Implications

Precision technologies, big data and data analytics have the potential to revolutionise pasture based ruminant production systems. Integrating precision technologies, big data and data analytics through a needs driven approach to solve the key drivers of pasture based systems will maximise the return for any investment and will ensure that these technologies are integrated within the systems. The likely benefits include increased efficiency, reduced costs and reduced environmental burden. At the same time, integrated databases and data structures that allow robust traceability and sustainability monitoring within systems can be developed.

Drivers of efficiency in pasture based systems

The relative cost of pasture as a feed source for livestock production compared with grass silage and concentrate was reported as 1:1.8: 2.4, respectively, by Finneran et al. (Reference Finneran, Crosson, Wallace, O’Kiely, Forristal and Shalloo2010). The competitive advantage of grass-based milk production thus lies in the ability to utilise grazed grass as the major feed source, significantly reducing costs per unit of production. Maintaining and increasing the competitiveness of grass-based systems thus requires growing and utilising more grass per hectare and converting that grass as efficiently as possible into valued products.

A number of factors influence costs of production and farm profitability in grass-based systems. Overall pasture utilisation, grazing season length, overall pasture management and level of supplementation are all strongly associated with profitability (Shalloo et al., Reference Shalloo, Dillon, Rath and Wallace2004; MacDonald et al., Reference Macdonald, Glassey and Rawnsley2010; Läpple et al., Reference Läpple, Hennessy and O’Donovan2012; Kelly et al., Reference Kelly, Shalloo, Geary, Kinsella, Thorne and Wallace2013; Ramsbottom et al., Reference Ramsbottom, Horan, Berry and Roche2015; Hanrahan et al., Reference Hanrahan, McHugh, Hennessy, Moran, Kearney, Wallace and Shalloo2018).

In addition to grass-based factors, cow health and fertility is also highly important. Across both pasture and high input total mixed ration based systems, the optimum calving interval is 365 days (Esslemont et al., Reference Esslemont, Kossaibati and Allcock2001). However, the relative importance of cow fertility is greater in seasonal dairy production systems which are more common on pasture based systems compared with non-seasonal systems (Veerkamp et al., Reference Veerkamp, Dillon, Kelly, Cromie and Groen2002; Shalloo et al., Reference Shalloo, Cromie and McHugh2014). Within pasture based systems, synchronising the demand for feed with the growth patterns of pasture is mainly achieved by planning the calving pattern such that the peak herd feed demand is matched with peak grass growth (Shalloo et al., Reference Shalloo, O’Donnell and Horan2007). In grass-based systems, the effect of sub optimal calving interval is thus magnified as feed supply and demand can become misaligned.

Boichard (Reference Boichard1990) highlighted that additional inseminations, veterinary and hormonal costs and a modification of current and subsequent lactations are also costs associated with sub optimal fertility. There are also indirect costs such as the reduction in expansion potential (scale inefficiency) and disease risks associated with purchasing stock (Shalloo et al., Reference Shalloo, Cromie and McHugh2014). Suboptimal fertility will also lead to an extended calving period, increasing the labour requirements for monitors cows about to calve and associated management (Shalloo et al., Reference Shalloo, Cromie and McHugh2014).

The four main costs associated with cow fertility are reduced grass utilisation efficiency, increased veterinary costs, increased culling (Esslemont et al., Reference Esslemont, Kossaibati and Allcock2001) and increased labour costs associated with longer calving period.

Overall, there are two major drivers of efficiency in pasture based systems. These are grass utilisation and cow health, and in particular within cow health, fertility. These are thus the areas precision technologies should focus on to maximise value in grass-based systems and sensors on the animal and sensors to measure grass can facilitate improved decision making.

Accurate grass growth and biomass measurement is one such area where significant adoption of existing technologies would likely be beneficial. Further development of existing technologies and the development of new technologies will allow for greater detail and inform pasture management through for example targeted use of inputs as is now common in arable contexts (Hanrahan et al., Reference Hanrahan, Geoghegan, O’Donovan, Griffith, Ruelle, Wallace and Shalloo2017). Understanding and monitoring animal’s behaviour, how she interacts with her environment and how her behaviour changes during times of different physiological states is likely to further improve the animal part of the systems performance. Combining both information sources or collecting information about how individual animals interact with variation in grasslands and vice versa, how grasslands react to animals, may also be a possibility in the medium to long term.

