There is rapid growth in precision livestock farming technologies for grazing livestock (Aquilani et al, 2022). Most wearable sensor systems are passive, providing data to livestock managers, with diagnostics for health, nutritional or reproductive states or events such as oestrus and calving. Animal location is often an important element of outdoor systems with Geographical Navigation Satellite Systems (GNSS, often referred to as GPS) creating map-based visualisations (Herlin et al, 2021).
Virtual fencing (VF) or virtual herding (VH) systems are a step up from passive sensors with active livestock management and precision livestock farming technology. VF systems use two-way digital communication between livestock managers and each animal via its wearable device. Stock managers can then actively, but remotely, manage boundaries for the herd, opening up or restricting movement in near real-time. One current commercial system goes a step further with signals prompting directional movement, to automate moves to and from pasture to milking parlour. Increasing quality of data and automation creates opportunities to reduce farm hardware such as fences and gates, reduce labour for moving stock and increase efficiencies around surveillance of livestock. However, there are undoubted welfare costs and risks to animals, leading to opposition by some commentators, e.g. RSPCA (2022), because they give electric shocks for crossing the virtual fence line.
Currently, there are four virtual fencing systems being commercialised globally with somewhat different technical and capability characteristics (Goliński et al, 2023). These are summarised in Table 1. The Nofence system for cattle, sheep and goats has rapid user uptake in Norway, UK and ROI. Vence, recently bought by Merck, is focussed upon cattle ranching in USA. The directional VH system Halter uses combined sound and vibration for milking dairy cows in New Zealand, where over 100 000 cows are wearing collars. The Agersens e-Shepherd, bought out by Gallaghers, the New Zealand-based electric fencing company, focuses upon cattle, but is yet to substantially launch commercially, having had a major re-design over the last few years.
Table 1. Summary of different commercial virtual fence systems
| VF System | Website | Target Species/system | Communication modes | Alert stimuli | Main regions where deployed | Collar numbers (estimated)1 |
|---|---|---|---|---|---|---|
| NoFence cattle | https://www.nofence.no/en-gb/ | Non-dairy cattle, beef cows, large growing stock | GSM (mobile phone) and locally Bluetooth2 | Sound | Much of Europe and North America | UK 8000 and global 60 000 |
| NoFence sheep and goat | https://www.nofence.no/en-gb/ | Adult sheep and goats, some farmers have used on dairy youngstock | GSM (mobile phone) and locally Bluetooth2 | Sound | Much of Europe and North America | c.15% of above are sheep/goat |
| Halter | https://halterhq.com/ | Dairy cows | LPWAN3 | Sound and vibration | New Zealand and recently Australia | Over 100 000 |
| Vence (now owned by Merck) | https://vence.io/ | Non-dairy cattle, beef cows, growing stock | LPWAN3 | Sound | USA and some parts of Australia | Thousands in N America, and a few large ranches in Australia |
| Gallaghers/eShepherd (formerly Agersens) | https://am.gallagher.com/en/new-products/eShepherd | Was beef cattle, but there may be dairy cow plans | LPWAN3 | Sound | Australia and New Zealand (trial marketing only) | Unknown |
Some of these estimates are directly from company staff, others are estimates from third party discussions.
2.Bluetooth connectivity – within field and within sight enables collars without GSM connection to be updated, e.g. to enable stock to be released from virtual paddock directly
3.LPWAN, Low Power Wide Area Network, sometimes protocols referred to as LoRa, requires gateway with aerial connecting to collars high on farm buildings or on a tower, gateway connects to local internet connections or directly by GSM.
The core capability is to give animals a collar-based electric shock for crossing a virtual boundary set remotely and managed via GNSS. Due to this shock element, many countries do not allow sales, with ongoing concerns about welfare and ethics of the product family. Rapid developments led the UK Government's independent Animal Welfare Committee to recently examine these novel technologies. The resulting publication ‘Opinion on the Welfare Implications of Using Virtual Fencing Systems to Contain, Move and Monitor Livestock’ (Animal Welfare Committee, 2022) considered early adopter experiences, published scientific information and evidence from commercial developers. The title referred to ‘contain, move and monitor’ and all are reflected in this article.
