Farmer-driven innovation: lessons from a case study of subterranean clover seed production

Farmers are often overlooked and undervalued as sources of innovation, but can be powerful drivers of ingenuity and development. We evaluate historical developments in the Australian subterranean clover seed-production industry as a case study of farmer-driven innovation. Subterranean clover seed machinery patents (75% of which were patented by farmers) are analysed using conventional innovation frameworks, such as the theory of inventive problem solving (TRIZ), to extract lessons for supporting farmer-driven innovation. The small scale of this industry, compared with mainstream cereal-cropping industries and the isolation of farmers, provides analogous lessons for agriculture in developing countries. Economic drivers are important in enabling farmer innovation and the value proposition for developing new inventions must be clear to justify the time and expense. Farmers are different from firms and their on-farm knowledge and experience can form an essential part of innovation. Drivers of innovation also differ, with farmers less likely to attempt to commercialize inventions. Farmers can also be hesitant to share their inventions, instead holding them as trade secrets in competitive industries. Support and collaboration are needed from government and researchers to assist in commercialization or dissemination of useful innovations and to prevent knowledge from being confined to a localized farmer or region. Advances in agriculture require farmer input in research and development, but the benefits will be greater if farmers are enabled to be drivers of innovation.


Farmer-driven innovation
Global agriculture faces the combined challenges of feeding the growing population while reducing its environmental footprint -particularly land use and greenhouse gas emissions (Schmidhuber and Tubiello, 2007;Godfray et al., 2010;FAO, 2018;Sadras et al., 2020). Agricultural innovation is essential to meet these challenges and intensify global agricultural production sustainably (OECD, 2013;MacMillan and Benton, 2014;Läpple et al., 2015;Gaffney et al., 2019). Recognition of these challenges, and of the economic, environmental and social benefits for overcoming them, is driving innovation in agriculture. Approximately US$26 billion was invested into agricultural food and technology (agrifood tech) startup firms in 2020 (AgFunder, 2021). In addition, there is significant investment from large, established corporations; for example, Bayer's Crop Science division plans to invest €25 billion (~US$30 billion) in research and development (R&D) over ten years (Bayer, 2019). This increasing investment from private firms and venture capitalists contrasts with stagnating (in real terms) public agricultural R&D investment from high-income countries such as the United States (Heisey and Fuglie, 2018).
However, R&D is not the sole driver of advancement in agriculture and farmers are often key sources of innovation (Biggs and Clay, 1981;Rhoades and Booth, 1982;Hoffmann et al., 2007;Sneddon et al., 2009;Sauer, 2017). On-farm innovation is overlooked (McKenzie, 2013;Bragdon and Smith, 2015;MacMillan and Benton, 2014;Läpple et al., 2015) and farmers are frequently considered only in terms of adoption of knowledge and technology, not its creation (Dolinska, 2017). The US Department of Agriculture estimates US$20 billion in annual private sector agriculture R&D investment, but counts $0 of farmer research in this total (MacMillan and Benton, 2014). This figure underestimates the contribution of on-farm R&D and there is significant untapped potential to harness farmers as sources of agricultural innovation.
As historical context can be a valuable tool for developing new solutions to existing challenges (Guzzomi et al., 2012;Díaz Lankenau and Winter, 2019), we present a history-based perspective of farmer innovation through a case study of subterranean clover (trifolium subterraneum L.) seed production in Australia. In particular, we look at the generation of patented machine designs for harvesting subterranean clover (subclover) seed and the factors affecting technology development and success. We posit that understanding the processes driving on-farm innovation and recognizing farmers as valuable sources of ingenuity could provide the tools to enable or empower farmers to contribute novel and grassroots innovation to assist meeting the challenges facing global agriculture.

Subterranean clover seed production
Subclover is Australia's most important and widely sown (over 29 million ha) annual pasture legume (Smith, 2011;Nichols et al., 2013;Ghamkhar et al., 2015). Subclover forms an important part of Australia's farming systems and has contributed to significant agricultural improvements over the last century (Donald and Williams, 1954;Puckridge and French, 1983;Smith, 2000;Peoples and Baldock, 2001). Subclover pastures increase soil fertility for subsequent cropping phases in addition to providing high-quality pasture for stocking phases (Puckridge and French, 1983;Nichols et al., 2012). This legume rotation system is receiving global attention as part of 'sustainable intensification' frameworks, which seek to address food security and climate change issues in global agriculture (Pretty and Bharucha, 2014;Pretty et al., 2018).
