Data and Methods
Study area
The study area covers the northern part of British Columbia, the Prairie Provinces (Alberta, Saskatchewan, and Manitoba), and Minnesota in the US. The two major data sources are the phenological observation of the provenance trial and the remotely sensed satellite data. The provenance trial series were established at Athabasca of Alberta (54°43’N, 113°17’W) (Figure 1), within which seedlings from 43 provenances (geographic origins) in the study area are planted and observed.
Provenance trial data
The provenance trials was established by Alberta Pacific Forest Industries Inc., Western Boreal Aspen Corporation and Alberta Pacific Forest Industries in 1998. The trial series were located at Athabasca of Alberta (54°43’N, 113°17’W), within which seedlings from 43 provenances from the study area are planted and observed. The trees are compared based on a randomized complete block design with 6 replications. Each provenance has a geographical coordinate (the longitude and latitude) of its origin recorded. The growth and phenological phases are recorded in order to select the proper provenances for tree breeding and reforestation. The growth, budburst and bud set are measured respectively. The bud set data for the preliminary research were collected in 2006, which were recorded in a scale from 0 to 1.0. The higher the score is, the earlier the bud set date is. More data will be collected with exact Julian date in the future.
Figure 1 A site image of the provenance trial.
With ArcGIS, a file-geodatabase is built to assemble 43 provenances (X, Y data) with bud set data as a point feature class with ModelBuilder (Environmental Systems Research Institute (ESRI), 2009). The data table of the bud set is as below attached with the daylight length, the geographic coordinates and elevation of each provenance location Table 1. The daylight length was referred to September 14th, 2006, one day before the measurement. Further observations will be done to record the exact Julian date of bud set event this year.
Table 1 Data table of the provenance trials.
Figure 2 An image of the bud before the budbreak in the provenance trial.
Brown-down data
The aspen leaf senescence (brown-down) data are prepared for validate the provenance trial data over a continental scale. These brown-down images were inferred with the Enhanced Vegetation Index (EVI) produced by the Moderate Resolution Imaging Spectroradiometer (MODIS) of NASA’s Terra satellite (Figure 3 and Figure 4), which have been validated with independent field survey. Each point of the data set represents a pixel of 500m by 500m. The annual brown-down date was calculated by the TIMESAT software package, which uses functions to fit the EVI values versus the phenological event date. The temporal EVI data has been widely used as an indicator of the onset of leaf senescence for deciduous trees (Zhang, et al., 2003). In this study, the date of annual leaf senescence (brown-down) was produced from 2001 to 2006 for the western boreal forest where the density of aspen distribution is over 40 %. The data table of the brown-down event is shown in Table 2. There are 88 sample units of brown-down pixels across the study area. The daylight length was referred to the daylight length of each sample unit on the brown-down date.
To evaluate the transferred aspen adaptation to Alberta climate, multi-scale spatial data are employed in the GIS modeling: West North America seed zones and ecosystems and the predicted ecoregions. The ecoregions are listed in Table 3. The ecosystems and seed zones polygons are overlaid with the bud set provenances and the brown-down points. The finest unit of ecosystems and seedzones are variants, which are applied as the units of the brown-down data. The climate variables are attached to the overlaid point attributes based on their geographic coordinates.
To evaluate the transferred aspen adaptation to Alberta climate, multi-scale spatial data are employed in the GIS modeling: West North America seed zones and ecosystems and the predicted ecoregions. The ecoregions are listed in Table 3. The ecosystems and seed zones polygons are overlaid with the bud set provenances and the brown-down points. The finest unit of ecosystems and seedzones are variants, which are applied as the units of the brown-down data. The climate variables are attached to the overlaid point attributes based on their geographic coordinates.
Figure 3 Global image of EVI in winter from NASA ( http://visibleearth.nasa.gov/view_rec.php?id=11870).
Figure 4 One of the satellite Tarra is scanning the earth with its swath (http://visibleearth.nasa.gov/view_rec.php?id=11870).
Table 3 Acronym and full names of ecoregions.
| doc001_provenancetrial.pdf |
Figure 5 A map of eco-regions in the study area.
Daylight length
The daylight length is the hourly difference between sunrise and sunset one day. Daylight length, or length of day, is the time each day from the moment the upper limb of the solar disk appears above the horizon during sunrise to the moment when the upper limb disappears. The day light length of each provenance is shown as above. The formula is as below (adapted from www. microsoft.com, November 2009). The formula provides a result of lower precisions, which is less than seven minutes different compared with the results produced by the National Oceanic and Atmospheric Administration (NOAA). The function of daylight length versus latitude and Julian date is as following:
Figure 6 Hours of daylight by four provenance regions. AB is the Alberta aspen provenance, SK is the Saskatchewan seed source, and MN is the Minnesota seed source.
Climate data
The climate data of the provenance trials are retrieved from two softwares the ClimatePP and the Climate WNA (Wang et al., 2006). Climatic variables are selected from the 1961-1990 normal data series, which shows the average trend of climate conditions. Also annual climate variables are retrieved for brown-down data from 2001 to 2006. These climate surface data of the provenances are joined to the point feature as well, including mean annual temperature (MAT, °C), frost-free period (FFP, days), the Julian date on which FFP ends (eFFP, day), autumn mean temperature (TAV_at, °C), August mean temperatures (TAV_08, °C), degree-days above 5°C- growing degree (DD>5, days) etc.
Table 4 Climate variables predicted.
Work flow chart
The work flow chart for the data analysis is as the figure below. The four types of data sources, such as the bud set data, the brown-down data, climate variables, daylight length are imported into file-geodatabases in ArcGIS. Then data tables for R and SAS imports as well as the maps are produced. By using the statistical packages such as R and SAS, the variance partitioning of autumn phenology versus the environmental variables, and the direct and indirect gradient analysis can be carried out.
Figure 7 A diagram of work flow.
