Abstract. Mimicking of natural disturbance for ecosystem management requires an understanding of the disturbance processes and the resulting landscape patterns. Since fire is the major disturbance in the boreal forest, three widely held beliefs about fire behavior and resulting landscape patterns are examined in light of the empirical evidence available. These beliefs are: (1) that there is a ‘natural’ fire frequency for boreal ecosystems; (2) that the landscape mosaic created by wildfire is generally one of small, younger patches embedded within a matrix of older forest; and (3) that forest flammability is largely controlled by fuel accumulation. Despite the apparently logical basis for such beliefs, they are not well supported by empirical evidence. This discrepancy is explained by problems such as failure to appreciate the relationship between number of fires and area burned and inappropriate extrapolations or generalizations from other regions and vegetation types. The most important implications for management are that the natural disturbance processes producing landscape patterns in the boreal forest generally operate at much larger scales than management units, and that humans may have more indirect (through landuse change) rather than direct (through fire suppression) effects on the frequency of wildfires.

The forested landscape consists of a mosaic of patches of different times since the last disturbance (i.e., different stand ages). Therefore, we can form a distribution of forest ages for the entire landscape (landscape age distribution). Studies of disturbance by fire in boreal and subalpine conifer forests have shown that the cumulative age distribution (landscape survivorship distribution) is best fit by a negative exponential model for which the parameter, the disturbance cycle, gives the time required to disturb an area equal in size to the study area. This distribution describes the rate at which parts of the landscape will survive disturbance, and consequently it tells us the percentage of the landscape that will survive to be old-growth forest. Empirical studies show that old forests make up a small proportion of the boreal and subalpine landscape. We introduce the concept of characteristic oldest age, which is a function of disturbance cycle and size of the study area. This landscape approach to old growth allows one to estimate the minimum area required to ensure the continued existence of some user-defined old-growth forest for any given disturbance cycle. Key words: old growth, disturbance cycle, ecosystem management, landscape age distribution, boreal forest, landscape ecology.

The southern edge of the boreal forest in central Saskatchewan, Canada, has had its forest composition changed in the first decades of this century, primarily by logging and escaped fires from adjacent agricultural clearance. Three timber berths were established in 1884 within and immediately adjacent to the present southern half of Prince Albert National Park (established in 1927). These timber berths were selectively logged for saw timber between 1900 and 1918. Between 1907 and 1918, an average of 70 trees per hectare were removed by selective logging. Most of these trees were white spruce (Picea glauca (Moench) Voss). Since logging companies were required to remove all merchantable trees with a basal diameter greater than 25 cm, it is estimated that between 28 and 54% of the canopy trees were removed. Between 1883 and 1942, 81% of the timber berths were burned two or more times by crown fires that spread through the study area from adjacent agricultural clearances 30 km or more away. By 1945, agricultural clearance was largely complete and the clearance-caused fires stopped. The changes in tree composition were determined by transition probabilities between forest surveys taken in 1883 and 1994. Forests subjected to short-interval, clearance-caused fires but no logging were significantly reduced in their abundance of sexually reproducing trees such as white spruce, but increased in trees with either vegetative reproduction (i.e., underground stems, not just basal sprouts) or serotinous cones, such as aspen (Populus tremuloides Michx.) and jack pine (Pinus banksiana Lamb.), respectively. Transition probabilities for forests experiencing both short-interval, clearance-caused fires and logging reveal an even more marked compositional change in this direction.

One approach to ecosystem management is to emulate the effects of natural disturbance in producing landscape patterns; this approach requires a spatial analysis of the pattern and an understanding of the processes producing the pattern. Forested landscapes exhibit mosaic patterns of both stand types and ages. This study investigates the spatial mosaic of stand ages produced by high‐intensity stand‐replacing fires in the mixed‐wood boreal forest of western Canada. A high‐resolution, accurately dated, time‐since‐fire map for a large (3461 km2) contiguous area is used to produce the landscape survivorship distribution in which both spatial and temporal changes in fire cycle are statistically tested. Spatial multivariate analysis of the time‐since‐fire map is also used to investigate the spatial assembly of the age mosaic. Significant changes in fire cycle can be explained by climatic change as well as land use change in the surrounding area. The shift from a short (15 yr) fire cycle to a longer (75 yr) cycle after 1890 in the northern half of the study area coincides with climatic change at the end of the Little Ice Age. In the southern half of the study area, the short fire cycle continues after 1890 due to the spread of human‐caused fires from the adjacent area which was settled and cleared for agriculture during the first half of the 20th century. Upon completion of settlement in 1945, the fire cycle becomes significantly longer due to the fragmentation of the once continuous forest that surrounded the study area and from which the majority of large fires propagated in the past. The different fire cycle histories of the two parts of the study area also explain the spatial mosaic pattern of stand ages, sizes, and shapes. The extended period of the short fire cycle through the first half of this century in the southern region results in it being dominated by younger, larger, oblong‐shaped polygons with irregular edges: characteristics that describe the shapes of large burns. The northern region has generally older and smaller, more circular, compact polygons that are the remnants of larger much earlier burns that have since been overburned. The polygons in the northern region are more similar in size and shape but less similar in age to adjacent polygons than are those in the southern region. Thus, this study shows how spatial heterogeneity in the landscape mosaic pattern can be characterized and related to the disturbance history of an area. Furthermore, it provides evidence of the impacts on the age mosaic due to forest fragmentation in surrounding areas.

No abstract