ARTICLE Received 3 Dec 2015 | Accepted 7 Jul 2016 | Published 18 Aug 2016 DOI: 10.1038/ncomms12489 OPEN Responses of sequential and hierarchical phenological events to warming and cooling in alpine meadows Xine Li1,*, Lili Jiang1,*, Fandong Meng1,*, Shiping Wang1,2,3,4,*, Haishan Niu5,*, Amy M. Iler6,*, Jichuan Duan7, Zhenhua Zhang2, Caiyun Luo2, Shujuan Cui1, Lirong Zhang1, Yaoming Li1, Qi Wang1,5, Yang Zhou1,5, Xiaoying Bao5, Tsechoe Dorji1, Yingnian Li2, Josep Peñuelas8,9, Mingyuan Du10, Xinquan Zhao2, Liang Zhao2 & Guojie Wang11 Organisms’ life cycles consist of hierarchical stages, from a single phenological stage (for example, flowering within a season), to vegetative and reproductive phases, to the total lifespan of the individual. Yet phenological events are typically studied in isolation, limiting our understanding of life history responses to climate change. Here, we reciprocally transfer plant communities along an elevation gradient to investigate plastic changes in the duration of sequential phenological events for six alpine species. We show that prolonged flowering leads to longer reproductive phases and activity periods when plants are moved to warmer locations. In contrast, shorter post-fruiting leaf and flowering stages led to shorter vegetative and reproductive phases, respectively, which resulted in shorter activity periods when plants were moved to cooler conditions. Therefore, phenological responses to warming and cooling do not simply mirror one another in the opposite direction, and low temperature may limit reproductive allocation in the alpine region. 1 KeyLaboratory of Alpine Ecology and Biodiversity, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China. 2 KeyLaboratory of Adaptation and Evolution of Plateau Biota, Northwest Institute of Plateau Biology, Chinese Academy of Sciences, Xining 810008, China. 3 CAS Center for Excellence in Tibetan Plateau Earth Science of the Chinese Academy of Sciences, Beijing 100101, China. 4 Naqu Integrated Observation and Research Station of Ecology and Environment, Tibet University and Institute of Tibetan Plateau Research of the Chinese Academy of Sciences, Lasa 850012, China. 5 University of Chinese Academy of Sciences, Beijing 100049, China. 6 Aarhus Institute of Advanced Studies, Aarhus University, Høegh-Guldbergs Gade 6B, Aarhus CDK-8000, Denmark. 7 Binhai Research Institute in Tianjin, Tianjin 300457, China. 8 CREAF, Cerdanyola del Vallès, Barcelona, Catalonia 08193, Spain. 9 CSIC, Global Ecology Unit CREAF-CEAB-CSIC-UAB, Cerdanyola del Vallès, Barcelona, Catalonia 08193, Spain. 10 Institute for Agro-Environmental Sciences, National Agriculture and Food Research Organization, Tsukuba 305-8604, Japan. 11 Oregon State University Agriculture and Natural Resource Program at Eastern Oregon University, La Grande, Oregon 97850, USA. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to S.P.W. (email: wangsp@itpcas.ac.cn). NATURE COMMUNICATIONS | 7:12489 | DOI: 10.1038/ncomms12489 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12489 I solated phenological events (for example, green-up, flowering) Total aboveground activity period have largely shaped our understanding of phenology as a bioindicator of climate change1–4; yet in reality, phenological events are inter-related and hierarchical. For example, flower bud Vegetative phase Reproductive phase formation, flowering and fruit/seed maturation compose the reproductive phase of angiosperms; furthermore, vegetative and Post-fruit reproductive phases compose the annual total aboveground Leaf out leaf Leaf colour Flower bud Flower Fruit activity period of plants in seasonal habitats (hereafter activity period; Fig. 1). Certain phenological stages within the plant life Figure 1 | Diagram of hierarchical phenological events measured in this cycle may limit later stages5,6 and therefore affect the study. Each phenological event was measured as the number days from responsiveness of other phenological stages to environmental start to end, or the duration of the event. The vegetative phase is the sum of change. Thus, assessing an entire sequence of phenological events the duration of individual-level vegetative events that occur before and after can reveal which individual stages have the largest effect on the reproductive phase. The activity period refers to the total length of the composite phenologies, and provide comprehensive insight into aboveground activity period. life history responses to climate change6–9. The degree to which plants respond to temperature changes by adjusting vegetative and reproductive phases should depend in reproductive phases and activity periods, but cooling significantly part on the effect of temperature on resource allocation to growth reduces both reproductive and vegetative phases and activity versus reproduction. In addition, temperature-dependent physio- periods. Our results show that the temperature sensitivities of logical processes can regulate the duration of plant life history flowering duration to warming and post-fruiting leaf duration to events. Plant growth and reproduction depend on the same cooling are greater compared with other phenological stages. internal resource pool10,11, and plants allocate a small percentage Thus, the responses of phenological stages to warming and of their total resources to sexual reproduction in arctic and alpine cooling do not simply mirror one another in the opposite regions12, presumably because they live in harsh environments direction, and low temperature may limit the reproductive with a short time window to complete their life cycles. Yet climate allocation in the alpine region. warming is resulting in a longer period of time that is conducive to growth and reproduction in many seasonal habitats13–16. Results Conversely, warmer environments may impose new constraints Temperature sensitivity of hierarchical phenological events. on the duration of phenological events if faster rates of Inter-related phenological events within each hierarchy were development occur under warmer temperatures8,17. Given the analysed using a repeated-measures multivariate analysis of harsh abiotic environment in alpine habitats, we expect that variance (MANOVA). The sensitivities of the activity period and warmer temperatures will