Global Change Biology Global Change Biology (2014) 20, 1473–1480, doi: 10.1111/gcb.12509
The influence of local spring temperature variance on temperature sensitivity of spring phenology 1 , S H U S H I P E N G 1 , 2 , I V A N A . J A N S S E N S 3 , X I N L I N 1 , TAO WANG1,2, CATHERINE OTTLE 1 B E N J A M I N P O U L T E R , C H A O Y U E 1 and P H I L I P P E C I A I S 1 1 Laboratoire des Sciences du Climat et de l’Environnement, CEA CNRS UVSQ, Gif-sur-Yvette 91191, France, 2CNRS and UJF Grenoble 1, UMR5183, Laboratoire de Glaciologie et Geophysique de l’Environnement (LGGE), 38041 Grenoble, France, 3 Department of Biology, University of Antwerp, Universiteitsplein 1, Wilrijk 2610, Belgium
Abstract The impact of climate warming on the advancement of plant spring phenology has been heavily investigated over the last decade and there exists great variability among plants in their phenological sensitivity to temperature. However, few studies have explicitly linked phenological sensitivity to local climate variance. Here, we set out to test the hypothesis that the strength of phenological sensitivity declines with increased local spring temperature variance, by synthesizing results across ground observations. We assemble ground-based long-term (20–50 years) spring phenology database (PEP725 database) and the corresponding climate dataset. We find a prevalent decline in the strength of phenological sensitivity with increasing local spring temperature variance at the species level from ground observations. It suggests that plants might be less likely to track climatic warming at locations with larger local spring temperature variance. This might be related to the possibility that the frost risk could be higher in a larger local spring temperature variance and plants adapt to avoid this risk by relying more on other cues (e.g., high chill requirements, photoperiod) for spring phenology, thus suppressing phenological responses to spring warming. This study illuminates that local spring temperature variance is an understudied source in the study of phenological sensitivity and highlight the necessity of incorporating this factor to improve the predictability of plant responses to anthropogenic climate change in future studies. Keywords: PEP725, spring phenology, temperature variance, temperature sensitivity Received 13 June 2013; revised version received 11 December 2013 and accepted 14 December 2013
Introduction Air temperature is the main environmental factor regulating the timing of spring events in temperate and boreal trees (e.g., Linkosalo et al., 2000), and these spring events are already showing an advancement that can be attributed to recent climatic warming (e.g., Schwartz & Reiter, 2000; Menzel et al., 2006a). But the magnitude and even direction of the advance rates are not uniform, which have been found to vary across within-species populations (e.g., Menzel et al., 2006b), species (e.g., Fitter & Fitter, 2002; Menzel, 2003; Dunnell & Travers, 2011), altitudes (e.g., Piao et al., 2011) and regions (e.g., Ahas, 1999; Schwartz & Reiter, 2000; Zhao & Schwartz, 2003; Ahas & Aasa, 2006). This large variation, on the one hand, is related to the fact that plant populations experience different degrees of climate change, e.g., a more rapid warming is occurring at high altitudes and latitudes. On the other hand, it can be attributed to their discrepancies in the response of the date of spring events to temperature (Menzel et al., 2006a; Bertin, 2008), which is expressed as the date of spring Correspondence: Tao Wang, tel. +33 1 6908 5221, fax +33 1 6908 7716, e-mail: [email protected]
© 2013 John Wiley & Sons Ltd
events change per degree Celsius of temperature change (or temperature sensitivity of spring phenology, days °C1). The variation in temperature sensitivity of spring phenology reflects its subjection to selection pressure (Bennie et al., 2010), which integrates local historic climate information thus evolving into its own optimal strategy under specific climate condition (Chabot & Hicks, 1982; Kikuzawa, 1989; Baliuckas & Pliura, 2003; Liang & Schwartz, 2013). In a fluctuating climate system, if spring phenology in trees has an instantaneous response to external temperature increase, the fatal consequences could be suffered possibly, i.e., due to elevated probability of tree exposure to frost (e.g., Bennie et al., 2010). Some mechanisms are thus developed to prevent the trees from sprouting or flowering at the ‘inappropriate’ time (Korner & Basler, 2010). To achieve this, the timing of spring events in the trees can only be responsive to temperature after fulfilling certain dose of low temperature (sufficient chilling, Murray et al., 1989; Morin et al., 2009; Schwartz & Hanes, 2010) or might evolve to be more dependent on other environmental cues (e.g., photoperiod, light quality). This leads to our hypothesis that the advancement of spring phenology 1473
1474 T . W A N G et al. responding to spring warming might decline at locations with large local variance in temperature. To test this hypothesis, data from within-species populations along an environmental gradient are mainly concerned. Previous studies (e.g., Kramer, 1995; Chuine et al., 2000; Berg et al., 2005; Vitasse et al., 2009, 2010) noticed stable temperature sensitivity among populations within a species, which is contrary to the diversified one found by others (e.g., Baliuckas & Pliura, 2003; Karlsson et al., 2003; Lu et al., 2006). However, the conclusion drawn from their studies is often based on a relatively small sample of population from the same species (Vitasse et al., 2010) and only several species being involved (e.g., Chuine et al., 2000). To avoid this, Pan European Phenological Database (PEP725) including many widely distributed species is used. The main objective of this study is to investigate influences of local spring temperature variance on spring phenological sensitivity to temperature at the species level.
Materials and methods
Pan European Phenology Database (PEP725) In 2004, the European funded ‘COST Action’ N°725 was running to establish a European Phenological data platform for climatological applications. The common dataset of COST725 hosted by ZAMG (ZentralAnstalt fur Meteorologie und Geodynamik), comprises 7.6 million data points in total from 15 countries plus International Phenological Gardens (IPG) from 7285 observation sites spanning from 1951 to 2000. In 2010, the successor project PEP725 (Pan European Phenology Database) funded by EUMETNET (EUropean national Meteorological Services NETwork) and ZAMG established an open access database, with plant phenology observations of 30 European countries (so far nine million of records collected) from 1951 onwards and develop quality checking procedures. In the PEP725 dataset, 19608 stations have been collected over 50 species from 30 European countries, covering
approximately 49% of Europe (http://www.pep725.eu). But most of the stations are concentrated in Germany (Fig. 1). In terms of mean annual climate, mean annual temperature (precipitation) ranges from 5 to 12 °C (from 550 to 1200 mm) over a majority of PEP725 stations. From a seasonality perspective, most PEP725 stations have a temperature seasonality (mean annual temperature range) of 16–25 °C and a precipitation seasonality (mean annual precipitation range divided by the mean annual precipitation) of 0.5–1.3. The PEP725 stations (mainly covering the continental Europe) have a mean climate typical of the temperate mid-latitudes but have a poor sampling in the relatively cold climates (e.g., north of 56 °N) and almost no sampling in warm and dry climates (e.g., subtropical, semiarid). Compared with the newly released phenological database across North America and Europe (NECTAR: Network of Ecological and Climatological Timings Across Regions), the PEP725 stations have a relatively limited temperature range and reduced precipitation and temperature seasonality. However, the PEP725 database includes many widely distributed species with broad ecological tolerances which may have phenologies that are phenotypically or genotypically plastic (e.g., Stinchcombe et al., 2004; Cook et al., 2012a). This characteristic well suits the purpose of this study. By contrast, NECTAR has relatively few sites (12 in total) and low replication of species across sites (Cook et al., 2012a), which, however, cannot satisfy our species selection criteria. The station information in PEP725 database includes latitude, longitude, altitude, and species. The species in this database can be grouped into two categories: wild and cultivated species (fruit trees and agricultural crops). In this study, we only focus on the wild species because the cultivated species could be highly genetically manipulated. The phenophases of plant species are assigned to a BBCH (Biologische Bundesanstalt, Bundessortenamt and Chemical industry) code (Meier, 2001). Among 28 phenophases recorded in the database, four spring phenophases are selected for this study (Table 1). For each spring phenophase, the stations having at least 25 of 61 years (1950–2010) are included in this study.
Climate datasets E-OBS is a daily gridded observational dataset for Europe with a spatial resolution of 0.25° grid for precipitation and temperature during the period 1950–2011. E-OBS displays an excellent correlation with the existing high-resolution regional gridded data for the United Kingdom and Alps that are based on much denser station networks (Hofstra et al., 2009), which forms the basis of this analysis. The temperature and precipitation fields Table 1 Spring phenophases in Pan European Phenological (PEP725) database
Fig. 1 Spatial distribution of stations from Pan European Phenological (PEP725) database.
