Boreal tree species diversity increases with global warming but is reversed by extremes

Tree species diversity is crucial for maintaining the health and function of forest ecosystems, particularly their role as a carbon sink in mitigating climate change. While global forest extent has fluctuated due to factors like forestry, agriculture, and wildfires, the impact of climate change on tree species diversity, especially at large spatial scales, remains poorly understood. Existing studies often rely on plot-level observations or coarse-resolution diversity maps, limiting their ability to capture the complex spatiotemporal dynamics of boreal forests. Boreal forests, covering 30% of the Earth's forested area, are particularly vulnerable to climate change due to their harsh climate and lower species diversity compared to tropical forests. Even small changes in boreal forest diversity can significantly affect carbon uptake and ecosystem stability. This study aims to quantify changes in boreal tree species diversity over recent decades, analyze the relationship between these changes and climate variables, and assess the impacts on the boreal ecosystem carbon cycle. The researchers developed a novel framework that combines field observations with high-resolution Landsat satellite data and machine learning techniques to produce spatially continuous and temporally dynamic maps of boreal tree species diversity. This approach offers a significant advancement over previous methods that rely on static maps and coarse spatial resolutions.

Previous research has highlighted the importance of tree species diversity in maintaining forest ecosystem functioning and carbon sequestration. Studies using plot-level observations have shown a positive relationship between biodiversity and productivity, contributing to ecosystem stability. Global forest extent has seen both gains and losses over the past decades, primarily driven by human activities such as deforestation and agriculture, but also influenced by climate change and rising CO2 levels. Climate change has been shown to cause continental-specific trade-offs in forest dynamics, influencing growth and mortality rates, and leading to shifts in tree species composition at local scales. However, the response of tree species diversity to these changes at a large spatial scale has remained largely unknown. Previous attempts at mapping global forest tree species diversity have used statistical relationships between diversity metrics and environmental variables. While these approaches have provided insights, they suffer from limitations in spatial resolution and temporal coverage. This study improves upon those limitations by using high resolution satellite imagery and machine learning to overcome the spatial and temporal limitations of previous studies.

The researchers developed a framework to estimate boreal tree species diversity using the Shannon diversity index (*H*′). They combined 5,312 field observations (including 190,516 trees and 254 species) with 55,560 Landsat images (covering 2000, 2010, and 2020). A deep learning approach, based on the Inception Time architecture, was used to train a predictive model using the field data. This model was then applied to the Landsat imagery to upscale the tree species diversity estimates across boreal forest ecosystems. The spatiotemporal dynamics of boreal tree species diversity were analyzed in relation to a variety of environmental factors including climate (temperature, precipitation), human population density, fire activity, and soil conditions. A boosted regression tree (BRT) model was used to assess the relative importance of these factors in explaining the spatial variability of tree species diversity. Finally, the association between spatiotemporal changes in tree species diversity and indicators of forest carbon (net primary production (NPP), kernel normalized difference vegetation index (kNDVI), vegetation optical depth climate archive Ku-band (VOD Ku-band), aboveground biomass (AGB), and temporal stability of biomass) was analyzed using multiple linear regression.

The study revealed a 12% average increase in boreal tree species diversity (*H*′) from 2000 to 2020 across the boreal forests. This increase encompassed 53% of all boreal forest areas, primarily in the eastern forest-boreal transition region, the Okhotsk-Manchurian taiga, and the Scandinavian-Russian taiga. The relationship between tree species diversity and temperature was positive but weakened at higher temperature changes, with extreme warming (>0.065 °C yr⁻¹) exhibiting a negative impact on diversity. Spatiotemporal increases in diversity were significantly associated with increased productivity and temporal stability of boreal forest biomass. Analysis of environmental determinants showed that mean seasonal temperature and precipitation were the most important predictors of spatial variability in diversity, followed by elevation and human population density. Temporal changes in diversity showed a negative response to increasing temperature and fire activity frequency, while the response to precipitation trends was more complex, being positive for minor trends but approaching zero for more extreme trends. A negative relationship was also observed between diversity and stand age. Finally, significant positive associations were found between diversity and forest carbon fluxes and stocks at both spatial and temporal scales.

The findings indicate a complex relationship between climate change and boreal tree species diversity. While moderate warming appears to promote diversity through various mechanisms (e.g., extended growing seasons, altered disturbance regimes), extreme warming can have detrimental effects, likely due to exceeding thermal tolerances of tree species and increased wildfire frequency. The strong positive association between increased diversity and higher carbon stocks and ecosystem stability suggests that preserving boreal tree species diversity is crucial for maintaining the functionality of these ecosystems as carbon sinks. The observed differences in the contribution of species richness and evenness to diversity changes across different boreal regions suggest the importance of considering regional context when studying the responses to climate change. The observed impact of stand age on both diversity and carbon cycling highlights the dynamic nature of boreal forests and the need for considering forest age in management and conservation strategies.

This study provides valuable insights into the effects of climate change on boreal tree species diversity and ecosystem functioning. The combination of high-resolution remote sensing data and advanced machine learning techniques enabled the creation of spatially and temporally detailed maps of tree species diversity. The findings emphasize the importance of considering both the magnitude and rate of climate change when assessing its impact on biodiversity and ecosystem services. Further research is needed to investigate the role of functional diversity and potential biome shifts, and to further refine the understanding of the interplay between climate change, biodiversity, and carbon cycling in boreal forests.

The accuracy of the diversity estimations may be influenced by factors such as data quality, lighting conditions, shadowing, and the presence of understory vegetation. While efforts were made to reduce these uncertainties through careful data processing and model development, some residual uncertainties may remain. The study relied primarily on satellite data for a relatively short time period (2000-2020), which might not fully capture long-term trends or the full range of climate variability. Finally, the causal relationships between climate change, diversity changes, and ecosystem functioning are complex and not fully disentangled in this study, requiring further investigation.