reduce constraints on allocation to vegetative and reproductive phases to temperature were sexual reproduction, in which case we expect longer reproductive significantly affected by temperature change direction (that is, phases and activity periods. Following this same logic, we expect warming and cooling) and/or species and their interaction reproductive phases to be shortened more so than vegetative (Tables 1 and 2). Warming prolonged the activity period and the phases under cooling, for a total reduction in activity periods. reproductive phase, while cooling shortened the activity period In this study we reciprocally transferred intact plant commu- and both vegetative and reproductive phases for all species nities in 1 m by 1 m blocks of soil at 3,200, 3,400, 3,600 and (Fig. 2a–c). Warming significantly prolonged the activity period 3,800 m above sea level in the Qilian Mountains of China. Our by an average of 15.7 days °C  1 across species (Fig. 2c), and the experiment provides insight into how plants may adjust reproductive phase accounted for on average 83% of the increase sequences of hierarchical phenological events within the life cycle in the duration of the activity period under warming (range (Fig. 1) to warmer temperatures expected under future climate 69–99% across species; Fig. 2d). In contrast, cooling shortened the change in addition to temperature anomalies, such as short-term activity period by an average of 14.8 days °C  1 across species cooling events18,19. In addition, by examining plastic responses in (Fig. 2c), and the reproductive phase accounted for on average phenological duration to both warmer and cooler temperatures, 44.8% of the decrease (range 20–68% across species) in the our study improves our understanding of the basic role of duration of the activity period under cooling (Fig. 2d). temperature in shaping the relationships among phenological events in the plant life cycle, a need that has been highlighted by the widespread effects of climate change on phenological events20. Temperature sensitivity of individual phenological event. Plant Intact soil blocks were transferred across all elevations in 2007. species, temperature change direction and their interaction all The phenology of six common perennial plants was monitored significantly affected the sensitivity of duration to temperature from 2008 to 2010 at all sites. These plants included two early- in the six individual phenological stages (Table 3). Warming flowering sedge species, Kobresia humilis (Kh) and Carex significantly prolonged the duration of the reproductive phase scabrirostris (Cs); two mid-summer flowering forbs Potentilla (11.1–15.1 days °C  1 across species) (Fig. 2b), which was mainly anserine (Pa) and P. nivea (Pn); and two mid-summer flowering caused by the prolonged duration of flowering for all plant species grasses, Poa pratensis (Pp) and Stipa aliena (Sa). We measured (which ranged from 7.4 to 11.3 days °C  1) (Fig. 3a–f). Longer the durations of six phenological stages in tagged individual fruiting also contributed to longer reproduction in four of the six plants of each species: leaf-out, flower bud, flowering, fruiting, species (0–5.7 days °C  1) (Fig. 3c–f) and longer flower budding post-fruiting leaf and leaf colouring21 (Fig. 1) with 3–4 day in four of six species (Fig. 3a,c,e,f), but these stages always con- intervals at each elevation. We report a metric of phenological tributed less than the flowering stage. Warming also significantly sensitivity to temperature changes from the reciprocal transplants prolonged the duration of the vegetative phase in three species for each phenological event: change in duration divided by (1.7–5.5 days °C  1) (Fig. 2a), which primarily derived from the absolute change in temperature between each donor and receptor prolonged duration of leaf colouring (1.5–4.4 days °C  1) site (Ddays °C  1). (Fig. 3a,c–f; Cs is the exception (Fig. 3b)). In contrast, cooling Here we find that warming significantly prolongs the duration significantly shortened the duration of the vegetative phase of flowering, which is primarily responsible for longer (4.6–9.6 days °C  1) (Fig. 2a), which resulted primarily from the 2 NATURE COMMUNICATIONS | 7:12489 | DOI: 10.1038/ncomms12489 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12489 ARTICLE Table 1 | Temperature sensitivity of duration change for total aboveground activity period. Df Sum Sq Mean Sq F value Pr(4F) Year 1 134 134 0.861 0.354 Species (S) 5 1,309 262 1.677 0.138 TCD 1 325,116 325,116 2,083.094 o0.001 S * TCD 5 7,163 1,433 9.179 o0.001 ANOVA, analysis of variance; TCD, temperature change direction (that is, warming and cooling). Repeated-measures ANOVA is used to test the effects of year, species and temperature change direction on sensitivity of duration change to temperature for the total aboveground activity period. Table 2 | Temperature sensitivity of duration change for vegetative and reproductive phases. df Pillai approx F num df den df Pr(4F) Year 1 0.00248 0.65 2 526 0.521 Species (S) 5 0.04017 2.16 10 1054 0.0181 TCD 1 0.81048 1,124.71 2 526 o0.001 S * TCD 5 0.17802 10.30 10 1054 o0.001 MANOVA, multivariate analysis of variance; TCD, temperature change direction (that is, warming versus cooling). Repeated-measures MANOVA analysis of sensitivity of duration to temperature for both vegetative and reproductive phases is used as the multivariate response and species, year and TCD as predictors. a Cooling b Vegetative phase Reproductive phase Warming Sa ** ** Sa ** ** Pp ** ns Pp ** ** Pn ** ns Pn ** ** Pa ** ** Pa ** ** Cs ** ns Cs ** ** Kh ** ns Kh ** ** –25 –20 –15 –10 –5 0 5 10 15 20 25 –25 –20 –15 –10 –5 0 5 10 15 20 25 Δdays °C–1 Δdays °C–1 c d Ratio of reproductive phase (RP) Total activity period and total activity period (TAP) Sa ** ** Sa ** Pp ** ** ** Pp ** ** Pn ** ** Pn ** ** Pa ** ** Pa ** ** Cs ** ** Cs ** ** Kh ** ** Kh ** ** –25 –20 –15 –10 –5 0 5 10 15 20 25 0.00 0.25 0.50 0.75 1.00 1.25 Δdays °C–1 Ratio of RP and TAP Figure 2 | Temperature sensitivity of duration change of phenological events. The panels include the vegetative phase (a), reproductive phase (b), activity period (c) and their ratios of reproduction to activity period (d) under warming and cooling conditions. A t-test