One of the strengths of subclover as a pasture plant is its ability to bury its seed, which protects the seed from desiccation and grazing by livestock, and establishes a seed bank for self-regeneration in future years (Loi et al., 2005;Nichols et al., 2012). However, harvesting this small seed (typically 6-10 mg) from below the soil surface presents significant difficulties compared with harvesting aerialseeding crops. Overcoming this challenge necessitated innovation to develop specialized machinery and techniques, and farmers played a key role in the development of the solutions that advanced this seed-production industry (Moss et al., 2022). This relatively small industry -the size of the Australian subclover seed-production is measured in the thousands of tonnes each year, while Australian wheat production is measured in millions of tonnes (ABARES, 2020, Australian Seeds Authority, 2020) -is analogous to agriculture in developing countries.
To provide context to innovations in the subclover seed industry, a brief historical account of the use of subclover in Australia follows. More detailed accounts are provided in Henzell (2007), Nichols et al. (2012) and Moss et al. (2022). Subclover seed was first harvested in small quantities around 1900 in Mt Barker, South Australia. From 1907, subclover seed began to be marketed commercially, which necessitated the development of systems to process the seed and overcome the difficulty of harvesting the buried burrs. The sowing of subclover became more widespread in the 1920s and 1930s as the plant became widely used in 'ley farming' systems, a uniquely Australian pasture-crop rotation system where cereal crops utilize the nitrogen fixed by the legume in the previous year (Underwood and Gladstones, 1979;Puckridge and French, 1983).
Rapid expansion in the area sown to subclover occurred from the 1950s to the 1970s. Key drivers included a wool boom, the opening up of new beef markets and government incentives for agricultural investment (Henzell, 2007). This was an era of large-scale pasture improvement, the so-called 'sub and super revolution', based largely on widespread sowing of subclover and the application of superphosphate (Underwood and Gladstones, 1979;Crofts, 1997;Henzell, 2007). The Horwood Bagshaw Clover Harvester was released in 1961, increasing harvest efficiency and becoming the most common subclover seed harvester (Moss et al., 2021b).
In the 1980s, a decline in subclover pasture productivity first became apparent, attributable to a combination of new plant diseases and insect pests, increased soil acidity, reduced fertilizer inputs and increased cropping frequency (Nichols et al., 2013). Increased international demand for synthetic fibres and a reduced demand for wool led to collapse of the wool industry in 1991, instigating a decline in sheep numbers (Henzel, 2007). Subsequent low prices for wool, relative to grains, continued this downturn and contributed to an increase in the area sown to crops at the expense of pastures. Recent trends, however, indicate an increasing domestic and export demand for beef and sheep meat (MLA, 2020) and dairy products (Dairy Australia, 2021) and have the potential to translate into increased demand for subclover seed.

Innovation frameworks
TRIZ, or the theory of inventive problem solving as it is often referred to in English, is a problemsolving framework developed to support the generation of innovative solutions (Altshuller, 1984). From a study of over 200,000 patents, the TRIZ proponents observed that systems evolve towards 'ideality'. Inventions increase ideality by improving an attribute of the system (Mann, 2001;Dieter and Schmidt, 2013). In its simplest form, TRIZ defines ideality as: ideality = useful effects of the system harmful effects of the e system Innovations and improvements progressively increase the ideality of the system over time. Central to TRIZ is the recognition that the strongest inventions emerge from addressing contradictionstrade-offs where improving one aspect of the system has a negative effect on another. Overcoming contradictions leads to high levels of innovation and paradigm shifts in the technology. The evolutionary trajectory of technology is often represented schematically through use of the S-curve, as depicted in Figure 1 (Christensen, 1992, Sunding andZilberman, 2001). The S-curve comprises a period of slow development in the technology's infancy, followed by a growth phase, which slows at maturity and eventually plateaus in the decline stage. This is not necessarily a permanent decline for the product and performance can continue to increase with innovation and new technology. In this scenario, a new period of growth is produced by new technology improving or replacing the previous system.