FLS LFU FFO FFF
Spring phenophases
No. of species
First leaf separated Leaf unfolding (first visible leaf stalk) First flowers open Full flowering
4 13 25 3
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
T E M P E R A T U R E S E N S I T I V I T Y O F S P R I N G P H E N O L O G Y 1475 in E-OBS are interpolated from over 2000 stations based on ECA&D (the European Climate Assessment and Data set; http://eca.knmi.nl) and other existing datasets (Haylock et al., 2008). We match the PEP725 phenology records to daily ECA temperature and precipitation at the geographically closest grid cell. Without specific mention, daily climate data from the gridded E-OBS product is used hereafter. Compared with the station climate data, the potential shortcoming concerned with daily E-OBS gridded climate dataset is that the local extreme events may be smoothed during the spatial interpolation process (Hofstra et al., 2009). Given that the local extreme climate (e.g., warm temperatures) could have an influence on spring phenological responses, we also use freely available blended station climate data from ECA&D (1950–2010) to cross-validate the results based on daily gridded E-OBS product. Note that not all of station data used for generating daily gridded E-OBS can be freely available. This would lead to the difference in phenological sensitivity calculation between gridded and station data in this study. If station climate data from ECA&D was used, the field locations in PEP725 database meeting the following criteria are included in this study. Firstly, the distances between the field locations and their nearest meteorological stations from climate station are below 40 km. Secondly, the field locations have both specific spring phenophase data record and yearly temperature record with no 30-day period missing more than 10 days at least 25 of 61 years during the period 1950–2010 (Fig. 1).
Calculating temperature sensitivity of spring phenology The temperature sensitivity of spring phenology is commonly computed as linear slope coefficient between spring phenology and mean preseason temperature (described below) on the interannual timescale (e.g., Menzel et al., 2006b). In this study, the same method is adopted to calculate temperature sensitivity of spring phenology (referring to spring phenophase date) for each station in PEP725 database. It is well recognized that the onset date of spring phenology responds primarily to integrated climate forcing (e.g., cumulative heat sums) (McMaster & Wilhelm, 1997; Leon et al., 2001). Additionally, we calculate the phenological sensitivity as the linear slope coefficient between spring phenology and preseason growing degree day (GDD) summation. The GDD and GDDsum are computed as: GDD ¼ maxðTmean GDDthresh ; 0Þ X GDD GDDsum ¼
ð1Þ ð2Þ
where Tmean is daily mean temperature, GDDthresh is the temperature threshold for a day to qualify as a GDD (0 °C in our case), and GDDsum is the sum of GDD over the preseason period. The same conclusions are obtained if 5 °C for GDDthresh was used (data not shown). Preseason refers to the average or integrated climatic conditions exposed to the plant prior to the phenological event taking place. For each station, the preseason period is chosen to be 60 days, which is specified to end at the date that is calculated by averaging spring phenological dates from all years.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
Mean preseason temperature and preseason cumulative precipitation is then computed by averaging and summing daily temperatures and precipitation in a 60-day preseason period, respectively. We also conduct sensitivity analyses to evaluate the impacts of different preseason period lengths (45 and 90 days). Without specific mention, the length of preseason period is defined as 60-day hereafter and phenological sensitivity to temperature is computed based on mean preseason temperature. In addition, we use the magnitude of phenological sensitivity to denote the strength of sensitivity.
Analyses At each station, multiyear averaged standard deviation of daily detrended air temperature in a 60-day period centered over mean onset date of spring phenology (60-day centered period), which is used as spring SD (or local spring temperature variance) hereafter, is employed to represent local spring temperature variance. The same conclusions are reached if the length of the centered period was changed to 45 and 90 days (data not shown). We perform both simple and partial correlation analyses between phenological sensitivity and spring SD across stations belonging to the same species. Partial correlation statistically removes the effects of cross-station variation in mean temperature and cumulative precipitation within the 60-day centered period. Note that we only consider the species in PEP725 database for which the number of the stations is larger than 10. Previous studies have indicated that minimizing the frost damage in temperate and boreal biomes is an important external factor in shaping plant phenological traits (e.g., Lockhart, 1983; Linkosalo et al., 2000; Saxe et al., 2001; Bennie et al., 2010). In this study, we hypothesize that the potential mechanism for local spring temperature variance impacts on phenological sensitivity is possibly linked to plant avoidance of frost damage. In other words, regions with large local spring temperature variance tend to have a high probability of frost risk, and the living plants might suppress temperature responses of spring phenology to minimize the frost risk. To test these assumptions, for each species, we firstly perform partial correlation analysis between indicator of the frost risk and spring SD after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60day centered period. Secondly, simple regression analyses are conducted between phenological sensitivity and indicator of the frost risk. Moreover, partial regression analyses are also performed for each species between phenological sensitivity and indicator of the frost risk after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. In this study, the frost risk is assessed based on spring frost frequency. The number of frost days, which are defined as daily minimum air temperatures fall bellow 0 °C over the 60-day centered period, is counted for each year. Spring frost frequency is then calculated through dividing mean frost days by the length of the 60-day centered period. The frost risk mentioned in this study should be distinguished from the extent of frost damage. For example, species are often exposed to more frost
1476 T . W A N G et al. events during early spring phases than during later phases but conversely the extent of frost damage could be greater at later phases (Sakai & Larcher, 1987; Augspurger, 2013). In addition, the use of a universal threshold (0 °C in this study) to determine spring frost frequency is a bit arbitrary because the frost thresholds could vary with species and phenological phases (e.g., Augspurger, 2013). Simultaneous gathering field observations of temperature, phenology and frost damage over long-term periods are thus necessary in the future.
Results
The relationship between temperature sensitivity of spring phenology and local spring temperature variance Figure 2 displays the distribution of temperature sensitivity of spring phenology from PEP725 database. All spring phenophases at a majority of the stations show a negative response to an increase in mean preseason temperature (Fig. 2). A 1 °C increase in mean preseason temperature induces an earlier shift of 3.6 (2.8–4.3), 2.9 (2.1–3.6), 4.2 (3.2–5.2), 4.4 (3.4–5.4) days for First Leaf Separated (FLS), Leaf Unfolding (LFU), First Flowers Open (FFO) and Full Flowering (FFF), respectively. Spring SD for each phenophase is also shown in Fig. 2. In PEP725 database, the linear regression coefficients between phenological sensitivity and spring SD (or local spring temperature variance) at the species level are displayed in Fig. 3. Our simple linear regression analysis show that there are more positive coefficients than the negative ones and most of negative ones are non significant (Fig. 3a). Partial regression analysis
Fig. 2 The boxplot distribution of spring phenological sensitivity (days °C1) and spring SD (°C) from all stations within each of the four spring phenophases [First Leaf Separated (FLS), Leaf Unfolding (LFU), First Flowers Open (FFO) and Full Flowering (FFF)] from Pan European Phenological (PEP725) database. See Table 1 for explanation of FLS, LFU, FFO and FFF. The bottom and the top of the box denote the 25th and 75th percentiles, respectively, and the bold line within the box is the 50th percentile (the median). Spring phenological sensitivity is calculated based on mean preseason (60 days) temperature.
Fig. 3 (a) Simple and (b) partial regression coefficients between spring phenological sensitivity (days °C1) and spring SD (°C) across all stations within each species from Pan European Phenological (PEP725) database. The species is labeled by species identity (ID) (see Table S1). Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. Simple regression (SR) and partial regression (PR) denote simple and partial regression coefficients, respectively. Note that spring phenological sensitivity is calculated based on mean preseason (60 days) temperature.
after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period also supports this finding (Fig. 3b). For example, 33 positive coefficients (of which 25 are significant) of 45 are found between phenological sensitivity and spring SD, and 12 negative coefficients (of which five are significant) are detected (Fig. 3b). This result is robust to the following three cases: the phenological sensitivity was calculated based on mean preseason temperature using other preseason periods (45 and 90 days) (Figure S1), the climate data from blended meteorological stations archived in ECA&D dataset (Figure S2) were used and phenological sensitivity was calculated based on preseason (60 days) GDD summation (Figure S3). © 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
T E M P E R A T U R E S E N S I T I V I T Y O F S P R I N G P H E N O L O G Y 1477
Based on PEP725 database, we identify that spring frost frequency is also positively correlated with spring SD (or local spring temperature variance) for a majority of species after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period (Fig. 4b). In addition, the simple correlation coefficient between spring SD and spring frost frequency is also displayed in Fig. 4a. As shown in Fig. 4b, 34 positive coefficients (of which 33 are significant) of 45 are found between spring frost frequency and spring SD, and the remaining 11 negative coefficients (of which two are significant) are detected. Again, these results are robust to the
following cases: the phenological sensitivity is calculated based on mean preseason temperature using other preseason periods (45 and 90 days), the climate data from blended meteorological stations archived in ECA&D dataset are used and phenological sensitivity is calculated based on preseason (60 days) GDD summation (data not shown). We also investigate the responses of phenological sensitivity to spring frost frequency. Many species from PEP725 database display a positive correlation between phenological sensitivity and spring frost frequency (Fig. 5a). For example, 30 positive coefficients (of which 21 are significant) of 45 are found, and the remaining 15 negative coefficients (of which five are significant) are detected. After statistically controlling for cross-station variation in mean temperature and cumulative
Fig. 4 (a) Simple and (b) partial regression coefficients between spring SD (°C) and spring frost frequency (%) across all stations within each species from Pan European Phenological (PEP725) database. Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60day centered period. Simple regression (SR) and partial regression (PR) denote simple and partial regression coefficients, respectively.
Fig. 5 (a) Simple and (b) partial regression coefficients between spring phenological sensitivity (days °C1) and spring frost frequency (%) across all stations within each species from Pan European Phenological (PEP725) database. Partial regression coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. Note that spring phenological sensitivity is calculated based on mean preseason (60 days) temperature.