was used for each phenological event and species, and the a was adjusted following a Bonferroni correction. The mean value of phenological sensitivity in each plot in each year was the unit of replication (n ¼ 45 for each t-test under warming and cooling conditions, respectively). Kh, K. humilis; Cs, C. scabrirostris; Pa, P. anserine; Pn, P. nivea; Pp, Poa pratensis; and Sa, S. aliena. Ddays was the difference between receptor site (that is, away site) and donor site (that is, home site). Positive values indicate a prolonged phenological stage and negative values a shortened phenological stage when transferred compared with donor site. Means±s.e. are shown in the figure. ‘*’, ‘**’ in the figure are the significant difference at 0.004 and 0.0008 levels between receptor site and donor site. All P values were o0.0001 except vegetative phase for Kh, Cs, Pn and Pp under warming, which were 0.031, 0.727, 0.002 and 0.101, respectively. NATURE COMMUNICATIONS | 7:12489 | DOI: 10.1038/ncomms12489 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12489 Table 3 | Temperature sensitivity of duration change for six individual phenological stages. df Pillai approx F num df den df Pr(4F) Year 1 0.01675 1.48 6 522 0.1821 Species (S) 5 0.33889 6.37 30 2,630 o0.001 TCD 1 0.82313 404.89 6 522 o0.001 S * TCD 5 0.53236 10.45 30 2,630 o0.001 MANOVA, multivariate analysis of variance; TCD, temperature change direction (that is, warming versus cooling). Repeated-measures MANOVA analysis of sensitivity of duration to temperature for the six individual phenological stages is used as the multivariate response and species, year and TCD as predictors. a Cooling Warming b Kh Cs FR ** ns ** ns RP FL ns ** ns ** FB ns * ns ns LC ns ** ns ns VP PFL ** ns ** ns LO ns ns ns ns 0 5 0 5 –5 10 15 –5 10 15 –15 –10 –15 –10 c d FR Pa * ** Pn * ** RP FL ** ** ** ** FB ** * ** ns LC ns ** ns * VP PFL ** ns ** ns LO ns ns ns ns 0 5 0 5 –5 10 15 –5 10 15 –15 –10 –15 –10 e Pp f FR ** Sa ** ** ** RP FL ** ** ** ** FB * ** * ** LC * ** ns ** VP PFL * ns ** * LO * ** ns ns 0 5 0 5 –5 10 15 –5 10 15 –15 –10 –15 –10 Δdays °C–1 Δdays °C–1 Figure 3 | Temperature sensitivity of duration change of individual phenological stages. The panels include six phenological stages of each species under warming and cooling (Ddays °C  1). VP, vegetative phase; RP, reproductive phase; LO, leafing-out; PFL, post-fruiting leaf; LC, leaf colouring; FB, fruit bud; FL, flowering; FR, fruit-set/seeding. Ddays was the difference between receptor site (that is, away site) and donor site (that is, homesite). Positive and negative values were prolonged and shortened phenological duration when transferred compared with donor site, respectively. Kh, K. humilis (a); Cs, C. scabrirostris (b); Pa, P. anserine (c); Pn, P. nivea (d); Pp, Poa pratensis (e); and Sa, S. aliena (f). A t-test was used for each phenological event and species, and the a was adjusted following a Bonferroni correction. Since 12 tests were presented in figures, the a level for each individual test was set to 0.05/12. Significant differences were reported at Po0.004 and 0.0008 levels in the text. The mean value of phenological sensitivity in each plot in each year was the unit of replication (n ¼ 45 for each t-test under warming and cooling, respectively). Means±s.e. are shown in a–f. ‘*’ and ‘**’ are significant difference at 0.004 and 0.0008 levels between receptor site and donor site. ns means no significant difference at 0.004 level. All P values were shown in Supplementary Table 1. shortening of the post-fruiting leaf stage for all species (5.1–10.1 Correlations among phenological events. There were significant days °C  1) (Fig. 3a–f). A shortened duration of fruit-set con- negative correlations between the sensitivity of the vegetative tributed to shorter reproductive phases (2.2–6.0 days °C  1) (VP) and reproductive phases (RP) to warming and cooling under cooling for all plant species (Fig. 3a–f), although for Pa and regardless of whether the analysis was pooled across all species Pn, shorter flowering stages contributed more than fruiting (Table 4) or conducted separately for each species (Fig. 3c,d). (Supplementary Tables 2–13; except for Pp (Supplementary 4 NATURE COMMUNICATIONS | 7:12489 | DOI: 10.1038/ncomms12489 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12489 ARTICLE Table 4 | Correlations among temperature sensitivities of phenological duration. Leaf-out Flower bud Flower Fruit Post-fruit leaf Leaf colour VP RP Warming Flower bud  0.12 Flower  0.60** 0.28** Fruit  0.14 0.55** 0.03 Post-fruit leaf  0.41**  0.25**  0.08  0.10 Leaf colour  0.18  0.07  0.04 0.05 0.05 VP 0.37**  0.31**  0.55**  0.14 0.37** 0.63** RP  0.48** 0.72** 0.70** 0.70**  0.17  0.01  0.50** AP  0.31** 0.62** 0.44** 0.71** 0.05 0.40** 0.09 0.82** Cooling Flower bud 0.18 Flower  0.38** 0.15 Fruit 0.06 0.28** 0.19* Post-fruit leaf  0.31**  0.26** 0.13 0.04 Leaf colour  0.29**  0.20  0.11  0.47**  0.29** VP 0.04  0.33**  0.16  0.26** 0.73** 0.21* RP  0.10 0.54** 0.66** 0.81** 0.01  0.41**  0.34** AP  0.06 0.25** 0.49** 0.56** 0.65**  0.21* 0.47** 0.67** AP, activity period; RP, reproductive phase; VP, vegetative phase. * and ** indicate significant correlations at Po0.0014 and 0.0003 levels, respectively. The temperature sensitivity of each phenological event is included under warming and cooling, across all three years of the study (2008–2010). All species were pooled for ease of interpretation, because results are largely consistent if each species is analysed separately (Supplementary Tables 1–12). The sample unit is the mean sensitivity value for each species in each plot and year (n ¼ 270 for warming and for cooling, respectively). Table 11) and Sa (Supplementary Table 13) under cooling). In Most studies understandably focus on the effects of warming other words, high sensitivities of RP to temperature correspond to on plant phenophases in response to climate change, but in our low sensitivities of VP to temperature and vice versa. Significant study we also address a need for more basic research on negative correlations were found between some individual stages relationships among inter-related phenological events23 by asking within VP, and between them and individual stages within RP whether responses to cooling