Animal status and behaviour sensors

Cow behaviour such as movement, location, rumination and resting are of importance for herd management (Delagarde and Lamberton, Reference Delagarde and Lamberton2015; Delagarde and Lemonnier, Reference Delagarde and Lemonnier2015) and heat detection is the most established use case of such information (Rutten et al., Reference Rutten, Veilthuis, Steeneveld and Hogeveen2013). Several different approaches have been developed to measure various cow behaviours.

Accelerometers are a common sensor used on cows and can measure behaviours such as feeding and rumination. The position of those systems on a cow varies with head, neck, ear and leg-mounted devices available (Bikker et al., Reference Bikker, Van Laar, Rump, Doorenbos, Van Meurs, Griffioen and Dijkstra2014; Borchers et al., Reference Borchers, Chang, Tsai, Wadsworth and Bewley2016). Acceleration measured by accelerometers, usually attached to the animal’s legs, can be used to measure locomotion activity (e.g. RumiWatch pedometre (Alsaaod et al., Reference Alsaaod, Niederhauser, Beer, Zehner, Schuepbach-Regula and Steiner2015) or IceTag pedometre (Ungar et al., Reference Ungar, Nevo, Baram and Arieli2017)). These measurements can give an insight into different events (e.g. heats, health and behaviour, etc.) as increased activity can be associated with oestrus events.

Other technologies for heat detection are mounting activity sensors including tail paint, scratch cards, KaMaRs (KaMaR) and HeatWatch (CowChips) (Holman et al., Reference Holman, Thompson, Routly, Cameron, Jones, Grove-White, Smith and Dobson2011). One of the most recent developments in using accelerometer data is in lameness detection with the IceTag3D™ (IceRobotics, Edinburgh, UK). They recently began offering lameness detection as function in late 2017 (IceRobotics, 2017).

As feeding behaviour differs when cows are grazing, the sensors which accurately measure feeding behaviour indoors are unlikely to be as effective with grazing cows. These sensors may thus require adaption and recalibration for grazing cows. Most commercial sensors that are used on farms for measuring grazing behaviour use an accelerometer around the neck (e.g. Heattime^® by SCR Engineers, Netanya, Israel (Molfino et al., Reference Molfino, Clark, Kerrisk and García2017) or MooMonitor+; Dairymaster, Tralee, Ireland (Werner et al., Reference Werner, Leso, Umstatter, Niederhauser, Kennedy, Geoghegan, Shalloo, Schick and O’Brien2017)).

One system for research uses a pressure band sensor in a halter to detect jaw movements (Zehner et al., Reference Zehner, Umstätter, Niederhauser and Schick2017). It is called the RumiWatch (Itin+Hoch GmbH, Liestal, Switzerland) and the recorded jaw movements are used to measure feeding behaviour (Werner et al., Reference Werner, Leso, Umstatter, Niederhauser, Kennedy, Geoghegan, Shalloo, Schick and O’Brien2017). The RumiWatch is capable of measuring detailed grazing behaviour such as rumination chews and grazing bites as well as rumination and grazing times (Werner et al., Reference Werner, Leso, Umstatter, Niederhauser, Kennedy, Geoghegan, Shalloo, Schick and O’Brien2017).

Understanding how an animal interacts with her environment helps to explain the individual animals grazing behaviour and will help to explain within paddock variation and its impact on performance. Cow localisation allows an analysis, spatial analysis of the individual animal to be completed. Recording grazing behaviour location and linking that to high resolution grass measurements could be useful for research and management purposes. There are already a number of techniques to determine animal position indoors. One such system is the Smartbow ear tag (Smartbow, Weibern, Austria). Smartbow works by triangulating the location of cow worn ear tags with low-frequency signals which are detected by multiple receivers in the barn allowing triangulation. It has been shown to have reasonable accuracy in determining cow location indoors (Wolfger et al., Reference Wolfger, Jones, Orsel and Bewley2017).

