between states. There appears to be two broad groups:

preference for non-lethal measures — this is more common in states where coyotes are considered a public nuisance problem rather than a farm production problem. The preference appears to be more focussed on targeted removal of individual problem coyotes that might challenge humans or attack animals, based on the philosophy that once problem coyotes are removed the remaining pack will revert to cautious and human avoiding behaviours.

targeted lethal measures — in a number of states where coyote predation is a more pressing agricultural problem (particularly those with the highest concentration of sheep farming such as Ohio, West Virginia and Virginia) there are coordinated programs of coyote control.

Knowlton et al.,(1999) concluded coyote depredation rates appear to be influenced by sheep management practices, coyote biology and behaviour, environmental factors, and depredation management programs. They found that successful control requires the use of multiple techniques within an integrated program and concluded:

local population reduction can provide temporary relief to sheep operators, but only until the local coyote population compensates for the removals and fills vacant territories;

the more focused removals are to the area of depredations, the shorter the duration of the effect; and

depredation relief resulting from a coyote population reduction program should be considered transitory, unless the removal program is maintained.

Various commentaries on coyote control in the United States indicate there is debate, among others, over the focus of programs including;

the true population of coyote and extent of control required;

the effect of take on the structure and territory packs and effects on livestock attack; and

the use of bounties to encourage take.

1.2.1Virginia Cooperative Coyote Control Program (VCCCP)

The USDA Wildlife Service operates control programs involving integrated pest management in Virginia, West Virginia and Ohio. These states have relatively higher densities of sheep farming to other states and more issues with coyote predation.

It is a cooperative program operated by the USDA Wildlife Services with costs shared 50:50 between the USDA Wildlife Service and the Virginia Department of Agriculture and Consumers Services.

In 2014, the USDA Wildlife Service provided direct control to nearly 200 livestock farms across 53 counties - 285 sheep, 81 calves, and 32 goats were verified killed by coyotes on those farms.

The program operates two forms of intervention:

Preventive control — occurs seasonally on farms with historical incidence of predation. This involves the removal of territorial coyotes between January and April before pups are born. In 2012-13, 160 landholders were assisted. Of the 87 landholders that had coyotes removed, none experienced subsequent predation in that year.

Corrective control — occurs at all times through the year in response to reported predation. In 2012-13, controls were implemented on 73 farms.

The program has adopted an integrated predation management strategy where any or all practical and legal methods are applied simultaneously or sequentially to prevent or reduce predation. Landholders can implement non-lethal (e.g. fencing, shed lambing and other husbandry) and some lethal methods (e.g. shooting, trapping and snaring) and usually request USDA Wildlife Service assistance when predation becomes overwhelming or preventative measures are warranted.

Since the program commenced in 1990, control measures have shifted from trapping to the use of CPEs which in 2013 accounted for 65 per cent of the coyotes killed by the USDA Wildlife Service in Virginia (Table ). Fox (2014a) reported the program killed 341 coyotes in 2013 and this represented a tiny proportion of an increasing coyote population and is dwarfed by the estimated 24,000 plus coyotes killed by hunters. Nonetheless Fox (2014a) emphasised the focus of the program was on dispatching problem coyotes and under taking preventative control:

Preventive control is implemented primarily from January through April on farms with historic predation. Preventive control strategies remove territorial coyotes before pups are born, which decreases the predatory behavior of coyotes during the lambing season (Wagner and Conover 1999). Of the 160 livestock producers assisted, 87 farms with historic coyote predation losses had coyotes removed to prevent predation. Of the farms receiving preventive control, 41 were sheep farms, 44 cattle farms, and 2 were goat farms. These farms had no livestock killed by predators in FY2013.

CPEs are more efficient than other measures: traps and snares must be inspected daily whereas CPEs and Livestock Protection Collars (LPCs) require inspection every 7 days (Louney, Houben and Eggborn 1997).

Livestock Protection Collars (LPC) are used in specific circumstances where CPE use is constrained as a result of use regulations and predation is a pressing problem. LPCs also contain restricted chemicals and their use is also regulated by the EPA. In 2013, LPCs use resulted in only one coyote removal in Virginia.
Table : Lethal methods used by Wildlife Services in Virginia 2013 (n=341)

Lethal method	Number of coyote removed
CPEs	220 (65%)
Snares	100 (29%)
Foot-hold traps	12 (4%)
Calling/shooting	8 (2%)
Livestock Protection Collar (LPC)	1 (<1%)

Source: Fox 2014a

Education and extension are an important component of the program with the USDA Wildlife Service providing advice to landholders on on-farm management practices. In 2013-14, more than 1,200 people attended 24 educational events run by the program on coyote ecology and damage management.

The VCCCP has resulted in a general reduction in the incidence of livestock loss to coyote predation and an increase in the number of coyotes removed from properties. The lagged pattern of predation closely maps the pattern of coyote removal in the previous year Figure .

Figure Number of sheep killed and coyotes removed in Virginia

(note below - sheep killed per farm on right hand y axis)

Source: prepared from data provided by Fox 2014

According to Fox 2014b, however, there is some evidence that removal of coyotes from the local pack group can have counter effects on predation such that increased removal increases predation over the short term as removal impacts pack structures, breeding pairs and learned scavenging behaviours.