are simply the opposite of responses under warming and cooling (Table 1), which may be indicative of to warming or whether events respond differentially. The tradeoffs among sequential events. However, positive correlations outcome depends on hierarchy level, which highlights the among individual reproductive stages shows that high sensitivities utility of considering phenological sequences within a in flowering are correlated with high sensitivities in fruiting and hierarchical framework. For the upper hierarchies of the activity flower budding when pooled across all species (Table 1). The period, reproductive phase and vegetative phase, it appears that exceptions to this pattern are significant negative correlations opposite responses are indeed observed under warming versus between sensitivities of fruit bud and flower durations for Kh cooling (longer versus shorter). However, cooler temperatures do under warming (Supplementary Table 2) and for Pa under both not simply have the opposite effect of warmer temperatures on warming and cooling (Supplementary Tables 6 and 7), and the duration of individual life history events, especially within the between sensitivities of flower and fruit durations for Pa vegetative phase. Therefore, it should not be assumed that effects (Supplementary Tables 6 and 7) and Pn under cooling of cooler temperatures on the duration of phenological events will (Supplementary Table 9), respectively. simply mirror responses to warmer temperatures in the opposite direction. The onset of reproduction signals a change in resource Discussion allocation from growth and survival to fecundity within the Our reciprocal transfer experiment revealed substantial plasticity plant life cycle24,25. Longer reproductive phases under warming in the duration of phenological events in six alpine plant species and shorter phases under cooling suggest that allocation to and allowed us to identify the individual stages that drive changes reproduction is probably increased and reduced according to in composite phenophases. In all three years of the study, moving temperature in these alpine plants. Our results are consistent with plants down in elevation to warmer temperatures prolonged the observation that alpine plants allocate a small proportion of activity periods of individuals in all six species, mainly because of their resources to sexual reproduction under harsh environmental longer reproductive phases; in particular flowering. Conversely, conditions12. Thus we find no evidence of a faster rate of moving plants up in elevation to cooler temperatures shortened development of reproductive structures under warmer both vegetative and reproductive phases, for shorter annual temperatures as seen in other studies8,17. Instead, our results activity periods in individuals of all six species. These patterns are support the hypothesis that reproductive allocation in these consistent despite the inclusion of three different plant life forms plants is generally constrained by cold-temperature conditions in in our study: sedges, grasses and forbs, which is suggestive of a the alpine environment. common response to the alpine environment. Unlike other Longer reproduction and activity periods under warmer experimental studies in high-elevation or high-latitude sites, we temperatures were primarily driven by longer flowering stages found no evidence of lag or build-up effects of warming on plant in all species (Fig. 2d). The flowering stage accounted for 70% of phenology14,22. Instead, sensitivity of duration to temperature the extension to the reproductive phase under warming and for was consistent across years in all species, despite interannual 58% of the extension to the activity period on average across variation in temperature (for example, 2008 was cooler than 2009 species. Advanced first flowering appears to be the main and 2010 by 0.6–0.7 °C on average)3. contribution to prolonged flowering duration under warming, NATURE COMMUNICATIONS | 7:12489 | DOI: 10.1038/ncomms12489 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12489 as opposed to later last flowering4. Some experiments have found post-fruit leaf stage. Leaf senescence is an integrated response of either longer flowering or maintenance of flowering duration leaf cells to endogenous developmental and external environ- under warmer temperatures, depending on the species17,26–28, mental conditions49, and our results are consistent with another whereas other experiments have found a shorter duration of study that found longer leaf coloration periods in deciduous trees flowering and other individual reproductive stages under under warmer fall temperatures50. Longer growing seasons often warming6,22,29. This variation across species in how the lead to increased primary production and carbon seques- duration of flowering responds to warming may in part reflect tration13,51,52. However, longer activity periods may not different warming methods (open-top chambers versus infrared translate into increased carbon sequestration in our study heaters versus transplants) in addition to ecosystem type— because senescence was the main driver of longer vegetative specifically the timing of abiotic extremes that may constrain phases and longer reproduction was the main driver of longer individual stages within a growing season. For example, activity periods under warming. Furthermore, the leaf-out stage divergence in entire reproductive phases and flowering stages at was only longer under warming in one species. The balance the species level has been detected in two habitats with extreme between respiration and photosynthesis under elevated abiotic conditions in the middle of the growing season—high temperatures will ultimately determine whether longer activity temperatures and drought—so that a mid-season gap in periods result in increased or decreased primary production53. reproduction appears at the community level29,30. In sum, Temperature is not the only factor that varied over the plastic responses to climate change are likely to include elevation gradient in our study and is only one of several factors phenological shifts away from parts of the season where abiotic that may affect plant phenology under climate change54. conditions are becoming more