It is also possible to localise animals via ultra-wideband technology (e.g. CowView; GEA Farm technologies GmbH, Germany (Tullo et al., Reference Tullo, Fontana, Gottardo, Sloth and Guarino2016)). In pasture based systems, the accurate measurement of the animal’s position used to be limited to global position system (GPS) based tracking systems (Williams et al., Reference Williams, Mac Parthaláin, Brewer, James and Rose2016), but other approaches that are less burdensome on batteries are being developed (e.g. using ground based triangulation with multiple base stations).

Lameness is also particularly relevant in larger pasture based dairy farms where cows may have to walk large distances between the milking parlour and pasture. Accelerometer based lameness detection discussed above is the most prominent. However, there are other approaches also in development such as walkover systems (e.g. StepMetrix^®; BouMatic, Madison, WI, USA) and camera based systems (Viazzi et al., Reference Viazzi, Bahr, Van Hertem, Schlageter-Tello, Romanini, Halachmi, Lokhorst and Berckmans2014; Fischer et al., Reference Fischer, Luginbuhl, Dellattre, Delouard and Faverdin2015). Camera based systems also have the potential for other uses such as automated body condition scoring (Halachmi et al., Reference Halachmi, Klopčič, Polak, Roberts and Bewley2013; Fischer et al., Reference Fischer, Luginbuhl, Dellattre, Delouard and Faverdin2015; Spoliansky et al., Reference Spoliansky, Edan, Parmet and Halachmi2016). To date, two automated body condition scoring systems have been made commercially which use image analysis based approaches (Delaval, Tumba, SE and Ingenera SA, Cureglia, CH).

Van De Gucht et al. (Reference Van De Gucht, Saeys, Van Nuffel, Pluym, Piccart, Lauwers, Vangeyte and Van Weyenberg2017) found in a survey that a sensor attached to the cow was preferred by Flemish farmers, followed by a walkover system and a camera system. Some studies already demonstrate positive results for automated lameness detection, but in many cases, the animal must be critically lame to be detected. Beer et al. (Reference Beer, Alsaaod, Starke, Schuepbach-Regula, Müller, Kohler and Steiner2016) showed that it is feasible to differentiate lame cows (>2.5 locomotion score) (1–5 scale) from non-lame cows with data gathered by a pedometer (RumiWatch). Despite significant development work and activity over many years, a September 2018 search of the literature for an independent validation of a commercially available lameness detection system returned no such studies.

Sensors specific to grassland

Focussing on pasture based milk production systems, the application potential of sensor technology is increasing. We now review sensor technologies that can aid grassland management including spectral sensors (terrestrial and remote), digitally enabled mechanical plate metres and ultrasonic sensor based plate metres.

Grassland farmers that rely on pasture as their primary feed source require accurate real-time measurement of pasture herbage mass and quality to optimise grazing and nutrition management. Each additional 1000 kg increase in pasture utilisation per hectare was recently reported to be associated with ~€173 greater profit per hectare (Hanrahan et al., Reference Hanrahan, McHugh, Hennessy, Moran, Kearney, Wallace and Shalloo2018).

With other crops such as maize and soybean, sensors are now routinely being placed in fields measuring the microclimate, spectral characteristics and small-scale differences in soil fertility to help improve crop management and productivity (Wolfert et al., Reference Wolfert, Ge, Verdouw and Bogaardt2017). However, sensors to aid pasture management and increase utilisation are still in development and with a few exceptions (e.g. Bluetooth-enabled plate metres), are not yet commercially available. As pasture utilisation is a key driver of profitability in such systems (Hanrahan et al., Reference Hanrahan, McHugh, Hennessy, Moran, Kearney, Wallace and Shalloo2018), these sensors, like pasture growth measurement tools, may offer particularly high returns to pasture based producers.

