Variation in wood density across South American tropical forests

Understanding spatial and temporal variation in forest biomass carbon stocks is critical for numerous applications and research questions, including national carbon stock inventories [e.g. ref. 1], assessments of forest responses and recovery from disturbance 2-4 , and investigation of climate feedbacks [e.g. ref. 5]. However, quantifying the distribution of aboveground live carbon stocks across the tropical forest biome remains challenging. Despite decades of fieldwork 6 and investment in satellite and airborne remote sensing to measure canopy structure with Lidar or vegetation volume through radar scattering 1,7 , there is still considerable uncertainty about the amount and distribution of aboveground carbon in tropical forests. Indeed, marked differences among recent global maps of biomass carbon 8-10 reflect the challenge of large-scale calibration and validation across tropical forests.

The challenge partly arises because remote-sensing approaches, which allow large-scale and spatially continuous measurements, cannot provide all the information available from ground-based surveys. Wood density is a fundamental determinant of tree biomass 11-13 , and estimating it requires skilled botanical surveys to identify trees.

Airborne and satellite remote-sensing methods provide measurements that allow estimates of tree height or volume, but not their identity or wood density 14 . While some inferences about taxonomic composition can be made from hyperspectral imagery [e.g. refs. 15, 16], this remains limited compared to what can be obtained by a ground-based botanical survey. Lack of wood density information can lead to marked discrepancies between remote sensed and ground-based estimates of aboveground biomass 17 , including spatial biases in aboveground biomass estimates of around 30% even within a single country 18 .

Future improvements in remote-sensing-based forest biomass maps therefore require improved knowledge of spatial variation in tree wood density. The need to tackle this huge challenge is especially important in South America. Not only are tropical rain forests here the most extensive in the world, but they also include many of the most productive and carbon-rich forests on Earth 19,20 and large carbon sinks and fluxes [e.g. 21-23 ]. The nature of the challenge is also most profound in South America, as ~40% of Earth's 73,000 tree species are found in forests here 24 . Amazonia alone is home to at least 15,000 25 , and beyond