extreme and toward parts of the Although higher elevation sites tended to be drier, the 3,600 m season where abiotic extremes are being attenuated. site had the highest soil moisture. Thus, our conclusions of the The response of flowering to warming may be attributed to effects of temperature are unlikely to be confounded by genetic and physiological mechanisms, and the genetic basis of concurrent changes in soil moisture because these two abiotic flowering is the most well understood plant life history phase31,32. factors do not precisely covary across our four sites. Although The onset of flowering occurs earlier under warming and later photoperiod can play a role in the regulation of alpine plant life under cooling in these six species3, which agrees with laboratory history events55, photoperiod does not vary across our sites that research on genetic pathways regulating how the timing of are separated by a maximum distance of 9 km. Increasing flowering responds to temperature in Arabadopsis thaliana33–37. atmospheric levels of CO2 can also extend the duration of life Thus, the responses of flowering duration in this study are history events under warming, and will most likely affect probably mediated through similar pathways at the genetic level vegetative stages directly28. Additionally, the timing of spring that determine the switch to allocation to reproduction. However, snowmelt often predicts the timing of flowering in high-elevation in contrast to what we find here, a recent gene regulatory model and high-latitude environments2,26, but in our study area the predicts shorter flowering durations under climate warming in ground is usually snow-free unless there is a large snow event56; Anemone halleri, a perennial relative of A. thaliana38. We suspect thus, the timing of snowmelt is unlikely to be an important this discrepancy is related to effects of warmer temperatures on phenological cue in our study. Our experimental protocol vernalization requirements35,36,38, which appear not to play a role controlled for plant community context and below-ground soil over the range of temperature variation in our study but could be communities, but other biotic factors may vary across the sites. important under continued climatic warming. For example, the activity and effectiveness of animal pollinators The population and community-level consequences of may have varied across the sites, but pollinators would only affect extended phenological stages under climate change will depend the reproduction of the two forb species, and tradeoffs between on biotic interactions with antagonists and mutualists, in addition pollinator visitation rates and pollinator effectiveness across to which type of interaction has a stronger effect on plant fitness. elevation gradients can compensate for such variation57,58. Longer flowering durations in individual plants may reduce the Finally, we focused on the duration of events at the individual likelihood of phenological mismatch with animal pollinators39–43 plant-level in this study, but the hierarchical nature of and may also compensate or buffer against potentially declining phenological events can be categorized in various ways. pollinator populations under climate change44,45. Longer For example, measuring the duration of individual leaves and flowering durations of individuals may also alter patterns of flowers would provide further insight into the consequences of assortative mating, with potential consequences for additive phenological responses to temperature change23,59. genetic variation within the population, depending on whether A main goal of ecological research is to understand what drives synchrony among individuals is maintained or increased46,47. the abundance and distribution of species in space and time, and Longer activity periods could increase exposure of plants to climate change presents ecologists with a unique opportunity to herbivory and disease in both vegetative and floral tissue. Indeed, advance basic research in temporal ecology. Rapidly shifting over longer time scales that span multiple plant generations, species’ phenologies under climate change have brought to light biotic interactions like the ones described above are likely to drive that we still need an improved understanding of how various adaptive responses in phenological duration48, which could differ abiotic and biotic factors affect the duration of life cycle events in from the short-term plastic responses reported in our study. order to predict species’ responses to climate change. Extended In contrast to the individual reproductive stages that largely phenological durations may or may not be adaptive under climate changed in concert with one another under warming and cooling, change, but in our study species, the duration of flowering, post- different individual vegetative stages accounted for the effects of fruiting leaf, and senescence should be particularly sensitive to warming and cooling on the duration of the vegetative phase climate change in the near future. (Table 1). In general, leaf senescence was extended under warming, whereas the post-fruit leaf stage was shortened under cooling (Fig. 3). Shorter post-fruit leaf stages are consistent with a Methods strategy of maintaining allocation to growth and reproduction Experimental site and data collection. The experimental site and design are described elsewhere in detail3,21. In brief, the experiment was conducted at Haibei under cooling, instead of allocating to tissue maintenance Alpine Meadow Ecosystem Research Station of the Chinese Academy of Sciences, following reproduction, although in two species the fruiting located at latitude 37°370 N, and longitude 101°120 E along a 3,200–3,800 m or flowering stages were reduced in a similar magnitude to the elevational gradient (3,200, 3,400, 3,600 and 3,800 m) on the south slope of the 6 NATURE COMMUNICATIONS | 7:12489 | DOI: 10.1038/ncomms12489 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12489 ARTICLE Qilian Mountains in Qinghai, China. The four sites included four different plant duration of the activity period to temperature (Table 1). The base package of communities within 9 km of one another (maximum distance between lowest and R-3.2.3 (ref. 61) was used to conduct