Long term monitoring of red needle cast. What drives episodic outbreaks on radiata pine in New Zealand?

Abstract

Background: Red needle cast (RNC) is a foliar disease of radiata pine (Pinus radiata D.Don) in New Zealand caused by Phytophthora pluvialis Reeser, W.L.Sutton & E.M.Hansen and, to a lesser extent, Phytophthora kernoviae Brasier, Beales & S.A.Kirk. Incidence and severity of RNC vary substantially between years. To investigate the impact of seasonal weather variables on this variation, RNC was assessed annually for ten years at radiata pine transects.

Methods: Fifty-three transects were established in 2015 in the Central North Island and Gisborne Region (east coast North Island) of New Zealand, with twenty-three monitored until 2024 (surviving harvest). The relationship between seasonal weather variables and RNC severity was analysed using two non-parametric statistical approaches: (1) correlation analyses (Spearman correlations, r_s, where positive values indicate an increase in RNC severity with an increase in the explanatory variable); and (2) binary recursive partitioning (with models trained on 85% of observations and tested on the remaining 15%).

Results: Disease expressed more consistently, and severity was generally greater, at Gisborne sites. Disease severity peaked in 2017 and 2023 in both regions. Autumn (March-May) variables tended to be prevalent amongst predictors of RNC severity. Autumn soil moisture index (calculated from cumulative rainfall and evapotranspiration) was the most strongly correlated variable for the Gisborne dataset (r_s = 0.70) and, along with vapour pressure, were the key partitioning variables in the recursive partitioning model. The strongest correlating variable for the Central North Island dataset was autumn potential evapotranspiration (r_s = -0.46) while the most important variable and first data partition was autumn vapour pressure. Model evaluation metrics indicated good performance: R² values were 0.63 and 0.68 for the Gisborne and Central North Island test datasets respectively, and mean absolute errors were 18.1 % and 7.8 % for the respective datasets.

Conclusions: The importance of autumn more than summer weather variables in determining disease expression differs from the findings of previous studies and indicates that conditions during periods of exponential epidemic growth may be as, or more, important than initial inoculum level in determining RNC severity. Proactive control activities may require long-term weather forecasting or frequent monitoring during this season.