Examples of technology that enable improved pasture utilisation include digitally enabled plate metres which streamline and automate aspects of collecting the data required to generate pasture budgets (French et al., Reference French, O’Brien and Shalloo2015). Distance measurement in the plate height is usually generated via ultra/micro-sonic sensors which measure the height of the plate resting on the canopy (Moeckel et al., Reference Moeckel, Safari, Reddersen, Fricke and Wachendorf2017). These have an added benefit in that the operator skill required is much less than for example visual assessments. Given the importance of measurement in managing and increasing pasture utilisation, plate metres can offer significant benefits for relatively modest capital investments. When combined with GPS localisation and app or server connectivity, this potentially allows significant and novel functionality.

A similar approach to a plate metre is to measure the height of the pasture canopy. However, while a plate metre presses down on grass to variable degrees based on grass density, this approach measuring the canopy height is less sensitive to pasture density and so less sensitive to herbage biomass.

Although plate metre approaches provide a good estimate of biomass, commercially available tools do not currently assess pasture quality. Be the platform handheld, drone or satellite-based, spectral analysis (terrestrial or remote) is a potential candidate approach for the estimation of grass quality and would be a potentially quick and efficient alternative to lab-based assessments (Sibanda et al., Reference Sibanda, Mutanga and Rouget2016). If successful, such information could allow farmer to allocate grass based on the feed quality available rather than specifically on a dry matter basis.

A common approach is to develop particular wavelengths to create indices such as Normalised Difference Vegetation Index (NDVI) which estimate if vegetation is present or not by assessing the ratio of red and near-IR wavelengths. The photosynthetically active range is between 400 and 700 nanometres which appear relatively dark in vegetation. Above 700 nanometres appears relatively bright as plants re-emit these wavelengths. Strong absorption above 700 nanometres would not aid photosynthesis and would result in overheating.

The challenges facing spectral approaches include background soil effects, atmospheric effects, grazing impact and heterogeneity of species and variation in pasture growth stages/proportion of senescent material in the canopy which disproportionately influence spectral data (Moeckel et al., Reference Moeckel, Safari, Reddersen, Fricke and Wachendorf2017). Further to this, most studies have focussed on measuring pastureland biomass in tropical savannahs via remote satellite sensing with less focus on temperate pasturelands (Moeckel et al., Reference Moeckel, Safari, Reddersen, Fricke and Wachendorf2017).

Moeckel et al. (Reference Moeckel, Safari, Reddersen, Fricke and Wachendorf2017) reported combining an ultrasonic sensor to measure grass canopy height, a handheld hyperspectral sensor data and satellite-derived spectral data. They found that the best predictions of biomass were early in the grazing season when pastures were more homogenous. As grazing season continued, issues such as pasture refusal post grazing resulted in variability in the sensor outputs which could not be accounted for when predicting biomass. The accuracy values achieved with biomass indicate that further development is required before practical application in pastureland management are possible (maximum R ²=0.52).

Successful non-satellite based spectral approaches to estimate pasture quality in temperate intensively managed pasturelands were not found in a search of the literature. There has been some progress using satellite data to estimate pasture biomass using Synthetic Aperture Radar. One study of Irish pasture found that grass biomass could be predicted with an R ² of up to 0.75 based on this approach (Ali et al., Reference Ali, Barret, Cawkwell, Green, Dwyer and Neumann2017). This prediction was only achieved periodically when conditions were right in ‘coherent’ paddocks. Highly coherent paddocks tended to have low biomass after mowing. Factors, such as wind flattening, tall grass and grazing/mowing complicated the interpretation of the satellite data, especially as biomass increased.

There is potential for technologies to streamline and potentially automate the collection of pasture data such as biomass and quality, calculate pasture availability and to integrate the data with other sources, for further use for example (French et al., Reference French, O’Brien and Shalloo2015). Improving fertiliser utilisation by targeting the areas with the most requirement based on previously recorded yields or spectral analysis, as well as protecting water courses are both potential applications in combination with GPS enabled spreaders.

There is the potential for precision technologies to significantly contribute to pasture management, and thus increase pasture utilisation. However, there are significant technical challenges to be overcome for that to occur. This is particularly true with regards to grass quality.

