Nevin Manimala Statistics

Upscaling xylem phenology: Sample size matters

Ann Bot. 2022 Aug 26:mcac110. doi: 10.1093/aob/mcac110. Online ahead of print.


BACKGROUND AND AIMS: Upscaling carbon allocation requires knowledge of the variability at the scales at which data are collected and applied. Trees exhibit different growth rates and timings of wood formation. However, the factors explaining these differences remain undetermined, making samplings and estimations of the growth dynamics a complicated task, habitually based on technical rather than statistical reasons. This study explored the variability in xylem phenology among 159 balsam firs (Abies balsamea (L.) Mill.).

METHODS: Wood microcores were collected weekly from April to October 2018 in a natural stand in Quebec, Canada, to detect cambial activity and wood formation timings. We tested spatial autocorrelation, tree size, and cell production rates as explanatory variables of xylem phenology. We assessed sample size and margin of error for wood phenology assessment at different confidence levels.

KEY RESULTS: Xylem formation lasted between 40 and 110 days, producing between 12 and 93 cells. No effect of spatial proximity or size of individuals was detected on the timings of xylem phenology. Trees with larger cell production rates showed a longer growing season, starting xylem differentiation earlier and ending later. A sample size of 23 trees produced estimates of xylem phenology at a confidence level of 95% with a margin of error of one week.

CONCLUSIONS: This study highlighted the high variability in the timings of wood formation among trees within an area of 1 km 2. The correlation between the number of new xylem cells and the growing season length suggests a close connection between the processes of wood formation and carbon sequestration. However, the causes of the observed differences in xylem phenology remain partially unresolved. We point out the need to carefully consider sample size while assessing xylem phenology to explore the reasons underlying this variability and to allow reliable upscaling of carbon allocation in forests.

PMID:36018569 | DOI:10.1093/aob/mcac110

By Nevin Manimala

Portfolio Website for Nevin Manimala