Winter and spring warming result in delayed spring phenology on the Tibetan Plateau

Abstract

Climate change has caused advances in spring phases of many plant species. Theoretically, however, strong warming in winter could slow the fulfillment of chilling requirements, which may delay spring phenology. This phenomenon should be particularly pronounced in regions that are experiencing rapid temperature increases and are characterized by highly temperature-responsive vegetation. To test this hypothesis, we used the Normalized Difference Vegetation Index ratio method to determine the beginning, end, and length of the growing season of meadow and steppe vegetation of the Tibetan Plateau in Western China between 1982 and 2006. We then correlated observed phenological dates with monthly temperatures for the entire period on record. For both vegetation types, spring phenology initially advanced, but started retreating in the mid-1990s in spite of continued warming. Together with an advancing end of the growing season for steppe vegetation, this led to a shortening of the growing period. Partial least-squares regression indicated that temperatures in both winter and spring had strong effects on spring phenology. Although warm springs led to an advance of the growing season, warm conditions in winter caused a delay of the spring phases. This delay appeared to be related to later fulfillment of chilling requirements. Because most plants from temperate and cold climates experience a period of dormancy in winter, it seems likely that similar effects occur in other environments. Continued warming may strengthen this effect and attenuate or even reverse the advancing trend in spring phenology that has dominated climate-change responses of plants thus far.

Fig. 1.

Vegetation types of the Tibetan Plateau according to the Atlas of the Tibetan Plateau (16) and locations of weather stations in the region. (Inset) The location of the Tibetan Plateau within Central Asia.

Fig. 2.

Average timing of the beginning of the growing season (BGS) on the Tibetan Plateau between 1982 and 2006. BGS dates varied by up to 2 mo between different parts of the Plateau.

Fig. 3.

Beginning (BGS; A and B) and end (EGS; E and F) of the growing season for steppe (A and E) and meadow (B and F) vegetation on the Tibetan Plateau between 1982 and 2006, derived from 15-d NDVI composites obtained from the Advanced Very High Resolution Radiometer (AVHRR) sensor. BGS dates advanced markedly between 1982 and the mid 1990s, before retreating significantly after that. EGS dates in the steppe region advanced progressively between 1982 and the mid 1990s, whereas variation in the meadow region was less consistent. Consistent increases in temperature (C and D) and gradual and steady change in snow melt dates (G and H; following a delay between 1985 and the early 1990s in the steppe environment) indicate that observed changes are not linear responses to temperature or snow cover changes. Lines in the graph represent 3-y running means.

Fig. 4.

Response of the BGS (A–D) and EGS (E–H) in steppe and meadow vegetation of the Tibetan Plateau between 1982 and 2006 to monthly temperatures, according to PLS regression. For the BGS, the variable importance plots (VIP; C and D) indicate that temperatures in both spring (May and June) and winter (October through March) were important for explaining the response of BGS dates (VIP values above 0.8). Model coefficients (MC) of the centered and scaled data showed that warm winter temperatures delayed spring phenology (positive coefficients), whereas warm spring temperatures advanced the BGS (negative coefficients) for both steppe (A) and meadow (B). Including both effects into phenological models could substantially enhance our understanding of climate-change effects on vegetation at temperate and cold locations. For the EGS in the steppe region, the VIP (G) indicates that temperatures during most months contributed to explaining the variation in EGS dates. MC values (E) showed that warm temperatures in all of the important months contributed to an advance of the EGS date. This finding indicates that additional heat allows plants to complete their growth cycle earlier rather than extending the period, during which vegetation is active. EGS dates of meadow vegetation were mainly determined by temperatures in spring and September (H). Although temperature influence in spring is somewhat inconsistent (F), warm conditions in September delay the EGS date. In contrast to steppe, meadow vegetation can thus make use of the extended period of favorable conditions.

Fig. 5.

Illustration of the NDVI ratio method used to model the BGS and the EGS. BGS is assumed to occur when the BGS-specific threshold [NDVI_ratio (BGS)] is exceeded and followed by three consecutive data points with increasing NDVI ratio. EGS occurs when the NDVI ratio falls below the NDVI ratio (EGS) threshold, followed by three consecutive points of falling NDVI.