The importance of large trees as carbon sinks in UK woodlands

Technical Summary: The Importance of Large Trees as Carbon Sinks in UK Woodlands

Problem Statement
Accurately mapping aboveground carbon (AGC) in temperate forests is critical for national greenhouse gas inventories and achieving net-zero targets. However, conventional methods for estimating AGC rely on size-to-mass allometric equations derived from destructive harvests of smaller trees. These datasets are often skewed toward smaller cohorts due to logistical and ethical constraints, rendering the resulting equations "not scale invariant." Consequently, applying these standard allometries to larger trees—particularly those in ancient or semi-natural woodlands—likely leads to significant underestimation of the carbon pool. The authors hypothesize that the contribution of large trees to the total AGC in UK woodlands is currently underestimated by existing inventory methods.

Methodology
The study was conducted at Kew Wakehurst (West Sussex, UK), a 211-hectare estate representing a mosaic of southern UK woodland types. The researchers selected three distinct habitat types for analysis:

1. Broadleaf Woodland: A Site of Special Scientific Interest (SSSI) dominated by Quercus robur, Fagus sylvatica, and Fraxinus excelsior.
2. Coppice Woodland: Actively managed on a seven-year rotation, primarily consisting of Corylus avellana stools with an Quercus robur overstorey.
3. Conifer Plantation: A mid-20th-century monoculture of Picea abies (Norway Spruce) on an Ancient Woodland Site (PAWS).

Data Acquisition and Processing:

• Terrestrial LiDAR (TLS): Data were captured using a RIEGL VZ-400i scanner. Scans were co-registered and processed into a British National Grid coordinate system.
• Segmentation and Modeling: Individual trees were isolated using the TLS2trees pipeline. Woody volume was estimated using TreeQSM (Quantitative Structure Models), which encloses point cloud data with geometric primitives (cylinders).
• Mass Calculation: Volume was converted to AGC using species-specific wood densities from the Global Wood Density Database and biome-specific biomass-to-carbon ratios. Uncertainty was propagated through volume, density, and conversion factors.
• Comparative Analysis: The TLS-derived AGC was compared against two conventional approaches:
  1. Bunce "Combined" Equation: A widely used allometric model for broadleaf trees.
  2. Woodland Carbon Code (WCC): The UK's standard method for carbon accounting, which utilizes specific allometric equations for broadleaf and conifer species.
• Inventory Validation: A subset of trees was manually measured (DBH and height) to validate TLS-derived metrics and serve as inputs for the comparative allometric models.

Key Results

• Disproportionate Carbon Storage: Large trees act as a disproportionate sink for AGC. In broadleaf and coppice woodlands, the largest 3–4% of trees (by DBH) account for approximately 50% of the total habitat AGC. Even in the conifer plantation, ~50% of AGC is stored in the largest ~25% of trees.
• Systematic Underestimation by Allometry:
  • Broadleaf & Coppice: The Bunce allometric equation underestimated AGC by 50–60% compared to TLS. The WCC method performed even worse, underestimating AGC by approximately 70% in ancient broadleaf woodlands and 50% in coppice woodlands.
  • Conifer: For Picea abies, TLS and Bunce allometry showed good agreement, though the WCC still underestimated AGC by 10–20%.
• Source of Bias: The primary driver of underestimation in existing allometries is the failure to capture the increased allocation of woody material to supporting large, spreading crowns. As tree size increases (DBH ≥ 30 cm), the ratio of stem to crown carbon shifts significantly (from 45:55 to 29:71), a morphological shift not captured in equations trained on smaller trees.
• Uncertainty: While individual tree uncertainty is high (mean ~41% for broadleaf), it reduces to ~27% when focusing on the largest trees that comprise 90% of the total AGC. For these large trees, uncertainty is dominated by wood density variability rather than volume estimation errors.

Key Contributions

1. Empirical Evidence of Bias: The study provides quantitative evidence that standard UK carbon accounting methods (specifically the WCC and Bunce equations) significantly underestimate AGC in woodlands containing large, mature trees.
2. Morphological Insight: The research identifies the shift in carbon allocation from stem to crown as a function of tree size as the biological mechanism behind the failure of scale-invariant allometries.
3. Methodological Validation: The paper demonstrates that terrestrial LiDAR, coupled with QSM, can accurately estimate AGC across diverse tree sizes and woodland structures without destructive harvesting.

Significance and Claims
The authors assert that the concentration of AGC in a few large trees implies a "gross underestimation" of reported national-level carbon stocks in the UK if current methods are applied broadly. They emphasize that existing allometric methods are "fit for purpose" for the specific woodland types and tree sizes they were designed for (e.g., managed plantations with smaller trees). However, they caution that applying these methods to "out-of-sample" woodlands—specifically those with large, ancient, or semi-natural trees—leads to systematic errors.

The paper concludes that new size-to-mass allometries must be developed using appropriately sized trees. Given the ethical and logistical impossibility of destructive harvesting for large trees, the authors propose a hybrid approach: using terrestrial LiDAR to estimate the "mass" component of size-to-mass equations, which can then be paired with conventional inventory measurements (like DBH) for scalable national reporting. Furthermore, they argue that woodland management and policy should prioritize the retention of large trees and the recruitment of future large individuals to maintain these habitats as significant carbon sinks.