Conceptualising the DAIS: Implications of the ‘Digitalisation of Agricultural Innovation Systems’ on technology and policy at multiple levels

Abstract As digitalisation transforms agriculture, the implications of cumulative innovation processes are essential to consider in order to mitigate risk and capitalise on opportunities. One project that involves imagining the future of the sector and aims to develop the necessary tools and infrastructure is the Australian Commonwealth Scientific and Industrial Research Organisation (CSIRO) Digiscape Future Science Platform (FSP). This paper explores the policy framework encompassing these technologies and elucidates considerations for future governance in Australia and beyond. Conceptually, we draw on Agricultural Innovation Systems (AIS) theorisation in the context of implications of digital technological development on policy. Methodologically, we utilise multi-level exploratory interviews with Digiscape FSP and prawn aquaculture value chain stakeholders. We argue society is at a critical point in time when the implications of digital agriculture need to be considered at broader regime and landscape levels. Questions around data privacy and ownership are prevalent in agricultural settings, although appropriate institutional guidance is lacking. Three propositions are made as a result of the analysis of scoping research involving innovation community stakeholders. We suggest that agricultural industries stand to benefit from the ‘Digitalisation of Agricultural Innovation Systems’ ex ante socio-technical transition, what we introduce as the DAIS, if forums for discussion and associated technological policy are enacted with emphasis placed on the underlying values of society-technology interaction.


Introduction
Digitalisation is pervasive in modern society and an increasing emphasis on the digitalisation of agriculture, both in terms of transformative potential and associated pressures, is evident in the foci of a growing number of major research organisations and organisational alliances (see for example; AgResearch Limited (2017), CSIRO (2017), Internet of Food and Farming 2020-IoF2020 (2017), Digital Agriculture Convergence Lab (2017)).In each instance, the digitalisation of agriculture and associated technologies are argued to be aligned with both incremental productivity gains and transformational change in Agricultural Innovation Systems (AIS).We introduce digitalisation; examining why it already has had, and will continue to have, increased impact on the agricultural sector; why the way in which a digital agricultural socio-technical transition is facilitated has implications for policy; and, why it is of urgency to examine.
Digitalisation involves the impacts of digital technology on everyday life -in societies generally, in interactions with our environment and in the functioning of economies -and is said to be one of the most significant trends globally at present (Leviäkangas, 2016).It has been argued that there is no single definition for digitalisation, it rather encompasses growth in human-computer or human-information and communication technologies interaction (Billon et al., 2010).The digitalisation of agriculture involves the development, adoption and iteration of digital technologies in the agricultural sector; what has been referred to as both digital agriculture (preferred in Australia and New Zealand) or smart farming (preferred in the European Union) in different spatial contexts (Robertson et al., 2016;Wolfert et al., 2017).The most often cited current impacts of the digitalisation of agriculture tend to involve the uptake of precision agricultural technologies that broadly reduce costs associated with inputs or work to increase yield and/or productivity (Aiello et al., 2019;Chen et al., 2015;Eastwood et al., 2017;Lindblom et al., 2017).For example, auto-steering tractors fitted with GPS units that utilise satellites to minimise overlap and driver fatigue in cropping industries, along with associated yield mapping and variable rate input application (Godoy et al., 2012;Zhang et al., 2015).Recent advancements have seen this technology applied autonomously, meaning the farmer is (technically) no longer required to spend hours in the tractor seat (Croplands, 2017).It has been argued that significant changes in agricultural systems have and will continue to be facilitated by new digital technologies and the associated convergence of multiple advances including: real-time monitoring, big data, Internet of Things, machine learning, cloud computing, and so on (Wolfert et al., 2017).
The study of AIS, or application of innovation systems conceptualisation to the agricultural sector, is a relatively recent body of scholarship that has rapidly gained traction in both the developed and developing world (Hall et al., 2003;Klerkx et al., 2010Klerkx et al., , 2009;;Knickel et al., 2009).Underpinning the study of AIS is an understanding that there is a relationship between the context of agricultural system characteristics and the stakeholders involved and the likely success, failure or otherwise of a given innovative idea (Schut et al., 2014).A national AIS comprises the legacy of technological, societal, economic and institutional interactions that provide the platform through, and extent to which, novel arrangements are possible (Fielke et al., 2018).In this context, innovation refers to the creation of something new, whether that be a product, process, practice or relationship (Turner et al., 2017).AIS farming allows for conceptualisation around the multi-level perspective (MLP) of socio-technical transitions which has been found to be useful in terms of framing multiple lines of action research (Fuenfschilling and Truffer, 2014;Ingram, 2018;Paschen et al., 2017).
This paper examines how digitalisation fits into AIS scholarship by exploring an ex-ante transition involving the 'Digitalisation of Agricultural Innovation Systems' (DAIS).The DAIS broadly involves the increasing influence of digital technology on agricultural outcomes that are currently satisfied by networks of human and technological actors within traditional AIS conceptualisation.Satisfied is used to imply human cognition (through both design and in use) is central to determining the particular values or qualities underlying the digital technologies developed (Flach et al., 2017).We argue that each novel digital technology should not be visualised as simply another individual, discrete, or independent technological innovation.Rather, the interactions between technological trends and the context within which they sit (for example the institutional environment) need to be assessed as component parts as AIS become further digitalised if they are to be fully appreciated.At some point the DAIS will result in existential questions for agricultural stakeholders world-wide by redefining labour, democracy and humanity (Scholz et al., 2018).
Transformational aspirations in terms of productivity and sustainability may be well-meaning but underlying these transformations are agricultural (and societal) values that need to be determined, articulated, reflected on, and will evolve over time with the acceptability (or not) of individual digital technologies (Eastwood et al., 2019;Fleming et al., 2018).Existing research has uncovered potential dichotomies of open versus closed data systems (Wolfert et al., 2017), examined the varying values of smart farming communities depending on their institutional context (Carolan, 2018), and questions have arisen about process-driven scientific advancement co-existing alongside data-driven machine learning algorithms.Along with these contextual questions, it seems we are moving rapidly (in historical terms) toward what once seemed like science fictional reality, for example fully automated (or farmerless) farms (e.g.Oluboyede (2017)).In fact, it was recently reported that the first crop was harvested without a human ever setting foot in the field in 2017 (OECD, 2018).This paper will delve into questions around the implications of the DAIS for policy and provide a timely contribution to social science in the digital agriculture space (Bronson and Knezevic, 2017;Leonard et al., 2017).We address calls for research, and the associated knowledge gaps, relating to the social implications of a digital agricultural future (Rissman et al., 2017) and an examination of the role of policy to both guide and be guided by the digitalisation of agriculture (Koch, 2017).We define policy as the formal institutions determining the legislative foundation of decision-making in society, in terms of what is appropriate conduct, what constitutes productive conduct, what are appropriate externalities, who should bear the cost of externalities, and so on (Antoniades, 2003).Policy is arguably one of the most important drivers in terms of the acceptability of action (or non action) within a given community (Wilson, 2013).
Our primary research question ties this novel DAIS conceptualisation to two components of the regime (policy and technology) through one research programme (Digiscape FSP) and one industry specfic value chain (prawn aquaculture).This exploratory work asks: How do innovation communities percieve technology and policy in relation to a sociotechnical transition toward the digitalisation of the Australian AIS?The digitalisation of agriculture, through a variety of converging technologies, is creating landscape scale questions that have the potential, through technical debt, to lock agricultural systems into pathways with increasingly unclear consequences (Digital Agriculture Convergence Lab, 2017).We argue that there are broader lines of questioning underlying the DAIS transition that elevate above individual niche innovation consideration.Higher level analysis/synthesis of the implications of new digital technologies will help in more strategically guiding the DAIS socio-technical transition (Eastwood et al., 2019;Fuenfschilling and Truffer, 2014).The remainder of the paper offers a more thorough explanation of the DAIS transition in the following conceptual framework, before the methodology, results, a discussion of those results, and conclusion follow.

