Abstract: Boreal forests, the largest terrestrial biome on Earth, are highly varied in local tree density. Despite the recent attempt of Crowther et al. (2015) to estimate tree density in boreal forests from a biome-level model, accurate estimates are still lacking, leaving the question how many trees there are in boreal forests unanswered. In this study, we compiled tree density data from 4,367 plots in North American boreal forest to address this question. We adopted random forest and Gamma regression models for estimating tree density of these plots and predicting boreal tree density of North America. Our results showed that the biome model underestimated boreal tree density by 32.3%, compared to a non-significant underestimation of our model that incorporated stand height as the most important predictor of the model. We estimated a total number of 275.0 billion trees in North American boreal forest compared to 211.2 billion estimated by the biome model. The underestimation of 63.8 billion trees was equivalent to missing 1,234 trees per person for the region. We further produced a boreal tree density map of North America, and projected tree density distribution in 2050 under the RCP2.6, 4.5 and 8.5 emission scenarios. This improved knowledge about boreal tree density contributes to understanding of the role of boreal forests in regulating forest ecosystem functions and informs adaptation and mitigation planning and policy-making.