statistical analyses. highest elevation). Although the dominant plant species vary across these four The effects of cooling and warming on sensitivity of duration for different communities, the six study species are commonly found at all four sites (for more phenological events were analysed by comparison to the controls (the plots details see3,21). At the centre of each site, HOBO weather stations (Onset Computer transferred within the same elevations). A t-test was used for each phenological Corporation, Cape Cod, Massachusetts, USA) were used to monitor soil event and species, and the a was adjusted following a Bonferroni correction (Figs 2 temperature at 5 cm depth and soil moisture at 20 cm depth. Data were sampled at and 3). Since 12 tests were presented both in Figs 2 and 3, the a level for each 1 min intervals, and 30 min averages were stored in the data logger. Annual average individual test was set to 0.05/12. Significant differences were reported at Po0.004 soil temperatures at 5 cm depth were 3.9, 2.5, 2.0 and 0.4 °C, and annual average and 0.0008 levels in the text. The mean value of phenological sensitivity in each soil moistures at 20 cm depth were 11.8, 11.3, 12.7 and 10.2% at 3,200, 3,400, 3,600 plot in each year was the unit of replication (n ¼ 45 for each t-test under warming and 3,800 m during the experimental period, respectively. Photoperiod is consistent and cooling, respectively). across the four sites because of their close proximity3,21. Simple correlation analysis was performed among the sensitivities of duration Twelve intact soil blocks (100  100 cm wide  30–40 cm deep) (30 cm depth change to temperature for all phenological events to determine their relationships at 3,800 m due to shallower soil layer) with attached vegetation from each elevation in direction and magnitude for each species and across all species during the were reciprocally transferred across the elevation gradient after the soils started to experimental periods. Since 36 tests were presented in Table 1, the significance level thaw in early May, 2007 (12 soil blocks per elevation  4 elevations, for a total of for each individual test was set to 1/36, a conservative correction for multiple 48 soil blocks or plots). There were three replicate transfers from each elevation, comparisons (that is, significant correlations were reported at Po0.0014 and and these intact soil blocks were fully randomized throughout the study site. Three 0.0003 levels in the text). blocks from each elevation were removed and then reinstated at the same site to produce experimental control blocks that had been handled as similarly as possible to those blocks moved to other elevations. Thus, there were six levels of warming Data availability. The data that support the findings of this study are available (transferred down) and six levels of cooling (transferred up). from the corresponding author upon request. Measurement of phenological events. Six common perennial plant species from these blocks that are common at each of the four sites were chosen for monitoring References of multiple phenological events at each elevation during the growing seasons of 1. Wolkovich, E. M. et al. Warming experiments underpredict plant phenological 2008 to 2010 (refs 3,21). The plant species observed were two early-spring responses to climate change. Nature 485, 494–497 (2012). flowering sedge species (K. humilis (Kh) and C. scabrirostris (Cs)), which flower 2. Iler, A. M., Høye, T. T., Inouye, D. W. & Schmidt, N. M. Nonlinear flowering before May, and four mid-summer flowering species (two grasses: Poa pratensis responses to climate: Are species approaching their limits of phenological (Pp) and S. aliena (Sa) and two forbs: P. anserine (Pa) and P. nivea (Pn)), which change? Philos. Trans. R. Soc. Lond. B Biol. Sci. 368, 20120489 (2013). flower between late June and July. Ten individuals for forbs and 10 stems for 3. Wang, S. P. et al. Asymmetric sensitivity of first flowering date to warming and graminoids for each plant species in each plot were marked during autumn 2007, cooling in alpine plants. Ecology 95, 3387–3398 (2014). so that individual plants could be followed throughout the growing season. The 4. Jiang, L. L. et al. Relatively stable response of fruiting stage to warming same individuals were monitored from 1 year to the next. Observations were made and cooling relative to other phenological events. Ecology 97, 1961–1969 at an interval of 3 or 4 days from early April to the end of October in each year, and (2016). recordings were made of the onset and end dates of six phenological events, listed 5. Sola, A. J. & Ehrlén, J. Vegetative phenology constrains the onset of flowering in here in chronological order: leaf emergence, flower budding, flowering, fruit-set for the perennial herb Lathyrus vernus. J. Ecol. 95, 208–216 (2007). forbs or seed-set for graminoids, post-fruit/seeding vegetative stage and leaf- 6. Post, E. S., Pedersen, C., Wilmers, C. C. & Forchhammer, M. C. Phenological colouring (number of days from first yellowing or browning of leaves to complete sequences reveal aggregate life history response to climatic warming. Ecology discoloration). Because phenology was assessed at 3–4 day intervals, onset and end 89, 363–370 (2008). dates of each phenological event were estimated; for onset, we used the average of 7. Donohue, K. Germination timing influences natural selection on life history the first date on which the event was recorded and the date of the prior census in which the event was not recorded (vice versa for end date). For example, if flow- charactersin Arabidopsis thaliana. Ecology 83, 1006–1016 (2002). ering was observed on 5 June for the first time, and the census before that on 1 June 8. Haggerty, B. P. & Galloway, L. F. Response of individual components of had no flowering, the estimated onset date was 3 June. The duration of each reproductive phenology to growing season length in a monocarpic herb. J. Ecol. phenological event was the number of days between the estimated starting and 99, 242–253 (2011). ending date of the event for each individual plant or stem. We calculated a 9. Sherry, R. A. et al. Changes in duration of reproductive phases and lagged plot-level average duration across the 10 individuals or stems for each phenological phenological response to experimental climate warming. Plant Ecol. Divers. 