Decision support systems for pasture based systems

Decision support systems (DSS) are available for systems that utilise grass as a major source of feed for dairy cows. Such technology facilitates the improved grassland management, delivering tangible benefits in terms of grass supply, quality and utilisation and ultimately profitability. The main features of the pasture management tools include the grass wedge, grass budget and the spring and autumn rotation planners. Some DSS use historical grass growth data to estimate pastureland production capacity and aid long-term management decisions like determining the most suitable stocking rates and plan paddock reseeding. These management decisions have been found to be strongly associated with the economic performance of pasture based dairies (Kennedy et al., Reference Kennedy, O’Donovan, Murphy, Delaby and O’Mara2007; Shalloo et al., Reference Shalloo, Creighton and O’Donovan2011; McCarthy et al., Reference McCarthy, Pierce, Delaby, Brennan, Fleming and Horan2013). One recent innovation has been to focus on the number of grazings achieved on individual paddocks. This is now a key performance indicator for Irish grassland systems. Such tools assist farmers in short- to medium-term grass management (MacDonald et al., Reference Macdonald, Glassey and Rawnsley2010).

Smartphone, and to a lesser extent personal computer, based applications are the platforms used. These provide greater flexibility to the user. Furthermore, internet-based architecture enables the collation of large quantities of data in central databases. This extends the data usefulness beyond farm level. Data collected on a wide number of farms can be used to benchmark performance of a single operation, but also for research purposes (e.g. animal and plant genetics, precision nutrition) and policy formulation (Hanrahan et al., Reference Hanrahan, Geoghegan, O’Donovan, Griffith, Ruelle, Wallace and Shalloo2017).

One example is Pastur’Plan, a new tool available in France (Delaby et al., Reference Delaby, Duboc, Cloet and Martinot2015) which combines two concepts. The first one is adapted from the Grass Wedge and the second one describes the balance between grass growth and demand according to different grazing options on a paddock by paddock basis. Pastur’Plan takes into account herd size and supplementation fed to calculate the grass demand of the herd. The weekly grass growth profile can be changed and updated by the user allowing the calculation of the grass available. Pastur’Plan helps the user (farmer or advisor) to anticipate the utilisation of their grass on farm at the paddock level.

This facilitates the testing of different scenarios to optimise grass utilisation. The objective of this tool is to stimulate better grazing management, improve grass utilisation and provide more confidence in grazing systems.

Another DSS in the area of pasture research is PastureBase Ireland (PBI) (Hanrahan et al., Reference Hanrahan, Geoghegan, O’Donovan, Griffith, Ruelle, Wallace and Shalloo2017). It builds on the concept of citizen science where data are collected by individuals while they go through their everyday lives (Conrad and Hilchey, Reference Conrad and Hilchey2011). Citizen science has proved successful in the past in many countries, forming the bedrock of biological recording in various large research projects, particularly in ecology and environmental sciences.

In the case of PBI, pastureland farmers are the citizen scientists collecting the pasture data that feed into a new research programme. This is being facilitated through the creation of discussion forums within the PBI system. These discussion forums allow for the communication and organisation of research and technology initiatives, increasing farmer engagement in grassland research.

Holistic farm solutions derived from the integration of sensor data, data interrogation and providing information to the end users through an appropriate medium and time framework are increasingly feasible (Figure 1; Shalloo et al., Reference Shalloo, O’Leary, McDonagh, Geoghegan, Werner and Leso2018). However, to date, literature about decision making with precision technologies is sparse, especially regarding pasture based dairy systems.

Figure 1 Schema of stages of decision making and the use of sensor information in dairy farm management. Adapted from cow focussed (Rutten et al., Reference Rutten, Veilthuis, Steeneveld and Hogeveen2013) to grass focussed. hDM=herbage dry matter.