The link between spring frost frequency and local spring temperature variance
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
1478 T . W A N G et al. precipitation over the 60-day centered period, the number of species from PEP725 database with positive coefficients has been reduced (Fig. 5b).
Discussion
Temperature responses of spring phenology Our analysis based on ground observations undoubtedly supports the finding that the responsiveness of spring phenology is dependent on the preceding mean preseason temperature (e.g., Badeck et al., 2004; Menzel et al., 2006a; Piao et al., 2006; Cong et al., 2013) and also on preseason integrated temperature forcing (McMaster & Wilhelm, 1997; Leon et al., 2001). Using COST725 dataset (roughly the same with PEP725, see ‘Materials and methods’), Menzel et al. (2006a) found that panEuropean temperature sensitivity of spring phenophases (leaf unfolding and flowering) is 2.5 days °C1, which is roughly comparable with our estimate (ca. 3.8 days °C1). A difference of 1.3 days response per degree C between these two estimates might be related to the difference in temperature dataset and the time period defined for spring season. We notice delayed responses of spring phenology to mean preseason temperature (positive phenological sensitivity) in PEP725 database (not shown). Recent studies (Cook et al., 2012b; Pope et al., 2013) have suggested that species are not advancing in their spring phenological behaviors because they respond more to lack of winter chill than increased spring heat. This is also found by Zhang et al. (2007) based on NDVI in more southern North America, which related this delay to insufficient chilling in the warm winter years (Murray et al., 1989). Based on NDVI of steppe and meadow in Tibet Plateau, Yu et al. (2010) found a delay of spring phenology due to warm winter conditions, though some debates have been stirred (Zhang et al., 2007; Chen et al., 2011; Luedeling et al., 2011; Shen, 2011; Yi & Zhou, 2011). However, we should not rule out the possibility that some plants might only need very minor chilling requirements, as exemplified by Lilac (Syringa) (Larcher, 2006).
Impacts of local spring temperature variance on phenological sensitivity In terms of PEP725 database, this study suggests that the strength of phenological sensitivity tends to be smaller for species that are located in climates with larger local spring temperature variance (or higher spring SD). This result is also found if the climate data from blended meteorological stations archived in ECA&D dataset was used and phenological sensitivity was calculated based on preseason (60 days) GDD summation.
The present study is the first reported case that species might be less likely to keep climatic warming at locations with more fluctuating climate systems. This probably reflects the adaptive responses of spring phenology to local climate (e.g., Lockhart, 1983; Lechowicz, 1984; Menzel et al., 2006a; Doi & Takahashi, 2008; Pudas et al., 2008). The likely mechanism for this behavior may be related to plant avoidance of frost risk. Our analysis indicates a widespread increased spring frost frequency with increasing local spring temperature variance at the species level (Fig. 4). This suggests that the probability of plant exposure to frost risk increases as the local climate becomes more fluctuating (higher spring SD). We also notice a decreased strength in phenological sensitivity with increasing spring frost frequency for many species (Fig. 5a), although the number of species with positive (and significant) coefficients has been reduced after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period (Fig. 5b). Our results indicate that plants in regions with high spring SD may adopt a relatively conservative approach to minimize the risk of experiencing frost damage through suppressing the response of spring phenology to external temperature increase. The selection pressure from frost risk could probably drive the plants to be evolved into different strategies (e.g., low to high chill accumulation, photoperiod) for spring phenology, an interesting question to be explored in the future experimental work. In addition, our analyses also indirectly supports the mainstream model framework that spring phenology simulations in land surface models need consider other cues (e.g., photoperiod, chill requirements) to avoid the potential frost damage imposed by large local temperature variability. The implications of this study are manifold. According to our analyses, spring phenology will be less advanced in response to spring warming in the regions characterized by more fluctuating climate system. Moreover, both mean and variance in temperature, which are two opposite forces in changing the frost risk (an increase in spring temperature variance tends to increase the frost risk), are likely to increase in future climate scenarios. If our results were extrapolated from space to time, the response of spring phenology to temperature will be decreased in the future. However, this space-for-time extrapolation further awaits careful examination in future studies when much more longterm data will be available (e.g., Rutishauser et al., 2008, 2009). Our analyses indicate that the spring phenological sensitivity can be shaped through the mechanism that the plants can decrease the risk of frost damage to actively growing parts through genotypic © 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
T E M P E R A T U R E S E N S I T I V I T Y O F S P R I N G P H E N O L O G Y 1479 plasticity. We should not rule out the possibility that the triggering of spring phenology may also be influenced by the plant’s ability in maximizing the growing season for carbon gain (e.g., Bennie et al., 2010). If this occurred, in response to the mean temperature increase in the future, the increased strength in spring phenological sensitivity to temperature can benefit the plant fitness. These two processes can be considered as two opposing forces in affecting the evolution of phenological sensitivity in the future climate change and the investigation into their relative roles is beyond the scope of this study and is further needed in the future. In addition, the present study does not explore multidecadal variability in the response of temperature sensitivity to spring SD because of relatively short time series in ground observations. The implications of this work also highlight the need for considering not only mean temperature (Cook et al., 2012b) but also local spring temperature variance in understanding variations in phenological sensitivity. It can further contribute to improving the predictability of plant responses to anthropogenic climate change that has been demonstrated to be related to plant performances (Cleland et al., 2012). Finally, we should be informed that the positive correlation of phenological sensitivity with spring SD at the within-species level is not equivalent to the unbounded linear decrease in the strength of phenological sensitivity with spring SD. For example, based on the PEP725 database, Cook et al. (2012a) found that the strength of phenological sensitivity firstly decreased and then seemed to have ceased as the mean flowering (leafing) date continues to increase at the inter-species level. In addition, the conclusions drawn from this study are mainly based on the PEP725 database, which mainly covers the continental temperate climate with few sites in the cold climates and almost no site sampling in sub-tropical and semiarid climates. Further testing of the generality of the relationship between phenological sensitivity and spring SD will require observations of more species across a wider array of environmental conditions.
Acknowledgements We acknowledge the funding by the French ANR CLASSIQUE project. We greatly thank the editor and two anonymous reviewers for their constructive comments on the manuscript.
References Ahas R (1999) Long-term phyto-, ornitho- and ichthyophenological time-series analyses in Estonia. International Journal of Biometeorology, 42, 119–123. Ahas R, Aasa A (2006) The effects of climate change on the phenology of selected Estonian plant, bird and fish populations. International Journal of Biometeorology, 51, 17–26.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
Augspurger CK (2013) Reconstructing patterns of temperature, phenology, and frost damage over 124 years: spring damage risk in increasing. Ecology, 94, 41–50. Badeck F, Bondeau A, Bottcher K, Doktor D, Lucht W, Schaber J, Sitch S (2004) Responses of spring phenology to climate change. New Phytologist, 162, 295–309. Baliuckas V, Pliura A (2003) Genetic variation and phenotypic plasticity of Quercus robur populations and open-pollinated families in Lithuania. Scandinavian Journal of Forest Research, 18, 305–319. Bennie J, Kubin E, Wiltshire A, Huntley B, Baxter R (2010) Predicting spatial and temporal patterns of bud-burst and spring frost risk in north-west Europe: the implications of local adaptation to climate. Global Change Biology, 16, 1503–1514. Berg H, Becker U, Matthies D (2005) Phenotypic plasticity in Carlina vulgaris: effects of geographical origin, population size, and population isolation. Oecologia, 143, 220–231. Bertin R (2008) Plant phenology and distribution in relation to recent climate change. Journal of the Torrey Botanical Society, 135, 126–146. Chabot B, Hicks D (1982) The ecology of leaf life spans. Annual Review of Ecology and Systematics, 13, 229–259. Chen H, Zhu Q, Wu N, Wang Y, Peng C (2011) Delayed spring phenology on the Tibetan Plateau may also be attributable to other factors than winter and spring warming. Proceedings of the National Academy of Sciences of the United States of America, 108, E93–E93. Chuine I, Belmonte J, Mignot A (2000) A modelling analysis of the genetic variation of phenology between tree populations. Journal of Ecology, 88, 561–570. Cleland E, Allen J, Crimmins T et al. (2012) Phenological tracking enables positive species responses to climate change. Ecology, 93, 1765–1771. Cong N, Wang T, Nan H, Ma Y, Wang X, Myneni RB, Piao S (2013) Changes in satellite-derived spring vegetation green-up date and its linkage to climate in China from 1982 to 2010: a multimethod analysis. Global Change Biology, 19, 881–891. Cook B, Wolkovich E, Davies T et al. (2012a) sensitivity of spring phenology to warming across temporal and spatial climate gradients in two independent databases. Ecosystems, 15, 1283–1294. Cook B, Wolkovich E, Parmesan C (2012b) Divergent responses to spring and winter warming drive community level flowering trends. Proceedings of the National Academy of Sciences of the United States of America, 109, 9000–9005. Doi H, Takahashi M (2008) Latitudinal patterns in the phenological responses of leaf colouring and leaf fall to climate change in Japan. Global Ecology and Biogeography, 17, 556–561. Dunnell K, Travers S (2011) Shifts in the flowering phenology of the northern great plains: patterns over 100 years. American Journal of Botany, 98, 935–945. Fitter A, Fitter R (2002) Rapid changes in flowering time in British plants. Science, 296, 1689–1691. Haylock M, Hofstra N, Tank A, Klok E, Jones P, New M (2008) A European daily high-resolution gridded data set of surface temperature and precipitation for 1950–2006. Journal of Geophysical Research-Atmospheres, 113, D20119. Hofstra N, Haylock M, New M, Jones P (2009) Testing E-OBS European high-resolution gridded dataset of daily precipitation and surface temperature. Journal of Geophysical Research-Atmospheres, 114, D21101. Karlsson P, Bylund H, Neuvonen S, Heino S, Tjus M (2003) Climatic response of budburst in the mountain birch at two areas in northern Fennoscandia and possible responses to global change. Ecography, 26, 617–625. Kikuzawa K (1989) Ecology and evolution of phenological pattern, leaf longevity and leaf habit. Evolutionary Trends in Plants, 3, 105–110. Korner C, Basler D (2010) Phenology under global warming. Science, 327, 1461–1462. Kramer K (1995) Phenotypic plasticity of the phenology of 7 European tree species in relation to climatic warming. Plant Cell and Environment, 18, 93–104. Larcher W (2006) Altitudinal variation in flowering time of lilac (Syringa vulgaris L.) in the Alps in relation to temperatures. Oesterr Akad Wiss Math-Natwiss Kl Sitzungsber Abt I, 212, 3–18. Lechowicz MJ (1984) Why do temperate deciduous trees leaf out at different times adaptation and ecology of forest communities. American Naturalist, 124, 821–842. Leon AJ, Lee M, Andrade FH (2001) Quantitative trait loci for growing degree days to flowering and photoperiod response in sunflower (Helianthus annuus L.). Theoretical and Applied Genetics, 102, 497–503. Liang L, Schwartz M (2013) Testing a growth efficiency hypothesis with continentalscale phenological variations of common and cloned plants. International Journal of Biometeorology, doi: 10.1007/s00484-013-0691-6. Linkosalo T, Carter T, Hakkinen R, Hari P (2000) Predicting spring phenology and frost damage risk of Betula spp. under climatic warming: a comparison of two models. Tree Physiology, 20, 1175–1182. Lockhart J (1983) Optimum growth initiation time for shoot buds of deciduous plants in a temperate climate. Oecologia, 60, 34–37.
