Yellow-cedar decline mapping at three spatial scales
Along with reconstructing mortality through the 1900s on the temporal scale discussed in the epidemology section, we are evaluating the yellow-cedar problem at three spatial scales:
- broad scale (~7x106 km2, southeast Alaska),
- mid-scale (~800 km2; e.g, medium-sized island), and
- fine scale (~1km2; small watershed).
Each spatial scale provides unique clues about the cause of yellow-cedar decline, and is also highlighted in the development of a management strategy for the species.
Broad scale
For the broadest scale, we developed a distribution map that depicts more than 2,500 locations totaling over 200,000 hectares (>1/2 million acres) of dead and dying yellow-cedar forests (Wittwer, 2004). This map was derived from sketch mapping from small aircraft, an approach that yields inexact locations and polygon boundaries. However, it is instructive to examine broad areas where decline is present or absent and relate any pattern to regional variation in climate.
Our distribution map captures occurrence of yellow-cedar mortality in Alaska, but not in adjacent British Columbia. Recently, we detected intensive areas of yellow-cedar decline about 150 km south into British Columbia where it often occurred at about 300 to 400 m elevation (Hennon et al., 2005). Surveys by the British Columbia Forest Service continue in an attempt to map the southern extent of the problem.
Download the cumulative cedar decline layer (shapefile) from southeast Alaska.
Snow Accumulation & Yellow-cedar Decline
In an early use of the map, we found the forest decline to align with warmer average winter temperature isotherms (Hennon and Shaw, 1994), an suggestion that climate was involved in the problem. Here, we compare the distribution of yellow-cedar decline to the first detailed model of snow accumulation zones in southeast Alaska (Fig. 4). The snow accumulation model developed courtesy of Dave Albert of the Nature Conservancy, is derived from PRISM data estimates of monthly temperature and precipitation (i.e., precipitation during months when mean temperature <+2ºC). Note that close association between occurrence of yellow-cedar decline and the lowest snow accumulation zone.
Mid scale
We produced a higher resolution mid-scale map of the Peril Strait area of Baranof and Chichagof Islands (~800km2) by delineating polygons of yellow-cedar decline on color infrared photographs rather than by aerial sketch mapping. We use this higher resolution map to evaluate the association of the decline problem with landscape position features including slope, aspect, and elevation. Our mapped polygons of yellow-cedar decline in this study area are concentrated at lower elevations: higher amounts below 150m, lesser amounts between 150 and 300m, and very little above 300m. Yellow-cedar decline occurs on all aspects within these zones, but more decline was mapped on warm aspects (south and southwest).
Currently, we are expanding this work to Mount Edgecumbe, a dominant volcano near Sitka with radial symmetry and fairly even slope gradients. The open canopy forests appear to extend close to timberline. These features help control confounding factors and allow us to detect the direct influence of elevation and aspect on snow and the decline problem.
Fine scale
Research at the small watershed scale is directed at an understanding of how forest conditions vary over local areas of a landscape. We established 129 vegetation plots on 100m grids at two small watersheds, Goose Cove on Baranof Island and Poison Cove on Chichagof Island, to measure live and dead trees and environmental variables (D’Amore and Hennon, 2006). Also, we used low altitude color infrared photographs to produce a forest classification based on canopy cover and the extent of tree death, resulting in four zones: bog, scrub, productive dead, and productive live. The hydrologic, soil, and temperature factors described below were then compared among these zones, as well as correlated with degree of tree death (expressed as percent basal area dead).
Goose Cove
click thumbnail to the left for an expanded view.
We classified the landscape at two small watersheds in Peril Strait according to canopy cover and tree mortality. Areas at the Peril Strait site are further classified by elevation because tree mortality appears to be limited to <150m elevation. HVD=High volume dead, HVL=High volume, live.
Poisen Cove
click thumbnail to the right for an expanded view. 
LiDAR (light detection and ranging)
We have explored and continue to use many types of imagery, including IKONOS, SPOT and image analysis techniques. Our plot system employs relatively small plots. We must rectify any imagery used to a high degree of spatial accuracy. A recent LiDAR acquisition has provided the much needed digital elevation model used for orthorectifying the imagery and micro-scale terrain analysis. The cedar decline seems to be directly related to hydrology and exposure. The high density LiDAR data at the micro-scale study sites has allow us to generated high resolution (1m) ground DEMs to analyze micro-scale hydrology and above ground structure DTMs to analyze micro-scale three dimensional canopy cover.
We have compared the canopy models to values of air and soil temperature, ultimately the key factors in the cause of cedar decline.
