Smart farming technology innovations – Insights and reflections from the German Smart-AKIS hub

Abstract

Digitalisation in agriculture is considered the fourth revolution in farming, which is expressed by a broad range of available digital technologies and data applications. Politicians and experts assume that smart farming technologies (SFT) have a strong potential to enhance economic performance of farming and will contribute to agricultural sustainability as they may increase precision of inputs to crops and soils based on site-specific needs, and link these aspects to farm management systems. This paper explores farmers' and other stakeholders’ perceptions and attitudes towards SFT in Germany with a multi-actor approach. Quantitative and qualitative data show that while there are generally positive attitudes, farmers are less enthusiastic with regard to expected positive effects of SFT for the environment. Also, there is still a number of adoption barriers on the technology level as well as due to an unfavorable institutional and infrastructural environment. Although a multi-actor approach was practiced, close cooperation of practitioners with developers were not frequently observed nor could they be easily supported through action-research. Notwithstanding, through the multi-actor approach, a comprehensive situational picture of SFT appraisal was composed and, a general raise of awareness among the respective AKIS actors generated.

Keywords:

• Smart farming technologies
• Innovation adoption
• Germany
• Multi-actor approach

1 Introduction

The European Union has made it its political aim to foster agricultural productivity and sustainability through increased generation and diffusion of innovations that enhance the sector’s overall competitiveness (EU SCAR, 2012, 2013). With this aim, two complementary policy measures have been implemented in the current policy period (2014–2020) that rely on the so-called “multi-actor approach.” These measures are (i) the European Innovation Partnership AGRI that is operationalised through the European Fund of Rural Development (EU, 2013) and (ii) a number of calls in the Horizon 2020 research programme aiming to bridge the research and innovation divide. The multi-actor approach is a recent instrument meant to render science more immediately relevant for practice by cultivating close science-practice collaborations and ensuring that results get translated (EIP Agri, 2018; Ingram et al., 2018). Conceptually, both measures have integrated the multi-actor approach by using an agricultural knowledge and innovation system (AKIS) paradigm. This paradigm assumes that whether or not an innovation is realised, depends on how effective the cooperation between different actors, and in particular between science and practice actors, is during an innovation process (EU SCAR, 2012, 2013; Wielinga et al., 2017). Consequently, these policy measures can inherently induce a shift in the role of scientists, from distant observation towards more direct interaction with practice, which possibly leads to dilemmas and tensions when bridging research and interventions for innovation implementation.

1.1 Digitalisation innovation as a means to increase agricultural sustainability

One of the most prominent fields of agricultural innovation is digitalisation, also considered the fourth revolution in farming (Walter et al., 2017). This process is driven by a rapid increase in the use of big data for further developing existing technologies for farming (e.g. tractor-based tools that rely on GNSS), apps for mobile devices and software that aim to link data on agricultural production processes (e.g. input quantity and timing) with information on farm-level work processes and quality management, along with other aspects, along the value chain (Wolfert et al., 2017). According to Fountas et al. (2015), four general types of technological applications can currently be distinguished: 1) recording and mapping technologies, which collect precise data for subsequent site-specific application, 2) tractor GPS and connected tools that use real time kinetics to appropriately apply variable rates of inputs and accurately guide tractors, 3) apps and farm management and information systems (FMIS) which integrate and connect with mobile devices for easier monitoring and management and 4) autonomously operating machines (e.g. weeding and harvesting robots). Although extremely diverse, this broad range of technologies is supposed to contribute to a “smarter” way of farming that benefits cultivation practices, crop yield and quality, and farm work, so that “it reduces the environmental and climate impact of farming, increase resilience and soil health and decrease costs for farmers” (COM, 2017:12). We refer to these technologies as smart farming technologies (SFT).

Farming will likely develop in a number of directions, including intensive, agro-ecological, organic, and others. Whichever trajectory farming will take, SFT might be utilized and integrated. However, the future for SFT might be associated with input-intensive farming. Within this context, technology developers, policy makers, and experts widely expect that SFT will contribute to agricultural sustainability because of their ability to increase precision of input application to crops and soils based on site-specific needs, and to directly connect management practices with farm management systems (Walter et al., 2017; COM, 2017; Müller, 2016), making farms more prepared to cope with labour shortage and climate change (Poppe et al., 2013). However, since the introduction of precision agriculture, these types of technologies have been criticised, emphasising the need for clear evidence of the economic and environmental benefits for farmers (Wolf and Buttel, 1996). Whether or not SFT will indeed contribute to environmental sustainability, perpetuate the “technological treadmill” (Levins and Cochrane, 1996), or increase or decrease farmers’ technological dependency (Carolan, 2017) could depend on how perceived barriers to SFT adoption – in particular risks – are addressed and dealt with during the innovation process.