1480 T . W A N G et al. Lu P, Yu Q, Liu J, Lee X (2006) Advance of tree-flowering dates in response to urban climate change. Agricultural and Forest Meteorology, 138, 120–131.
Yu H, Luedeling E, Xu J (2010) Winter and spring warming result in delayed spring phenology on the Tibetan Plateau. Proceedings of the National Academy of Sciences of
Luedeling E, Yu H, Xu J (2011) Replies to Shen, Chen et al., and Yi and Zhou: linear regression analysis misses effects of winter temperature on Tibetan vegetation. Proceedings of the National Academy of Sciences of the United States of America, 108, E95–E95. McMaster GS, Wilhelm WW (1997) Growing degree-days: one equation, two interpretations. Agricultural and Forest Meteorology, 87, 291–300. Meier U (Ed.). (2001) Growth Stages of Mono- and Dicotyledonous Plants. BBCH Mono-
the United States of America, 107, 22151–22156. Zhang X, Tarpley D, Sullivan J (2007) Diverse responses of vegetation phenology to a warming climate. Geophysical Research Letters, 34, L19405. Zhao T, Schwartz M (2003) Examining the onset of spring in Wisconsin. Climate Research, 24, 59–70.
graph, 2nd edn. Federal Biological Research Centre of Agriculture and Forestry. Germany. Available at: http://www.bba.de/veroeff/bbch/bbcheng.pdf Menzel A (2003) Plant phenological anomalies in Germany and their relation to air temperature and NAO. Climatic Change, 57, 243–263. Menzel A, Sparks T, Estrella N et al. (2006a) European phenological response to climate change matches the warming pattern. Global Change Biology, 12, 1969–1976. Menzel A, Sparks T, Estrella N, Roy D (2006b) Altered geographic and temporal variability in phenology in response to climate change. Global Ecology and Biogeography, 15, 498–504. Morin X, Lechowicz M, Augspurger C, O’ Keefe J, Viner D, Chuine I (2009) Leaf phenology in 22 North American tree species during the 21st century. Global Change Biology, 15, 961–975. Murray M, Cannell M, Smith R (1989) Date of budburst of 15 tree species in Britain following climatic warming. Journal of Applied Ecology, 26, 693–700. Piao S, Fang J, Zhou L, Ciais P, Zhu B (2006) Variations in satellite-derived phenology in China’s temperate vegetation. Global Change Biology, 12, 672–685. Piao S, Cui M, Chen A, Wang X, Ciais P, Liu J, Tang Y (2011) Altitude and temperature dependence of change in the spring vegetation green-up date from 1982 to 2006 in the Qinghai-Xizang Plateau. Agricultural and Forest Meteorology, 151, 1599–1608. Pope K, Dose V, Da Silva D, Brown P, Leslie C, DeJong T (2013) Detecting non-linear response of spring phenology to climate change by Bayesian analysis. Global Change Biology, 19, 1518–1525. Pudas E, Leppala M, Tolvanen A, Poikolainen J, Venalainen A, Kubin E (2008) Trends in phenology of Betula pubescens across the boreal zone in Finland. International Journal of Biometeorology, 52, 251–259. Rutishauser T, Luterbacher J, Defila C, Frank D, Wanner H (2008) Swiss spring plant phenology 2007: extremes, a multi-century perspective, and changes in temperature sensitivity. Geophysical Research Letters, 35, L05703. Rutishauser T, Schleip C, Sparks T et al. (2009) Temperature sensitivity of Swiss and British plant phenology from 1753 to 1958. Climate Research, 39, 179–190. Sakai A, Larcher W (1987) Frost Survival of Plants: Responses and Adaptation to Freezing Stress. Springer-Verlag, New York, USA. Saxe H, Cannell M, Johnsen B, Ryan M, Vourlitis G (2001) Tree and forest functioning in response to global warming. New Phytologist, 149, 369–399. Schwartz M, Hanes J (2010) Continental-scale phenology: warming and chilling. International Journal of Climatology, 30, 1595–1598. Schwartz M, Reiter B (2000) Changes in north American spring. International Journal of Climatology, 20, 929–932. Shen M (2011) Spring phenology was not consistently related to winter warming on the Tibetan Plateau. Proceedings of the National Academy of Sciences of the United States of America, 108, E91–E92. Stinchcombe J, Weinig C, Ungerer M et al. (2004) A latitudinal cline in flowering time in Arabidopsis thaliana modulated by the flowering time gene FRIGIDA. Proceedings of the National Academy of Sciences of the United States of America, 101, 4712–4717. Vitasse Y, Delzon S, Dufrene E, Pontailler J, Louvet J, Kremer A, Michalet R (2009) Leaf phenology sensitivity to temperature in European trees: do within-species populations exhibit similar responses? Agricultural and Forest Meteorology, 149, 735–744. Vitasse Y, Bresson C, Kremer A, Michalet R, Delzon S (2010) Quantifying phenological plasticity to temperature in two temperate tree species. Functional Ecology, 24, 1211–1218. Yi S, Zhou Z (2011) Increasing contamination might have delayed spring phenology on the Tibetan Plateau. Proceedings of the National Academy of Sciences of the United States of America, 108, E94–E94.