Conceptual framework
This section frames the original conceptual contribution of this paper in relation to existing theory.We build upon AIS literature to explore a transition toward the DAIS premised on the aspirations of multiple large-scale publicly funded digital agriculture research programs.This is not to discount significant private investment in digital agriculture, rather the focus on some of the recently initiated and publicly funded research programmes provides a more relevant analysis for national policy discussion around agricultural innovation because these programmes are publicly, as opposed to purely commercially, funded and directed.
This paper draws on four core foundations in conceptualising the DAIS transition, which are also used to frame our analytical approach: the multi-level perspective (MLP); AIS; responsible research and innovation; and, innovation ecosystems (Table 1).Innovation ecosystems involve networks of innovation communities interacting at various levels in search of a common goal, with similarities to MLP recognition that a shared vision is required to facilitate the influence of a niche innovation upward (Paschen et al., 2017;Pigford et al., 2018;Walrave et al., 2019).Valuable themes from AIS and responsible research and innovation scholarship include involving stakeholders, anticipating consequences, responding to change, and aiming to build innovation capacity (Bronson, 2018;Eastwood et al., 2019;Klerkx et al., 2012c;Owen et al., 2012).Each of these conceptual frameworks were useful building blocks that contributed to the development of the ex-ante DAIS transition.Utilising the MLP in a real-time transition addresses recent criticisms that the concept is only applied retrospectively and lacks utility, as we are using this thinking to frame an ongoing applied social science research project (Sorrell, 2018).  1 but elsewhere, with the push to find space and become a player in the digital agricultural game evident across the globe (Bronson and Knezevic, 2016;Carolan, 2018;Walter, 2016).

Research programmes claiming to contribute to the DAIS sociotechnical transition
In terms of conceptually framing the DAIS, an interesting (and potentially contradictory) point in regard to each of the three research programmes are their multi-facited and ambitious aims (Schut et al., 2016;Turner et al., 2017).While of obvious importance for funders, all three research programmes espouse a simultaneous focus on both incremental and transformational innovation in the agricultural sector (Nagji and Tuff, 2012).For example, the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Digiscape Future Science Platform (FSP) initiative aims to both support next generation decisionmaking and transform agricultural industries and environmental outcomes (CSIRO, 2017).The IoF2020 (2017)IoF, 2017IoF2020 (2017) initiative is argued to have revolutionary potential to increase the sustainability of value chains yet another priority is that efficiency is optimised.The AgResearch Limited (2017) Digital Agriculture initiative aims to both identify the barriers to taking up new digital technologies on and off farm and develop a technology roadmap to support the industry's transition, assumedly involving the DAIS.While a number of other digital agriculture or smart farming research programmes could have been included, the aim of highlighting the three in Fig. 1 is to visualise the converging technological development elevating the proposed DAIS transition beyond niche level outcomes.
While each of the aformentioned research programme aims are challenging in their own right, in many cases the core values underlying incremental productivity or sustainability gains versus transformational or systemic changes might be directly at odds or involve very different innovation capabilities and capacities (Turner et al., 2017).The sectoral foci of the respective research programmes is also a common feature and such an approach to research needs to be managed to avoid exacerbation of scientific silos (Polk, 2015).For example, the addition of cross-cutting research workflows increases the likelihood of network development within the project teams with brokers likely to spread learning between project teams (Boari and Riboldazzi, 2014;Klerkx et al., 2012a).The conceptual visualisation of the DAIS transition, from a reportedly disconnected Australian AIS to a digital AIS (Robertson et al., 2016) is shown in Fig. 2. The explicit examples referred to further in the remainder of this paper are included in Fig. 2 in italics.The subsequent method and results lead to a discussion of the potential implications of the DAIS on agricultural policies that formally dictate agricultural system boundaries, along with associated informal institutions, attitudes and norms.