The technologies
Figure 1 highlights the main generic components of VF. This article will aim to cover these capabilities as a family, though will necessarily mention individual products. In summary, all animals in the grazing group wear ‘sat-nav’ collar units that give warning sounds to each animal as it moves into close proximity (a few metres) of the virtual boundary. An electric shock is given after further movement forward. The full boundary may be multi-layered with repeated sound and then shock pairings if the animal proceeds forward, effectively stopping when the animal has fully escaped. These negative stimuli decrease or stop as the animal turns away. Any escaped, or partially escaping animal, can return to the virtual paddock with no negtive stimuli, it is a strictly one-way barrier. A virtual paddock is set for the grazing livestock group from the user's smartphone or laptop, transmitted to each collar over system communication routes. Core activities of collar units, communications and visualisations are controlled from each company's computer servers in a ‘back-office’ management system, with communication to/from each collar, the ‘back office’ and the user's computer or smartphone. The collars can continue to function autonomously through patchy or faulty communication coverage. For the user, each commercial system provides maps and controls within smartphone apps that receive information about the animal and its collar/virtual fenceline, also enabling rapid changes or removal of boundaries or livestock movements. For the Halter system alone, directional vibration signals provide further prompts, e.g. to come in for milking. Maps with VF boundaries and ‘pins’ (with ID) for each animal's location are app features which can be interrogated about each animal's behaviour and can link to animal database information. Being virtual, all this data can be shared ‘live’ amongst farmstaff, and with other interested parties, such as veterinarians. Historic data (yesterday, last week, last year) can be accessed.
More than just a fence replacement
Of the three aspects of the Animal Welfare Committee (2022), ‘contain, move, monitor’, the containment element has been most researched, and only recently are monitoring elements being independently studied and published (e.g. Versluijs et al, 2023). There is little information available in the scientific literature about the ‘move’ element, or virtual herding. The monitoring capabilities, still in further development, differentiate VF systems from being simple replacements for electric fencing. The technical operation of remotely managed virtual fencing is dependent upon GNSS access and communication to and from the manager. On-board accelerometers provide further high quality monitoring data. Combined monitoring capabilities enable VF system developers to provide further added value for producer buyers. These behavioural diagnostics and pasture management elements are likely to expand further because of competition with other precision livestock farming technologies.
Do virtual fencing systems work effectively?
All the VF systems continue to evolve so that commercial systems are not the same as those in earlier research articles. In terms of simple effectiveness as a fence, there is good evidence (Lomax et al, 2019; Campbell et al, 2020; Boyd et al, 2022; Hamidi et al, 2022). Many publications show how animals quickly learn and the virtual fenceline becomes effective and numbers of shocks diminish (e.g. Hamidi et al, 2022). All companies emphasise the need for simple training phases when all livestock experience the combination of warning signals and electric shock to create associative learning. Thus each animal, after training, can manage predictable and controllable warning and shock signals (Lee et al, 2018).
Some instances are described where VF is not fully successful, often where limits are tested. Failures were noted, for example, where only one-third of the sheep flock was collared (Marini et al, 2020) or gaps between cattle groups in different virtually fenced paddocks were too small (Verdon et al, 2021). High levels of success for virtual boundaries set according to manufacturer's guidance has been a clear finding from many tens of research publications.
The electric shock
The autonomous electric shock within all VF/VH systems is the element that elicits a prime concern for animal welfare. Many national and regional authorities have legislation that prevents use (RSPCA, 2022; Staahltoft et al, 2023).
It is difficult as humans to understand how animals experience pain from electrical systems, whether standard electric fences, prods or goads or VF systems. Whiting (2016) stated that ‘what an individual non-human animal experiences as pain is unknowable, but presumably is in some way similar to human pain’. Grumett and Butterworth (2022) more recently and specifically considered the welfare and ethical issues of VF along with standard electric fences. Whilst identifying that all electric shock systems, fixed wire or VF, were painful, these authors did not identify any particular thresholds for acceptable power use, though noted the principle of minimal power to achieve their objectives. In the Animal Welfare Committee's (2022) opinion, it was noted that to be ethically justifiable, all aversive stimuli must bring some clear welfare benefit that is not realistically deliverable by a non-aversive method. This report also described examples of opportunities for welfare benefits, which are illustrated further in this article. The capability of the trained animal to predict (through warning sounds) and control (by turning away) future shocks is critical to minimising the cost of the electric shock element to each animal.