4, event and species. Note that unlike the reproductive stage, the vegetative phase is 23–35 (2011). not continuous and is the sum of the duration of each of the three vegetative stages 10. Nicotra, A. B. Reproductive allocation and the long-term costs of reproduction in in Fig. 1. Leaf colouring represents the switch from growth to dormancy, so we Siparuna grandflora, a dioecious neotropical shrub. J. Ecol. 87, 138–149 (1999). classify it as a vegetative stage that determines the end of the activity period 11. Aĺvarez-Cansino, L., Zunzunegui, M., Barradas, M. C. D. & Esquivias, M. P. (consistent with ref. 56). The reproductive phase excluded the event of seed Gender-specific costs of reproduction on vegetative growth and physiological ripeness because this was difficult to monitor in the field. The activity period performance in the dioecious shrub Corema album. Ann. Bot. 106, 989–998 represents the number of days from the earliest date on which a leaf was observed (2010). to the latest date on which a leaf retained colour. No data were obtained at 3,600 m 12. Molau, U. Relationships between flowering phenology and life-history in 2010 because mice destroyed the plots. strategies in tundra plants. Arct. Alp. Res. 25, 391–402 (1993). 13. Myneni, R. B., Keeling, C. D., Tucker, C. J., Asrar, G. & Nemani, R. R. Increased plant growth in the northern high latitudes from 1981 to 1991. Nature 386, Data analysis. The response variable in our analysis is a measure of the sensitivity 698–702 (1997). of phenological duration to temperature changes from the reciprocal transplants: 14. Arft, A. M. et al. Responses of tundra plants to experimental warming: meta- change in duration per change in °C (Ddays °C  1). For example, if the difference analysis of the international tundra experiment. Ecol. Monogr. 69, 491–511 in mean flowering duration between a receptor and donor site is 5 days, this (1999). is divided by the temperature difference between the two sites, say 5 °C, for a 15. Oberbauer, S. F. et al. Phenological response of tundra plants to background sensitivity of 1 day °C  1. Each species-level response is an average across 10 climate variation tested using the International Tundra Experiment. Philos. individuals per plot, and this calculation was performed for each phenological Trans. R. Soc. Lond. B Biol. Sci. 368, 20120481 (2013). event for each species in every plot. 16. CaraDonna, P. J., Iler, A. M. & Inouye, D. W. Shifts in flowering phenology Phenological responses to warming and cooling within each hierarchy level reshape a subalpine plant community. Proc. Natl Acad. Sci. USA 111, were analysed with the method repeated-measures MANOVA (Tables 2 and 3), except total activity period, which was analysed with a repeated-measures ANOVA 4916–4921 (2014). (Table 1) because it is a single event. MANOVA investigates whether there are 17. Gugger, S., Kesselring, H., Stöcklin, J. & Hamann, E. Lower plasticity exhibited statistically significant mean differences among groups in a linear combination of by high- versus mid-elevation species in their phenological responses to several dependent variables60. Several phenological responses within each hierarchy manipulated temperature and drought. Ann. Bot. 116, 953–962 (2015). level are correlated with one another (Table 4 and Supplementary Tables 2–13), 18. Mayle, F. E. & Cwynar, L. C. Impact of the Younger Dryas cooling event upon indicating that a MANOVA is necessary to test for an overall effect of predictors on lowland vegetation of Maritime Canada. Ecol. Monogr. 65, 130–154 (1995). the multiple phenological responses. Pillai’s statistic (the default) was applied to the 19. Sirocko, F., Brunck, H. & Pfahl, S. Solar influence on winter severity in central MANOVA output. Only minor differences in results were observed with the use of Europe. Geophys. Res. Lett. 39, L16704 (2012). other statistics (Wilks and Hotelling–Lawley). Repeated-measures ANOVA was 20. Wolkovich, E. M., Cook, B. I. & Davies, T. J. Progress towards an used to estimate the contributions of the factors year, species and temperature interdisciplinary science of plant phenology: building prediction across space, change direction (two levels, that is, warming and cooling) to the sensitivity of the time and species diversity. New Phytol. 201, 1156–1162 (2014). NATURE COMMUNICATIONS | 7:12489 | DOI: 10.1038/ncomms12489 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12489 21. Wang, S. P. et al. Timing and duration of phenological sequences of alpine 50. Archetti, M., Richardson, A. D., O’Keefe, J. & Delpierre, N. Predicting climate plants along an elevation gradient on the Tibetan plateau. Agric. For. Meteorol. change impacts on the amount and duration of autumn colors in a New 189-190, 220–228 (2014). England forest. PLoS ONE 8, e57373 (2013). 22. Hoffmann, A. A. et al. Phenological changes in six Australian subalpine plants 51. Nemani, R. R. et al. Climate-driven increases in global terrestrial net primary in response to experimental warming and year-to-year variation. J. Ecol. 98, production from 1982 to 1999. Science 300, 1560–1563 (2003). 927–937 (2010). 52. Piao, S. L. et al. Net carbon dioxide losses of northern ecosystems in response to 23. Bolmgren, K. & Eriksson, O. Are mismatches the norm? Timing of flowering, autumn warming. Nature 451, 49–52 (2008). fruiting, dispersal and germination and their fitness effects in Frangula alnus 53. Richardson, A. D. et al. Influence of spring and autumn phenological (Rhamnaceae). Oikos 124, 639–648 (2015). transitions on forest ecosystem productivity. Philos. Trans. R. Soc. Lond. B Biol. 24. Bazzaz, F. A. & Grace, J. Plant Resource Allocation (Academic Press, 1997). Sci. 365, 3227–3246 (2010). 25. Roff, D. A. Life History Evolution (Sinauer Associates, 2002). 54. Cleland, E. E., Chiariello, N. R., Loarie, S. R., Mooney, H. A. & Field, C. B. 26. Price, M. V. & Waser, N. M. Effects of experimental warming on plant Diverse responses of phenology to global changes in a grassland ecosystem. reproductive phenology in a subalpine meadow. Ecology 79, 1261–1271 (1998). Proc. Natl Acad. Sci. USA 103, 13740–13733 (2006). 