Methodology
The primary contribution of this paper involves iterative social research conducted within the CSIRO Digiscape FSP.Our analysis involves a multi-level case study approach with in-depth interviews conducted at the research programme (Digiscape FSP stakeholders) and niche application domain (prawn aquaculture value chain stakeholders) levels (Kläy et al., 2015;Yin, 2014).As this paper is reporting on a platform that is currently in the early stages of development our methodology can be considered a form of initial project scoping to serve the purpose of both contributing reflexivity to the action learning within Digiscape, as well as reporting thematically on systemic change (Beers and van Mierlo, 2017).This aligns with what Pant and Odame (2017) describe as reflexive process monitoring to combine elements of Fig. 1.Three research programmes attempting to capitalise on the Digitalisation of Agricultural Innovation Systems (DAIS) and their components as reported on respective web pages (AgResearch Limited, 2017;CSIRO, 2017;IoF2020, 2017.Note: the particular research programme (Digiscape FSP) and case study domain (aquaculture) focused on in this paper are highlighted in bold.

Table 1
Conceptual foundations utilised to build the ex-ante DAIS transition thesis.

The digiscape future science platform
The Digiscape FSP investment is led by Australia's CSIRO and aims to harness the digital revolution for farmers and land managers (CSIRO, 2017).This challenge is addressed through use case applications of digital technologies in industry based projects along with cross-cutting socio-technical work flows that simuiltaneously develop common digital infrastucture for use into the future.As such, the Digiscape FSP is one of the research programmes with ambitions to pursue the DAIS, with CSIRO an incumbent organisation aiming to exploit an opportunity for the Australian agricultural sector (van Mossel et al., 2019).Each of the Digiscape FSP use cases could be conceptually linked to niche level innovation communities with interaction amongst relevant individual networks also contributing learning to the Digiscape FSP researcher innovation community and up to broader innovation ecosystem stakeholders through programme level learning and technological development (Walrave et al., 2019).Importantly, as recently recognised elsewhere in the energy sector, both human and technological actors interact in these innovation ecosystems (Kolloch and Dellermann, 2019) -see Fig. 2 notes.For the remainder of this paper we focus on human actor perceptions across and within this developing innovation community to determine both use case and regime level implications.

The Australian prawn aquaculture industry
To delve deeper into the perspectives of a relevant use case industry we also examined one of the value chains that is the focus of an application domain, in the Australian prawn aquaculture industry.As recently explained by Joffre et al. (2018), given the farming of animals it entails, it is appropriate to employ innovation system conceptualisation on the prawn aquaculture sector as an extension of traditional AIS research.The prawn aquaculture value chain is a unique example of a young industry (beginning in the 1980s) with different drivers to more established Australian agricultural industries such as the cropping, beef, sheep, or horticulture sectors (Australian Prawn Farmers Association, 2015).Due to consistent warm water requirements, approximately 95% of the 900 ha of prawn farms in Australia are in Queensland, the remaining 5% in New South Wales (Australian Prawn Farmers Association, 2015).All prawns farmed in Australia are consumed domestically which is in contrast to the export focus of most other industries.There is an emerging focus on the role of digital technologies in the industry, with the most recent industy-wide symposium themed smart farming (Australian Prawn Farmers Association, 2018).Prawn aquaculture in Australia is high-risk, capital intensive and requires significant technical expertise contributing 15-20% of total prawn production -roughly $85 million per annum -the remainder of prawn production is wild caught (AgriFutures Australia, 2017).Recent outbreaks of white spot disease have resulted in farm operations closing in some regions.
We interviewed 8 individuals from various points of the prawn aquaculture value chain in order to both provide information back to the Digiscape FSP aquaculture project team and build broader innovation platform awareness about how digital innovation might be percieved within the sector.This value chain analysis provided a mechanism to explore the perspectives of multiple stakeholders with different roles and perspectives within this particular domain.It is included as a case study of one niche innovation project that the Digiscape FSP is trying to influence (Fig. 2) such that consideration is given to factors beyond technological development 'action/s' (Beers et al., 2014).