The investigation by Coccia (2017Coccia ( , p.1050) into sources of technological innovation has the central hypothesis that 'relevant and consequential problems/needs of consumers induce problem solving activities of firms (by learning processes and acts of insight) that generate incremental and radical innovations in markets, ceteris paribus'. His framework is in line with S-curve theory; an initial technological paradigm or radical innovation occurs in response to a relevant problem or need and subsequent problems/needs result in further incremental innovations over time until there is a new radical innovation that results in a paradigm shift and substantial progress. Here we expand Coccia's hypothesis to the agricultural context where farmers, rather than firms, drive innovation.
However, farmers also act as consumers having to solve problems that affect themselves, in addition to addressing problems of outside consumers. Singh and Fleming (2010) examine whether lone inventors are more or less likely to invent breakthroughs. The Australian subclover seed case study provides an opportunity to examine this question in an on-farm context, where seed producers were often geographically isolated from each other with limited communication between them (until recent advances in communication technology and especially the advent of social media).

Data analysis
We apply these innovation frameworks to analyse technological development in subclover seed production and identify the key innovations that drove the industry forward. This is primarily achieved through analysis of the 48 Australian patents related to subclover seed harvesting technology, which span 1924 to 1992 when the last patent was granted. To classify the technological type represented by each patent, we adapt the standard five-tier TRIZ innovation ratings (Dieter and Schmidt, 2013) into three levels to offer greater clarity: Level 1: Minor corrections made to an existing system by well-known methods.
Level 2: Substantial improvement in an existing system that resolves a basic behaviour compromise by using knowledge from the same technology area; the improvement typically involves adding a component or subsystem.
Level 3: Paradigm shift solutions based on application of a new principle to eliminate basic performance compromises.
Patent data are classified into these innovation levels. Patents are also classified by inventor profession and locations, recorded from information specified in the patent filing. Where these details are not available in patent records, primarily after 1955, inventor information is sought through other literature -mainly Moss et al. (2022) and Smith (2000). Professions are classified as: farmer, engineer (including related trades such as blacksmith), firm (patents filed by corporations) and researcher. Patents are classified by invention type on the basis of method or purpose: thresher or huller, rake, seed cleaner, pneumatic harvester, brush harvester and sheepskin roller.
The cumulative number of patents over time is plotted in frequency charts to display trends in patent activity to analyse technology development and reveal S-curve patterns (Daim et al., 2006, Park et al., 2013 based on inventor occupation (where noted), invention type and TRIZ innovation level. A stacked column chart is used to present the relative proportions of inventor professions and innovation levels across the patent data. Adapting the recent methods of Díaz Lankenau and Winter (2019), who use a flow diagram to represent the contextual factors shaping technological development in tractors in the United States, a flow diagram is used to depict the contextual factors driving the evolution of subclover seed-production technology.

Results and discussion
Farmers have been very powerful sources of innovation and have driven progress in the subclover seed industry. Some 75% of identified subclover patents were patented by farmers ( Figure 2B). With the exception of threshing technology, all of the paradigm-shifting advancements in seed production were the result of farmer innovation. In contrast, over a similar time period (1900-1950s), major developments in tractor technology in the US were dominated by firms rather than farmers; for example, International Harvester and Ford (Díaz Lankenau and Winter, 2019). Figure 2 depicts the cumulative number of Australian subclover production patents by inventor profession, invention type and innovation classification level in panels A, C and D respectively. A full list of Australian subclover seed-production patents, with their innovation classifications and inventor details, is provided in Tables 1 and 2 of the Appendix.

Drivers of innovation
The factors driving innovation in the subclover seed industry meet the Coccia (2017) framework for problem-driven innovation: relevant and consequential problems induce problem-solving activities and generate incremental and radical innovation. However, in the case of technology development for subclover harvesting, it was primarily farmers and not researchers or firms that developed the new solutions. The economic and social context of this industry is important in the progression of innovation. Although driven by a consequential problem, the value proposition of undertaking innovation must be readily apparent to farmers because they are less able than firms to invest time and capital into research activities. The contextual factors driving the evolution of subclover seedproduction technology are depicted in Figure 3.
