Research Interest I:

Fire, Climate and Vegetation Systems Reconstructed Using Paleoecology

I have interests in modeling interrelations between the fire, climate and vegetation systems, and in developing methods to reconstruct histories of fire, climate and vegetation conditions over the past 1000 years.

Prospective students could develop any of the ideas described below using data I have from Alberta, using a variety of publicly available GIS and paleo databases, or using data they collect from a geographic area of specific interest.

Interrelations between fire, climate and vegetation systems

My research has investigated four relationships between fire, climate and vegetation as they are exhibited over time and space.

Climate-induced variations in fire frequency

There is concern that human-caused global warming will increase fire frequency, that this will release carbon that has been locked up in wetland biomass, and thereby amplify the warming. To address this question I have used tree-ring methods to determine how fire frequency has changed in response to past climatic oscillations. I have found that, despite an increase in global temperatures over the past 150 years, fires during this time period have become less frequent in the subalpine forests of southern Alberta (Johnson and Larsen 1991) and the boreal forests of northern Alberta (Larsen 1996, 1997). The reason for this appears to be the increase in precipitation that has accompanied this past global warming.

Climate-induced variations in vegetation

There is also some concern that human-caused global warming may cause a rapid change in the geographic range of forest types. In northern Alberta it is believed that valuable softwood timber will be replaced by less valuable aspen. To address this question I reconstructed forest composition over the last 600-800 years using fossil pollen records with a five-year resolution. The pollen record I created for Rainbow Lake A shows that the typical white spruce dominance was replaced by aspen between 1250 and 1550 AD (Larsen and MacDonald 1998a). This change may represent a delayed conversion to the more warm-tolerant aspen in response to the Medieval Warm Period of 900 to 1300 AD, and then a delayed reversion back to the more cool-tolerant white spruce when the warm period ended. These results suggest that forest composition might not change for several hundred years following a climate change.

Fire-induced variations in vegetation

The traditional belief that forest succession follows an orderly pathway has recently been proven untrue by researchers that employed tree-ring reconstructions of succession in replicate 100 m2 plots. To test for variation in successional sequences in one site over multiple fires, I used pollen analysis to reconstruct 600-800 year time periods at a 5-year resolution. The pollen records come from small lakes that record pollen from a several hectare area and thus just show successional events occurring after large local fires. I found that at this spatial scale there is an orderly temporal sequence to the succession that occurs between any two fires, but that the abundance of each plant type varies with each succession. Thus, both the traditional and new ideas are true, depending on the spatial scale of the study.

Spatial variations in fire and vegetation

The relations between climate and fire that I have discovered might lead to the expectation that a regional change to a drier climate would lead to regional change in vegetation with species that require long intervals between fires becoming locally extinct. However, I have found that different portions of the landscape experience different fire frequencies (Larsen 1997). Fires are less frequent in areas with many waterbodies since they stop the spread of fire. The delayed loss of white spruce dominance from Rainbow Lake A likely reflects fire protection the many lakes around it provided (Larsen and MacDonald 1998a).

I have also shown that spatial variations in fire frequency can be modeled using a Geographic Information System, spatial coverages of present-day environmental conditions and less than 50 years of historical fire records (Pew and Larsen 2001). This model could be used by fire managers to predict the likelihood of future fire occurrence in different portions of the landscape.

Reconstructing histories of fire, climate and vegetation

I have made advances in four different methods of reconstructing the ecological past.


The ability to reconstruct pre-historic fire frequency is difficult in areas where large intense fires kill most of the vegetation in their path and thus leave few records of the fire”Ēs occurrence. I have helped develop and refine methods of analyzing tree-rings and lake-sediments to address this problem. My broad background in fire reconstruction allowed me to coauthor a review paper on this topic that will be published in what should be a standard reference book for paleoenvironmental studies (Whitlock and Larsen 2001).


I found that annual tree-ring widths in northern Alberta were more strongly related with seasonal averages of fire-weather indices than with mean monthly climate (Larsen and MacDonald 1995). The tree-ring records I created were thus useful for relating to the life-table reconstructions of decadal-scale variations in fire frequency with variations in the summer average in the fire weather. In Stuart-Williams et al. (submitted) we have shown that oxygen isotopes from the calcite fraction of annual laminae can provide a record of mean summer temperature with a sub-decadal resolution. This 227-year record shows that the temporal pattern of climate change in the central boreal forest was no different from that in the arctic. I have submitted a NSF grant to extend this record to the past 2000 years. I have recently found that changes in the pollen accumulation rate of all terrestrial plants evidences centennial-scale climate cycles that have been observed in arctic regions (Larsen and Morris in prep.). This finding refutes the general belief that, because of inertia in vegetative responses to climate change, pollen records are only useful for reconstructing climate changes that occur over periods of 1000 years or more.

Sediment mixing

Since paleoenvironmental research is best performed on lake sediments that are little mixed, and especially on ones with annual laminations, I compiled and developed heuristics that related the surface area and maximum depth of a lake with the degree of sediment mixing (Larsen and MacDonald 1993). These heuristics, though not deterministic (Larsen et al. 1998), aid the identification of lakes with minimal mixing and the exclusion of ones with excessive mixing. Although massive (non-laminated) sediments are mixed, I have found that if sampled at an approximate 5-year resolution they can still exhibit a relatively unsmoothed stratigraphy that evidences fire events and forest succession (Larsen and MacDonald 1998b). This usefulness appears to be lost in shallower lakes as their sediments are more mixed (Larsen 1994).

Sediment aging

Radioisotopic methods allow massive sediments to be absolutely dated with a decadal-scale error over the last 150 years, but with only a centennial scale error during less recent periods. One method that has been suggested to allow sub-centennial scale dating of massive sediments is the slotting of a record of geomagnetic secular variation contained in its sediments with an absolutely dated record of them from annually laminated sediments. This method has not been applied to recent lake sediments because it has not been possible to recover undisturbed recent sediments. I applied two methods for obtaining undisturbed sediments and have shown that sub-centennial scale changes in the geomagnetic declination are recorded in lake sediments, and that the changes from laminated sediments could potentially be used to absolutely age massive (non-laminated) sediments (Larsen et al. 2000).