Trained stock rapidly manage the fenceline using warning sounds rather than electric shocks, with rapid reductions in shocks per animal (Campbell et al, 2017). Keshavarzi et al (2020) reported that livestock observe herdmate responses to either warning sound or shock and hear neighbouring stock receiving the warning sound, so creating sophisticated learnt responses to the virtual fenceline.
Applications of virtual fencing and herding systems in practice
The farming system, the goals of the managing farmer and the level of self-dependency of the livestock affect how these VF systems operate. Different farming systems have different risks and challenges for animal and livestock manager and, unsurprisingly, VF system setups will have different types and levels of risks, alongside different farm management objectives.
Two contrasting farming contexts are illustrated, each with different opportunities for changing labour use, grazing efficiency and infrastructure costs resulting from different strategies for VF operation. The Animal Welfare Committee (2022) report provides further narratives about these two approaches, particularly for UK contexts. Table 2 summarises the welfare issues, alongside observations on VF use, that may arise.
Table 2. Summary of welfare risks and opportunities associated with virtual fencing (VF) and virtual herding systems
| Welfare issue (I), risk (R) or opportunity (O) | Production context | Comment |
|---|---|---|
| Electric shock during training phase (I) | Training | Essential that all livestock receive shock to associate with warning signals. |
| Electric shock from VF challenges or escapes (I,R) | All | Animals making choice to escape virtual boundary and then experience the shock(s). |
| Electric shock given by error (R) | All | GNSS error, or mapping/operator errors may give animals shock when in wrong place, wrong time. |
| Neck collars themselves, risks of rubbing or being caught in trees or wire fences (I,R) | All | All collars, including cow bells have these risks. Animal Welfare Committee (2022) advised that multiple collars (e.g. a VF collar and an accelerometer collar) should not be worn together. Good design and pre-sale testing should mitigate. |
| Real-time remote data (O) | All | Check on animals, and on VF activity, without visual observation or as a triage opportunity using 1) locations, 2) location changes and 3) activity levels. |
| Alerts of shocks, escapes (O) | All | Capability to check that any virtual fence border breaches are problematic for animal or management and quickly rectify either physically or remotely/virtually. |
| Location mapping (O) | Mainly extensive | Find animals quickly for physical checks, including counts and checks for any outlier animals. Identify outlier animals for physical checks. |
| Create a boundary without physical equipment, where the boundary materials (e.g. barbed wire) are a risk (O,R) | Mainly extensive | Removing/reducing risks (to livestock, wildlife and farm dogs) created by wire, especially barbed wire.Enable livestock to escape relatively easily (such as in snow or in dog attacks), but potential to put livestock in more risky scenarios as a result of capability. |
| Create boundaries where human-based risks e.g. open public roads or to separate stock from human access routes (O) | Mainly extensive | Many Commons are used by multiple stock herds/flocks and this may create issues depending upon herding characteristics. Cannot be used for horses/ponies. There is increasing wariness of livestock and human interactions, especially cattle attacks in open access areas, footpaths across fields. |
| Using VF to fence off ‘danger’ areas (O) but any VF breakdown might put stock at greater danger as now in danger area (R) | Mainly extensive | Stock will be virtually fenced away from areas of danger within a paddock, using virtual exclosures or excluded fully by paddock boundary. However, the technology may encourage access to grazing areas previously not used and thus animals may be exposed to greater risk, even if partially protected. |
| With greater remote, virtual data, then visual inspections might be reduced (R) | All | The balance between greater remote data, more easily achieved physical inspections and risks of reduced inspections will depend upon the managers. As a tool this is in the hands of managers to use responsibly. |
| Virtual herding removes/reduces risks/challenges in human and dog-based musters (O) | Dairy with Halter and extensive with others | For Halter, cattle move at their own speed and are not at risk of being chased at the back of the herd. In extensive systems, the ‘open-up in front, close off behind’ should be more relaxed, a slower pace. But any unseen risks (such as a flooded gully or a broken gate) might create risks to stock injury and lame or injured stock may be isolated or left behind. Little narrative or experimentation in scientific or grey literature. |