Supporting Information Additional Supporting Information may be found in the online version of this article: Figure S1. Simple regression (SR) (a) and (c) and Partial regression (PR) coefficients (b) and (d) between spring phenological sensitivity (days °C1) and spring SD (°C) across all stations within each species from Pan European Phenological (PEP725) database. Phenological sensitivity is calculated based on mean preseason temperature. The preseason period is defined as 45 days in (a) and (b), and 90 days in (c) and (d). Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation within the 60-day centered period. The species is labeled by species identity (ID) (see Table S1). Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. Figure S2. Simple regression (SR) (a) and partial regression (PR) coefficients (b) between spring phenological sensitivity (days °C1) and spring SD (°C) across all stations within each species from Pan European Phenological (PEP725) database. The preseason period used to calculate spring phenological sensitivity is defined as 60 days. Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. The species is labeled by species identity (ID) (see Table S1). Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. The daily climate data is derived from freely available climate stations in ECA&D. Figure S3. Simple regression (SR) (a) and partial regression (PR) coefficients (b) between spring phenological sensitivity (days °C1) and spring SD (°C) across all stations within each species from Pan European Phenological (PEP725) database. Note that the phenological sensitivity is calculated as the linear coefficient between spring phenology and the sum of growing degree day (GDDsum) over the preseason period (60 days). Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. The species is labeled by species identity (ID) (see Table S1). Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. The daily climate data are derived from daily gridded E-OBS. Table S1. Summary of wild species from Pan European Phenological (PEP725) database.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
The influence of local spring temperature variance on temperature sensitivity of spring phenology 1 , S H U S H I P E N G 1 , 2 , I V A N A . J A N S S E N S 3 , X I N L I N 1 , TAO WANG1,2, CATHERINE OTTLE 1 B E N J A M I N P O U L T E R , C H A O Y U E 1 and P H I L I P P E C I A I S 1 1 Laboratoire des Sciences du Climat et de l’Environnement, CEA CNRS UVSQ, Gif-sur-Yvette 91191, France, 2CNRS and UJF Grenoble 1, UMR5183, Laboratoire de Glaciologie et Geophysique de l’Environnement (LGGE), 38041 Grenoble, France, 3 Department of Biology, University of Antwerp, Universiteitsplein 1, Wilrijk 2610, Belgium
Abstract The impact of climate warming on the advancement of plant spring phenology has been heavily investigated over the last decade and there exists great variability among plants in their phenological sensitivity to temperature. However, few studies have explicitly linked phenological sensitivity to local climate variance. Here, we set out to test the hypothesis that the strength of phenological sensitivity declines with increased local spring temperature variance, by synthesizing results across ground observations. We assemble ground-based long-term (20–50 years) spring phenology database (PEP725 database) and the corresponding climate dataset. We find a prevalent decline in the strength of phenological sensitivity with increasing local spring temperature variance at the species level from ground observations. It suggests that plants might be less likely to track climatic warming at locations with larger local spring temperature variance. This might be related to the possibility that the frost risk could be higher in a larger local spring temperature variance and plants adapt to avoid this risk by relying more on other cues (e.g., high chill requirements, photoperiod) for spring phenology, thus suppressing phenological responses to spring warming. This study illuminates that local spring temperature variance is an understudied source in the study of phenological sensitivity and highlight the necessity of incorporating this factor to improve the predictability of plant responses to anthropogenic climate change in future studies. Keywords: PEP725, spring phenology, temperature variance, temperature sensitivity Received 13 June 2013; revised version received 11 December 2013 and accepted 14 December 2013
Introduction Air temperature is the main environmental factor regulating the timing of spring events in temperate and boreal trees (e.g., Linkosalo et al., 2000), and these spring events are already showing an advancement that can be attributed to recent climatic warming (e.g., Schwartz & Reiter, 2000; Menzel et al., 2006a). But the magnitude and even direction of the advance rates are not uniform, which have been found to vary across within-species populations (e.g., Menzel et al., 2006b), species (e.g., Fitter & Fitter, 2002; Menzel, 2003; Dunnell & Travers, 2011), altitudes (e.g., Piao et al., 2011) and regions (e.g., Ahas, 1999; Schwartz & Reiter, 2000; Zhao & Schwartz, 2003; Ahas & Aasa, 2006). This large variation, on the one hand, is related to the fact that plant populations experience different degrees of climate change, e.g., a more rapid warming is occurring at high altitudes and latitudes. On the other hand, it can be attributed to their discrepancies in the response of the date of spring events to temperature (Menzel et al., 2006a; Bertin, 2008), which is expressed as the date of spring Correspondence: Tao Wang, tel. +33 1 6908 5221, fax +33 1 6908 7716, e-mail: [email protected]
© 2013 John Wiley & Sons Ltd
events change per degree Celsius of temperature change (or temperature sensitivity of spring phenology, days °C1). The variation in temperature sensitivity of spring phenology reflects its subjection to selection pressure (Bennie et al., 2010), which integrates local historic climate information thus evolving into its own optimal strategy under specific climate condition (Chabot & Hicks, 1982; Kikuzawa, 1989; Baliuckas & Pliura, 2003; Liang & Schwartz, 2013). In a fluctuating climate system, if spring phenology in trees has an instantaneous response to external temperature increase, the fatal consequences could be suffered possibly, i.e., due to elevated probability of tree exposure to frost (e.g., Bennie et al., 2010). Some mechanisms are thus developed to prevent the trees from sprouting or flowering at the ‘inappropriate’ time (Korner & Basler, 2010). To achieve this, the timing of spring events in the trees can only be responsive to temperature after fulfilling certain dose of low temperature (sufficient chilling, Murray et al., 1989; Morin et al., 2009; Schwartz & Hanes, 2010) or might evolve to be more dependent on other environmental cues (e.g., photoperiod, light quality). This leads to our hypothesis that the advancement of spring phenology 1473
1474 T . W A N G et al. responding to spring warming might decline at locations with large local variance in temperature. To test this hypothesis, data from within-species populations along an environmental gradient are mainly concerned. Previous studies (e.g., Kramer, 1995; Chuine et al., 2000; Berg et al., 2005; Vitasse et al., 2009, 2010) noticed stable temperature sensitivity among populations within a species, which is contrary to the diversified one found by others (e.g., Baliuckas & Pliura, 2003; Karlsson et al., 2003; Lu et al., 2006). However, the conclusion drawn from their studies is often based on a relatively small sample of population from the same species (Vitasse et al., 2010) and only several species being involved (e.g., Chuine et al., 2000). To avoid this, Pan European Phenological Database (PEP725) including many widely distributed species is used. The main objective of this study is to investigate influences of local spring temperature variance on spring phenological sensitivity to temperature at the species level.
Materials and methods
Pan European Phenology Database (PEP725) In 2004, the European funded ‘COST Action’ N°725 was running to establish a European Phenological data platform for climatological applications. The common dataset of COST725 hosted by ZAMG (ZentralAnstalt fur Meteorologie und Geodynamik), comprises 7.6 million data points in total from 15 countries plus International Phenological Gardens (IPG) from 7285 observation sites spanning from 1951 to 2000. In 2010, the successor project PEP725 (Pan European Phenology Database) funded by EUMETNET (EUropean national Meteorological Services NETwork) and ZAMG established an open access database, with plant phenology observations of 30 European countries (so far nine million of records collected) from 1951 onwards and develop quality checking procedures. In the PEP725 dataset, 19608 stations have been collected over 50 species from 30 European countries, covering
approximately 49% of Europe (http://www.pep725.eu). But most of the stations are concentrated in Germany (Fig. 1). In terms of mean annual climate, mean annual temperature (precipitation) ranges from 5 to 12 °C (from 550 to 1200 mm) over a majority of PEP725 stations. From a seasonality perspective, most PEP725 stations have a temperature seasonality (mean annual temperature range) of 16–25 °C and a precipitation seasonality (mean annual precipitation range divided by the mean annual precipitation) of 0.5–1.3. The PEP725 stations (mainly covering the continental Europe) have a mean climate typical of the temperate mid-latitudes but have a poor sampling in the relatively cold climates (e.g., north of 56 °N) and almost no sampling in warm and dry climates (e.g., subtropical, semiarid). Compared with the newly released phenological database across North America and Europe (NECTAR: Network of Ecological and Climatological Timings Across Regions), the PEP725 stations have a relatively limited temperature range and reduced precipitation and temperature seasonality. However, the PEP725 database includes many widely distributed species with broad ecological tolerances which may have phenologies that are phenotypically or genotypically plastic (e.g., Stinchcombe et al., 2004; Cook et al., 2012a). This characteristic well suits the purpose of this study. By contrast, NECTAR has relatively few sites (12 in total) and low replication of species across sites (Cook et al., 2012a), which, however, cannot satisfy our species selection criteria. The station information in PEP725 database includes latitude, longitude, altitude, and species. The species in this database can be grouped into two categories: wild and cultivated species (fruit trees and agricultural crops). In this study, we only focus on the wild species because the cultivated species could be highly genetically manipulated. The phenophases of plant species are assigned to a BBCH (Biologische Bundesanstalt, Bundessortenamt and Chemical industry) code (Meier, 2001). Among 28 phenophases recorded in the database, four spring phenophases are selected for this study (Table 1). For each spring phenophase, the stations having at least 25 of 61 years (1950–2010) are included in this study.
Climate datasets E-OBS is a daily gridded observational dataset for Europe with a spatial resolution of 0.25° grid for precipitation and temperature during the period 1950–2011. E-OBS displays an excellent correlation with the existing high-resolution regional gridded data for the United Kingdom and Alps that are based on much denser station networks (Hofstra et al., 2009), which forms the basis of this analysis. The temperature and precipitation fields Table 1 Spring phenophases in Pan European Phenological (PEP725) database
Fig. 1 Spatial distribution of stations from Pan European Phenological (PEP725) database.
FLS LFU FFO FFF
Spring phenophases
No. of species
First leaf separated Leaf unfolding (first visible leaf stalk) First flowers open Full flowering
4 13 25 3
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
T E M P E R A T U R E S E N S I T I V I T Y O F S P R I N G P H E N O L O G Y 1475 in E-OBS are interpolated from over 2000 stations based on ECA&D (the European Climate Assessment and Data set; http://eca.knmi.nl) and other existing datasets (Haylock et al., 2008). We match the PEP725 phenology records to daily ECA temperature and precipitation at the geographically closest grid cell. Without specific mention, daily climate data from the gridded E-OBS product is used hereafter. Compared with the station climate data, the potential shortcoming concerned with daily E-OBS gridded climate dataset is that the local extreme events may be smoothed during the spatial interpolation process (Hofstra et al., 2009). Given that the local extreme climate (e.g., warm temperatures) could have an influence on spring phenological responses, we also use freely available blended station climate data from ECA&D (1950–2010) to cross-validate the results based on daily gridded E-OBS product. Note that not all of station data used for generating daily gridded E-OBS can be freely available. This would lead to the difference in phenological sensitivity calculation between gridded and station data in this study. If station climate data from ECA&D was used, the field locations in PEP725 database meeting the following criteria are included in this study. Firstly, the distances between the field locations and their nearest meteorological stations from climate station are below 40 km. Secondly, the field locations have both specific spring phenophase data record and yearly temperature record with no 30-day period missing more than 10 days at least 25 of 61 years during the period 1950–2010 (Fig. 1).
