On-farm experimentation in the Australian winegrape sector : approaches and opportunities for change

Farmers worldwide conduct experiments on their own farms to test new practices or ideas. Their on-farm experimentation (OFE) provides information that supports operational decisions and generates knowledge about their farming systems. Grapegrowers in Australia also use OFE to evaluate alternative practices against various business goals. While some researchers perceive that growers' trials lack rigour associated with scientific methods, little has been reported about why and how growers conduct their own trials. Therefore, the main objectives of this research were to investigate trial behaviours of grapegrowers and viticultural consultants in Australia and to explore the need to facilitate the adoption of new trial approaches.

The first part of the thesis reviews farmers' experimentation across agricultural industries in different counties and existing alternative approaches to OFE. The review provides insights on why and how farmers conduct trials, the outcomes of their OFE and factors influencing their trial behaviours. Also, some farmers were found to be interested in changing trial methods to obtain more reliable results for confident decision-making. Scientific methods based on small-plots and analysis of variance are unlikely to suit most farmers due to resource requirements and logistical issues related to trial work. Such methods also have limitations in generating meaningful information for operational decisions at the 'landscape scales' at which farmers make decisions. Spatially distributed experimentation offers a solution by taking advantage of precision agriculture technologies (PA) such as remote sensing and geographic information system (GIS). However, spatial approaches typically require large amounts of measurements and/or GIS skills which might make them unsuitable to farmers who do not have access to sensors or GIS skills.

We then investigated why and how grapegrowers and consultants in Australia conduct experiments and the need for them to change trial approaches. Semi-structured interviews with 35 growers and 8 consultants across Australia revealed that they mainly trial to learn about alternative practices, to gain knowledge and to enhance confidence in management decisions. Trial approaches are diverse and not associated with business or personal attributes. Growers value conclusive results for confident decision-making; however, they face difficulties in OFE, including time, labour constraints, a lack of efficient and objective measures of crop responses, and spatial variability that confounds trial results. While growers value their experimentation, they indicated a desire for robust results. This desire and challenges associated with OFE suggest a need for more efficient trial approaches that can generate more useful information to support growers' decisions.

Given the need for new approaches, we explored a simple strip approach to OFE whereby a single crop row in a vineyard block is positioned according to a useful covariate to a given response variable to infer likely treatment effects across the entire block. This approach uses moving window analysis to reveal spatial variation in treatment effects and the covariate along the trial row. This analysis can be performed with the aid of a spreadsheet and without the need to use GIS software. Simulated strip trials using data from previous experiments demonstrated that a strip in which the range of variation in a useful covariate is close to that encountered in the whole block can reveal likely crop responses across the entire block, thus contributing to informed decision-making.

We next evaluated the strip trial approach through four case studies with grapegrowers and consultants in Australia in a participatory manner. In each case, the strip approach was applied to the grower's own trial by collaborating with the grower and consultant. Participatory experimentation was effective for them to assess the strip approach according to their needs and situations. The researcher also gained deep insights on the complexity of vineyard experimentation and issues of applying spatial approaches to OFE within a commercial operation. The participatory approach to research, though time-demanding, was valued highly by all participants. It enabled rigorous and relevant trial results and enhanced the connection between the researcher and the farming businesses. Therefore, more genuine collaboration among researchers, farmers and consultants is recommended to produce valuable outcomes.

Through these case studies, we also explored the perceptions of participating growers and consultants about the strip approach. The theory of planned behaviour was used as an analytical framework to identify factors influencing their intention to adopt the approach. The findings revealed that growers and consultants perceived several advantages of the strip approach over their own methods. Key barriers to uptake include resource constraints for data collection, a lack of GIS skills and knowledge to apply and analyse spatial data for OFE, and a lack of capacity for efficiently applying results to differential management of vines in a block. A role for agricultural consultants in assisting growers with OFE using spatial approaches was identified; however, major obstacles for consultants to adopt the approaches are a lack of GIS skills and 'consultant inertia' regarding changing their trial behaviours.

These findings highlight a need for support that can facilitate the adoption of spatial approaches by growers and consultants. This thesis highlights the importance of vineyard experimentation to wine businesses while providing insights on the trial behaviours of grapegrowers and consultants in Australia and the challenges they face during experimentation. While robust results are desired, time and labour availability limits their capacity to collect sufficient data. Spatial variability in the vineyard also confounds the interpretation of trial results. Spatial approaches to OFE can account for the impact of variability on crop responses and generate meaningful information to support informed decision-making. These advantages were perceived by growers and consultants; however, the adoption of spatial approaches will likely need easy access to reliable spatial data and training of growers and service providers in how to apply and analyse spatial data in an OFE context. Efficient implementation of spatial approaches also needs automated measurements of different viticultural variables, tools for easy data analyses and efficient application of trial results for differential management of crops in a block.