A few studies indicate that European farmers are hesitant to adopt SFT at large (Reichardt and Jurgens, 2009; Long et al., 2016; Kernecker et al., 2019). Moreover, in their exploration of the factors influencing adoption and diffusion, studies have frequently been limited to considering only a small selection of factors, may it be farmers’ age, education, technological affinity (Tey and Brindall, 2012; Paustian and Theuvsen, 2017), innovations’ characteristics and whether the technology fits to the farm (Kutter et al., 2011), or farm-related factors such as economic strengths (Long et al., 2016 and references therein), full-time farming (Daberkow and McBride, 2003), farm size (Kutter et al., 2011), or proximity to SFT vendors (Reichardt and Jürgens, 2009). Another limitation is that many of these studies are conducted with empirical surveys that are not representative of the heterogeneous farming community throughout different regions or countries. For example, the studies by Reichardt and Jürgens (2009), and Reichhardt et al. (2009) rely on surveys conducted solely with farmers attending Agritechnica, the biggest European Agricultural Technology Exhibition, which automatically preselects for farmers who are looking for information on agricultural technology. In other cases, samplings tend to be small or targeting experts only (Kutter et al., 2011; Busse et al., 2014). Thus, detailed knowledge about how SFT are perceived, what SFT characteristics respond to farmers’ needs and innovation adoption, is sparse. This knowledge gap is one obstacle for SFT to contribute to more resilient agriculture, as it was expressed by the EU Commission in their declaration on “the future of food and farming”(COM, 2017:12).

1.2 Project framework and objectives of the paper

As a response to the challenge of improving agricultural digitalisation, the” smart-AKIS” project (http://www.smart-akis.com) was designed as an EU horizon 2020 Coordination and Support Action (CSA). Its aim was to address this knowledge gap by using a multi-actor approach to study SFT innovation and to enrich SFT innovation processes. Given the political expectations related to technological progress for sustainable agriculture, and the premise that using a multi-actor approach in a project could effectively improve innovation processes, we designed and implemented a research project to systematically explore:

•	Stakeholders’, and in particular farmers’, perceptions of SFT as a practical solution to advance farming performance and agricultural sustainability;

•	Stakeholders’ participation in SFT development to ensure that technological progress is implemented in a user-oriented and socially inclusive way;

•	The strengths, weaknesses, and added value of using a multi-actor approach for developing and disseminating SFT innovations; and

•	Scientists’ roles and interventions in such multi-actor settings, aimed at interactive innovation development and dissemination support based on the example of SFT.

Here, we make use of the data collected with research activities in the German smart-AKIS hub. The aim of this paper is consequently twofold: (a) to present a multi-faceted picture of the current state of SFT innovation in Germany, and discuss findings in comparison with the latest evidence from other European countries, and (b) to demonstrate the methodological strengths and challenges of the multi-actor approach applied for improving innovation processes, including innovation development, adoption, and dissemination. Conclusions will be drawn at two levels that correspond with our aims. First, we address the current state of knowledge and further research needs for SFT innovation in Germany with a particular focus on the role of the AKIS and related dynamics. We then illustrate and critically discuss the benefits of using an action-oriented, multi-actor research approach. In particular, we create transparency about the German hub’s activities and discuss their impacts in terms of action-research.

2 Conceptual background

The conceptual background of the above-mentioned policies (EU H2020 CSA, European Innovation Partnership AGRI) relies on a combination of agricultural knowledge and innovation system theories (Ison, 2012; Touzard et al., 2015), and a multi-level governance approach towards transformative processes in societies (Geels, 2002). For the case we present here, we use the procedural understanding of the AKIS concept, which means that the core interest is on the interactions and linkages among the manifold actors to support improved problem solving and innovation (Leeuwis and van den Ban, 2004; Klerkx et al., 2012; Knierim et al., 2015). Empirical studies have illustrated that innovations towards more sustainable forms of agriculture emerge from targeted cooperation among heterogeneous actors (Moschitz et al., 2015; Schmid et al., 2016) and can be successfully promulgated through concerted interventions from actor networks at either local, regional, national, and/or international levels (Beers and Geerling-Eiff, 2013). From a social science perspective, such politically induced, interactive network settings that involve researchers and practitioners from diverse fields, can also be considered a “real world laboratory” implemented with a transdisciplinary research approach (Renn, 2018). In other words, for social scientists, a multi-actor approach towards innovation development is simultaneously research on and support to innovation processes within a knowledge system context.

The study of innovation processes includes both the individual level of an innovation’s adoption, and the collective consideration of and decision-making in regard to an innovation’s development, implementation, and diffusion process within a social system (Rogers, 2003; Karnowski, 2017). In this study, successful SFT innovations are investigated using a static lens on the one side, which considers (i) the innovation’s attributes, (ii) farm characteristics, and farmer characteristics, attitudes, and perspectives, (iii) information sources and channels in play, and (iv) the influences coming from the surrounding social system (Hoffmann et al., 2009:100). Secondly, we consider innovation as a process, taking into account the innovation’s development over time, and the interactions and communication flows within actor networks that occur during the process (Wielinga et al., 2017). However, there is no formula for the influencing factors’ concurrences, but rather contingencies (Albrecht, 1963; Hoffmann, 2007). Empirically, a systemic and situational conceptualisation of individual behaviour change has been substantiated for environmentally innovative agricultural practices (e.g. Siebert et al., 2006; Burton, 2014), although this has not been applied to SFT innovation processes yet. Most importantly, however, the majority of studies on agricultural technology adoption neglect farmers’ perceptions, attitudes and, interests regarding the technologies, thus omitting their subjective perspectives. Investigating farmers’ perceptions and application of SFT, and in particular how their active participation in innovation and knowledge networks could be improved, provide new insights into how and why farmers adopt SFT or not. Previous studies have emphasized the key role that farmers’ participation and connectedness to networks and communities of practice have for innovation generation and adoption (Leeuwis and van den Ban, 2004; Klerkx et al., 2010). This is attributed to farmers exchanging and sharing trusted knowledge (Läpple et al., 2016). However, given that they are part of heterogeneous networks and communities involving multiple actors (besides farmers), a lack of interaction with these other actors can hamper their participation in and access to innovation processes (Dolinska and D’Aquino, 2016), including SFT adoption.