Exploratory qualitative research
In total 24 semi-structured interviews were conducted from August to November 2017, whereby individuals were asked about their experience with digitalisation and the implications of technological innovation on policy and more broadly on the industries they were involved with.By analysing the perceptions of key individuals involved with the Digiscape FSP, and stakeholders along the value chain in one of the use case industries involved (prawn aquaculture), we came to conceptualise the DAIS transition.The Digiscape researcher interviews provide abstraction from a direct industry but are closer to the action, in terms of the development of the sensor networks, digital tools, and technological infrastructure that will support the envisaged DAIS transition.As such, the two groups provide consciously chosen and varying perceptions on the influence of digitalisation on Australian agricultural industries.This exploratory method allowed us to include responses from stakeholders as broad as technology funders, initial developers, data scientists, users, testers, and social researchers on the input side of technology development (addressing calls from Ballon et al. ( 2019)) for diverse views).The method also allowed for a prawn aquaculture value chain analysis of policy influencers, industry scientists, business managers, input suppliers, and marketers all of which have the potential to either benefit or be cut out of the value chain to Fig. 2. Visualising innovation ecosystem interactions contributing to an ex-ante DAIS socio-technical transition.Note: Individual nodes can represent both human and non-human (ie technological) actors.Fig. 2 was built by combining various concepts and does not predict either adaptation (maintenance of current institutional incumbents) or transformation (complete systemic alteration) at this point in time, rather the transition eludes to the increasing influence of human-digital technology interaction in the agricultural sector (Ingram, 2015(Ingram, , 2018;;Robertson et al., 2016;Smink et al., 2015;Walrave et al., In press).some extent due to technological displacement and social change.Such a broad collection of perceptions contributed to our thinking that digitalisation could (or could not depending on the interplay of existing institutional logics) impact (strategically or inadvertently) any stakeholder in the broader Australian AIS.
Contacting the key individuals working in the Digiscape FSP research programme was relatively straightforward due to an awareness on behalf of the authors of their collegues involved; the prawn aquaculture interviews proved more difficult.Snowball sampling was used to garner the perceptions of at least one individual at each point of the supply chain adjacent the prawn farm (see Fig. 3) (Bernard, 2000).This involved enquiring at the end of each interview whether the interviewee was aware of anyone either side of them in the supply chain that might be interested in being interviewed, after which they either provided contact details directly or via email.Each of the 24 interviews (16 Digiscape and 8 prawn aquaculture value chain) went for between 30 and 60 min and were subsequently professionally transcribed.The resulting data were divided into key themes following a two-stage thematic analysis and subsequent workshopping of the findings within the Social Dimensions of Digiscape project team (CSIRO, 2018).

Results
The results are structured in five sections whereby the first three include data from the Digiscape FSP interviewees in an effort to understand policy relevance to digital technological development generally, the uncertainty around policy in relation to the DAIS transition, and the identification of social learning in regard to what the policy implications of the DAIS meant for respondents.The final two sections include data from the prawn aquaculture interviewees, with particular relevance to policy and finally the digitalisation of technologies utilised in the industry more generally.

Policy relevance to technological development in the Digiscape FSP
There was broad recognition that respondents were familiar with policy within the context that their research projects, as part of the broader Digiscape FSP, operated.For example, recognition of state and national policy relevance of research and technological outputs was explicit: The science that comes out of the project could inform water quality policies and regulations.We might find that things are not as bad as people thought or things are a lot worse than people thought (DI3).Specifically, in relation to the prawn aquaculture value chain that informs the sectoral unit of analysis, an interviewee made links to development within the industry as a result of their previous experience: We've identified that there's this great opportunity for expansion [of prawn aquaculture] but partially it hasn't happened because of policy around granting leases.Policy and our outputs might be linked because what we are hoping to do is to facilitate -to remove some of the barriers for expansion (DI8).
In one case there was explicit recognition of the higher level international policy implications of project impacts: Carbon is totally tied in with international policy, the opportunity we can create to generate low cost carbon [abatement] will profoundly shift the way Government sets its national indicative targets under international treaties, and that will have a role in how Australia negotiates international policy on climate… so by creating through carbon markets those opportunities to drive abatement, build livelihoods or social outcomes and drive environmental outcomes, we can create opportunities (DI14).
These comments indicate that current and future policy settings need to be considered in order for at least some of the research projects within the broader FSP to maximise their impact: We've pretty much designed ourselves to live in the current policy settings and be robust to changes in those policy settings in two, three years' time (DI5).
Taken together, the policy environment within which the research project technological developments are embedded are seen to be critical to the success (or not) of the resulting niche innovation/s.Elevating to the DAIS transition level, particularly in terms of the broader implications of digitalisation on agriculture, resulted in less certainty.

Practical uncertainty around implications of the DAIS transition
While there was recognition, where relevant, that policy settings would influence the impacts of individual research projects and vice versa, the broader discussion around digitalisation and directionality of the policy/technology development relationship was more ambiguous.For example, the following statements indicate a lack of certainty regarding how policy would be affected: Now quite where the intersections of the social and the technical with the policy are I don't have a line of sight to.That's an interesting question (DI5).
So whether the project drives policy or policy dictates success -long term success of the project -I'm not sure about (DI7).
Even when there was certainty around there being an impact on policy in the digital agriculture space there remained gaps in terms of exactly what these impacts might entail: Yeah, certainly, definitely we'll be having policy… something to say about the agricultural data/digital context… [and implications for the] policy landscape (DI2).
The more that we try to implement policy around data and those sort of things -I think that's often an excuse to give it a go and try to sort through those things… who is using and who is actually accessing data (DI10).
In one rather developed sector case, however, the interviewee envisaged the implications of the application of their project: But at the Federal Government level, we will be able to provide far more timely information about agricultural production, [more so] than ever before.That means it could have impacts on how drought funding is distributed (DI12).
So while policy was widely recognised as aligned to research outcomes and ultimate technological impact, and in many cases explicit links to policy impacts were made, in terms of actually envisioning outcomes only one stakeholder had imagined how their project would influence drought funding policy in Australia.Due to the relatively developed nature of the project where there were clear policy implications, it could be argued that actual influence of the other Digiscape research projects on policy (and vice versa) will develop over time.The uncertainty regarding impacts on policy was further evident when respondents reflected on the implications of digitalisation and the broader agricultural systems they were embedded in.

