Thursday, August 9, 2007

PS 64-112: Reconstructing fire regimes: Multivariate analysis of Holocene charcoal records from lake sediments

Elizabeth A. Lynch, Luther College, Sara C. Hotchkiss, University of Wisconsin, and Randy Calcote, University of Minnesota.

Charcoal sieved from lake sediments is routinely used to reconstruct fire histories in conifer ecosystems dominated by crown fires.  Our objective is to develop strategies for analysis of late-Holocene charcoal records in oak-pine ecosystems where fire regimes include both frequent surface fires and crown fires.  We compiled 1000-4000 yr    long records of charcoal accumulation rates (CHAR; # fragments >125µm/cm^2 /yr) from 10 small lakes across a 450 km^2 sand plain in northwestern Wisconsin.  Using stratigraphically constrained cluster analysis we identified periods with distinctive patterns of CHAR. Within the record for each site CHAR values were standardized to the period 200-700 yr BP, allowing comparison of records across sites. Cluster analysis based on the median and interquartile range of the standardized CHAR values, the ratio of graminoid cuticle:total CHAR, and the average frequency of CHAR peaks produced three primary types of charcoal regimes. Type 1 is characterized by low peak frequencies and occurred primarily in the most recent several centuries across most of the sand plain.  Type 2 has high peak frequencies and occurred over the past 1000 yr in the central sand plain on coarse sandy soils dominated by jack pine, as well as before ~800 cal yr BP in some northern sites. Type 3 has very high CHAR values with large variability from sample to sample and was typical at several sites before ~1000 cal yr BP. Over the past 4000 yr fire regimes exhibited a trend toward regime types with lower amounts of charcoal and fewer peaks, with the exception of two sites on the central sand plain that have had high peak frequencies throughout the past 1000 yr. Our results suggest that characteristics of CHAR assemblages can be used to classify fire regimes and show spatial and temporal variability in relation to landscape context and climate.