Operating in the multi-actor setting, we use an action research framework to clarify scientists’ roles in the research process. Action research is characterised as “a participatory process concerned with developing practical knowing in the pursuit of worthwhile human purposes. It seeks to bring together action and reflection, theory and practice, in participation with others, in the pursuit of practical solutions to issues of pressing concern to people and more generally the flourishing of individual persons and their communities” (Reason and Bradbury, 2008:4). More precisely, our project’s approach require actors to take a position on how project objectives were determined, and how research was designed and implemented so that the results can be interpreted appropriately. Project activities consisted of classical empirical field studies, and participatory research involving consortium partners from various professional fields, and interactive workshops and (semi-) public events with a broader range of interested stakeholders. Within this context, we use an action research concept to clarify our epistemological perspective and refer to Checkland and Holwell (1998) who highlight action research as an iterative and reflexive integration of classical research elements into real-life situations. Classical research is characterised by a researcher’s framework of ideas which are applied to an area of concern with the help of methods and thereby produce results and learning outcomes Fig. 1. In an action research approach (Fig. 1), the researcher enters into a “real world problem situation” instead of an “area of concern”, which means that actions are not only shaped by ideas and methods but also by interactions in the field. In such settings, findings are obtained by both classical empirical research and by using and reflecting upon results of researchers’ interventions.

Fig. 1 Generic research components and their arrangement in the action research frame (Checkland and Holwell, 1998:15).

In this paper, the presented concepts constitute the framework of ideas of the researchers, as documented at the outset of the project in a deliverable (Kernecker et al., 2017). Further explanations of how the action research was implemented, are presented in the methods section.

3 Methods and material

3.1 The empirical case

Our empirical material originates from the smart-AKIS field research and in particular from the German smart-AKIS project hub, which was established in cooperation between the DLG (German agricultural society, an independent professional farmers’ association at the national level) and ZALF (an independent public research body). From early 2016 to mid-2018, the two partners cooperated in empirical research, regional innovation workshops (RIW) and other public events, aiming to identify farmers’ needs and interests with regard to digital and smart farming technologies, and to investigate factors affecting the generation, adoption, and diffusion processes of SFT in the larger AKIS. A mixed-methods approach was chosen (Creswell and Plano-Clark, 2007), combining instruments from both classical empirical and participatory research in (i) a farmers’ survey, (ii) expert consultation and (iii) a series of multi-actor workshops in combination with further (semi-) public communicative activities.

3.2 Methods of data collection and analysis

Both a standardised survey and a semi-structured interview scheme were elaborated in cooperation between the scientific partner (ZALF) and other partners in the consortium (among them the German DLG). Research questions were based on findings from the literature on technology adoption and were also based on the interests of the project partners, specifically those from farmers’ and advisory organisations, companies, and other consortium members. To ensure that SFT were understood and communicated coherently, the survey included images of the four SFT types: (1) recording and mapping technologies, 2) tractor GPS and connected tools, 3) apps and farm management and information systems (FMIS), and 4) autonomously operating machines. The survey was implemented by the Smart-AKIS partners in seven European countries (the UK, the Netherlands, France, Spain, Greece, Serbia, and Germany). Data was predominantly quantitatively processed and analysed by ZALF (Kernecker et al., 2017).

To complement the survey, semi-structured expert interviews were conducted with actors from science, industries and the wider public (administration, media etc.) (Newing et al., 2010). A software supported transcription and qualitative content analysis were undertaken, applying a deductive category assignment (Dresing and Pehl, 2015; Mayring, 2010, 2014). The results were summarised in a report (Borges et al., 2017). As many experts made statements with an international rather than a national perspective, we do not refer here to German experts only, but present excerpts of the aggregated findings.

As a third empirical component, 3 multi-actor workshops were held during 2017–2018 in different locations throughout Germany: Saxony-Anhalt (RIW1), Bavaria (RIW2), and Saxony (RIW3). Each workshop followed an agenda that was jointly decided upon by the German hub and targeted a range of different actors (farmers, technology providing companies, members of the agricultural administration, agricultural advisors, etc.). Most participants were contacted personally via phone or email by the DLG partner and a few through the ZALF network. The program of each of the workshops followed a project-related, predefined scheme and focused on the initial phases in an innovation process and related specific requirements for integrating SFT into farming. Specifically, the focus of each of the RIW were (i) identifying and confirming farmers’ needs and main challenges for SFT adoption, (ii) identifying the particular needs of small-scale farmers or small-structured farms and presenting potential solutions; and (iii) inviting innovators (researchers, start-ups, consultants) and funders to create potential funded project groups. Methodologically, the workshops were composed of (i) targeted inputs from scientists and experts in the subject matter (out of the team of authors); (ii) presentations on SFT from technology providers, (iii) facilitated and documented group discussions, and (iv) participatory ranking and assessments with all workshop participants.