| Virtual herding acts as goad to any tail-enders in virtual muster or shocks any animals left behind by data communication errors (R) | All | Software can mitigate this by preventing any ‘backfence’ from approaching livestock in a virtually and remotely managed movement. Physical observation of virtual herding may overcome many risks. |
| Breakdown of system, e.g. failures of GNSS ‘system’, company ‘back office’ or long/medium term IoT or GSM communications may cause system breakdowns (R) | All | Farm systems with no (or reduced) fences or gates may be challenging to revert back to traditional systems. Physical external boundaries are essential to cover breakdowns due technology failures or human failures. |
| A relationship breakdown between technology supplier and user may involve supplier turning off the system (e.g. over billing or welfare concern) (R) | All | Withdrawal of the service may be quite complex. Some collars systems the collars are owned, in other scenarios they are leased. Subscriptions to providers cover communications and all ‘back office’ operations and withdrawal of operations would prevent collars being updated or they are turned off. Stock out in the field wearing collars may no longer have virtual boundaries. There is little case history of problems. |
| Over-confidence in systems may lead to failures, that create problems (R) | All | Using VF to keep cows and ewes in oestrus away from bulls, rams may not work leading to unwanted pregnancies.Escapes into dangerous areas, such as gorges or forestry or across watercourses, may create challenges for escapees, though GNSS location should aid rapid alerts and recovery. |
| ‘External’ scrutiny (O) | All | Shared access amongst farming staff and managers, with other interested parties (e.g. farm consultant or veterinarian) may provide oversight as well as back-up cover, so that welfare oversight, and actions, are not performed by a single person.At least one company reserves the right to turn off the system if welfare abuses are detected/suspected, though how this works in practice is unclear. |
Intensive grazing management
Strip or paddock grazing systems require close control of stock and often use standard electric fencing systems for internal divisions, and managing and rationing grass supply. Using standard wire or electric fencing in all frequent-move systems requires much labour input, with much stock checking at pasture. VF systems can change how these systems operate. Dairy (with the Halter product), beef (with the other three products) and sheep/goats (Nofence) are all designed to function in small paddock or strip grazing systems. The evidence (Confessore et al, 2022; Staahltoft et al, 2023) shows that VF is effective and comparable in efficacy terms with electric fencing (Campbell et al, 2019). How fenceline breaches occur and are resolved provide an apt comparison showing VF is not a like-for-like replacement. For standard wire-based systems, whether electric or plain wire, often one or two animals initially breach the fence. After the first escape, the wire fence may partially drop and often much or all of the herd/flock quickly advance through. In practice, breakdowns of standard fences create extra work and re-herding of animals. Both the breach and herding of returns creates risks of heightened stress and injury. For VF, instead of a single boundary barrier for the whole herd, each individual has its own fenceline. A breach by any individuals has no impact upon the VF boundary for the rest. So, one escapee does not lead to mass breakouts. Considerable narrative (e.g. Wallington, 2021) confirms that individuals or small groups that ‘escape’ return to the larger herd, avoiding isolation and seeking to remake social contact. As all VF systems are one way, then return through the boundary, whether a full or partial breach, has no warning or shock stimuli for returnees. Alerts can also be sent to smartphones. Combined with stock locations, judgement can be made to do nothing, check things later or venture out for physical inspections. For physical fence breaches, knowledge of any breach is often delayed and requires stockperson attendance.
In terms of practical management, external fixed fences, particularly keeping livestock within the overall farm boundary, or away from risks such as public roads are still seen as important. But internal strips or paddocks become highly flexible in size and timing, stock can be moved to new strips or paddocks by remotely extending the virtual boundary, and once one animal finds that VF boundaries have changed and new pasture is available, the herd/flock observes the change and moves through.
