**4. Discussion**

#### *4.1. Spatial–Temporal Patterns of Plant Phenology*

The time of the SOS experienced a significant downtrend (*slope* < 0), but the LOS increased over time (*slope* > 0) during 2001–2018 in the Yellow River Basin (Figure 5j,l). The research showed that with the increase of annual mean precipitation, temperature, relative humidity, shortwave radiation, soil temperature, and soil moisture in the Yellow River Basin during the SOS and LOS period, the time of SOS and LOS became earlier and longer (Figures S3 and S5). The favorable water and heat environment provided important resources for vegetative growth [37,38]. Water supply determines whether the photosynthesis occurs normally with an adequate CO2 concentration and sufficient light [24,39]. Meanwhile, water is also an indispensable intermediary used to ensure

nutrient substance transport [24,40]. Therefore, the increased humidity, precipitation, and soil moisture played a crucial role in the advance of the timing of the SOS and the prolonged nature of the LOS. In addition to water, temperature is also an indispensable factor in vegetative growth. An increase in temperature could facilitate vegetative growth if plants do not encounter water limitation [4,40]. Furthermore, climate warming can stimulate the enzymatic activities involved in photosynthesis [24,41], accelerate the mineralization and decomposition of organic matter [42], and extend the length of the vegetative growing season [7,43]. In general, the improvement of water and heat conditions visibly promoted the growth of vegetation in the TRHR.

The multi-year (2001–2018) average time of the SOS, EOS, and LOS in the TRHR presented a discrepant geographical pattern. In general, the time of the plant SOS was earliest in the Lancang and western Yellow river basins, while the time of the plant EOS was latest in the middle of the Yangtze River Basin. The duration of the vegetation growing season was longest in the Lancang and western Yellow river basins (Figure 5a–c). This phenomenon is closely related to the distribution of climate in the TRHR. The TRHR's climate is dominated by the East Asian monsoon, because the Himalayan Mountain Range obstructs the Indian monsoon [44,45] and causes a gradual reduction in precipitation, relative humidity, and soil moisture from southeast to northwest (Figures S2–S5). Likewise, vegetative growth is easily affected by climate change in the TRHR as also supported by previous studies [46,47]. Furthermore, the time of the plant SOS was delayed with an increase in elevation, but the times of the EOS or LOS were advanced or shortened, respectively, with an increase in elevation (Figure 6a–c). Possible reasons include the following: the ecosystems of high elevation areas are fragile, and the vegetative growth is vulnerable to extreme weather, such as extreme low temperature and frost. Another possibility is that the perennial snowfall occurring in high elevation regions causes low temperatures, which weaken the activity of soil microorganisms [14]. In contrast, with an increase in slope, the time of the SOS and EOS are in advance, while the LOS is prolonged (Figure 6d–f). The main reason for this result is that the areas with high slopes were mainly concentrated in the Lancang and the south part of the Yellow river basins, which have lower elevations (Figure S1). This provides reliable water and temperatures to guarantee the normal operation of vegetative photosynthesis. Finally, the vegetation of shady slopes started growing earlier in the growing season, ended later, and so had the longest growing season (Figure 6g–i), mainly due to the strong illumination and high temperature that accelerated soil organic matter mineralization and caused sunny slopes to have less soil moisture [48]. However, the shady slopes have soft solar radiation, moist soil, less moisture evaporation, and higher soil fertility [49,50].

#### *4.2. The Response of the Plant Phenology to Climate Change*

Our results illustrate that the variations in soil resources (e.g., pH and soil total N) that support vegetative growth, together with the climatic conditions that were suitable for vegetative growth, co-explained the phenological differences in plants from different basins. Specifically, the Yangtze River Basin is affected by the East Asian monsoon and elevation (Figure S1a), insufficient water supply, and relatively low temperatures, and low levels of soil nutrients constrained the growth of vegetation at the start of the growing season (Figures S2–S3). The air and soil temperatures are relatively low with less precipitation and soil moisture from April to May, which is not enough to support the transport of nutrients in plants, soil nutrient absorption by roots, and photosynthesis [24,51]. With the increased temperature, winter snow, permafrost, and glaciers have begun to melt slowly, and mineralization and decomposition of organic matter are accelerated [52,53] so that warmer temperatures provide plants with earlier opportunities to germinate [32]. However, the Yellow and Lancang river basins have higher temperatures and more shortwave radiation and precipitation than other areas owing to the lower elevation (Figures S1–S5). The increase in precipitation significantly increased soil carbon and N content, making it easier for plants to absorb nutrients owing to an increase in leaf stomatal conductance

and photosynthesis [24,54]. This may explain why temperature is more important for seed germination than water at the start of the growing season in the Yangtze River Basin, while water and heat are equally important for seed germination in the Yellow and Lancang river basins. In addition, the Yangtze River Basin lies in a high elevation area, which has thin air, strong solar radiation, and a long sunshine season (Figure S3). At high elevations, the decomposition of soil litter slows, which promotes the accumulation of organic matter due to the low temperatures caused by snow cover, which, in turn, slows the activity of soil microorganisms [14]. Thus, the growing season ends relatively late in the Yangtze River Basin. An interesting question arises: why were the soil factors having a greater impact on LOS in the Yangtze River Basin when compared with the Yellow and Lancang river basins (Table 1 and Figure 7)? Here, we propose one explanation. The precipitation, air temperature, relative humidity, soil moisture, and soil temperature showed a decreasing trend from southeast (the Yellow River Basin) to northwest (the Yangtze River Basin) in the TRHR because of the influence of the monsoon and elevation (Figures S2–S5). This situation led to low air temperature and less precipitation in the Yangtze River Basin, which does not provide enough energy for the growth of plants. At this time, the melting of permafrost and glaciers and the mineralization of soil organic matter provide energy for plant growth. However, the Yellow and the Lancang river basins have high air temperature, soil temperature, precipitation, and soil moisture, which can provide sufficient energy for plant growth.

#### *4.3. Limitations of the Current Study*

Despite the achievements in this study, large uncertainties still exist. In addition to NDVI, multiple vegetation indices can be used to reflect vegetation dynamics, such as EVI and LAI [1,34]. Note that the calculated plant phenology results may be vary based on the differences in resolution and quality of datasets using different vegetation indices. Furthermore, the present smoothing methods of remote sensing time series data have grea<sup>t</sup> differences in the model structure, which may result in grea<sup>t</sup> differences among the extracted plant phenology results [55,56]. Meanwhile, although the smoothing method used for the remote sensing time series is the same, different smoothing parameters also cause different results. Although the guidelines for some smoothing methods suggested using default parameter values when they were proposed, the best parameter values may vary because of the different growth trajectories of vegetation at specific sites, which lead to a difference in plant phenology in various regions and with different vegetation types [57,58]. Moreover, many methods can be used to extract plant phenological information, and they all have a certain level of applicability. Therefore, different methods may lead to different conclusions regarding the same question [56]. As mentioned above, it is necessary to further check whether the plant phenological results calculated from different datasets, smoothing methods of remote sensing time series, smoothing parameters, and phenology extraction methods provide the same or similar results and to improve the credibility of the results. Furthermore, there are many factors that affect plant phenology. Some changes in phenology are caused by climatic and soil factors; other decisive factors have shown effects on plant phenology, such as flash floods and extreme drought. Hence, more attention should be paid to the relationship between plant phenology and natural disasters in future studies.
