Figure 1.
Map of study area showing phenology observations sites.
The rectangular area in the upper right inset elevation map with major highways corresponds to the study area extent shown on the main map. Vegetation information is derived from the Normalized Difference Vegetation Index (NDVI) computed from July 28, 2010 Landsat TM image (Vegetation: NDVI = 0.5–0.7, Dense vegetation: NDVI >0.7).
Figure 2.
Historical changes in January–May mean air temperature (shown here with the fitted trend line) and timing of first leaf emergence for 3 tree species monitored in Hohhot arboretum during 1963–2010.
Lines in phenology graphs connect contiguous data points and illustrate significant data gaps in the time-series.
Figure 3.
Graphical representation of ANCOVA analysis for the leaf emergence phenophase.
Historical data are depicted by blue star symbols and spatial data by red crosses. Linear fits (either a single or two parallel lines) are selected based on ANCOVA. X axis is the growing degree days accumulated to a specified day of year.
Table 1.
Phenology basic statistics.
Figure 4.
Comparison of mean AGDD (∑>5) accumulation (upper panel) and the associated standard deviation (lower panel) for the period of March,31– May,30.
Historic data is the average of years for which phenology observations are available during 1963–2010, and spatial temperature data is the average of two years (2010–11) shown for three major land uses separately.
Table 2.
Analysis of covariance (ANCOVA) testing differences between linear regression fits to historical and spatial phenological observations.
Table 3.
Comparison of best phenological regression models for combined and historical data.
Table 4.
Comparison of regression fits to historical data and to the time-series of predictions (1963–2009) from phenology models developed based on combined data.
Figure 5.
Absolute slopes (Y-axis) of linear fits to phenological time-series (1963–2009) predicted based on phenology regression models from combined data and those constructed from historical records.
All slopes are negative and significant at p<0.05. Numbers along the X-axis are sample sizes (top = combined and bottom = historical) used to develop phenology models. They decrease according to the elimination of one historical data point at a time, starting from the earliest. * indicate significance (p<0.05) of slope difference based on ANCOVA analysis. Phenophases are same as in Table 1.
Figure 6.
Scatterplots illustrating hypothetical situations when spatial data (star symbols) are combined with historical data (squares).