How many lake-sediment cores do we need to characterize regional fire-regime changes using macro-charcoal records?

Background/Question/Methods

           Macroscopic charcoal in lake sediments records fires that occur within 10-20 km from the lake. An adequate spatial array of sites is necessary to characterize regional fire-regime shifts.  However, no quantitative analysis has been made to evaluate how many charcoal records are sufficient for inferring fire-regime shifts across a region.  Although more records should yield more reliable results, at some point increasing the number of records may not significantly increase the information content. We created a series of charcoal-accumulation-rate (CHAR) composites by randomly selecting varying numbers of records from two Alaskan ecoregions (Copper River Basin, CRB, and Yukon Flats, YF).  Each of these composites was then correlated with (1) the composite record from all sites in each ecoregion (n=24 for CRB; n=14 for YF) and (2) a Northern Hemisphere temperature reconstruction (Moberg et al. 2005). We further examined the average standard error of each composite to assess how additional records affect the variability within fire history reconstructions. Finally, we examined various characteristics of the nearby landscape (e.g. land cover, connectivity), study lake (e.g. coring depth, watershed area), and charcoal records (e.g. CHAR, fire return interval) to assess how these characteristics affect the reliability of fire-history inferences.

Results/Conclusions

            The temporal patterns are highly variable among the individual charcoal records.  For example, the correlation values (r) between four randomly selected records and the regional CHAR composite record range from -0.12 to 0.87. Increasing the number of records in the composite reduces variability and strengthens the correlation with both the regional CHAR composite and temperature until a plateau is reached.  Many records are required to reach the plateau (90% of the maximum correlation) with both the regional charcoal composite (YF: n=23, CRB: n=34) and temperature (YF: n=15, CRB: n=44).  However, strong correlations to the regional CHAR were still observed with fewer records, particularly in the YF [YF: n=14, r=0.87 (95% CI=0.69-0.96); CRB: n=24, r=0.79 (95% CI=0.49-0.94)]. This pattern reflects (1) that CHAR values are typically higher for the records from YF than those from the CRB, and (2) that the strength of the correlation between each record and the regional composite increases with the mean CHAR value of the record (r=0.34, p=0.04).  Other characteristics of the nearby landscape, study lake, and charcoal records do not significantly affect the relationships of the individual charcoal records with either the regional charcoal composite or the temperature record.