Finally, the partners separately or jointly presented smart-AKIS findings during exhibitions, fairs, innovation network events and similar occasions in order to sensitise for SFT innovation and to mobilise stakeholders for participation in project activities.

3.3 Sampling

Sampling was done separately for each of the different components. For the farmers’ survey, a purposive scheme was designed combining a) the dominant cropping systems per country, including arable farming, orchards, field vegetable and vineyards (based on a selection for investigations in all 7 EU regions of the project), b) a range of representative farm size classes, and c) adopters and non-adopters of SFT from a wide range of ages. We aimed to conduct surveys with 5–15 farmers from the relevant size classes for each cropping system, and thereby reflect the heterogeneity of European cropping systems to capture the technological needs, ideas, interests, and perceptions from a stratified farming population. As the sampling of farmers was restricted by consortium partners’ resources for field work, a bias towards organisationally related farmers cannot be excluded. We collected surveys from a total of 287 farmers from across Europe. Here, we rely on the 27 responses from arable German farmers (Table 1).

Table 1 Stratification of farmers surveyed; Number of adopter (A) and non-adopters (NA) are listed according to farm size class, and both country (above) and cropping system (below).

	0-2 ha		2-10ha		11-50ha		51-100ha		101-200ha		201-500ha		over 500 ha
	A	NA	A	NA	A	NA	A	NA	A	NA	A	NA	A	NA	Total A	Total NA	Total
France	0	0	2	8	3	7	3	6	8	6	4	0	2	0	22	27	49
Germany	0	0	0	1	1	0	3	1	2	1	5	0	14	0	25	3	28
Greece	0	9	14	33	3	7	0	2	0	0	0	0	0	0	17	51	68
Serbia	1	3	2	17	1	6	1	5	0	0	0	0	0	0	5	31	36
Spain	0	0	0	1	4	10	2	1	5	4	1	0	0	0	12	16	28
Netherlands	0	0	2	1	12	2	6	5	9	3	4	0	0	0	33	11	44
UK	0	0	0	2	0	1	1	0	4	0	6	1	19	0	30	4	34
Total	1	12	20	63	24	33	16	20	28	14	20	1	35	0	144	143	287
Arable	1	3	2	10	6	6	10	15	21	11	18	1	29	0	87	46	133
Orchards	0	8	6	21	7	4	0	0	0	0	0	0	0	0	13	33	46
Vegetables	0	0	0	9	3	7	4	4	5	2	2	0	6	0	20	22	42
Vineyards	0	1	12	23	8	16	2	1	2	1	0	0	0	0	24	42	66
Total	1	12	20	63	24	33	16	20	28	14	20	1	35	0	144	143	287

The selection of experts for consultative interviews was based on previous projects’ experiences and on recommendations of some project partners, such as members of CEMA, the European Agricultural Machinery Association, and colleagues from each of the hubs. Here, we present aggregated insights from all 22 experts (10 from industry, 4 from practice and 8 from research) and highlight selected responses from the 3 German experts (2 from research, one from practice).

For the RIWs, particular attention was paid to a mixed actor composition in order to reach cross-cutting discussions and multiple perspectives on the issues addressed. This aim was largely achieved for the first and the second workshop, while the third workshop had an” innovation development” focus and therefore a reduced and selective sample of participants (Table 2).

Table 2 Participants in the regional workshops (RIW) according to professional affiliation.

Actor group	RIW 1	RIW 2	RIW3
Researchers	17	28	4
Industry	24	19	2 (start-ups)
Farmers	13	15
Advisors/consultants	12	5	1
Funding party			2
Total	66	67	9

4 Findings

In the following, we present quantitative and qualitative findings from the German smart-AKIS hub and contrast these with cross-cutting findings from all seven case-studies which were described in more detail by Kernecker et al. (2019).

4.1 German farmer sample

In Germany, a total of 27 farmers participated in the survey, among which 25 had mainly arable crops and 2 farmers produced mainly open field vegetable crops (Fig. 2). Farm size was quite large in general, with 14 farms having more than 500 ha, and 5 others having between 200 and 500 ha (Table 1). Such sizes are far beyond the German average size of 59 ha for full-time farms (BMEL, 2015:47). In this regard, the sample is quite unique. Both farm size and cropping patterns were quite similar for the German and the UK hub, while the other participating countries reached a higher variation of farms, covering at least 3 cropping systems and a more equal representation of the different size classes (Kernecker et al., 2019). Like the majority of all participating farms, 15 out of the 27 German farms were family farms, most of them run at full-time or almost full-time. Lastly, the participating farms reflect the spatial diversity of Germany, being located in 10 out of the 13 non-urban states (“Länder”) which allows us to at least partly capture the regional diversity with the sample.

Fig. 2 Total number of farmers surveyed per country, grouped according to the different cropping systems.