Innovation occurred from farmers' desire to overcome a problem and recognition of the benefit from doing so. However, these farmers were not full-time inventors and their approach likely differed significantly from that of research entities or firms. Many inventions were the result of over ten years of work and significant personal investment (Moss et al., 2022). As opposed to a systematic research effort to overcome a problem, most innovations stemmed from on-farm observations and experience. For example, in subclover harvesting seed burrs sticking to wool inspired the sheepskin roller method (i.e., wool-covered drums rolled over a paddock to collect seed burrs) and the observation of seed blowing in the wind led to the development of a suction harvester (Moss et al., 2022). These farmer insights, born from years of experience and trial and error, were key drivers of innovation. The innovative nature of the sheepskin roller is noteworthy considering it was first patented in 1933, nearly a decade before Swiss engineer George de Mestral's similar insight about burdock seeds clinging to fabric in 1941, leading to the paradigm-shifting development of the Velcro hook and loop fastening system (Holyoak et al., 1995).
All Level 3 inventions were patented prior to 1934 ( Figure 2). These innovations introduced a new principle to subclover production technology, but were often adapted from other industries. Threshers and brush systems in particular were well known in similar agricultural sectors and it is unsurprising that subclover-specific adaptions were devised early on. Sheepskin and suction harvesters were derived from elements of existing systems, but were applied in much more novel ways that stemmed from farmer experience. Subsequent advancements in technology came from Level 2 inventions that combined different systems to overcome problems and increase performance.
The Horwood Bagshaw (HB) clover harvester has been the most significant of these inventions in the industry. This suction style harvester, manufactured by Horwood Bagshaw Ltd, but based on a farmer's design, remains the most common subclover seed harvester used today (Moss et al., 2021b). Suction harvester designs prior to the HB were limited in their adoption by lack of commercial support. The original patent, filed by farmer Ronald Earnshaw, had little effect on the industry until commercialized by Horwood Bagshaw Ltd. However, commercial manufacturers were hesitant to be involved in the subclover seed industry in this period. Earnshaw's original Details are provided for Level 3 patents and patents related to the Horwood Bagshaw Clover Harvester. No relevant patent activity has occurred since 1992. Figure 3. Progression of subclover seed-production technology and key economic or historical events that influenced innovation design was rejected by firms prior to being developed by Horwood Bagshaw Ltd. Firms believed there was insufficient demand by seed producers to justify the research, development and manufacture of complex harvesting equipment (Avery et al., 2001;Hassall & Associates, 2001: Moss et al., 2022. This lack of commercial support inhibited innovation and highlights the importance of industry support and collaboration with farm innovators in creating widespread benefit, as farmers often lack the resources to commercialize inventions themselves.

Innovation plateau
As shown in Figure 2, there are clear successive S-curve trends in subclover patents with periods of initial high activity decreasing over time before the next upswing. From the 1960s there was a shift in the level of innovation and a significant decrease in patent activity through the subsequent decades. This coincided with the patent and commercial release of the HB (see Figure 2D). Although this harvester represented a paradigm shift in the industry, its release also began a period of reduced innovation. The release of the HB harvester was perfectly timed at the heart of the 'sub and super revolution' (a period of widespread subclover pasture expansion) and it became very popular with seed producers. This period of rapid adoption of subclover by graziers created a need and tangible incentives to improve seed production. However, instead of leading to increased innovation, there was a marked reduction in patent activity in this period. There are several explanations for this decrease. First, seed producers were interested in increasing seed production and a great deal of effort went into improving the HB. Numerous HB modifications, representing incremental improvements, were developed by farmers (Boyle, 1995;Moss et al., 2021a). These modifications were not patented (they lacked the 'novel' step required for a patent) and therefore are not reflected in patent data. The most significant of these modifications was the tandem drive, which allowed multiple HBs to run behind a single tractor, but the majority of modifications were only minor improvements (Moss et al., 2021a).
