Arctic permafrost soils hold vast organic matter reserves (~1000 PgC in the top 3 m), potentially contributing significantly to rising atmospheric CO2 and the carbon-climate feedback. Arctic air temperatures are warming at twice the global average rate (0.6 °C per decade), accelerating microbial decomposition of soil organic matter (SOM) and releasing greenhouse gases (CO2, CH4, N2O). Simultaneously, rising temperatures cause increased drought, higher vapor pressure deficits, and lightning, leading to more frequent and intense tundra fires. Fires profoundly disturb high-latitude ecosystems, altering surface energy balance, soil hydrodynamics, reducing soil carbon stocks (including ancient permafrost carbon), increasing nutrient losses, and shifting plant and microbial community compositions. These effects can persist for decades. While the impacts of fire on tundra plant communities are relatively well-understood (e.g., shrubs and graminoids recover rapidly from seed banks, while cryptogams recover slowly), the role of belowground microbial communities in post-fire recovery and the interplay between microbial and plant communities remain less clear. This study leverages the mechanistic model 'ecosys' to investigate how tundra fire interacts with long-term climate changes (warming, increasing CO2, elevated precipitation) and focuses on the 2007 Anaktuvuk River Fire in Alaska, using data from this well-studied event for model parameterization and benchmarking. The research addresses three key questions: the long-term impacts of fire under climate change, the role of the belowground microbial community in plant community recovery, and how recovery differs between early 21st-century graminoid-dominated and late-century shrub-dominated ecosystems.

Existing literature demonstrates the significant impacts of tundra fires on aboveground vegetation, with varying recovery timescales for different plant types. Graminoids exhibit rapid recovery, while shrubs take longer, and cryptogams are often decimated. Belowground, fire directly impacts microbial communities through heat-induced mortality and indirectly through changes in nutrient availability and carbon sources. Bacteria generally recover faster than fungi after fire, although mycorrhizal fungi can show resilience due to their association with resprouting shrubs. The recovery of microbial communities is crucial for organic matter decomposition and nutrient cycling, driving vegetation recovery. However, the sequence of events leading to ecosystem steady-state post-fire remains challenging to determine empirically. Furthermore, shrub expansion under warming climates alters belowground communities, as shrubs produce litter with higher carbon-to-nitrogen ratios, favoring fungi over bacteria. This is significant because fungi and bacteria play distinct roles in soil carbon cycling. The interaction between climate, fire, vegetation, and microbial community composition creates complex feedback loops in the tundra carbon cycle.

The study utilizes the ecosys model, a mechanistic model simulating interdependent physical, hydrological, and biological processes in ecosystems. This model incorporates carbon, water, nitrogen, and phosphorus dynamics in plants and soils and has been successfully applied to numerous high-latitude ecosystems. The model was parameterized and benchmarked against data from the 2007 Anaktuvuk River Fire, including plant community metrics, soil carbon, and site physical factors. Model performance was evaluated using normalized Root Mean Square Error (RMSE). The model simulations explored the long-term effects of fire under an RCP8.5 climate scenario, considering fires of varying severities (mild, moderate, severe) initiated in both the early (2007) and late (2080) 21st century. The study analyzed changes in carbon and nitrogen cycling, soil moisture and temperature, and belowground microbial community structure and nutrient cycling. Transfer entropy analysis was used to identify the most influential factors driving changes in ecosystem net primary productivity (NPP).

Model testing against the 2007 Anaktuvuk River Fire data showed good agreement (RMSE = 0.037) in simulating NPP, PFT-specific NPP, and soil carbon stocks. Under the baseline RCP8.5 scenario (no fire), NPP more than doubled by 2100, primarily due to shrub expansion. Despite increased ecosystem respiration, the soil became a stronger carbon sink. Fires significantly reduced soil carbon stocks through combustion, with the severity of loss depending on fire intensity. Soil carbon recovery differed between fires ignited in 2007 (graminoid-dominated) and 2080 (shrub-dominated), with faster recovery in the later simulations. Post-fire NPP decreased more strongly in late-century fires. In early-century fires, graminoids led recovery, while shrubs re-established more slowly but ultimately dominated the community by 2100. Late-century fires showed rapid graminoid dominance followed by rapid shrub growth. Transfer entropy analysis revealed that nutrient availability and uptake were crucial for post-fire NPP recovery, alongside soil moisture and temperature. Fire deepened the active layer, causing long-term increases in soil moisture and temperature. Fire reduced saprotrophic fungi, creating a niche for heterotrophic bacteria, which accelerated organic matter decomposition and nutrient release. Nitrogen fixation increased post-fire, but its contribution to overall nitrogen recovery was less significant than heterotrophic bacterial mineralization. Diazotrophic bacteria showed the most substantial increases in abundance and distribution post-fire, while nitrogen-cycling microbial groups experienced declines in later decades. Early-century fires caused long-term deficits in nitrogen, while late-century fires resulted in faster recovery and even increased nitrogen fixation.

The model results indicate that tundra ecosystems will remain a carbon sink in the 21st century under RCP8.5, mainly due to increased NPP driven by shrub expansion and enhanced nutrient cycling. However, fires significantly impact this carbon sink capacity, with long-term effects on soil carbon stocks and vegetation composition. The rapid recovery of graminoids post-fire highlights their resilience, while the slower shrub recovery suggests different adaptation strategies. Nutrient availability, driven largely by microbial activity, plays a key role in post-fire ecosystem recovery. The shift from fungal to bacterial dominance in the microbial community accelerates nutrient cycling but might have long-term implications for soil carbon dynamics. The enhanced nitrogen cycling post-fire underscores the critical role of microbial communities in nutrient turnover and their response to disturbances. The faster recovery in late-century fires reflects the higher initial nutrient levels and more open nutrient cycles under warmer conditions.

This study demonstrates that microbial processes strongly control tundra ecosystem recovery post-fire. Early-century fires have long-lasting effects, shaping vegetation and nutrient cycles, while late-century fires, under a warmer climate, lead to faster recovery due to higher initial nutrient levels. The findings emphasize the necessity of integrating microbial dynamics into models to accurately predict tundra carbon cycle responses to climate change and fire frequency increases. Future research should focus on further refinement of microbial representations in ecosystem models and investigations of the long-term implications of increased fire frequency on tundra ecosystem stability.

The study relies on a mechanistic model, which involves inherent simplifications and parameter uncertainties. While the model was benchmarked against observations from the Anaktuvuk River Fire, extrapolating these results to other tundra regions may require careful consideration of site-specific factors. The study primarily focused on nitrogen cycling; incorporating other nutrient cycles (e.g., phosphorus) in a more comprehensive manner might provide additional insights. The transfer entropy analysis identifies important directional influences but does not establish direct causal relationships. More empirical data on microbial community dynamics and nutrient cycling following tundra fires are needed to validate and further refine the model.