Roughly half of the German farmers belonged to the group of young farmers (age class 20–29), and they all had either a technical or university education. In this regard, again, there were similarities to the farmers surveyed in the UK, while in the samples from the other countries, farmers had more diverse educational backgrounds. Almost all of the German arable crop farmers used SFT, and only 2 farmers were non-adopters, which makes the German sample stand out from the rest. The use of mobile phones was common among all groups, including German farmers, where only one farmer did not have a mobile phone. In all countries except Serbia, farmers documented their data by using either manual or digital or both methods. German farmers managed their farm data digitally (7 farmers), by hand (2 farmers), and both digitally and manually (18 farmers). Only in the Netherlands and the UK, there was a larger proportion of farmers who documented their farm data solely digitally and a correspondingly slim proportion of farmers who documented their data only by hand.

In sum, the German sample is characterized by large farms, one dominant cropping system, and a relatively young, highly educated group of farmers which have almost all already some experience with SFT, in particular with tractor-based GPS and associated tools. Although this homogeneity of the sample can partly be attributed to the way the interviewees were recruited, which relied on the DLG farmers’ network and a field day event among ZALF partners, there are interesting parallels to the Dutch and the UK sample which suggest a certain socio-structural and farming system related predisposition for SFT.

4.2 Farmers’ attitudes, expectation, and perceptions of SFT

All farmers across all participating countries had positive attitudes regarding technology in general, and in this regard, German farmers did not differ from the other groups. Most farmers were active in seeking solutions through their own, farm-level experimentation: Throughout Europe, 78% of the farmers do experiment on their farms, either by themselves (30%), with other farmers (8%), or with researchers or advisors (26%). Furthermore, 70% of the farmers had pro-actively sought out information on SFT. Over half of the farmers said they visited trade fairs more than once a year, and about 30% said they visited once a year. These figures were reflected in the German sample: 6 out of the 27 farmers stated that they do not experiment on their farm, only 4 farmers had not sought out information on SFT, and all but 2 farmers attend trade fairs and exhibits at least once a year.

With a number of closed questions, we explored farmers’ expectations of SFT to support their farms’ performance and environmental sustainability. As a point of reference for the survey questions, we asked farmers to keep in mind one of the pre-defined SFT types that were presented in images, and to agree or disagree with statements on what that particular SFT could do for farming performance and environmental impact. 21 out of 27 farmers referred to GPS related technologies. All responses were largely positive, in particular, those to more general statements, for example, “SFT is useful for farming” (Fig. 3). A more cautious attitude was expressed with regard to SFT’ potential to improve environmental or work comfort conditions with 8 and 11 people disagreeing. In general, the more specific questions had more contested responses – both among German farmers and also in the total European sample.

Fig. 3 German farmers’ expectations of SFT potential for general on-farm considerations.

Furthermore, we asked farmers to rank societal challenges related to agricultural production that make or could make SFT more important for their farms. These questions revealed more explicit expectations that farmers had for SFT to fulfil certain environmental requirements. About 80% of the German farmers (21 out of 27 farmers) considered using SFT to reduce N use as important or very important. Soil conservation was also considered of high importance, as was regulation compliance and biodiversity conservation. Reducing herbicides and harvest losses were the least important, on average, for German farmers (Fig. 4). The German farmers’ ranking differed considerably to the total sample in terms of attributing importance to the SFT to overcome challenges in general (absolute mean values). Obviously, German farmers’ perception of SFT’ potential to effectively deal with challenges reflects the current issue of nitrate emission into water bodies, and related recent changes in fertilizer regulation, expanding upon the Nitrate Directive. However, the general expectations for SFT are relatively modest compared to the total sample.

Fig. 4 German farmers' ranking of farm challenges that could make SFT a promising means to increase sustainability.

In sum, we observe a high interest in and a generally positive attitude towards SFT among a well-informed and SFT experienced sample of German farmers. Judgements from this same group were somewhat more cautious when survey statements became more specific with regard to SFT potential for certain farming outcomes. In particular, technologies’ contributions to increase farm-level environmental sustainability were moderately rated.

Out of the 27 German farmers, 19 provided feedback about barriers to SFT adoption highlighting costs (8 mentions) and compatibility (6 mentions) predominantly, mirroring the total sample in regard to the barriers they perceived (Kernecker et al., 2019). Other key statements addressed the need to consolidate functions, thereby adapting the SFT to address not only single farming issues, but taking the holistic nature of farming into account. Moreover, one farmer mentioned that new SFT are not particularly needed, but that manufacturers and developers should focus on optimizing existing SFT. Specifically, the constant onslaught of new SFT slows adoption, since farmers need time to become adapted to devices. In regard to barriers to SFT adoption, German farmers differed from the European sample, since access through information and infrastructure was not mentioned – only cost.