The success of the program has been in spite of evidence that the coyote population is continuing to rise. The rise in the coyote population is evidenced by:

the substantial increase in the removal of coyotes by the program and by private hunting. For example, in 1993 less than 1000 coyotes were removed annually by private hunters and by 2013 close to 24,000 coyotes were removed annually by private hunters; and

49 new farms were assisted with livestock protection by the program in 2012-13.

2.

3. Assessment of efficacy

3.1Key points

3.2Framework for assessing efficacy and examining potential improvements

The Wild Dog Control Program uses a range of lethal control measures with the aim of reducing dog attacks and livestock losses. This is summarised in Figure . We focus on key lethal control measures given absence of detailed data on the distribution and level of use of non-lethal controls. We also include analysis of key output measures including estimates of dog take — dogs trapped and estimated bait take.

In the following sections we examine the performance of the Wild Dog Control Program with reference to inputs, outputs and outcomes – both at an overall level and at a regional level. We do this by considering performance measures outlined in the Directions Statement for the Wild Dog Management Program in Victoria and our own analyses.

Our assessment of the efficacy of the program is limited to the period 2012/13 to 2014/15. Differences in approaches to recording of data pre 2012 mean that pre 2012 data are not directly comparable across inputs and outcomes or across years.

While our analysis at an overall program level will be based on annual data, our analysis at a regional level will be more disaggregated and have the following features:

the data are rearranged into seasonal periods (i.e. summer, autumn, winter, spring). This better aligns with the periods of pack reproduction cycles such as mating, and whelping and farm system activities such as lambing (i.e. autumn and spring); and

the period of analysis is from summer 2013 to spring 2015 (2 years of quarterly data). In some cases, it is from autumn 2014 to spring 2015 since summer 2013 does not have data for December 2013.

Therefore, while we are able to make comparisons across three years of data at an aggregate level, our regional analysis is restricted to two years of data. This is because the Dogbytes database only contains reliable data from summer 2013/14 onwards.

Figure : Key program inputs, outputs and outcomes

3.3Overall program performance

3.3.1Overall program inputs: control measures

The current approach to wild dog management comprises a mix of control measures with ground baiting (community and DELWP) and trapping constituting the large majority of activity over the most recent three years (Figure ). Noticeably, there has been an increase in community ground baiting and government ground baiting over this three-year period. This has been accompanied by a decline in DELWP trapping. Shooting contributes a very small amount of the overall control measures and is not illustrated in Figure .

The increase in baiting has resulted from a shift in program focus over the last three years from reactive public service provision toward more proactive public service provision and improved community engagement.

Note that our assessment of statistical significance for overall program measures considered t-stats at the 95 per cent confidence level (2-tailed).

Figure : Change in program effort 2012/13 – 2014/5



Source: Marsden Jacob estimates based on DEDJTR and DELWP data

The change in program focus and increase in baiting has also been made possible by an increase in government program expenditures which have increased from close to $4.5 million in 2012-13 to just under $5 million in 2014-15.

Figure : Change in program expenditures

The overall increase in baiting and increased involvement by the community in baiting over the period 2012/13 to 2014/15 is consistent with the relevant strategic objectives outlined in the Directions Statement and performance against these objectives – relevant for strategic objective 3 and 6 (Box 1).

It is difficult to gauge trends in the effectiveness of the two key control measures (baits and traps) over the last four years as the data does not exhibit a statistically significant trend upwards or downwards. In part, this is likely to be the result of the short time period over which we have estimated trends. Effectiveness is illustrated in Figure for baits and in performance against Direction Statement strategic objective 4 for traps (Box 2).

Effectiveness of inputs in producing outputs is defined as the ratio of number of baits taken by wild dogs or dogs trapped to the number of baits laid or traps set.

It is important to consider the effectiveness of deploying baits and traps in killing dogs in the context of overall program effectiveness as there is an uncertain relationship between the numbers of dogs killed and livestock attacked using the data available to us.

It is also important to note that as dogs are killed over time through control activities, it is possible that the effectiveness of control measures will fall as it becomes increasingly more difficult to kill the remaining dogs. Indeed, it is possible that some dogs learn to avoid some control measures.

Figure : Baits effectiveness – overall

1. 
   Effectiveness is a measure of baits taken by wild dogs per bait laid. Effectiveness in the graph only refers to DELWP ground baiting as data on effectiveness is not available for community baiting. It is assumed that if a bait is taken by a wild dog the dog is killed.
   
2. 
   Effectiveness has been calculated as the proportion of baits taken to baits laid (sourced from DELWP data) multiplied by the proportion of baits taken by dogs. The proportion of baits taken by dogs was estimated by DELWP at 10 per cent on average. This is discussed more in the notes to Table . Therefore, effectiveness over the period 2012/13 to 2015/16 (part) only varies across years because of variations in the proportion of baits taken to baits laid since the proportion of baits taken by dogs has been fixed at 10 per cent as there is no actual observable data available for this variable.
   
3. 
   Partial 2015/16 data (first six months of this period) has been included to measure 2015/16 effectiveness. Number of baits laid has not been included as the year has not been completed.