The sophistication of the Halter system, with its combined sound, vibration, and electric shock providing directional guidance for milking dairy cows, creates the potential removal of many internal fence lines and gates that currently parcel up grazing land. With reduced need for fences, physical gates, or even tracks, then the internal landscape of dairy farms may change, with significant savings. Beyond the need for external fixed boundaries, and appropriate water points, the flexibility of virtual fencing should cover many grazing management scenarios on lowland farms. The challenge may be to prove clear benefits for farm management and profitability with technology that has high costs.
Extensive grazing, including common grazing
Livestock often graze very large areas with poor, or absent, divisions and external boundaries, and there is little opportunity to manage grazing location at even modestly precise scales. Conventional walls and fences need continuous, and expensive, upkeep and in many cases are either impractical (e.g. high altitude through snow damage), not allowed (for rented land and Commons) or judged too expensive. Inspection of stock has very variable characteristics and is often time-consuwming. With conventional boundary systems it is often impractical to focus grazing even if judged useful for farming and/or nature conservation objectives. Some farmers will not graze areas because of dangers to the stock; wet boggy areas, areas prone to flooding, cliffs or even proximity to the next farm's bull. VF can overcome many of these issues. For these extensive systems VF offers opportunities to easily, and flexibly, manage grazing at a finer scale, keeping livestock well away from risk areas (which may be seasonal or temporary) with virtual exclosures or to focus them onto particular areas. Management of stock location may ease risks to animals and to humans, for example keeping stock away from away from open public roads running through common grazing land and open moorlands in areas such as National Parks. As the virtual paddocks do not have gates, as VF requires continuous polygons, risks from gates being mistakenly left open reduces. There is potential in extensive grazing to create multiple smaller VF paddocks within the larger area, including these local issues. Shelter and water access need to be factored in, but there is potential to manage increased or reduced local grazing pressures, to meet management aims. This capacity for relatively fine-scale management suits many conservation grazing regimes by nature conservation bodies (e.g. Wright, 2021; Wallington, 2021; O'Donoghue 2022), but also suits grant-based schemes where farming is still the priority of the manager. VF systems can provide historical data to show grant bodies that rules or prescriptions were followed, but also to enable year-on-year fine-tuning. A second capability comes from having livestock within virtual paddocks so physical inspections can be easier and quicker, as stock should be both where they have been placed, but also precisely located using real-time apps. This provides opportunities to reduce time searching, to be more certain that all are checked and overall improving the value/effectiveness of physical checks. Remote virtual inspections of location or movement patterns provide opportunities to be both reassured and alerted for active management action.
Moving stock with virtual herding
Virtual herding or movement operates somewhat differently in the different systems. Halter and its directional vibration/sound provides alerts to individual cows to actively trigger stock movement in addition to VF functions. The other systems rely on changes by the manager of VF lines to enable stock movement to occur. Stock are simply allowed to move to new areas of grazing in virtual strip or paddock systems, by remotely opening up new adjacent areas, with stock finding the new section through absence of any warning alerts (Campbell et al, 2021; Hamidi et al, 2022). Technically this requires a new or changed paddock to be loaded remotely to all collars. Problems with communications with poor signals may potentially create some risks for stock left behind, causing distress.
Actively moving or herding extensive stock into a particular area, for example prior to cattle handling, requires more digital management and time as there is no ‘new grass’ incentive, and multiple paddock moves may be required. For all systems including Halter, the Animal Welfare Committee (2022) report noted that any ‘back’ fence should not be used as a potential goad equivalent. Whether intentionally or accidently giving animals sound or electric stimuli may alarm them and becomes a ‘goad’. However, early adopters who ‘open up in front, close off behind’ using mapped locations of all animals have described it working well (e.g. Haarland, personal communications). In extensive systems, these types of manoeuvres appear best taken over days, rather than hours. Many practical farmers will readily accept that physical gathers on foot, or with vehicles and possibly dogs, can sometimes be challenging for stock, creating alarm and distress, along with risk of injury to animals and people. The Halter system takes this virtual herding approach to a higher level with milking time herd movements all being automated. Furthermore, arguments are made that enabling each cow to move at their own pace, rather than being driven, is better for the cows as well as labour-saving. A further level of sophistication is that within-field drafting of small numbers of cattle can be undertaken, for example automating a move into collecting areas. As noted earlier, there is no independent peer-reviewed literature for Halter describing these features or their potential impacts for welfare. Large numbers of New Zealand dairy farmers (now over 100 000 collars) provides evidential support for effectiveness from practitioners.