4.3 Experts’ perspectives – qualitative findings

In regard to farmers’ and farms’ characteristics and how they relate to SFT innovation processes, all experts considered age as a crucial determinant for adoption. Experts argue that older generations tend to suspend the farms’ technological equipment, and a disruption of technology is expected with the generation change. Referring to farmers’ level of education, opinions were split. While northern European experts emphasized the importance of further education towards innovative technology adoption, southern European experts addressed a general lack of education as a hindering factor. Farmers’ general interest in new technologies was broadly confirmed. While various economic arguments were prominent as fostering factors, the interest in more environmentally friendly or sustainable operations was considered as a non-primary need but a welcomed side-effect of the technologies. Some of the experts didn’t consider structural factors, including farm size, as decisive alone but only in combination with a particular cropping system. On the other hand, several interviewees of the practice group emphasized that economic farm size does matter because of farms’ capacity to invest and the resulting economy of scales. Nevertheless, some experts suggested that in the long run, this scale advantage may become obsolete as a drift towards autonomous machines and robotics with a more flexible scope of application is expected.

Furthermore, experts stated that farmers are not primarily interested in technology, but rather in tools for farming. They believed that farmers’ attitudes and values play a major role for their decision making, since farmers think systemically and are generally well informed of both challenges and potential solutions. Moreover, experts stated that farmers think broadly, so that when farmers consider using or adopting SFT, they consider all technological possibilities to address their needs. For example, the limited availability of labour force may become more and more a constraint for farm management, and SFT could, therefore, be a possibility to replace manpower. However, experts denied that this was farmers’ intentions and instead highlighted a deficit of skilled workers that can operate SFT. One of the experts addressed not only the divide in the technological markets between large and small farms, but also between conventional and organic farms, and lamented that there were not single technologies applicable to all and any types of farms. On a similar note, experts addressed the high value that farmers attribute to their autonomy and stated: “farmers hate black boxes (…) a farmer would only accept technology when the technology is telling him why it suggests something” (research expert). Consequently, farmers tend to avoid lock-ins with technology companies, a particularly pertinent issue for small-scale farmers. The interface between the SFT manufacturer or developer and the agricultural user also caught experts’ attention and it was frequently observed that “industry is offering something that doesn’t fit to the needs of farmers” (research expert), thereby requesting that technology development and practice targeted real practical problems in SFT innovation processes. Obviously, there are communication challenges and one problem “in smart farming technologies is that a farmer has to learn the vocabulary of the programmer” (research expert).

Besides farm-level factors and farmers’ characteristics, experts referred to a number of further influencing factors from the direct and the wider social environment. A strong emphasis was put on access to impartial information and client-tailored advice, which was perceived as lacking all over Europe. Secondly, supportive political instruments were acknowledged. Thirdly, societal expectations and rising demand for quality products was considered as supporting SFT innovation adoption, especially in northern Europe. And, as a further crucial point for SFT dissemination, rural areas’ infrastructures for digitalisation were emphasised.

4.4 German regional innovation workshops

The regional innovation workshops (RIW) confirmed and expanded upon findings from the farmer survey and expert interviews regarding stakeholders’ perceptions of SFT to deal with on-farm challenges. Specifically, the invited product presentations from private companies in RIW 1 and 2 corresponded with the anticipated interests of the participants, notably farmers. Also, both workshops served to crosscheck, complement, and differentiate previous findings. SFT’ relative advantage for large farms was both confirmed, and questioned at the same time (“Digital platforms are a great help for farms, regardless of the size of the farm”, participant RIW 1). Barriers to SFT adoption and dissemination that were discussed at the RIW 1 and 2, allowed us to identify four main constraints: missing standards, hardware development, mobile/digital infrastructure, and communication and information (Erdle, 2018). A lack of standardization at device interfaces to ensure rapid and secure exchange of data between different systems was widely criticised. Also, there is a mismatch between hardware developments and the speed of software solution development. For example, software, data analysis, and information supply are ahead of machine steering (seed densities, tilling technology). However, there is not yet sufficient infrastructure available for the exchange and transportation of large amounts of data required for using SFT in agriculture, which is particularly true for rural areas in Eastern Germany. Another complaint was that information on SFT for farmers is not neutral. Finally, a number of farmers cannot reproduce the advantages of using SFT, especially when the costs of SFT exceed the direct visible benefit. While during RIW 2, small-scale farmers tended to confirm both interests in and similar problems to cope with SFT, the specific challenge they formulated, is to keep up with the learning and training requirements that are connected to the application of new technologies, because they usually have less time availabilities. On the other hand, they addressed specific needs and opportunities for SFT, which included work assistance through automated steering systems, or an improved section control when working on small fields with irregular shapes (Pfeiffer et al., 2018).

The workshops also yielded more concrete feedbacks on particular technologies. For example, while tools for fertilizer and pesticide application are adopted frequently due to their direct effect on crop production, digital platforms are viewed with more scepticism because there is no visible direct benefit for the production system. Furthermore, the question arose whether it was SFT supply or demand shaping what farmers adopt and how. SFT developers lamented that farmers did not have enough ideas for SFT. On the other hand, farmers lamented that they want SFT developers to supply solutions that work and are relevant, and that instead SFT providers took a “wait-and-see” position for what is demanded in practice. Furthermore, consultants and farmers agreed that the current vocational education does not meet the needs for using SFT appropriately in farming. Simultaneously, basic knowledge about soil-plant-climate interactions and basic plant production seems to be getting lost.