Second order uncertainty around implications of the DAIS for agricultural policy
In questioning researchers on the policy implications of their work, the broad uncertainty in the following two cases was linked to a lack of individual expertise in the policy arena: There's obviously policy angles around that.But I don't really -it's not something I've thought about (DI13).I don't know anything about policy around data ownership.So I have no idea if we're going to be having an impact on that (DI3).
Similarly, there was also conscious recognition of thinking about implications for policy, but an in-depth understanding of those considerations were not the role of the researcher in question: Yes I have thought about it at a policy level.But I am not a policy wonk and nor am I driven by it (DI12).
Taken further, two interviewees provided thoughts that suggest their ontological drivers are being questioned by the digital technologies they are developing: [Are we just] collecting more information or do we change how we operate on the land?That's where I don't know (DI1).
I think one of the concepts I've certainly been listening too a fair bit is you've got to give up something to get something back.We don't seem to be uncomfortable in terms of every time we use our phones we're probably giving up something about ourselves, and the way we go about doing things in the agriculture sector I think they haven't quite grasped [those data privacy issues might be coming] (DI10).
The Digiscape researcher interview data suggest that while there is uncertainty around the extent or directionality of policy implications of the DAIS transition, there was recognition that there would be ontological implications for individual researchers and their perceptions of their projects.Understanding and becoming comfortable with the unknown, within innovation communities and the broader ecosystem, will become increasingly important in order to strategically manage niche innovations to influence regimes (Hermans et al., 2013).This argument supports existing scholarship concerning the importance of being able to embrace the unknown in multi-inter-transdisciplinary research programmes (Bammer, 2012;Fleming and Howden, 2016).

Policy influence on prawn aquaculture
To add value to the analysis beyond the Digiscape research programme, the following two sections delve into a particular research project domain, with perceptions of stakeholders in the prawn value chain gathered as a baseline on the relevance of policy in the industry (more generally), and current and future technological implications.In terms of policy, perceptions varied from negative impacts on the industry to more nuanced understandings of the benefits of regulation for stakeholders involved in prawn aquaculture.Negative comments revolved around the fairness of pollution discharge policies, particularly when compared to more established agricultural pursuits: It's something that the prawn farmers feel very deeply, in that everyone thinks that they're destroying the world… nobody alive today had a grandfather who was a prawn farmer.And so all these other industries, they're part of the national psyche; we've never questioned them.And yet a new industry comes along and is easily picked on (AI1).
It was so ridiculous at one stage, that they were proving that water coming out of the farm had better quality than water coming in.It was, no, we want the same as what's coming in (AI4).
It's just ludicrous the discharge conditions that we have got and they vary from one farm to another and it just hog ties the industry when you have got sugar cane, cattle properties that are putting out far more (AI8).
As well as discharge restrictions the multiple government departments (federal, state and local) having requirements for prawn farm development and operation also contributed to frustration: Do you really want us to develop or do you not want us to develop, because one agency is telling you to develop but other agencies are trying to block us everywhere they can (AI8).
The expansion of the industry is hugely restricted by government policy, particularly because you've got different levels of state and federal policy interacting and then overarching organisations doing things completely differently (AI2).
Whilst negative sentiment exists in respect to the policy requirements to develop and operate a prawn farm, the benefits of such restrictions for the industry were also recognised by interviewees: It is an advantage for our industry to have good regulation because it protects that clean green image.It's not easy to grow prawns but a risk would be that if a dozen companies decided to do what we're doing and all of a sudden there was 10 times the product dumped on the market, we would have issues with price and supply or demand (AI3).
You don't have to walk far to get primary industry being critical of government regulation.I think they'd much prefer to have this regulated environment than have the seemingly unregulated environment of prawn farms in south-east Asia; I can see the benefits from a biosecurity level (AI8).
Of particular relevance to this paper, in terms of links to AIS thinking, one interviewee considered policy to provide an innovation imperative: Applying regulations to an industry gives them the emphasis to innovate to work around it and to come up with new techniques for processing water and so forth.So that's actually the strength of the industry in that by forcing them to do it they will actually do it… so we end up with all the new industries being quite sustainable and yet the ones that we've got, because it's established practise, because everyone has done it since their grandfather, we can't change it.And that's just the way the world is.And it's a difficult thing for government to deal with (AI1).
While there were both positive and negative views on the implications of varying levels of policy on prawn aquaculture, barriers to industry development were also noted to result in protection in terms of managing product demand and biosecurity risk.