Not specifically related to pasture but related are animal health databases. Veterinary use of medicines (especially antibiotics) is regulated in most countries with the recording of information regarding diagnosis and treatments being a common requirement. The quantity and type of medicines administered to individual cows could potentially be stored in a central database. However, in most countries, although this information has been systematically recorded on paper, it generally has not been recorded electronically and has not been collated centrally (Egger-Danner et al., Reference Egger-Danner, Fuerst-Waltl, Obritzhauser, Fuerst, Schwarzenbacher, Grassauer, Mayerhofer and Koeck2012). One example country where a significant effort has been put into standardising and aggregating animal health data relating to dairy cows is Austria. In Austria, the recording of animal health treatments was made a legal requirement in 2002 but the form and coding of this recording were not standardised or aggregated until 2006 (Egger-Danner et al., Reference Egger-Danner, Fuerst-Waltl, Obritzhauser, Fuerst, Schwarzenbacher, Grassauer, Mayerhofer and Koeck2012).

In relation to animal genetics, there are a number of databases that are critical to increasing herd genetic merit (e.g. International Committee for Animal Recording Interbull database; Irish Cattle Breeding Federation database). Such data can be collected by public companies or associations such as in Ireland and Italy but also by private genetic companies. Information regarding animal performance including milk yield, fertility, longevity, animal health, administered treatments, calving, etc. is routinely recorded. Precision technologies offer significant potential to inform breeding programmes through the automated collection of phenotypic traits such as activity and behaviour, which may be of economic value to select for. An example might include monitoring activity data that is collected by devices on individual animals as well as diagnoses of illness. Many of the traits that these devices collect information on are heritable, and it is desirable to record it to improve future breeding programmes.

The robustness of data collected by sensors is still perceived as a priority and thus most research focusses on sensing accuracy and systems validation. Decision making remains pivotal, but underdeveloped, to the successful application of precision technologies. Decision support and automation is the basis of value creation from precision technologies for the farmer (Hostiou et al., Reference Hostiou, Fagon, Chauvat, Turlot, Kling-Eveillard, Boivin and Allain2017). Though accurate collection of more data is a dominant theme in precision technology, using and combining extant sources of data better through improve decision support and automation should also be a priority.

Precision technologies appropriateness and farm attributes

For farmers, it is difficult to accurately estimate the economic benefit for most technologies in advance of adoption. The risk of expensive technologies delivering poor financial returns to farmers is significant. One of the few studies that investigated this found that sensors did not deliver a discernible return to Dutch dairy farmers between 2008 and 2013 (Steeneveld et al., Reference Steeneveld, Hogeveen and Oude Lansink2015). They compared three categories of dairy farms, conventional milking systems with no sensor technology, conventional milking systems with sensors (usually cow activity sensors) and automatic milking systems (AMS). The AMS systems were significantly less profitable than conventional farms with sensors with profit per 100 kg of milk dropping from €3.86 to €1.31.

The reduction in profit was mostly attributable to higher depreciation costs and, contrary to expectations, only modest recorded reductions in labour costs (Steeneveld et al., Reference Steeneveld, Hogeveen and Oude Lansink2015). They found that no statistically significant difference in productivity (before finance costs) could be discerned between conventional systems before and after installing sensor systems. This indicates, on average, a neutral or very modest return on investment for the average farmer in the study.

Sensor technology has advanced since that study and likely reduced in cost significantly. Increasing performance of activity sensors (Steeneveld et al., Reference Steeneveld, Rutten, Oude Lansink and Hogeveen2017; Werner et al., Reference Werner, Leso, Umstatter, Niederhauser, Kennedy, Geoghegan, Shalloo, Schick and O’Brien2017) and the application of advanced techniques such as machine learning have likely added additional value (Borchers et al., Reference Borchers, Chang, Proudfoot, Wadsworth, Stone and Bewley2017). Returns may have also improved as technologies and knowledge about them improve.

There may, however, be significant variation in the returns achieved. This variation can be attributed to variation in farms, farmers and technologies invested. Consequently, there will continue to be a proportion of farmers investing in technologies which are not achieving good financial returns. Replicating Steeneveld et al.’s (Reference Steeneveld, Hogeveen and Oude Lansink2015) study in both indoor and pasture-based dairying systems would be advantageous.

