Research

Forest Ecosystems

Key Findings

The Climate Impacts Group's (CIG) research on climate and Pacific Northwest (PNW) forest ecosystems has made important contributions to our understanding of climate influences on forest productivity, forest fires, and forest resource management. The CIG's use of tree-ring chronologies has also been key in identifying past climate variations and trends in the PNW. Key findings from this research include the following.

Detailed the Influence of Climate Variations on PNW Tree Growth and Forest Ecosystems:

• The growth of mountain hemlock (Tsuga mertensiana) and subalpine fir (Abies lasiocarpa), two of the most common high-elevation trees in the PNW, responds strongly to variations in the Pacific Decadal Oscillation (PDO). Growth of these species is positively correlated with the PDO index at snowy, treeline sites (where length of growing season is limiting); growth is negatively correlated with the PDO index at lower elevation and drier sites (where soil moisture is limiting). The effect of PDO on snowpack is clearly the limiting factor to long-term growth in these environments (Peterson and Peterson 2001, Peterson et al. 2002). High PDO index values (i.e., warm PDO years) tend to correspond to winters with lower than average snowpack while low PDO index values (i.e., cool PDO years) tend to correspond to winters with higher than average snowpack.
• Ponderosa pine (Pinus ponderosa), the dominant tree species at lower elevations on the east side of the Cascade Range, responds strongly to the PDO. Growth of this species is positively correlated with the PDO index throughout its range, indicating a strong response to soil moisture as a limiting factor.
• A biophysical environmental analysis of distribution of dominant tree species in northern Washington was conducted on >10,000 vegetation plots on federal lands. The ecological niche of most species was limited by one and occasionally two environmental parameters with relatively simple quantitative relationships. This information provides accurate algorithms for modeling the response of vegetation to climatic change (McKenzie et al. 2003(a), McKenzie et al. 2003(b)).

Placed 20th Century PNW Climate and Streamflow in Broader Context Through Paleo-climate Research:

• Multiple tree-ring chronologies and coral chronologies were used to calculate a more accurate reconstruction of PDO than is possible when using only a single data type. Drawing on five published climate reconstructions extending as far back as 1600, we found that the PDO has probably been a mode of climate variability since at least 1600. The variability of the PDO has fluctuated over time, with the period of 1840 to 1920 (roughly) having lower interdecadal variance than other times. This multi-bioproxy technique holds great promise for even better reconstructions of PDO and hydrological time series in the PNW (Figure 1)(Gedalof et al. 2002).

click image to enlarge



Figure 1 Comparison of the reconstructed (black) and observed (gray) PDO index. The reconstructed chronology in this figure is restricted to 1840-1990, the time common to the five climate reconstructions used for the composite reconstruction. Note the period of reduced variability from 1840 to approximately 1920. The time series is of the leading principal component, scaled to match the mean observed October to March PDO index (R=0.64).

• A network of tree-ring chronologies was used to reconstruct streamflow for the Columbia River back to 1750. This analysis showed that droughts during the 1840s and 1920-30s were far more prolonged and severe than those that occurred since 1950 and that the potential for multi-year droughts has probably been underestimated by regional planners. Because water management strategies were generally developed during a period characterized by a lack of severe multi-year droughts, this research suggests that these strategies may be inadequate to fully protect against possible events (Gedalof et al. in review).

Characterized Relationships Between PNW Climate and Forest Fire Frequency and Extent:

• Specific atmospheric circulation (i.e., weather) patterns are associated with four spatial patterns of 20th century variability in PNW forest fires. Blocking high pressure ridges over western North America were the most prominent feature associated with large wildfires, especially in the Cascade Range (Gedalof et al. in review).
• Forest fire activity is generally higher during warm phases of the PDO. Forest fires were more extensive across the PNW during the 1925-45 warm phase PDO than during the cool phases before and after that (Figure 2) (Mote et al. 1999). The resurgence of fire activity in the late 1980s was consistent with the recurrence of warm phase PDO. Some of the decline in forest fire activity during the cool phase after 1945 and some of the increase in the most recent warm phase may be related to effects of fire suppression programs. Through drought-related wildfire activity, the PDO may influence broader fluctuations in forest structure, composition, and function.


Figure 1 Annual forest fire burned area index for Washington and Oregon. The burned area index represents the area burned each year, normalized by the area monitored for forest fire activity during each year. Vertical dashed lines indicate PDO transition years; open diamonds, cool phase PDO; solid diamonds, warm phase PDO.

• Forest fires in PNW states tend to burn more acreage during persistent warm phases of PDO. Such associations are not found with ENSO for recent fires (Mote et al. 1999), but did apparently exist during the historical period 1650-1900 (Hessl et al. 2004).
• In contrast to a common view that past forest management practices are solely responsible for a recent spate of years with very large forest fires, we found that in most western states, the area burned by wildfire in a given year was very strongly influenced by that year's summer climate. In particular, large fire years are much more likely to occur during warm dry summers and future warming - even at the low end of projected climate scenarios - may lead to at least a doubling in average area burned. The implications of more frequent, extensive fires include an increased probability of losing local populations of species dependent on late-seral habitat (McKenzie et al. 2004). (Press coverage of these findings...)

Related Work

Other forest ecosystem/climate-related research findings at the University of Washington include:

• Analysis of lake sediments in the North Cascade Range reveal good correlations between charcoal deposits, pollen, and macrofossils during the past 10,000 years. Fire has been highly variable over this time although the mean frequency has not varied greatly during different climates. Vegetation distribution and abundance has varied considerably, with pulses of establishment keyed to individual large fires. These data provide an empirical baseline for the natural range of variability in large-fire occurrence (Prichard, S.J. 2003. Spatial and Temporal Dynamics of Fire and Vegetation Change in Thunder Creek Watershed, North Cascades National Park, Washington. PhD dissertation, University of Washington, Seattle).
• Carbon distribution in subalpine forest ecosystems of the Cascade Range is highly variable at small spatial scales, with much larger carbon storage in tree islands than in adjacent meadows. Carbon stored in subalpine forests likely has a long residence time in cold conditions. Clearcut logging in these forests can affect the distribution of soil carbon, depending on the post-harvest treatment (burning versus no burning). In general, leaving logging slash and minimizing burning will encourage long-term storage of carbon (Sanscrainte et al. 2003; Sanscrainte et al. 2003)