Calculating temperature sensitivity of spring phenology The temperature sensitivity of spring phenology is commonly computed as linear slope coefficient between spring phenology and mean preseason temperature (described below) on the interannual timescale (e.g., Menzel et al., 2006b). In this study, the same method is adopted to calculate temperature sensitivity of spring phenology (referring to spring phenophase date) for each station in PEP725 database. It is well recognized that the onset date of spring phenology responds primarily to integrated climate forcing (e.g., cumulative heat sums) (McMaster & Wilhelm, 1997; Leon et al., 2001). Additionally, we calculate the phenological sensitivity as the linear slope coefficient between spring phenology and preseason growing degree day (GDD) summation. The GDD and GDDsum are computed as: GDD ¼ maxðTmean GDDthresh ; 0Þ X GDD GDDsum ¼
ð1Þ ð2Þ
where Tmean is daily mean temperature, GDDthresh is the temperature threshold for a day to qualify as a GDD (0 °C in our case), and GDDsum is the sum of GDD over the preseason period. The same conclusions are obtained if 5 °C for GDDthresh was used (data not shown). Preseason refers to the average or integrated climatic conditions exposed to the plant prior to the phenological event taking place. For each station, the preseason period is chosen to be 60 days, which is specified to end at the date that is calculated by averaging spring phenological dates from all years.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
Mean preseason temperature and preseason cumulative precipitation is then computed by averaging and summing daily temperatures and precipitation in a 60-day preseason period, respectively. We also conduct sensitivity analyses to evaluate the impacts of different preseason period lengths (45 and 90 days). Without specific mention, the length of preseason period is defined as 60-day hereafter and phenological sensitivity to temperature is computed based on mean preseason temperature. In addition, we use the magnitude of phenological sensitivity to denote the strength of sensitivity.
Analyses At each station, multiyear averaged standard deviation of daily detrended air temperature in a 60-day period centered over mean onset date of spring phenology (60-day centered period), which is used as spring SD (or local spring temperature variance) hereafter, is employed to represent local spring temperature variance. The same conclusions are reached if the length of the centered period was changed to 45 and 90 days (data not shown). We perform both simple and partial correlation analyses between phenological sensitivity and spring SD across stations belonging to the same species. Partial correlation statistically removes the effects of cross-station variation in mean temperature and cumulative precipitation within the 60-day centered period. Note that we only consider the species in PEP725 database for which the number of the stations is larger than 10. Previous studies have indicated that minimizing the frost damage in temperate and boreal biomes is an important external factor in shaping plant phenological traits (e.g., Lockhart, 1983; Linkosalo et al., 2000; Saxe et al., 2001; Bennie et al., 2010). In this study, we hypothesize that the potential mechanism for local spring temperature variance impacts on phenological sensitivity is possibly linked to plant avoidance of frost damage. In other words, regions with large local spring temperature variance tend to have a high probability of frost risk, and the living plants might suppress temperature responses of spring phenology to minimize the frost risk. To test these assumptions, for each species, we firstly perform partial correlation analysis between indicator of the frost risk and spring SD after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60day centered period. Secondly, simple regression analyses are conducted between phenological sensitivity and indicator of the frost risk. Moreover, partial regression analyses are also performed for each species between phenological sensitivity and indicator of the frost risk after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. In this study, the frost risk is assessed based on spring frost frequency. The number of frost days, which are defined as daily minimum air temperatures fall bellow 0 °C over the 60-day centered period, is counted for each year. Spring frost frequency is then calculated through dividing mean frost days by the length of the 60-day centered period. The frost risk mentioned in this study should be distinguished from the extent of frost damage. For example, species are often exposed to more frost
1476 T . W A N G et al. events during early spring phases than during later phases but conversely the extent of frost damage could be greater at later phases (Sakai & Larcher, 1987; Augspurger, 2013). In addition, the use of a universal threshold (0 °C in this study) to determine spring frost frequency is a bit arbitrary because the frost thresholds could vary with species and phenological phases (e.g., Augspurger, 2013). Simultaneous gathering field observations of temperature, phenology and frost damage over long-term periods are thus necessary in the future.
Results
The relationship between temperature sensitivity of spring phenology and local spring temperature variance Figure 2 displays the distribution of temperature sensitivity of spring phenology from PEP725 database. All spring phenophases at a majority of the stations show a negative response to an increase in mean preseason temperature (Fig. 2). A 1 °C increase in mean preseason temperature induces an earlier shift of 3.6 (2.8–4.3), 2.9 (2.1–3.6), 4.2 (3.2–5.2), 4.4 (3.4–5.4) days for First Leaf Separated (FLS), Leaf Unfolding (LFU), First Flowers Open (FFO) and Full Flowering (FFF), respectively. Spring SD for each phenophase is also shown in Fig. 2. In PEP725 database, the linear regression coefficients between phenological sensitivity and spring SD (or local spring temperature variance) at the species level are displayed in Fig. 3. Our simple linear regression analysis show that there are more positive coefficients than the negative ones and most of negative ones are non significant (Fig. 3a). Partial regression analysis
Fig. 2 The boxplot distribution of spring phenological sensitivity (days °C1) and spring SD (°C) from all stations within each of the four spring phenophases [First Leaf Separated (FLS), Leaf Unfolding (LFU), First Flowers Open (FFO) and Full Flowering (FFF)] from Pan European Phenological (PEP725) database. See Table 1 for explanation of FLS, LFU, FFO and FFF. The bottom and the top of the box denote the 25th and 75th percentiles, respectively, and the bold line within the box is the 50th percentile (the median). Spring phenological sensitivity is calculated based on mean preseason (60 days) temperature.
Fig. 3 (a) Simple and (b) partial regression coefficients between spring phenological sensitivity (days °C1) and spring SD (°C) across all stations within each species from Pan European Phenological (PEP725) database. The species is labeled by species identity (ID) (see Table S1). Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. Simple regression (SR) and partial regression (PR) denote simple and partial regression coefficients, respectively. Note that spring phenological sensitivity is calculated based on mean preseason (60 days) temperature.
after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period also supports this finding (Fig. 3b). For example, 33 positive coefficients (of which 25 are significant) of 45 are found between phenological sensitivity and spring SD, and 12 negative coefficients (of which five are significant) are detected (Fig. 3b). This result is robust to the following three cases: the phenological sensitivity was calculated based on mean preseason temperature using other preseason periods (45 and 90 days) (Figure S1), the climate data from blended meteorological stations archived in ECA&D dataset (Figure S2) were used and phenological sensitivity was calculated based on preseason (60 days) GDD summation (Figure S3). © 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
T E M P E R A T U R E S E N S I T I V I T Y O F S P R I N G P H E N O L O G Y 1477
Based on PEP725 database, we identify that spring frost frequency is also positively correlated with spring SD (or local spring temperature variance) for a majority of species after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period (Fig. 4b). In addition, the simple correlation coefficient between spring SD and spring frost frequency is also displayed in Fig. 4a. As shown in Fig. 4b, 34 positive coefficients (of which 33 are significant) of 45 are found between spring frost frequency and spring SD, and the remaining 11 negative coefficients (of which two are significant) are detected. Again, these results are robust to the
following cases: the phenological sensitivity is calculated based on mean preseason temperature using other preseason periods (45 and 90 days), the climate data from blended meteorological stations archived in ECA&D dataset are used and phenological sensitivity is calculated based on preseason (60 days) GDD summation (data not shown). We also investigate the responses of phenological sensitivity to spring frost frequency. Many species from PEP725 database display a positive correlation between phenological sensitivity and spring frost frequency (Fig. 5a). For example, 30 positive coefficients (of which 21 are significant) of 45 are found, and the remaining 15 negative coefficients (of which five are significant) are detected. After statistically controlling for cross-station variation in mean temperature and cumulative
Fig. 4 (a) Simple and (b) partial regression coefficients between spring SD (°C) and spring frost frequency (%) across all stations within each species from Pan European Phenological (PEP725) database. Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60day centered period. Simple regression (SR) and partial regression (PR) denote simple and partial regression coefficients, respectively.
Fig. 5 (a) Simple and (b) partial regression coefficients between spring phenological sensitivity (days °C1) and spring frost frequency (%) across all stations within each species from Pan European Phenological (PEP725) database. Partial regression coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. Note that spring phenological sensitivity is calculated based on mean preseason (60 days) temperature.