Finally, the RIW findings deviated from survey and interview findings by contesting the assumption that farm size matters for adoption. While survey findings clearly indicated that across Europe, SFT adoption has increased with farm size until now, and several experts confirmed the trend and likelihood of larger farms to be the SFT adopters, the second RIW on small farm structures in southern Germany clearly revealed that participating farmers did not agree. Farmers and other workshop participants suggested that rather than size, adoption was hindered due to lack of education and information related to both SFT and agronomy for farmers, advisors, and SFT providers. In this sense, farmers’ uptake of SFT was influenced by their trust in and risks associated with SFT use, which was one reason for their pursuit to communicate with other AKIS stakeholders (Pfeiffer et al., 2018).

The RIW 1 and 2 did not generate direct interactions related to SFT innovations among the various participating actors, even though this was a goal for the workshops. On the contrary, a certain reservation of both the industry partners and farmers to discuss innovative ideas was observed. This can be explained by the potentially competitive situation that resulted from having a range of private entrepreneurs jointly in the workshops. The third RIW was planned to overcome this dilemma and dedicated to defining and developing two particular projects for innovative technologies with selected business and research partners. Consequently, the number of participants in RIW 3 was considerably lower and the diversity of actors clearly reduced compared to the first ones (Table 1). Also, the input of the ZALF-DLG team was targeted towards financing and project development measures and instruments only. In this regard, the results with regard to the factors enhancing SFT innovation generation and adoption were limited to economic challenges in the phase of prototype development.

4.5 Multi-actor approach in SFT innovation processes

The smart-AKIS multi-actor approach in Germany was implemented through the continued cooperation between ZALF and DLG and through the multi-actor composition of the three regional workshops. The ZALF-DLG cooperation was effective in data production and knowledge generation, in the outreach to the AKIS actor groups, and with regard to SFT awareness creation in several farmers’ communities. Nevertheless, the cooperation was also demanding in terms of communication time, and in particular, data collection was hampered by different expectations and methodological approaches, so that in this regard, the shared work was not efficient. This science-practice cooperation was perceived as gradually improving, and finally matured during the last year of project. The cooperation was operationalized when DLG utilized data from the survey to stimulate workshop input. Since the project ended, the German hub or ZALF-DLG cooperation no longer exists, so the smart-AKIS approach did not allow a lasting science-practice partnership to be built up.

The RIWs used the multi-actor approach, and provided insight to the complexity of the SFT innovation processes, yielded considerable additional information regarding farmers’ needs and science-practice cooperation, and therefore allowed us to learn new lessons about the innovation system. The success of the RIWs was clearly due to the specific focus, both on topics, goals, and regions, which can be attributed to the DLG’s excellent links to the practitioners within the AKIS. Still, a high level of engagement of both DLG and ZALF was necessary to mobilize participants and to carry out an interactive programme. In contrast to assumptions made in the project proposal, the workshops’ multi-actor settings did not directly lead to establishing or further developing SFT innovations.

Aside from their professional profiles, the authors were strongly involved in dissemination activities such as an interactive presentation to the Agritechnica audience, presenting to interested farmers at other events, interacting with other groups within the German EIP-agri networks, reaching out to the wider public interested in science, and creating awareness for the web-based technology portal. These engagements were inherent to the project’s character, and were personally rewarding. However, they were surprisingly demanding in time and resources and had little relevance to the scientific professional profile.

5 Discussion

5.1 Current knowledge and further research needs for SFT innovation

In the framework of the innovation-supportive research project smart-AKIS, which applied a multi-actor approach, we studied the factors influencing farmers’ SFT adoption in Germany. Findings partly confirm and partly expand on and differentiate current knowledge on farmers’ and farming systems’ characteristics that are favourable to SFT adoption. First, the farming system that we studied is highly familiar with SFT, mostly due to the prevalence of GPS-based technologies in arable farming. This phenomenon of farming system- specific technology choices was even more evident for the other smart-AKIS regional cases, where large differences between regions, farm structures, and cropping systems reflected how SFT could respond to farmers’ needs and farming objectives (Kernecker et al., 2019). To our knowledge, such particular relationships have not yet been addressed elsewhere in the literature and would merit further systematic exploration. Secondly, the socio-demographic characteristics of the German sample, which included young, well-educated farmers who are responsible for large, arable crop farms often with > 500 ha, largely correspond to what has been formerly reported for Germany and other EU countries (Paustian and Theuvsen, 2017; Pierpaoli et al., 2013). This trend is ascribed to scale effects and related economic profitability (see Barnes et al., 2019 and references therein).

In regard to farmers’ perceptions, we were able to expand upon previously unaddressed factors, including farmers’ attitudes towards SFT for dealing with societal challenges. We observed that positive expectations of SFT potentials are widely expressed by experts and stakeholders from the wider AKIS (Busse et al., 2014; Antle et al., 2017; Walter et al., 2017), including experts’ views that were presented here. The obvious reservation of the German farmers towards SFT performance in moderating farms’ external (i.e. environmental) effects, is in contrast to their general technology-affirmative attitude. This result is interpreted as farmers’ realistic assessment of technological pros and cons, which is however weakened by the deplored lack of information, training, and access to advice on SFT innovations. Farmers’ attitudes and perceptions have been widely studied in relation to adoption of agri-environmental schemes or conservation management practices (e.g. Morris and Potter, 1995; Defrancesco et al., 2008; Mills et al., 2017; Werner et al., 2017). However, based on our review of the literature, how technologies are truly viewed as a means to achieve agricultural sustainability has not been studied.