Summary of welfare opportunities and risks
Table 2 summaries whole system welfare issues. The collars themselves and the electric shock element pose direct welfare risks and costs to animals. The shocks need to be experienced by stock during training to associate shock and warning alerts, but thereafter, using the non-aversive warnings, all animals can avoid future shocks. Removing or reducing physical wire systems reduce the risk that wire creates for livestock and wildlife and removes/reduces the welfare risk for conventional systems to trap animals such as for floods, snowstorms and predator attacks.
Mapped locations and other sensor systems (such as on-board accelerometers) can provide more information on herd/flock and individual animal state. The relationship between the livestock, the grazing system and context and the managing people is changing. More information, better stock locational data and avoidance of risky zones all provide opportunities for better animal welfare and risk reduction. Better, flexible management in relation to third party people, in their cars or walking their dogs for example, may improve management of these potential conflict scenarios.
Remote virtual checks and appropriate alerts provide opportunities to use visual inspection with appropriate timings and discernment. There are risks of virtual data availability reducing physical inspections and that un-seen welfare issues may occur.
A number of authors have aimed to directly investigate and report upon welfare issues, identifying welfare impacts implicitly in the title or key words of their paper. There have been specific comparisons between VF-based treatments and standard electric fence systems. Different techniques/metrics from solely behavioural to some physiological metrics conclude that welfare measures are not compromised or at least there are no differences compared to electric fences (Brunberg et al, 2015; Campbell et al, 2019; Lomax et al, 2019; McSweeney et al, 2020; Aaser et al, 2022; Sonne et al, 2022; Hamidi et al, 2022). Published case studies (Wallington, 2021; O'Donoghue, 2022) provide examples of technical problems but also of system-level welfare gains, and overall support a balance of positive animal welfare. There is little evidence that the electric shock element of virtual fencing systems is problematic, and animals adapt quickly and calmly to VF systems and there are no measurable long-term differences compared to standard electric fence systems. Evidence of welfare value of remote data, providing location and activity data to remotely observe or act upon alerts, though, is virtually absent from the scientific literature.
The future
In less than a decade, VF systems have evolved from ideas to sophisticated commercial systems enabling containment, herding and considerable monitoring capabilities. This is providing endusers with unparalleled information in near real-time for actively managing grazing livestock. There are legislative barriers to uptake in certain countries and an overall wariness over animal welfare. The additionality enabled by on-board sensors provides ever increasing scope to become comprehensive livestock management systems. In standard farming systems, the current high costs of VF require the need for clear economic benefits, from combinations of hardware savings (e.g. wire fences) or labour or production benefits. Evidence of any system-level impacts upon welfare or production statistics though is very limited. With pressures on labour availability and profitability, and continued interest in labour-saving, data-rich systems seem likely to continue to expand if cost–benefit can be achieved.
KEY POINTS
- Commercial virtual fence systems have rapidly increasing uptake in dairy, beef and sheep and goats, with active deployment particularly in the UK, Norway, New Zealand, USA and States of Australia where allowed.
- Animal welfare is an on-going area of concern, with regulations in some nations and regions not allowing virtual fence deployment.
- Electric shocks from collars are welfare costs to each animal, which can control these through associated learning of warning signals from close proximity to GNSS-set virtual boundaries, but virtual fencing can easily reduce risks from danger areas, including reducing contact with people and their dogs.
- Real-time data on location and activity provide opportunities for frequent virtual observations and quicker and more effective physical observations, particularly in extensive systems.
- Management of strip or paddock grazing systems in intensive grasslands, more controlled focussed extensive grazing, and managing animal movements and gathers creates opportunities for reduced labour and improved grassland management.