Spatial heterogeneity of first flowering date in Beijing's main urban area and its response to urban thermal environment

Phenology - the rhythm of periodic plant life cycle events - was significantly shaped by urban climate, with flowering as one most sensitive phenophase. Apart from the widely noticed urban-rural phenological discrepancy caused by heat island effect, driven by the aggravating spatial unevenness of urban thermal environment, the spatial heterogeneity of flowering time was also found within the urbanized area of some metropolitans, bringing multiple impacts on urban ecology, landscape and public health. This research aimed to reveal the intraurban spatial variation and response characteristics of Beijing's trees flowering phenology that remained largely unclear before. We analyzed the spatial heterogeneity pattern of the first flowering date (FFD) for 42 deciduous woody species in Beijing's main urban area (MUA), and explored the species-specific phenological response to local thermal environment. The sample plots were set in 9 green spaces distributing from urban center to northwest suburb in Beijing's MUA, the FFD data was collected by ground-based phenological observation, and local thermal environment was measured with land surface temperature (LST) retrieved from MOD11A1 products. The main results are as follows: (1) A significant spatial variation for FFD existed among 9 sample plots and the maximum spatial difference of FFD reached 6.76 ± 1.77 days in average, FFD showed an overall delay trend from urban center in 2^nd Ring to outskirts beyond 5^th Ring with 3^rd Ring as a critical line for significant phenological difference. (2) The FFD of 35 species was found to be negatively correlated with [Formula: see text] (average of daily mean LST above 0 °C before mean FFD) in the sample plot (p < 0.05) with a response sensitivity of 2.99 ± 0.87 days/°C, which reflected the significant impact of LST variation during flower development period. Furthermore, the spatial difference and response sensitivity of FFD for a specific species were found to be negatively associated with its mean FFD value (p < 0.05), i.e., the flowering time of early-blooming species tended to be more sensitive to thermal environment variation compared with late-blooming ones. This research illustrated how flowering phenology responded to the heterogeneous intraurban thermal environment in Beijing's MUA, which can improve our understanding of the vegetation dynamics in a constantly changing urban environment. And as a critical indicator of trees' climate vulnerability assessment, the species-specific phenological response sensitivity could also guide species selection in urban forest construction.

Changes in flowering phenology of woody plants from 1963 to 2014 in North China

Existing evidence demonstrates that the first flowering date (FFD) of most plant species became earlier in response to temperature increase over the past several decades. However, the studies on changes in flowering duration (FD) were limited. By using the non-parametric Theil-Sen estimator, this study investigated the temporal trends in 127 time series of FFD, end of flowering date (EFD), and FD of 97 woody plants from 1963 to 2014 at three sites (Harbin, Beijing, and Xi'an) in North China. The relationship between flowering phenophases and temperature was analyzed using two phenological models. The results showed that most of FFD and EFD time series exhibited an apparent advancing trend. Among them, trends of 52.0% (40.9%) of FFD (EFD) time series were significant (P < 0.05). FFD and EFD time series (95.3 and 89.8%, respectively) responded negatively and significantly to preseason temperature (P < 0.05). The direction of FD changes varied among sites and species. On average, a shortening trend of FD was observed at Harbin (-0.51 days decade^-1), with 7.5% of species significantly. However, FD on average extended by 0.42 and 0.93 days decade^-1 at Beijing (24.5% significantly) and Xi'an (28.9% significantly), respectively. The regression models could simulate the interannual changes in FFD and EFD with the mean goodness of fit (R²) ranging from 0.37 to 0.67, but fail to simulate the changes in FD accurately (R² ranging from 0.09 to 0.18). The growing degree day model could improve the R² for simulating FFD and EFD except for FD. Therefore, more phenological models need to be tested, and more drivers of FD need to be further investigated.