Technological influence on prawn aquaculture value chain
In regard to the implications of technology (including digitalisation) on the prawn aquaculture industry interviewees were also both pessimistic and optimistic.For example, pessimism fell into the categories of a lack of value, privacy concerns, and environmental constraints.One interviewee was not at all convinced that the promise of digital technologies were living up to expectations: From our end we have looked at trying to automate our ordering and documentation procedures to simplify it all.It became im-possible… Everything we do is electronic through the [retailers] system.They send us our invoice, we confirm it and we do a delivery docket and our invoice on their system.So the automation on that has been their end.But has it worked?They spend more time fixing the system than they do what it was like beforehand, to be honest (AI8).
Arguably, these perceptions indicate a weakness in the process of technological development whereby a lack of user-centred design can lead to the belief that the system would have been better off without any such alteration (Lacoste and Powles, 2016;McCown et al., 2009).There was also confirmation of data privacy concerns from previous work (Carolan, 2018;Jakku et al., 2016), specifically in regard to taxation implications: I would doubt that [prawn farmers would be open to sharing data] completely, because simply the fact that you have got farms that have got one or two sites or more and if you have got a public view on all of that, including the Australian Taxation Office… it will make one farm look like he's made an enormous profit and a loss on the other one isn't going to be connected… It is still farming and to have that sort of visibility I think would be quite dangerous from their point of view (AI8).
Of further significance in terms of reasons digital technologies may not have a major impact on prawn aquaculture were the environmental constraints of prawn farms.The corrosive saltwater along with the humid and stormy spatial locations which were required in order to maintain high enough water temperature all year round were seen as barriers to investment in in-situ digital monitoring technologies.Connectivity was also seen to be a barrier: The use of iPads and phones we have at the ponds to do real-time data input.Wi-fi is a bit of an issue as far as salt water and corrosion and what not, so real-time monitoring, sending information back via Wi-fi is still a little bit of a challenge, more an environmental challenge rather than technical challenge (AI7).
There's not a farmer's management system in operation anywhere I don't think, but they're probably thinking about it.I guess that gets onto the next question of how feasible it is to install these sorts of things in a fairly hostile environment (AI1).
While the existing reasons digital technologies may not dramatically alter the prawn aquaculture industry were obvious to some interviewees, there was also recognition that digitalisation was coming to the industry, and in some instances was already present.The following four interviewees described digital technologies already improving the efficiency or productivity of their agricultural endeavours through automation, connectivity, monitoring and predictive modelling: Everything runs through a [digital] accounting system, we sell our feed through our website.We're looking at the solutions for a customer portal.Where people log in and are able to see what feed is what, where it's at, we're working on a lot of those kind of things but it's a slow process (AI2).
Just speaking about our business it's quite advanced in that regard.So not just the admin side of things but also our production is virtually all digitalised these days.So things like feed monitoring, water quality monitoring, and feeds are done via machines that can be downloaded or done by iPads.We have a company server so all the information's stored and is accessible centrally.There's room for making it more advanced digitally.For things like 24 h a day remote monitoring.We're still in the R&D phase of that one but virtually everything we do these days is done digitally rather than on paper (AI4).
We have a product tracing program so everything's barcoded and scanned and stored and the same when it goes out so that when it goes to the customer, everything's traceable, digitally (AI5).
We have even started to use a lot of data now to do modelling.The farm guys use their daily data to do real-time modelling on when to harvest to try and predict two, three, four months down the track when they should start harvesting to optimise profit.So, whether they start pulling out prawns that are smaller in size and let the rest grow up or let them stay in the ponds for a larger size before they start harvesting (AI7).
Along with the digital technologies already present there was also recognition that there was a drive to increase digitalisation within the prawn aquaculture industry, coupled with excitement about what might be possible in the not-too-distant future: I think farmers are trying to push for better ways to have their data captured and stored because at the moment I think it just relies on people's memory to record the data in these areas (AI6).
There's a lot of appetite for that in our team here but it's just resourcing it and making it work right.But there's a lot of people doing those kind of processes already (AI3).
If we can use real-time monitoring to measure eight, nine different parameters and then have some advanced modelling to interpret all those data and the trends then I think that can be extremely powerful and certainly be one of those quantum leaps forward.I think really understanding the fine relationships between all the different water body parameters that might be otherwise overlooked just through lack of time or from the complexity of the data (AI2).
These prawn aquaculture value chain findings suggest that to date policy within the industry is considerd both a hindrance and to be beneficial and that despite some disagreement, further digitalisation seems to be inevitable, with associated increases in efficiency and productivity articulated.The results raise some key discussion points surrounding the DAIS transition, prawn aquaculture, and considerations for agricultural policy more broadly.

Discussion
The results indicate that there are diverging opinions on who will and who will not be influenced by digitalisation.For example, due to their roles in developing the digital technologies of the future (for examples see Fig. 3), Digiscape interviewees were optimistic in regard to the technological development but cautious regarding the implications for policy and society more generally.Prawn aquaculture interviewees were open to various digital technologies improving their productivity but those that were interested in such technology seemed to be those that embraced sectoral change in such a young industry.The fact that the most recent industry symposium was premised on the potential of smart farming technologies and still sponsored by corporations and organisations already embedded within the industry, suggests that digitalisation is on the way (Australian Prawn Farmers Association, 2018).These findings align with those of Rotz et al. (2019)) in that already powerful incumbents are perhaps most likely to capture gains through the DAIS due to the privilege their existing position provides them.

Of prawns
While there was mention of cases whereby technological advancements had reduced the efficiency of movements through the value chain, the potential beneficial outcomes of digitalisation on productivity were widely recognised.These perceptions support propositions and existing assumptions that digitalisation can contribute to industry economic growth and systemic change into the future (Leonard et al., 2017;Walter, 2016;Wolfert et al., 2017).
A critical examination of the prawn aquaculture results also indicates that there were no significant concerns around the need for policy to co-evolve alongside digital technologies, policy discussions focused on the hurdles in place to start a prawn farm and more practical concerns.For example, the duplication of bureaucratic processes was found to be frustrating, although there was some understanding that policy acted to protect commercial interests, such as during biosecurity outbreaks and in terms of supply management.In this case, industry discussions around the governance of digitalisation moving forward were immature (Leonard et al., 2017).More nuanced views of the implications of digitalisation were elucidated above this specific industry level case at the Digiscape FSP innovation community level.

Of policy
Digiscape interviewees explicitly recognised the link between technologically driven change through digitalisation and appropriate governance.For Digiscape interviewees, however, there remained questions around the mechanisms that might appropriately facilitate digital agricultural development.These results further support arguments that (at both the niche and broader regime level) digital agricultural governance is in an immature state in Australia with uncertainty around potential implications a common theme throughout interviewee responses (Leonard et al., 2017).Comparatively, lessons from the Australian experiment in the DAIS transition will have implications for other agriculturally competitive export nations with minimal trade barriers, for example Canada, New Zealand and the United States of America.Each of these regions governance structures, along with rather different agricultural regimes (e.g. the EU or developing world), will determine the implications of the DAIS on their constituents.The diversity of action, reflection, and adaptation throughout such a transition process, however, will benefit from lessons arising in other contexts.These findings also support arguments that the development of a dedicated innovation systems research agenda, that moves beyond innovation for innovation's sake in an attempt to strategically guide transitions in various places and spaces, is required (Schlaile et al., 2017).