Second, the HB's patent provided it with protection and restricted other inventions. This conforms with the theory that intellectual monopolies act to stifle innovation by providing too much power to existing patent holders (Boldrin et al.;2008, Chu et al., 2012. The HB would also have acted as a commercial monopoly, being the only suction harvester on the market produced by an established manufacturer. To be commercialized, other inventions would have needed to compete with the incumbent HB, something manufacturers were reluctant to do. Third, complacency is likely another factor in the decline of innovation in the subclover seed industry. While high demand for seed during the 'sub and super revolution' would have provided incentives to innovate, the marginal benefits might not have outweighed the capital costs necessary to develop the new technologies. The HB was effective and subclover seed harvesting could be highly profitable, with anecdotes of farmers paying off their farm mortgages after just a single harvesting season (Moss et al., 2022). Farmers continued to work on new equipment and techniques, but with harvesting activities already so financially lucrative, there was less incentive to invest time, money and effort in developing new innovations. This, in particular, would have reduced the likelihood of paradigm-shifting innovations, which require higher development costs and longer timeframes for farmers to develop.
Innovation stagnated until the 1980s and then declined further as economic conditions deteriorated and subclover seed demand dropped. Farmers still had an interest in improving seed production, but this focused on utilizing modern technology already on farm (e.g., cereal combines) and on guarding trade secrets rather than commercialization (Moss et al., 2021b). Financial incentives were not great enough to support high levels of innovation and consequently there was to be no significant advance in subclover seed-production systems since the 1960s. This contrasts with earlier periods of subclover seed production, where strong economic incentives supported high levels of innovation. However, stagnating conditions inhibited innovation in the last several decades, highlighting the need for collaboration to support farmer invention.

Isolation and collaboration
A continuous feature of technological development in the subclover case study has been the isolation of farm inventors, which both encouraged and impeded innovation. The vast majority of patents were filed by a single inventor (Figure 4). Single inventor patents are well represented in paradigm-shifting Level 3 innovations as well as in lower level incremental improvements, and there is no evidence that multiple-inventor patents were more likely to result in a technological breakthrough. This illustrates that lone inventors can be powerful sources of agricultural innovation. The predominance of single inventors may have been out of circumstance rather than choice. In 75% of patents with multiple inventors, all listed inventors on the patent had the same surname, suggesting they were related (Moss et al., 2022). If this is the case, then <5% of non-firm patents were the result of collaboration among non-family inventors. This suggests significant barriers to collaboration exist for farmers, probably resulting from geographical isolation, lack of communication technology and a highly competitive industry that disincentivizes farmers from sharing ideas.
There does not appear to be notable collaboration among inventors of subclover seed harvesting techniques, even in similar locations. The location and year of each identified patent is shown in Figure 5. While groupings of patents appear in similar locations and timeframes, we found no evidence to indicate this was the result of inventors working together or sharing ideas. Instead, these clusters of activity are likely to be attributable to high subclover seed production in these regions, creating conditions favourable to invention and patent protection. For example, Western Australian farmers Earnshaw and Moore both patented suction harvest systems in 1949. Despite their properties being located <100 km from each other (Appendix, Table 1, Bokal and Benjinup) and working on similar systems for years in close proximity, they did not have knowledge of each other's inventions and developed their harvesters independently (Moss et al., 2022). At this time, there were limited means for farmers to share information and there would also have been reluctance to do so in order to obtain a competitive advantage. These factors resulted in reduced knowledge transfer in the industry.
Success in the subclover seed harvesting industry necessitated a high level of inventiveness from farmers. However, there has also been a clear duplication of effort, with farmers devoting time and resources to overcoming the same problem, without sharing knowledge gained or lessons learned. Limited communication and reduced knowledge sharing created barriers to commercializing useful inventions, incentivizing farmers to focus on developing and retaining a competitive advantage rather than a commercial product. So, while isolated farmers may have been driven to invent, the lack of communication and collaboration has had negative consequences for the overall seed-production industry. Singh and Fleming (2010) hypothesize that diversity of experience across team members increases inventive outcomes. As a result of their isolation, subclover seed harvesting inventors have utilized their own diversity of experience as skilled fabricators, mechanics and trouble shooters, in addition to being farmers. There is a business imperative for farmers to be knowledgeable about their machinery repair, fabrication and servicing (especially in isolated and sparsely populated rural locations of Australia), as elucidated by recent 'right to repair' arguments (Fitzgerald, 2008;Shah, 2018;Sirois-Leclerc, 2018). This encourages farmers to tinker with equipment and is also a driver of innovation (Hatta, 2020). Overall, this diversity of experience and the mechanical aptitude common among farmers has had a positive effect on inventive outcomes.
