National-scale datasets systematically underestimate vegetation recovery in Australian carbon farming projects

Tim Moore, Andrew O'Reilly-Nugent , Kenneth Clarke, Cathleen Waters

Limiting global warming below 2C requires nature-based climate solutions which are expected to supply more than a third of cost-effective climate mitigation by 2030, while prioritising avoiding fossil fuel emissions. Regenerating native forests under the Australian Government’s Australian Carbon Credit Unit (ACCU) program are delivering large-scale carbon storage across approximately 3.4 million hectares. Projects using the human-induced regeneration (HIR) method aim to restore native forests through improved land management, generating ACCUs that underpin both legislated emissions reduction and voluntary decarbonisation targets. Scientific rigour must underpin the integrity of the ACCU program. Constructive critiques of carbon crediting programs allow refinement over time, strengthening climate action4. However, flawed analyses that lack necessary analytical rigour can undermine investment decisions and diminish real outcomes when they impact critical policy decisions.
. 
Macintosh et al. (2024; hereafter Macintosh) contend that HIR activities are having limited influence on changes in woody vegetation cover in Australia. Macintosh analysed the National Forest and Sparse Woody Vegetation Dataset (NFSWVD) and, elsewhere, Woody Cover Fraction (WCF) to compare vegetation trends between credited HIR areas and adjacent comparison areas. However, Macintosh provides no valid empirical support to validate the data sets they applied were fit for purpose, a standard scientific convention.  Here, we provide evidence that their assessment relies on two untested assumptions:

• that publicly available, national-scale datasets can accurately detect and quantify vegetation cover at the scale of individual projects, and 


• that adjacent comparison areas represent valid experimental controls. 

We applied high quality reference data, collected on HIR projects as standard practice, as empirical evidence that the national-scale datasets (NFSWVD & WCF) used by Macintosh  systematically underestimate regeneration success on HIR projects in the Australian rangelands, and are therefore not fit for purpose as used by Macintosh. We also demonstrate that Macintosh's experimental design has significant fundamental design failures that undermines their findings.