The DAIS ex-ante socio-technical transition?
Recent calls for a shift away from the over simplification of digital agriculture as a silver bullet to agricultural productivity and sustainability concerns espouse the need for responsible research and innovation (Bronson and Knezevic, 2017;Eastwood et al., 2019) and appropriate governance in the context of agriculture 4.0 (Rose and Chilvers, 2018).Typical transition management timeframes of a human generation (20 or more years) should be considered when thinking about the DAIS (Rotmans et al., 2001).Our analysis supports these calls, with uncertainty suggesting that engagement in leadership and governance of the DAIS will be critical in determining the outcomes for years to come.As Bronson and Knezevic (2017) explain, the ethical implications of power relations on data acquisition, citing the example of powerful social media companies, can determine who benefits as a result of data availability and analytics -with recent developments in the Facebook data sharing scandal and implications for British and American democracy one such example (Duffy, 2018).Digital governance will need to co-evolve with digital technological development in the Australian context such that a discussion concerning how we distribute benefits fairly and equitably is had, contingent on value or productivity increased as opposed to legislative loopholes or rentseeking behaviour of technological elites (Bronson and Knezevic, 2016).There is a need to engage stakeholders in policy forums that consider the impact of pervasive digitalisation in the agricultural sector to gather and organise thoughts, co-develop approaches, and also inform parties of developments in technology and policy (Linkov et al., 2018).Some of this work has begun in Australia (see Leonard et al. (2017)) but long-term and ongoing negotiation about what is or is not acceptable will be required in such a socio-technical transition (Jakku et al., 2019).The ontological question regarding what it means to be productive as AIS digitalise is ambiguous when comparing the intangibility and ubiquity of some on-line digital technologies and the human need to consume and fuel ourselves with the product of agricultural enterprise (Carolan, 2017;Fleming and Howden, 2016).
The questions raised by conceptualisation and interrogation of the DAIS will challenge existing agricultural policy infrastructure in ways that are both understandable and unknowable, and as such the transition will require adaptive governance and guidance toward common principles (Chaffin and Gunderson, 2016).For example, the broad policy implications of data privacy are already front of mind in regard to international scholarship (Bygrave, 2014;Carolan, 2018;Jakku et al., 2016;Rissman et al., 2017;Wolfert et al., 2017).There are three important propositions that are made on the basis of analysis of the results that combined help to answer the primary research question at this point in time: How do different innovation communities percieve technology and policy in relation to a socio-technical transition toward the digitalisation of the Australian AIS? 1) There is uncertainty around the broad policy implications of the DAIS socio-technical transition -as such there is an opportunity to develop a shared vision for digital agriculture in Australia.2) There is a need to engage various agricultural stakeholders with policy forums to initiate discussions around appropriate facilitation and consequences of the DAIS.
3) Such engagement and visioning processes will increase reflexivity and transparency regarding the implications of the DAIS and mitigate risks of ad hoc decision-making.
To apply these findings to the conceptual contribution of this paper, DAIS theorisation can form a bridge between innovation community and broader ecosystem actors (such as policy-makers or governance stakeholders) as it helps explain the convergent implications of individual digital technologies (Jakku and Thorburn, 2010;Kimble et al., 2010;Klerkx et al., 2012b).Thinking about the DAIS could prompt discussion regarding the implicatons for society, and building on broader existential theorisation occuring elsewhere (Dunlop, 2016;Rifkin, 2013).As Tumbas et al. (2019) explain 'organi[s]ations need to enact digital practices, and do so by considering basic cultural elements, mindsets and rituals that characteri[s]e professional backgrounds involved in the process'.
Reflexivity and anticipation will be required as policy and technology co-evolve to manage likely technological and social change as a result of digitalisation.Considering responsible research and innovation principles through the technological development process will help in strategically negotiating the transition toward a digital agricultural innovation system (Bronson, 2018;Eastwood et al., 2019;Fleming et al., 2018).Similarly, recognising the value of open and honest dialogue through agricultural innovation community and ecosystem theorisation also has the potential to manage expectations (Pigford et al., 2018).The DAIS transition conceptualisation begins to articulate (Fig. 2) the implications of a transition toward digital AIS of the future, although as existing work has argued, Australia is only now on the cusp of such a transition process (Leonard et al., 2017;Robertson et al., 2016).
Importantly, we are at a critical point in time whereby the initial conditions of the DAIS are being negotiated.As such, consideration of the drivers of digital technological development to influence agricultural land use need to be debated openly.In Australia, the future development of digital technologies will reshape production, values and understanding in the land-sector.If stakeholders are aware of the values and conditions prescribed by data underlying digital technologies, the technologies have the potential to be more transparent and trustworthy.While there is currently uncertainty around the real policy implications of the DAIS, particularly from those embedded in the agricultural research space, there is a need to engage with policy forums to determine the acceptability of human-machine interaction as machines begin to think, learn and make decisions from their own data-driven experiences through unsupervised machine learning (Brynjolfsson and McAffee, 2017).
The DAIS transition concept can be used to engage stakeholders as the expections of applied digital agricultural research and development strives to meet the impact imperative that has been promised, in terms of expected economic returns on investment from digital agricultural technologies (Leonard et al., 2017).Internationally, work to understand the role of digital platforms to help provide credit access in the developing world (Agyekumhene et al., 2019), and increasing linkages with design-oriented thinking (Berthet et al., 2019) are expanding our understanding of human-technology interaction and the social, organisational and institutional barriers to innovation (Munthali et al., 2019).It is important for international research stakeholders to strategically reflect on and anticipate the potential implications of their work beyond the immediately obvious outcomes of increasing the economic competitiveness of export orientated agricultural production systems.Technological developments are already intermediating network interaction, for example farmers utilising whatsapp group messaging or twitter hashtags (Kelly et al., 2017).It is hoped that the conceptual contribution of this paper will help AIS stakeholders visualise existing and potential agricultural knowledge, advice and value chain networks in a more holistic manner and imagine the cumulative and interconnected implications beyond individual technologies.
The DAIS transition concept links AIS research to increasingly ubiquitous digitalisation in a manner such that reflexive and responsible research and innovation can be further considered in an Australian (and international) context anywhere the digital and the agricultural intersect.Consideration of the rules that are set during maturing of the DAIS transition could introduce an experimental form of adaptive governance, possibly even utilising digital technologies (Chaffin and Gunderson, 2016).In the Australian context, work has already made explicit governance challenges created by the DAIS, with distinct recommendations around how to create the space for a more 'sustainable 'digital agricultural future (Barry et al., 2017).Promisingly, there are significant parallel lessons that can be applied from the rapid evolution of AIS scholarship in the context of the DAIS, with innovation capacities and capabilities being transferrable to the context of the digital landscape, specifically the ability to experiment, take risks, and adapt (Turner et al., 2017).Further, the roles of innovation intermediaries reported in existing work in connecting, involving, and addressing conflicts when working in blurred or unknown spaces will influence outcomes of the DAIS (Agogué et al., 2017).