The success of the subclover seed-production inventors occurred prior to modern advances in communication technology. Singh and Fleming (2010) acknowledge that heroic lone inventors may have been common in the 19th and early 20th centuries, but changes in technology and organization have increased the advantages of collaboration. Social media in particular have increased farmers' ability to communicate and share knowledge. While it is clear that some level of isolation is beneficial for driving on-farm innovation, collaboration and shared knowledge are also important factors for maximizing the benefits of innovation across an industry. Particularly in remote agricultural settings and non-industrialized regions, communication and collaboration are important in supporting commercialization and wide adoption of farmer inventions.

Commercialization
As inventions became more complex in the subclover seed industry, it became increasingly important to have commercial support for the innovation to have widespread positive impact. Lack of commercial support inhibited innovation and contributed to the decline in the industry. Therefore, collaboration with industry, government and research is a key component to support innovation in agriculture. Sadras et al. (2020) advocate multidisciplinary research to improve return on agricultural R&D investment. Multidisciplinary approaches also increase the ability to translate scientific discoveries into realized productivity improvements (Passioura, 2020). This 'translational research' approach is a core TRIZ principle for achieving high levels of innovation (Altshuller, 1984). Although collaboration between research and industry can support high levels of innovation, on-farm stakeholders must still be a key part of this process to maximize benefits (McKenzie, 2013). The Rhoades and Booth (1982) 'farmer-back-to-farmer' model for generating agricultural technology stresses that applied research must begin and end with the farmer. This includes farmer participation in setting priorities (Sadras et al., 2020). The subclover case study Figure 5. Location of each identified subclover seed-production patent Note: The year an invention was patented is indicated by the colour scale. There are distinct groupings of patents, notably in the Adelaide region and southern Western Australia. All inventors were located in either seed-producing regions or capital cities. The area of contemporary subclover pastures is shaded (adapted from Donald, 2012). The land used to grow subclover as a seed crop is significantly smaller. The pasture area reflects current figures and would have been smaller at the time inventions were patented. has shown just how innovative farmers can be, and why farmers should be utilized in generating agricultural innovation.
Farmers cannot afford to conduct large research projects themselves, but frameworks exist to involve farmers directly in funding R&D. Australia, Canada, the US and the UK have created R&D organizations funded by levies on agricultural output and controlled, at least partially, by farmer representatives (Sadras et al., 2020). These organizations collectively invest hundreds of millions in R&D annually. In Australia, the Grains Research and Development Corporation (GRDC) is the largest of these organizations, with annual revenue of A$200 million (Sadras et al., 2020). The high farmer involvement in these organizations and collaboration with research and industry allow them to create and leverage new innovations. These rural development and research corporations were key in supporting innovative farmers in developing conservation agriculture systems in Australia (Bellotti and Rochecouste, 2014). Recent examples of GRDC-funded innovations are the Harrington seed destructor and the weed chipper. The Harrington seed destructor (Walsh et al., 2012), which devitalizes weed seeds that exit grain harvesters in the chaff fraction, has had a significant impact on global grain production. The evolved technology is now commercially available in an integrated version compatible with many combine harvester makes and models and is demonstrating weed control success internationally. The weed chipper is the first fallow mechanical weeder compatible with large-scale cropping systems (Walsh et al., 2020). The success of the weed chipper is attributed to collaboration with stakeholders in farming, research (agronomy and engineering), manufacturing and industry.
Farmer-funded organizations like these could be valuable in accelerating agricultural innovation in small industries and developing countries. There are relevant and significant agricultural problems that need to be addressed, but solutions may not have sufficient market size to attract investment from firms. Farmer-funded organizations can fill the gap left by firms, and align those who pay for R&D with those who benefit from it. However, collecting agricultural tax or levies in developing countries can be challenging because of the absence of standard account-keeping and are often effective only in large-holdings (Rajaraman, 2004). While a levy approach may not work in all contexts, systems are needed to support on-farm innovation and enable the commercialization or dissemination of useful inventions.