The link between spring frost frequency and local spring temperature variance
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
1478 T . W A N G et al. precipitation over the 60-day centered period, the number of species from PEP725 database with positive coefficients has been reduced (Fig. 5b).
Discussion
Temperature responses of spring phenology Our analysis based on ground observations undoubtedly supports the finding that the responsiveness of spring phenology is dependent on the preceding mean preseason temperature (e.g., Badeck et al., 2004; Menzel et al., 2006a; Piao et al., 2006; Cong et al., 2013) and also on preseason integrated temperature forcing (McMaster & Wilhelm, 1997; Leon et al., 2001). Using COST725 dataset (roughly the same with PEP725, see ‘Materials and methods’), Menzel et al. (2006a) found that panEuropean temperature sensitivity of spring phenophases (leaf unfolding and flowering) is 2.5 days °C1, which is roughly comparable with our estimate (ca. 3.8 days °C1). A difference of 1.3 days response per degree C between these two estimates might be related to the difference in temperature dataset and the time period defined for spring season. We notice delayed responses of spring phenology to mean preseason temperature (positive phenological sensitivity) in PEP725 database (not shown). Recent studies (Cook et al., 2012b; Pope et al., 2013) have suggested that species are not advancing in their spring phenological behaviors because they respond more to lack of winter chill than increased spring heat. This is also found by Zhang et al. (2007) based on NDVI in more southern North America, which related this delay to insufficient chilling in the warm winter years (Murray et al., 1989). Based on NDVI of steppe and meadow in Tibet Plateau, Yu et al. (2010) found a delay of spring phenology due to warm winter conditions, though some debates have been stirred (Zhang et al., 2007; Chen et al., 2011; Luedeling et al., 2011; Shen, 2011; Yi & Zhou, 2011). However, we should not rule out the possibility that some plants might only need very minor chilling requirements, as exemplified by Lilac (Syringa) (Larcher, 2006).
Impacts of local spring temperature variance on phenological sensitivity In terms of PEP725 database, this study suggests that the strength of phenological sensitivity tends to be smaller for species that are located in climates with larger local spring temperature variance (or higher spring SD). This result is also found if the climate data from blended meteorological stations archived in ECA&D dataset was used and phenological sensitivity was calculated based on preseason (60 days) GDD summation.
The present study is the first reported case that species might be less likely to keep climatic warming at locations with more fluctuating climate systems. This probably reflects the adaptive responses of spring phenology to local climate (e.g., Lockhart, 1983; Lechowicz, 1984; Menzel et al., 2006a; Doi & Takahashi, 2008; Pudas et al., 2008). The likely mechanism for this behavior may be related to plant avoidance of frost risk. Our analysis indicates a widespread increased spring frost frequency with increasing local spring temperature variance at the species level (Fig. 4). This suggests that the probability of plant exposure to frost risk increases as the local climate becomes more fluctuating (higher spring SD). We also notice a decreased strength in phenological sensitivity with increasing spring frost frequency for many species (Fig. 5a), although the number of species with positive (and significant) coefficients has been reduced after statistically controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period (Fig. 5b). Our results indicate that plants in regions with high spring SD may adopt a relatively conservative approach to minimize the risk of experiencing frost damage through suppressing the response of spring phenology to external temperature increase. The selection pressure from frost risk could probably drive the plants to be evolved into different strategies (e.g., low to high chill accumulation, photoperiod) for spring phenology, an interesting question to be explored in the future experimental work. In addition, our analyses also indirectly supports the mainstream model framework that spring phenology simulations in land surface models need consider other cues (e.g., photoperiod, chill requirements) to avoid the potential frost damage imposed by large local temperature variability. The implications of this study are manifold. According to our analyses, spring phenology will be less advanced in response to spring warming in the regions characterized by more fluctuating climate system. Moreover, both mean and variance in temperature, which are two opposite forces in changing the frost risk (an increase in spring temperature variance tends to increase the frost risk), are likely to increase in future climate scenarios. If our results were extrapolated from space to time, the response of spring phenology to temperature will be decreased in the future. However, this space-for-time extrapolation further awaits careful examination in future studies when much more longterm data will be available (e.g., Rutishauser et al., 2008, 2009). Our analyses indicate that the spring phenological sensitivity can be shaped through the mechanism that the plants can decrease the risk of frost damage to actively growing parts through genotypic © 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
T E M P E R A T U R E S E N S I T I V I T Y O F S P R I N G P H E N O L O G Y 1479 plasticity. We should not rule out the possibility that the triggering of spring phenology may also be influenced by the plant’s ability in maximizing the growing season for carbon gain (e.g., Bennie et al., 2010). If this occurred, in response to the mean temperature increase in the future, the increased strength in spring phenological sensitivity to temperature can benefit the plant fitness. These two processes can be considered as two opposing forces in affecting the evolution of phenological sensitivity in the future climate change and the investigation into their relative roles is beyond the scope of this study and is further needed in the future. In addition, the present study does not explore multidecadal variability in the response of temperature sensitivity to spring SD because of relatively short time series in ground observations. The implications of this work also highlight the need for considering not only mean temperature (Cook et al., 2012b) but also local spring temperature variance in understanding variations in phenological sensitivity. It can further contribute to improving the predictability of plant responses to anthropogenic climate change that has been demonstrated to be related to plant performances (Cleland et al., 2012). Finally, we should be informed that the positive correlation of phenological sensitivity with spring SD at the within-species level is not equivalent to the unbounded linear decrease in the strength of phenological sensitivity with spring SD. For example, based on the PEP725 database, Cook et al. (2012a) found that the strength of phenological sensitivity firstly decreased and then seemed to have ceased as the mean flowering (leafing) date continues to increase at the inter-species level. In addition, the conclusions drawn from this study are mainly based on the PEP725 database, which mainly covers the continental temperate climate with few sites in the cold climates and almost no site sampling in sub-tropical and semiarid climates. Further testing of the generality of the relationship between phenological sensitivity and spring SD will require observations of more species across a wider array of environmental conditions.
Acknowledgements We acknowledge the funding by the French ANR CLASSIQUE project. We greatly thank the editor and two anonymous reviewers for their constructive comments on the manuscript.
References Ahas R (1999) Long-term phyto-, ornitho- and ichthyophenological time-series analyses in Estonia. International Journal of Biometeorology, 42, 119–123. Ahas R, Aasa A (2006) The effects of climate change on the phenology of selected Estonian plant, bird and fish populations. International Journal of Biometeorology, 51, 17–26.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
Augspurger CK (2013) Reconstructing patterns of temperature, phenology, and frost damage over 124 years: spring damage risk in increasing. Ecology, 94, 41–50. Badeck F, Bondeau A, Bottcher K, Doktor D, Lucht W, Schaber J, Sitch S (2004) Responses of spring phenology to climate change. New Phytologist, 162, 295–309. Baliuckas V, Pliura A (2003) Genetic variation and phenotypic plasticity of Quercus robur populations and open-pollinated families in Lithuania. Scandinavian Journal of Forest Research, 18, 305–319. Bennie J, Kubin E, Wiltshire A, Huntley B, Baxter R (2010) Predicting spatial and temporal patterns of bud-burst and spring frost risk in north-west Europe: the implications of local adaptation to climate. Global Change Biology, 16, 1503–1514. Berg H, Becker U, Matthies D (2005) Phenotypic plasticity in Carlina vulgaris: effects of geographical origin, population size, and population isolation. Oecologia, 143, 220–231. Bertin R (2008) Plant phenology and distribution in relation to recent climate change. Journal of the Torrey Botanical Society, 135, 126–146. Chabot B, Hicks D (1982) The ecology of leaf life spans. Annual Review of Ecology and Systematics, 13, 229–259. Chen H, Zhu Q, Wu N, Wang Y, Peng C (2011) Delayed spring phenology on the Tibetan Plateau may also be attributable to other factors than winter and spring warming. Proceedings of the National Academy of Sciences of the United States of America, 108, E93–E93. Chuine I, Belmonte J, Mignot A (2000) A modelling analysis of the genetic variation of phenology between tree populations. Journal of Ecology, 88, 561–570. Cleland E, Allen J, Crimmins T et al. (2012) Phenological tracking enables positive species responses to climate change. Ecology, 93, 1765–1771. Cong N, Wang T, Nan H, Ma Y, Wang X, Myneni RB, Piao S (2013) Changes in satellite-derived spring vegetation green-up date and its linkage to climate in China from 1982 to 2010: a multimethod analysis. Global Change Biology, 19, 881–891. Cook B, Wolkovich E, Davies T et al. (2012a) sensitivity of spring phenology to warming across temporal and spatial climate gradients in two independent databases. Ecosystems, 15, 1283–1294. Cook B, Wolkovich E, Parmesan C (2012b) Divergent responses to spring and winter warming drive community level flowering trends. Proceedings of the National Academy of Sciences of the United States of America, 109, 9000–9005. Doi H, Takahashi M (2008) Latitudinal patterns in the phenological responses of leaf colouring and leaf fall to climate change in Japan. Global Ecology and Biogeography, 17, 556–561. Dunnell K, Travers S (2011) Shifts in the flowering phenology of the northern great plains: patterns over 100 years. American Journal of Botany, 98, 935–945. Fitter A, Fitter R (2002) Rapid changes in flowering time in British plants. Science, 296, 1689–1691. Haylock M, Hofstra N, Tank A, Klok E, Jones P, New M (2008) A European daily high-resolution gridded data set of surface temperature and precipitation for 1950–2006. Journal of Geophysical Research-Atmospheres, 113, D20119. Hofstra N, Haylock M, New M, Jones P (2009) Testing E-OBS European high-resolution gridded dataset of daily precipitation and surface temperature. Journal of Geophysical Research-Atmospheres, 114, D21101. Karlsson P, Bylund H, Neuvonen S, Heino S, Tjus M (2003) Climatic response of budburst in the mountain birch at two areas in northern Fennoscandia and possible responses to global change. Ecography, 26, 617–625. Kikuzawa K (1989) Ecology and evolution of phenological pattern, leaf longevity and leaf habit. Evolutionary Trends in Plants, 3, 105–110. Korner C, Basler D (2010) Phenology under global warming. Science, 327, 1461–1462. Kramer K (1995) Phenotypic plasticity of the phenology of 7 European tree species in relation to climatic warming. Plant Cell and Environment, 18, 93–104. Larcher W (2006) Altitudinal variation in flowering time of lilac (Syringa vulgaris L.) in the Alps in relation to temperatures. Oesterr Akad Wiss Math-Natwiss Kl Sitzungsber Abt I, 212, 3–18. Lechowicz MJ (1984) Why do temperate deciduous trees leaf out at different times adaptation and ecology of forest communities. American Naturalist, 124, 821–842. Leon AJ, Lee M, Andrade FH (2001) Quantitative trait loci for growing degree days to flowering and photoperiod response in sunflower (Helianthus annuus L.). Theoretical and Applied Genetics, 102, 497–503. Liang L, Schwartz M (2013) Testing a growth efficiency hypothesis with continentalscale phenological variations of common and cloned plants. International Journal of Biometeorology, doi: 10.1007/s00484-013-0691-6. Linkosalo T, Carter T, Hakkinen R, Hari P (2000) Predicting spring phenology and frost damage risk of Betula spp. under climatic warming: a comparison of two models. Tree Physiology, 20, 1175–1182. Lockhart J (1983) Optimum growth initiation time for shoot buds of deciduous plants in a temperate climate. Oecologia, 60, 34–37.