We used the procedural AKIS concept (Klerkx et al., 2012) to study the performance of SFT innovation processes, which allowed us to systematically explore actors’ linkages, their effective cooperation and co-design. Through the RIW, we found evidence of a lack of effective interaction and knowledge flows, in particular between farmers and technology providers. Also, there was an almost complete lack of engagement from the agricultural advisors’ group, a gap confirmed by some of the experts. Weak links between AKIS actors have been recognized as a bottleneck for effectively implementing agricultural technology systems (van Crowder and Anderson, 1997). Such fragmented AKIS is apparently not as prolific as e.g. ‘learning innovation networks for sustainable agriculture’ (LINSAs) (Moschitz et al., 2015), and one barrier to innovation is the missing support function of providing access to impartial information and advice (Faure et al., 2019), which can be attributed to the regionalised, highly pluralistic extension system in Germany (Knierim et al., 2015).

5.2 Benefits of using a multi-actor approach and action research

Methodologically, the multi-actor approach yielded a diversity of perspectives on SFT innovation that emerged from the various research settings that we used in this study. In particular, we found that RIW were successful in that they provided a platform for the many different actors to engage in a dialogue regarding SFT innovation processes, even if in these settings company representatives were not prepared to negotiate in public. In sum, the multi-actor approach (i) yielded complementary data, which portrayed the current state of SFT adoption and diffusion in Germany; (ii) created awareness for SFT innovation processes among diverse actors, and (iii) mobilized a few actors in project development. However, looking closer at the multi-actor settings we engaged in, we have to conclude that apparently, they were not conducive enough to close divides between the various private entrepreneurial actors. We ascribe this to the characteristics of the SFT in question, to their non-triability, reduced observability and limited relative advantages (Rogers, 2003, Erdle, 2018), and, to the incontestable barriers of the infrastructural environment, such as missing technological standards, mismatches in hardware and software developments and deficits of digital infrastructures, notably in rural areas (Walter et al., 2017; Erdle, 2018). Most obviously, both such hindering factors cannot be overcome in a project setting, thus challenging the assumptions behind the political instruments used by the EU.

As social scientists cooperating with different actors involved in the exploration and support of (SFT) innovations, our interventions in ‘real-life’ settings resulted in an expansion of roles, which was challenging - in particular, the time allotted to communication with project partners and the wider AKIS network. Taking on roles as facilitator, informant or process designer, and engaging in related activities required openness, a learning attitude, and high flexibility (Schmid et al., 2016). Thus, this kind of action research was personally enriching and rewarding, however, insights and professional stimulations in a strict social science sense were mostly of general character and did not delve deep conceptually. In sum, acknowledging and working with perspectives from other sectors, including industry, practice, and policy, generated more questions than answers with regard to the design of effective multi-actor innovation processes. Moreover, participation in such a project made us question how well disseminating SFT and other innovations fits to scientists’ professional profile, or if it is too close to the market or politics (see, for example, Jain et al., 2009). Further reflexive research and work to consider how the strengths of the multi-actor approach can be combined with a scientifically more rewarding action research format is necessary.

6 Conclusion

With this study, we present a topical overview of smart farming technology (SFT) adoption and its influencing factors in Germany, as perceived by a range of stakeholders, in particular different farmer groups, experts and multi-actor constellations. Farmers’ perceptions and judgements, although obtained from different intentionally constituted samples, are quite complementary. Farmers, both in a survey and during regional innovation workshops (RIW), had a positive view towards SFT in general, but perceived a broad range of barriers when it came to implementation. Partially, these barriers could be overcome by a better adjustment of technologies to farmers’ needs and farm conditions, which however would require additional activities from both, farmers and technology providers. On the other hand, an improved enabling environment would greatly improve the favourable adoption conditions, in particular focusing on better access to SFT related information, training and advisory services and to reliable digital infrastructure.

The multi-actor approach that we used, allowed to obtain information on SFT innovation processes that was relevant to more diverse actors, and provided insight to the inconsistencies in the expectations, goals related to SFT innovations, and deficits in the SFT related AKIS. As a sum, the science-practice cooperation between the public research body and the professional farmers’ organisation to implement the German smart-AKIS hub, was successful in sensitising and mobilising professional actors. However, the project’s time was much too short to evaluate possible impacts, and with its end, the cooperation ended, so that its effective success can hardly be assessed.

Finally, the multi-actor approach in this type of funding measure, clearly impacted on the scientists’ roles, and expanded upon their professional profile, leading away from traditional duties to being coordinators, facilitators, and conducting outreach for disseminating information on SFT innovations. Whether or not these activities go beyond the scope of participatory, multi-actor research, and more into marketing and representing industry is unclear – at least this tension highlights the fine line separating these worlds.

Acknowledgements

The authors acknowledge the smart-AKIS project consortium. This work was funded by the EU horizon 2020 Program, grant agreement n° 696294. The authors are grateful for constructive comments and valuable hints from two anonymous referees and the guest editor.