It is likely that farmers generally have less motivation to commercialize their ideas than firms. In previous decades, the high price of subclover seed provided financial incentive to innovate, but there are several instances of subclover harvesting inventions, both patented and unpatented, where commercialization was not attempted (Moss et al., 2022). The prospect of developing techniques to provide a competitive advantage in seed production would have been sufficient in some cases, particularly in a highly competitive and protected industry (Avery et al., 2001;Moss et al., 2021b). Farmers tend to be less likely than firms to document or patent their inventions (Rhoades, 1989) and can be reluctant to bear the cost of patent protection, relying instead on secrecy to prevent others from seeing and copying their ideas. Many unpatented and unadvertised subclover seed harvester inventions developed from the desire of farmers to make their own jobs easier (and the urge to tinker and improve practices/equipment) rather than to pursue commercialization. A drawback of this approach is that the benefits of the innovation can become restricted to its inventor or region. Frameworks are needed to support farmers to share or commercialize useful inventions.
Promoting knowledge sharing among farmers can be challenging, as farmers are competitors in the same industry. However, collaboration does not have to rely on altruism. Knowledge sharing can result in net positive benefits for the industry, which will benefit individual farmers as members of that industry. Innovation is likely to proceed more efficiently when networks of informal know-how sharing are encouraged. Networks of knowledge sharing among competitors have enabled important discoveries in industries from steam engines to semiconductors (Pedraza-Fariña, 2016). These networks should be supported; shared knowledge can still be protected. Research collaboration is becoming increasingly common among firms, universities and institutes, but is navigated with intellectual property protection (Slowinski et al., 2006). Governments and research and development corporations can provide financial and logistical support to farmers to protect or commercialize their inventions. This will secure farmers' intellectual property, allowing valuable innovations to have widespread benefit.

Lessons and recommendations to support future agricultural innovation
The subclover seed-production industry offers a number of lessons for agricultural innovation in the future. The high proportion of farmer invention reflects the innovativeness of farmers in this industry, but also highlights the reluctance of larger firms to invest unless there is significant market size. These lessons are particularly applicable to small-scale farms and the developing world. Constraints in the developing world often lead to radically different solutions from those found in resource-rich countries (Mattson and Wood, 2014;Mattson and Winter, 2016). Farms in the US, where a significant amount of R&D takes place, are 100 times larger than 80% of all farms globally (Díaz Lankenau and Winter, 2019) and Australian farms are significantly larger than US farms (Australian Bureau of Statistics, 2016; USDA, 2021). However, half the world's food is produced by 1.5 billion small-scale farmers and in nonindustrialized nations, 80% of food is produced by small-scale farmers (Bragdon and Smith, 2015). Significant positive impacts in global agriculture can be realized if innovation from these farmers is supported and focused.
To maximize farmer innovation, the drivers must be understood. There must be a relevant problem and perceived benefit in overcoming it. This aligns with current economic and management technology theory (Coccia, 2017), but the approach of farmers is different from that of firms. Favourable economic conditions will encourage farmers to invest in innovation. However, the costs and time of developing an invention can be an obstacle to farmers, which can be overcome with research and industry collaboration. The absence of this support can inhibit farmer innovation, particularly in small markets.
Farmers are also less likely than firms to develop inventions for commercialization, instead aiming to improve their own production and competitive advantage. Industry engagement is crucial to increase commercialization of useful inventions and enable maximum benefit. However, farmers can be hesitant to share ideas and knowledge, and keep inventions as closely guarded trade secrets. Knowledge sharing amongst farmers, industry and research can contribute positively to innovation. Therefore, frameworks and incentives are needed to encourage farmers to communicate and collaborate. External support for intellectual property protection can encourage farmers to share or commercialize inventions while still maintaining ownership.
Farmer-funded research and development corporations are one avenue of support for farmers to enable innovation and reduce the amount of useful knowledge confined to an individual or a region. While isolation was a key driver of innovation in subclover seed harvesting, collaboration will be key to creating new advances in a more complex and connected future. Harnessing the creative power of farmers has the potential to accelerate agricultural innovation and contribute to solving the considerable challenges facing global agriculture.