1480 T . W A N G et al. Lu P, Yu Q, Liu J, Lee X (2006) Advance of tree-flowering dates in response to urban climate change. Agricultural and Forest Meteorology, 138, 120–131.
Yu H, Luedeling E, Xu J (2010) Winter and spring warming result in delayed spring phenology on the Tibetan Plateau. Proceedings of the National Academy of Sciences of
Luedeling E, Yu H, Xu J (2011) Replies to Shen, Chen et al., and Yi and Zhou: linear regression analysis misses effects of winter temperature on Tibetan vegetation. Proceedings of the National Academy of Sciences of the United States of America, 108, E95–E95. McMaster GS, Wilhelm WW (1997) Growing degree-days: one equation, two interpretations. Agricultural and Forest Meteorology, 87, 291–300. Meier U (Ed.). (2001) Growth Stages of Mono- and Dicotyledonous Plants. BBCH Mono-
the United States of America, 107, 22151–22156. Zhang X, Tarpley D, Sullivan J (2007) Diverse responses of vegetation phenology to a warming climate. Geophysical Research Letters, 34, L19405. Zhao T, Schwartz M (2003) Examining the onset of spring in Wisconsin. Climate Research, 24, 59–70.
graph, 2nd edn. Federal Biological Research Centre of Agriculture and Forestry. Germany. Available at: http://www.bba.de/veroeff/bbch/bbcheng.pdf Menzel A (2003) Plant phenological anomalies in Germany and their relation to air temperature and NAO. Climatic Change, 57, 243–263. Menzel A, Sparks T, Estrella N et al. (2006a) European phenological response to climate change matches the warming pattern. Global Change Biology, 12, 1969–1976. Menzel A, Sparks T, Estrella N, Roy D (2006b) Altered geographic and temporal variability in phenology in response to climate change. Global Ecology and Biogeography, 15, 498–504. Morin X, Lechowicz M, Augspurger C, O’ Keefe J, Viner D, Chuine I (2009) Leaf phenology in 22 North American tree species during the 21st century. Global Change Biology, 15, 961–975. Murray M, Cannell M, Smith R (1989) Date of budburst of 15 tree species in Britain following climatic warming. Journal of Applied Ecology, 26, 693–700. Piao S, Fang J, Zhou L, Ciais P, Zhu B (2006) Variations in satellite-derived phenology in China’s temperate vegetation. Global Change Biology, 12, 672–685. Piao S, Cui M, Chen A, Wang X, Ciais P, Liu J, Tang Y (2011) Altitude and temperature dependence of change in the spring vegetation green-up date from 1982 to 2006 in the Qinghai-Xizang Plateau. Agricultural and Forest Meteorology, 151, 1599–1608. Pope K, Dose V, Da Silva D, Brown P, Leslie C, DeJong T (2013) Detecting non-linear response of spring phenology to climate change by Bayesian analysis. Global Change Biology, 19, 1518–1525. Pudas E, Leppala M, Tolvanen A, Poikolainen J, Venalainen A, Kubin E (2008) Trends in phenology of Betula pubescens across the boreal zone in Finland. International Journal of Biometeorology, 52, 251–259. Rutishauser T, Luterbacher J, Defila C, Frank D, Wanner H (2008) Swiss spring plant phenology 2007: extremes, a multi-century perspective, and changes in temperature sensitivity. Geophysical Research Letters, 35, L05703. Rutishauser T, Schleip C, Sparks T et al. (2009) Temperature sensitivity of Swiss and British plant phenology from 1753 to 1958. Climate Research, 39, 179–190. Sakai A, Larcher W (1987) Frost Survival of Plants: Responses and Adaptation to Freezing Stress. Springer-Verlag, New York, USA. Saxe H, Cannell M, Johnsen B, Ryan M, Vourlitis G (2001) Tree and forest functioning in response to global warming. New Phytologist, 149, 369–399. Schwartz M, Hanes J (2010) Continental-scale phenology: warming and chilling. International Journal of Climatology, 30, 1595–1598. Schwartz M, Reiter B (2000) Changes in north American spring. International Journal of Climatology, 20, 929–932. Shen M (2011) Spring phenology was not consistently related to winter warming on the Tibetan Plateau. Proceedings of the National Academy of Sciences of the United States of America, 108, E91–E92. Stinchcombe J, Weinig C, Ungerer M et al. (2004) A latitudinal cline in flowering time in Arabidopsis thaliana modulated by the flowering time gene FRIGIDA. Proceedings of the National Academy of Sciences of the United States of America, 101, 4712–4717. Vitasse Y, Delzon S, Dufrene E, Pontailler J, Louvet J, Kremer A, Michalet R (2009) Leaf phenology sensitivity to temperature in European trees: do within-species populations exhibit similar responses? Agricultural and Forest Meteorology, 149, 735–744. Vitasse Y, Bresson C, Kremer A, Michalet R, Delzon S (2010) Quantifying phenological plasticity to temperature in two temperate tree species. Functional Ecology, 24, 1211–1218. Yi S, Zhou Z (2011) Increasing contamination might have delayed spring phenology on the Tibetan Plateau. Proceedings of the National Academy of Sciences of the United States of America, 108, E94–E94.
Supporting Information Additional Supporting Information may be found in the online version of this article: Figure S1. Simple regression (SR) (a) and (c) and Partial regression (PR) coefficients (b) and (d) between spring phenological sensitivity (days °C1) and spring SD (°C) across all stations within each species from Pan European Phenological (PEP725) database. Phenological sensitivity is calculated based on mean preseason temperature. The preseason period is defined as 45 days in (a) and (b), and 90 days in (c) and (d). Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation within the 60-day centered period. The species is labeled by species identity (ID) (see Table S1). Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. Figure S2. Simple regression (SR) (a) and partial regression (PR) coefficients (b) between spring phenological sensitivity (days °C1) and spring SD (°C) across all stations within each species from Pan European Phenological (PEP725) database. The preseason period used to calculate spring phenological sensitivity is defined as 60 days. Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. The species is labeled by species identity (ID) (see Table S1). Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. The daily climate data is derived from freely available climate stations in ECA&D. Figure S3. Simple regression (SR) (a) and partial regression (PR) coefficients (b) between spring phenological sensitivity (days °C1) and spring SD (°C) across all stations within each species from Pan European Phenological (PEP725) database. Note that the phenological sensitivity is calculated as the linear coefficient between spring phenology and the sum of growing degree day (GDDsum) over the preseason period (60 days). Partial correlation coefficients are calculated after controlling for cross-station variation in mean temperature and cumulative precipitation over the 60-day centered period. The species is labeled by species identity (ID) (see Table S1). Significance levels are indicated as **, * and o representing P < 0.01, P < 0.05 and P < 0.1. The daily climate data are derived from daily gridded E-OBS. Table S1. Summary of wild species from Pan European Phenological (PEP725) database.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 20, 1473–1480
