Determination of climatic predictors influencing seed production in seed orchards of Korean red pine based on different regression models

Abstract Pinus densiflora is a pine species native to the Korean peninsula, and seed orchards have supplied material needed for afforestation in South Korea. Climate variables affecting seed production have not been identified. The purpose of this study was to determine climate variables that influence annual seed production of two seed orchards using multiple linear regression (MLR), elastic net regression (ENR) and partial least square regression (PLSR) models. The PLSR model included 12 climatic variables from 2003 to 2020 and explained 74.3% of the total variation in seed production. It showed better predictive performance ( R 2 = 0.662) than the EN (0.516) and the MLR (0.366) models. Among the 12 climatic variables, July temperature two years prior to seed production and July precipitation after one year had the strongest influence on seed production. The time periods indicated by the two variables corresponded to pollen cone initiation and female gametophyte development. The results will be helpful for developing seed collection plans, selecting new orchard sites with favorable climatic conditions, and investigating the relationships between seed production and climatic factors in related pine species.