Boreal forest and tundra biomes are key components of the Earth system because the mobilization of large carbon stocks and changes in energy balance could act as positive feedbacks to ongoing climate change. In Alaska, wildfire is a primary driver of ecosystem structure and function, and a key mechanism coupling high-latitude ecosystems to global climate. Paleoecological records reveal sensitivity of fire regimes to climatic and vegetation change over centennial-millennial time scales, highlighting increased burning concurrent with warming or elevated landscape flammability. To quantify spatiotemporal patterns in fire-regime variability, we synthesized 27 published sediment-charcoal records from four Alaskan ecoregions, and compared patterns to paleoclimate and paleovegetation records. Biomass burning and fire frequency increased significantly in boreal forest ecoregions with the expansion of black spruce, ca. 6-4 thousand years before present (yr BP). Biomass burning also increased during warm periods, particularly in the Yukon Flats ecoregion from ca. 1000-500 yr BP. Increases in biomass burning concurrent with constant fire return intervals suggest increases in average fire severity (i.e., more biomass burning per fire) during warm periods. Results also indicate increases in biomass burning over the last century across much of Alaska that exceed Holocene maxima, providing important context for ongoing change. Our analysis documents the sensitivity of fire activity to broad-scale environmental change, including climate warming and biome-scale shifts in vegetation. The lack of widespread, prolonged fire synchrony suggests regional heterogeneity limited simultaneous fire-regime change across our study areas during the Holocene. This finding implies broad-scale resilience of the boreal forest to extensive fire activity, but does not preclude novel responses to 21-century changes. If projected increases in fire activity over the 21^st century are realized, they would be unprecedented in the context of the last 8,000 years or more.

All 27 lake-sediment charcoal records were developed with virtually identical methods, originally published by: Higuera et al. 2007, 2009, 2011 a/b, Kelly et al. 2013 and Barrett et al. 2013. Sediment cores were collected from small (< 10 ha), deep (> 5 m) lakes with simple basin shapes and minimal inlets or outlets, using a polycarbonate tube fitted with a piston and/or a modified Livingstone coring device. Core tops were sampled in the field at continuous 0.5-1.0 cm intervals, and in the laboratory, cores were sampled in contiguous 0.25-0.5 cm intervals, with 0.5-3.0 cm³ samples taken for charcoal analysis. Samples were treated with sodium metaphosphate, oxidized with sodium hypochlorite or hydrogen peroxide, and sieved to isolate macroscopic charcoal (> 150-180 um). Charcoal particles were counted at 10-40x with a stereomicroscope, and CHAR (pieces cm^-2 year^-1) was derived as the product of charcoal concentrations (pieces cm^-3) and sediment accumulation rates (cm yr^-1). Sediment accumulation rates were based on age-depth models, developed from ²¹⁰Pb (in 26 of 27 records) and AMS ¹⁴C (in all records) dating of terrestrial macrofossils or concentrated charcoal particles. All records underwent a decomposition and peak analysis procedure by their original authors using the CharAnalysis software (github.com/phiguera/CharAnalysis), which separates background (“noise”) from foreground (“signal”) patterns in CHAR to identify statistically significant peaks. We made no changes to the chronologies or peak identification analyses from the original publications. We excluded (portions of) records with a signal-to-noise index < 3 from metrics that rely on peak identification (Kelly et al. 2011).

To interpret changes in Holocene fire activity in the context of millennial-scale vegetation change, we rely on four previously published pollen records from each ecoregion (Tinner et al. 2006, Higuera et al. 2009, 2011a, Kelly et al. 2013).

To evaluate potential relationships between fire activity and millennial-scale climatic change, we rely on a recent summary of paleoclimate in northwestern North America (Kaufman et al. 2016) spanning the entire Holocene, and a tree-ring based air temperature reconstruction from 1200 yr BP to present (Wiles et al. 2014). The composite temperature reconstructions from Kaufman et al. (2016), includes midge-inferred July air temperature (Clegg et al. 2011, Irvine et al. 2012), pollen-inferred air temperature (Szeicz et al. 1995, Bunbury and Gajewski 2009, Viau and Gajewski 2009), and temperature anomalies from other proxies (McKay et al. 2008).

See detailed description of data and analyses here: https://doi.org/10.5281/zenodo.3736520.