Drops in needle production are early-warning signals of drought-triggered dieback in Scots pine

[Key message] Reconstruction of needle dynamics reveals prolonged drought legacy effects on crown condition that represent early warnings of drought-induced dieback in Scots pine.

[Abstract] Understanding the mechanisms of drought-induced forest dieback and tree mortality is a priority for predicting forest responses to climate change. However, long-term information on drought impacts on crown dynamics and how it relates to tree water and carbon economy is still lacking. Comparing declining and non-declining Scots pines at the same site, we quantified primary and secondary growth and intrinsic water-use efficiency (WUEi). Further, we reconstructed the needle production and the annual total number of needles using the Needle Trace Method. Here, we reconstructed the apical needles through the assessment of 2460 longitudinal sections. We fitted partial least squares regression models to assess climate influence on primary and secondary growth and needle dynamics, and then linear mixed models using climate covariates and discerning between health status in the 1975–2016 period. Finally, we analyzed drought legacy effects applying a Bayesian hierarchical framework. We detected the highest differences between health tree classes in the annual total number of needles during the warm-dry 1980s. Declining trees responded more negatively to climate than non-declining conspecifics and showed a higher variability of drought legacy effects. We found significant drought legacy effects in the annual total number of needles up to 3 years after a drought. The warm, dry 1980s, that preceded the dieback, showed the highest δ13C values since the 1970s. Declining trees showed higher WUEi than non-declining trees from the 2000s onwards. Our study sheds light on the major role of needle dynamics in dieback episodes and illustrates how past drops in needle production may be interpreted as early warnings of drought-induced dieback in Scots pine.

Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. GS-B was supported by a Postdoctoral grant (IJC2019- 040571-I) funded by MCIN/AEI/10.13039/501100011033. This work was supported by project PROWARM (PID2020-118444GA-100) funded by MCIN/AEI/10.13039/501100011033, and the Instituto de Estudios Turolenses (IET).

Peer reviewed