Evaluating cane pruning decision criteria and the identification of grapevine pruning styles

Abstract

Winter pruning is the highest yearly expenditure in the typical New Zealand vineyard budget, yet few attempts have been made to bring quantitative measurement tools into its management. The research presented here constitutes first steps towards this end, in tandem with University of Canterbury researchers working towards an artificially intelligent pruning robot. In pursuit of information regarding cane pruning preferences and decision-making criteria, a two-part survey was conducted in the regions of Marlborough, Hawke’s Bay, Waipara and Central Otago. Part One of this survey asked participants to rate a set of already-made cane pruning decisions for one (cv.) Sauvignon Blanc vine. Participants rated these decisions on 24 individual pruning criteria and also provided two overall assessments. One of these overall assessments was recorded before participants rated the decisions on the 24 individual criteria, and a second overall assessment was recorded after such time. All ratings were collected via Qualtrics software, either online or via the Qualtrics offline survey application. Part Two of the survey asked participants to indicate, with highlighter pens on paper, their own preferred pruning decisions for the same vine. Linear Models, based on the relationship between the individual criteria and overall assessments (Part One), have revealed spur and cane position to be the dominant influencing factors in the pruning of the subject vine. Participant first impressions, as measured by the first overall assessment (before the individual criteria ratings), were almost exclusively reflective of participant attitude towards spur and cane position. The dominance of position was corroborated by Correspondence Analysis of the preferred pruning decisions (Part Two), which suggested that the decision to modify the vine’s cane or spur position was a fundamental point of divide within participant responses. In a related finding, results from Principal Component Analysis (Part One) have suggested that overall impressions were a heavy influence throughout the course of participant responses to Part One of this study. By extension, this finding suggested that attitudes towards position, which were strongly linked to participant overall assessments, permeated into participant attitudes towards other aspects of the presented decision set (Part One). Generally speaking, the dominance of a single group of decision-making criteria calls for further investigation into how pruning is conceptualised as a task. Results from this study suggest that there exists a somewhat broad, non-specific, view of whether or not a particular set of spur and cane selections are acceptable. This finding, while perhaps not immediately impactful for practitioners, has considerable implication for the design of future pruning research, as well as for the evaluation of artificially intelligent pruning. This research also reports the detection of pruning preference (Part One and Part Two) groups, based on region and organisational role. Correspondence Analysis and Multiple Correspondence Analysis (Part Two) have revealed that participants from Hawke’s Bay and, particularly, Central Otago tended towards a decision to restructure the subject vine by not leaving a spur from its existing right half. This was in contrast to those participants from Marlborough and Waipara who tended towards a maintaining of the current vine configuration. Aside from these differing propensities to restructure the vine, several regions were associated with unique spur and cane selections. It is unclear at present whether regional differences are due to social influences, regional viticulture conditions, cultivar familiarity, or some unknown factor. Participants also differed in their preferences when grouped based upon their organisational role. Those participants identifying exclusively as labourers were considerably less likely to restructure the vine, compared to those participants identifying as supervisors, managers, or proprietors. Managerial implications of this finding are discussed, with several potential remedies explored.