TL;DR
This study reveals that xylem hydraulic vulnerability, specifically the point at which a tree loses 50% of its water transport capacity (P50), is the primary driver of year-to-year fluctuations in both earlywood and latewood growth across the Northern Hemisphere. These findings are crucial for improving our understanding of how forests will respond to increasing climate variability and for refining models that predict future carbon sequestration.
What Problem Does This Paper Address?
Forests play a critical role in the global carbon cycle, but their ability to sequester carbon can fluctuate significantly from year to year due to climate variations. Understanding these interannual growth variations (IGV) is essential for predicting future carbon storage. However, it's been unclear how different parts of a tree's wood – earlywood (formed in spring) and latewood (formed in summer) – vary in their growth responses across different regions and what factors drive these differences, especially in the face of amplified climate extremes like droughts. This paper aims to fill this knowledge gap by investigating the spatial and temporal patterns of earlywood and latewood IGV and identifying the key environmental and plant-trait drivers.
Methods
The researchers employed a multi-faceted approach:
- Tree-Ring Data Analysis: They analyzed 596 tree-ring chronologies from across the Northern Hemisphere, spanning from 1901 to 2013. They calculated the interannual variability of earlywood (EWcv) and latewood (LWcv) using the coefficient of variation (CV) for each site.
- Climate Data Integration: Monthly climate data (temperature, precipitation, potential evapotranspiration, vapor pressure, solar radiation) were collected and processed to derive variables like climatic water deficit (CWD) and vapor pressure deficit (VPD). These were analyzed for different growth periods (earlywood, latewood, and the preceding non-growing season).
- Plant Trait and Stand Property Data: They incorporated data on key plant functional traits (xylem water potential at 50% loss of hydraulic conductivity - P50, canopy height, maximum rooting depth) and stand properties (tree density, available water storage capacity, cation-exchange capacity, ecological memory).
- Boosted Regression Tree (BRT) Models: This non-parametric machine learning technique was used to identify and quantify the relative importance of various predictors (climate, hydraulic traits, stand properties, foliar traits) in explaining EWcv and LWcv. BRT models are robust to non-linear relationships and interactions between variables.
- Temporal Analysis: To assess how the influence of these factors has changed over time, the study period was divided into two sub-periods (1901-1945 and 1946-2013), and BRT models were re-run for each.
Datasets
- Tree-Ring Chronologies: 596 paired earlywood and latewood width chronologies from the International Tree-Ring Data Bank (ITRDB). These covered 50 tree species across 6 main genera and were collected from sites between 15°N and 75°N latitude.
- Climate Data: Gridded monthly data from CRU TS v. 4.04 (temperature, precipitation, potential evapotranspiration) and CRUNCEP (solar radiation) for 1901-2013.
- Plant Functional Traits: Data for P50, canopy height, and maximum rooting depth were sourced from global datasets (Choat et al., 2012; Simard et al., 2011; Fan et al., 2017).
- Soil Properties: Harmonized World Soil Database v.1.2 for cation-exchange capacity (CEC) and available water-storage capacity (AWC).
- Tree Density: Global map of tree density (Crowther et al., 2015).
- Foliar Traits: Data for foliar nitrogen concentration (Nm), foliar phosphorus concentration (Pm), and specific leaf area (SLA) from the TRY database.
Key Insights and Ideas
- Aridity Drives Variability: The study strongly supports the hypothesis that drier regions exhibit higher interannual variability in both earlywood and latewood growth. This makes sense, as trees in arid environments are more directly exposed to water stress and its year-to-year fluctuations.
- Earlywood vs. Latewood Divergence: While both are more variable in dry regions, earlywood and latewood show different spatial patterns and sensitivities. Earlywood seems more influenced by long-term aridity, while latewood appears more susceptible to within-season water availability, suggesting different temporal windows of vulnerability.
- Hydraulic Vulnerability is King: The most striking finding is the dominant role of xylem hydraulic vulnerability (P50) in mediating growth variability. This highlights that a tree's ability to withstand water stress and avoid embolism is a more critical factor for year-to-year growth fluctuations than many climate or stand properties.
- Increasing Importance of P50 Over Time: The study reveals that the influence of P50 on growth variability has increased over time, particularly in arid regions. This suggests that as climate change intensifies water stress, a tree's intrinsic hydraulic safety margin is becoming even more critical for its survival and growth consistency.
Key Results
- Spatial Patterns: High EWcv and LWcv were predominantly found in arid regions, especially the southwestern United States, while lower variability was observed in boreal regions. Eurasia showed a more west-east gradient for EWcv.
- Aridity Gradient: EWcv showed a significant decreasing trend with increasing aridity index (AI), indicating less variability in wetter regions. LWcv also decreased with increasing aridity, but with more fluctuations in wetter areas.
- Top Predictors: P50, rooting depth, tree density, CEC, CWDPN_CV (climatic water deficit in the preceding non-growing season), and TMPLP_CV (temperature variability in the latewood growth period) were identified as the most important factors influencing growth variability.
- P50 Dominance: P50's median relative contribution to LWcv was 21.1% in arid regions and 18.2% in humid regions, significantly higher than its contribution to EWcv (9.2% and 8.5%, respectively).
- Temporal Shift: The relative contribution of P50 to EWcv in arid regions significantly increased by 4.1% from the 1901-1945 period to the 1946-2013 period. Similarly, its contribution to LWcv in humid regions increased by 0.5%.
Limitations and Future Work
The authors acknowledge several limitations:
- Data Availability for Hydraulic Traits: Global datasets for hydraulic traits like P50 are still limited, especially across diverse species and spatial scales, which can introduce uncertainties in model validation.
- Spatial Mismatch: There's a potential mismatch between the fine-scale tree-ring data and the coarser grid-scale measurements of plant traits and climate.
- Unexplained Variance: The BRT models, while powerful, leave some variance unexplained. This "unexplained variance" could be attributed to factors not included in the analysis, such as pests, pathogens, or non-structural carbohydrate dynamics, which are crucial for understanding forest stability and carbon feedback.
Future work should focus on:
- Species-Specific Analyses: More detailed, species-specific studies are needed to fully understand how hydraulic traits influence growth variability across different functional groups (e.g., angiosperms vs. gymnosperms).
- Expanded Data Networks: Expanding field observation networks and utilizing higher-resolution remote sensing products will be crucial for better understanding tree adaptations and improving carbon sequestration predictions.
- Integrating Other Factors: Future models should aim to incorporate the effects of pests, pathogens, and non-structural carbohydrates to provide a more complete picture of forest dynamics.
Summary updated