Such a study could also establish the drivers of variation in investment returns and aid the provision of advice to farmers. Advice and knowledge arising could have multiple applications. First, it would guide vendors to better target their research, development and marketing of technologies. Second, farmers would have more insight as to which combination of technologies and farm situations deliver the best returns.

Why some Dutch dairy farmers benefited financially whereas others did not, from investing in precision technologies (Steeneveld et al., Reference Steeneveld, Hogeveen and Oude Lansink2015), has to date not been investigated. Farmers will be rightfully cautious about technologies until this is clarified. In being cautious, they will likely be overlooking technologies which offer significant economic benefits. This is because these beneficial technologies are currently hard to distinguish from those that do not. This is a major issue as the adoption rate of precision technologies remains low (Steeneveld et al., Reference Steeneveld, Rutten, Oude Lansink and Hogeveen2017). In light of this, increased knowledge of the drivers of adoption rates can also be used to increase average returns.

While the drivers of good returns (as opposed to just adopting or not) from technology investment are relatively unknown. There is, however, a wealth of literature on the motivations of farmers to invest in a technology or adopt specific management practices which also like to predict returns (Garforth et al., Reference Garforth, McKemey, Rehman, Tranter, Cooke, Park, Dorward and Yates2006; Garforth, Reference Garforth2010; Jones et al., Reference Jones, Sok, Tranter, Blanco-Penedo, Fall, Fourichon, Hogeveen, Krieger and Sundrum2016).

Advisory services may wish to consider developing knowledge and expertise so that independent sources of advice and resources are available farmers to improve the returns farmers make from investments, and so drive the adoption of beneficial technology. These could include centralised information sources on technology efficacy, standard operating procedures for using the technology and budgeting tools designed with technology investment appraisal in mind (O’Leary et al., Reference O’Leary, Ahrend, McDonagh, O’Brien and Shalloo2018).

There is a risk that technologies developed in other contexts such as indoor systems will be marketed to farmers in pasture based systems where their utility is marginal and their efficacy reduced. A focus on technologies that meet clear needs at the farm level such as pasture quality assessment is advisable.

We venture that the key need of pasture based farming systems is transforming sustainability from a broad ill-defined concept to a clear and measurable construct. Within sustainability, we further specify the following as priority areas for precision technology research, development and innovation: labour shortages (automation/labour efficiency), management ability (improving farmer capacity for an increasingly complex technology reliant role), economic resilience (adding value to products, reducing costs and increasing efficiencies) and environmental, animal health and welfare benchmarking.

We further introduce the concept of ‘Need Driven’ development to agriculture precision technologies (Kruger and Cross, Reference Kruger and Cross2006). These are a set of principles to guide and expedite the achievement of precision technologies potential in pasture based livestock ruminant production systems. Need driven technologies are solutions that are delivered in ways that are comprehensible and relevant to the farmer. In addition, they generate data that can inform major research programmes (e.g. animal genetics, grass genetics and performance evaluation). We contrast this to ‘Solution Driven’ development where developers start with a technology (solution) and try to find additional applications for the technology. Solution based approaches are potentially valuable but are not necessarily focussed on the areas of most value to farmers.

Needs driven precision technology should in contrast meet pressing needs such as providing mechanisms to quantify the sustainability of pasture based ruminant production and increase farm profitability. These needs can be addressed through the integration of existing databases, capturing new data and the application of best practice in data science to generate the most value from the data.

Technological solutions which can increase the utilisation of pasture and the productivity and fertility of seasonal calving herds are likely to add significant value to pasture based systems. These should provide real benefits in profitability, sustainability and resilience by providing informed, real-time solutions to the farmer through an appropriate medium.

In order for big data and smart technologies to realise their potential in grassland agriculture, there is a requirement for the different industry stakeholders to work together and to develop business models that facilitate and encourage the different stakeholders to integrate their solutions within an overall platform.

Although many researchers and companies already embrace approaches similar to Need Driven development, some do not, and by naming and defining this paradigm in a grass-based context – we hope it may add value and clarity.