Conclusions and avenues for future research
This paper utilised aspects of four existing conceptual frames to argue that we are undergoing an international socio-technical transition toward the DAIS (Table 1 and Fig. 2).Regime level themes of policy and technology were explored within the Digiscape FSP and the prawn aquaculture innovation communities leading to three key findings.Firstly, there is uncertainty around the policy implications of digital technologies in the context of agricultural industry, research and development stakeholders.Secondly, that industry, research and development stakeholders should be engaged in discussion around the implications of a DAIS transition.Finally, that negotiation and awareness raised in these forums will mitigate risk and increase the flexibility of transition pathways into the future.The DAIS transition is likely a once in a (human) generation occurrence that will lead to drastic changes in agricultural values and means of production, in a manner similar to the industrial revolution, green revolution, or neoliberal restructuring of centuries past.While the convergence of digital technologies on agricultural and land based industries are exciting, and the associated possibilities for transformation and empowerment are recognised in Australia and around the world, it is important to strategically set up policy infrastructure that can cope with socio-technical change.
Such an analysis of a limited component of the wider AIS does not acknowledge the various, simultaneously evolving, networks of actors in the digitalisation space.For example, this paper neglects the influence of the private sector on digital innovation in agriculture, where discussions of institutional power relations to digital technology implications are just as important, if not more so (Carbonell, 2015).In defence of this paper, however, is the use of an exploratory case study approach to scope in-depth perceptions on the relevance of the conceptualised DAIS transition in Australia.By limiting the scope of this work to one industry (prawn aquaculture), one research programme (Digiscape FSP) and two regime components (policy and technology) we will be able to monitor changes in interactions over time as the research programme, and other DAIS influences evolve.
Our contribution raises important lines of enquiry for future research: What is the likely outcome of digital technological innovation in agriculture in the absence of appropriate policy to protect those at the production end of the value chain?Are there general principles that could be followed for adaptive policy in order to minimise risk in the agricultural sector in the face of rapid technological change?And, what are the ethical implications for action researchers (such as ourselves) in this digitally driven innovation space?Future work will aim to investigate these lines of enquiry.

Fig. 1
Fig. 1 represents examples of three large digital agriculture initiatives in an attempt to highlight how industry based niche innovations form the majority of the applied research work currently underway.The broader questions around data ownership, information access, power relations, meaningful agricultural activities, open or closed networks, and so on are framed by the research programmes and associated regimes and landscapes within which they sit.The DAIS transition aligns with the rapid prioritisation of digital agriculture initiatives in not just the three research programme examples in Fig.1but elsewhere, with the push to find space and become a player in the digital agricultural game evident across the globe(Bronson and Knezevic, 2016;Carolan, 2018;Walter, 2016).In terms of conceptually framing the DAIS, an interesting (and potentially contradictory) point in regard to each of the three research programmes are their multi-facited and ambitious aims(Schut et al., 2016;Turner et al., 2017).While of obvious importance for funders, all three research programmes espouse a simultaneous focus on both incremental and transformational innovation in the agricultural sector(Nagji and Tuff, 2012).For example, the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Digiscape Future Science Platform (FSP) initiative aims to both support next generation decisionmaking and transform agricultural industries and environmental outcomes (CSIRO, 2017).The IoF2020 (2017)IoF, 2017IoF2020 (2017) initiative is argued to have revolutionary potential to increase the sustainability of value chains yet another priority is that efficiency is optimised.The AgResearchLimited (2017) Digital Agriculture initiative aims to both identify the barriers to taking up new digital technologies on and off farm and develop a technology roadmap to support the industry's transition, assumedly involving the DAIS.While a number of other digital agriculture or smart farming research programmes could have been included, the aim of highlighting the three in Fig. 1 is to visualise the converging technological development elevating the proposed DAIS Where we would like to feed into national policy is ultimately to do multi-year climate forecasts.The big issue with multi-year climate forecasts is drought.That's what has the big impacts when it comes along (DI9).So in Australia it is more around trying to improve access to technology to improve profitability.At the Australian government level they are interested in the trade perspective of what the sectors are producing and how much, policy flows off that (DI1).

Fig. 3 .
Fig. 3. Visualising the Digiscape FSP and prawn aquaculture innovation community interviewees.Note: Each interviewee is represented by a solid rectangle, dashed boxes involve stakeholders in the prawn aquculture value chain that were not interviewed, codes DI and AI refer to Digiscape Interviewees and Aquaculture Interviewees respectively in the quotations that follow.