27. Dunne, J. A., Harte, J. & Taylor, K. J. Subalpine meadow flowering phenology 55. Keller, F. & Körner, C. The role of photoperiodism in alpine plant development. responses to climate change: integrating experimental and gradient methods. Arct. Antarct. Alp. Res. 35, 361–368 (2003). Ecol. Monogr. 73, 69–86 (2003). 56. Dorji, T. et al. Plant functional traits mediate reproductive phenology and 28. Reyes-Fox, M. et al. Elevated CO2 further lengthens growing season under success in response to experimental warming and snow addition in Tibet. Glob. warming conditions. Nature 510, 529–262 (2014). Chang. Biol. 19, 459–472 (2013). 29. Sherry, R. A. et al. Divergence of reproductive phenology under climate 57. Kearns, C. A. Anthophilous fly distribution across an elevation gradient. Am. warming. Proc. Natl Acad. Sci. USA 104, 198–202 (2007). Mid. Nat. 127, 172–182 (1992). 30. Aldridge, G., Inouye, D. W., Forrest, J. R. K., Barr, W. A. & Miller-Rushing, A. J. 58. Bingham, R. A. & Orthner, A. R. Efficient pollination of alpine plants. Nature Emergence of amid-season period of low floral resources in a montane meadow 391, 238–239 (1998). ecosystem associated with climate change. J. Ecol. 99, 905–913 (2011). 59. Forrest, J. R. K. & Thomson, J. D. An examination of synchrony between insect 31. Andrés, F. & Coupland, G. The genetic basis of flowering responses to seasonal emergence and flowering in Rocky Mountain meadows. Ecol. Monogr. 81, cues. Nat. Rev. 13, 627–639 (2012). 469–491 (2011). 32. Obeso, J. R. The costs of reproduction in plants. New Phytol. 155, 321–348 60. Scheiner, S. M. & Gurevitch, J. (eds) Design and Analysis of Ecological (2002). Experiments (Oxford University Press, 2001). 33. Blázquez, M. A., Ahn, J. H. & Weigel, D. A thermosensory pathway controlling 61. R Development Core Team. A Language and Environment for Statistical flowering time in Arabidopsis thaliana. Nat. Genet. 33, 168–171 (2003). Computing (R Foundation for Statistical Computing, 2015). 34. Kumar, S. V. et al. Transcription factor PIF4 controls the thermosensory activation of flowering. Nature 484, 242–245 (2012). 35. Lee, J. H. et al. Role of SVP in the control of flowering time by ambient Acknowledgements This work was supported by projects from the National Science Foundation of China temperature in Arabidopsis. Genes Dev. 21, 397–402 (2007). (41230750), National Basic Research Program (2013CB956000), the Strategic Priority 36. Liu, J. Z., Feng, L. L., Li, J. M. & He, Z. H. Genetic and epigenetic control of Research Program (B) of the Chinese Academy of Sciences (XDB03030403), National plant heat responses. Front. Plant Sci. 6, 267 (2015). Science Foundation of China (31272488 and 31470524) and the European Research 37. Wigge, P. A. et al. Integration of spatial and temporal information during floral Council Synergy grant SyG-2013-610028 IMBALANCE-P. A.M.I. was funded by a induction in Arabidopsis. Science 309, 1056–1059 (2005). COFUND-Marie Curie fellowship. 38. Satake, A. et al. Forecasting flowering phenology under climate warming by modelling the regulatory dynamics of flowering-time genes. Nat. Commun. 4: 2303 (2013). Author contributions 39. Memmott, J., Craze, P. G., Waser, N. M. & Price, M. V. Global warming and the X.E.L., L.L.J., F.D.M., S.P.W., H.S.N. and A.M.I. contributed equally to the paper. S.P.W. disruption of plant-pollinator interactions. Ecol. Lett. 10, 710–717 (2007). and X.Q.Z. designed the research; J.C.D., Z.H.Z., C.Y.L., Y.N.L., M.Y.D. and L.Z. per- 40. Hegland, S. J., Nielsen, A., Lazaro, A., Bjerknes, A. L. & Totland, O. How does formed the research; X.E.L., F.D.M., H.S.N., L.R.Z., Y.M.L. and X.Y.B. analysed the data; climate warming affect plant-pollinator interactions? Ecol. Lett. 12, 184–195 and X.E.L., L.L.J., F.D.M., S.P.W., H.S.N., A.M.I., J.P. and G.J.W. wrote the paper. (2009). 41. Thomson, J. D. Flowering phenology, fruiting success and progressive deterioration of pollination in an early-flowering geophyte. Philos. Trans. R. Additional information Soc. Lond. B Biol. Sci. 365, 3187–3199 (2010). Supplementary Information accompanies this paper at http://www.nature.com/ 42. McKinney, A. M. et al. Asynchronous changes in phenology of migrating naturecommunications broad-tailed hummingbirds and their early-season nectar resources. Ecology 93, 1987–1993 (2012). Competing financial interests: The authors declare no competing financial interests. 43. Iler, A. M. et al. Maintenance of temporal synchrony between syrphid flies and Reprints and permission information is available online at http://npg.nature.com/ floral resources despite differential phenological responses to climate. Global reprintsandpermissions/ Change Biol. 19, 2348–2359 (2013). 44. Burkle, L. A., Marlin, J. C. & Knight, T. M. Plant-pollinator interactions over How to cite this article: Li, X. E. et al. Responses of sequential and hierarchical 120 years: loss of species, co-occurrence, and function. Science 339, 1611–1615 phenological events to warming and cooling in alpine meadows. Nat. Commun. 7:12489 (2013). doi: 10.1038/ncomms12489 (2016). 45. Høye, T. T. et al. Phenology of high-arctic butterflies and their floral resources: species-specific responses to climate change. Curr. Zool. 60, 243–251 (2014). This work is licensed under a Creative Commons Attribution 4.0 46. Fox, G. A. Assortative mating and plant phenology: evolutionary and practical International License. The images or other third party material in this consequences. Evol. Ecol. Res. 5, 1–18 (2003). article are included in the article’s Creative Commons license, unless indicated otherwise 47. Franks, S. J. The unique and multifaceted importance of the timing of in the credit line; if the material is not included under the Creative Commons license, flowering. Am. J. Bot. 102, 1401–1402 (2015). users will need to obtain permission from the license holder to reproduce the material. 48. Elzinga, J. A. et al. Time after time: flowering phenology and biotic interactions. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ Trends Ecol. Evol. 22, 432–439 (2007). 49. Lim, P. O., Woo, H. R. & Nam, H. G. Molecular genetics of leaf senescence in Arabidopsis. Trends Plant Sci. 8, 272–278 (2003). r The Author(s) 2016 8 NATURE COMMUNICATIONS | 7:12489 | DOI: 10.1038/ncomms12489 | www.nature.com/naturecommunications