Water stress assessment on grapevines by using classification and regression trees

Area of Science:

• Viticulture and Enology
• Plant Physiology
• Agricultural Science

Background:

• Vineyard environment and winemaking significantly impact grape development and composition.
• Water stress can enhance phenol synthesis and grape quality, provided it doesn't limit yield.
• Identifying key management parameters for water stress and grape quality is crucial.

Purpose of the Study:

• To identify critical parameters for managing water stress and optimizing grape quality in viticulture.
• To utilize multivariate statistical analysis for predicting vineyard vulnerability and wine quality.

Main Methods:

• Employed classification and regression trees (CART) for multivariate statistical analysis.
• Models were built by recursively partitioning data space, represented as decision trees.
• Identified decisive variables for predicting vineyard vulnerability and wine quality parameters linked to water stress.

Main Results:

• In Priorat AOC, Carignan grapevines showed high water potential and abscisic acid concentration early on, enabling mesoclimate classification.
• Leaf water potential (LWP) and stem water potential (Ts) at the pea-size stage differentiated warm and cold vineyard areas.
• Decision tree models effectively predicted vulnerable vineyards and quality parameters associated with water stress.

Conclusions:

• Multivariate statistical methods, particularly decision trees, are effective for analyzing complex viticultural data.
• Leaf water potential and temperature during early growth stages are key indicators for vineyard management and climate differentiation.
• This approach aids in identifying vineyards susceptible to water stress and managing for optimal wine quality.