Multi-scale approach to Euro-Atlantic climatic cycles based on phenological time series, air temperatures and circulation indexes

Science of the Total Environment 593–594 (2017) 253–262

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Multi-scale approach to Euro-Atlantic climatic cycles based on phenological time series, air temperatures and circulation indexes Luigi Mariani a,b,⁎, Franco Zavatti a a b

Lombardy Museum of Agricultural History, via Celoria 2, 20133 Milano, Italy Department of Agricultural and Environmental Sciences, Università degli Studi di Milano, via Celoria 2, 20133 Milano, Italy

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• CIRCULATION(A) → TEMPERATURE(B) → PLANT PHENOLOGY(C) is the causal chain considered. • Main aim was to analyze how spectral peaks of in A affects B, which in turn imprints C. • Problem approached with suitable methods of spectral analysis • Teleconnection with ENSO was also explored. • Results highlights phenological peaks influenced by macroscale circulation indexes.

a r t i c l e

i n f o

Article history: Received 22 December 2016 Received in revised form 7 March 2017 Accepted 20 March 2017 Available online xxxx Editor: D. Barcelo Keywords: Spectral analysis Grapevine and cherry phenology European temperatures AMO NAO ENSO

a b s t r a c t The spectral periods in North Atlantic Oscillation (NAO), Atlantic Multidecadal Oscillation (AMO) and El Nino Southern Oscillation (ENSO) were analyzed and has been verified how they imprint a time series of European temperature anomalies (ETA), two European temperature time series and some phenological series (dates of cherry flowering and grapevine harvest). Such work had as reference scenario the linear causal chain MCTP (Macroscale Circulation → Temperature → Phenology of crops) that links oceanic and atmospheric circulation to surface air temperature which in its turn determines the earliness of appearance of phenological phases of plants. Results show that in the three segments of the MCTP causal chain are present cycles with the following central period in years (the % of the 12 analyzed time series interested by these cycles are in brackets): 65 (58%), 24 (58%), 20.5 (58%), 13.5 (50%), 11.5 (58%), 7.7 (75%), 5.5 (58%), 4.1 (58%), 3 (50%), 2.4 (67%). A comparison with short term spectral peaks of the four El Niño regions (nino1 + 2, nino3, nino3.4 and nino4) show that 10 of the 12 series are imprinted by periods around 2.3–2.4 yr while 50–58% of the series are imprinted by El Niño periods of 4–4.2, 3.8–3.9, 3–3.1 years. The analysis highlights the links among physical and biological variables of the climate system at scales that range from macro to microscale whose knowledge is crucial to reach a suitable understanding of the ecosystem behavior. The spectral analysis was also applied to a time series of spring – summer precipitation in order to evaluate the presence of peaks common with other 12 selected series with result substantially negative which brings us to rule out the existence of a linear causal chain MCPP (Macroscale Circulation → Precipitation → Phenology). © 2017 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: Lombardy Museum of Agricultural History, via Celoria 2, 20133 Milano, Italy. E-mail address: [email protected] (L. Mariani).

http://dx.doi.org/10.1016/j.scitotenv.2017.03.182 0048-9697/© 2017 Elsevier B.V. All rights reserved.

254

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

1. Introduction The midlatitudes of our planet are the theater of the peculiar causal chain MCTP that links macroscale atmospheric and oceanic circulation to surface air temperature which in its turn determines the timing of phenological phases of plants (Lieth, 1974). The global nature of the MCTP causal chain was recently stated by Reid et al. (2016) that working on time series of flowering dates of three cherry species in three remote locations of the North hemisphere (Japanese cherry - Prunus jamasakura, at Kyoto, Japan; cherry - Prunus avium - at Liestal, Switzerland; Yoshino cherry Prunus x yedonensis in the tidal basin, Washington D.C., US) detected the presence of the signature of the shift in surface air temperatures happened in the 80's of the XXth century and caused by an abrupt change in macroscale circulation (Mariani et al., 2009a). Moreover Zeng et al. (2014) stated the profound ecological consequences of the MCTP causal chain with reference to the increase of the seasonal amplitude of the CO2 cycle. More problematic is rather the analysis of the MCPP causal chain because it is remarkably weakened both by (i) the relevance of mesoscale processes in precipitation genesis (Barry and Carleton, 2001) and (ii) the effect of precipitation on plant phenology that is largely mediated by the soil water budget. On this latter aspect it should be noted that (a) soil water excess delays the spring heating of soils (Zhang and Zuo, 2011) with effects on timing of phenological phases, (b) soil water shortage affects the phenological timing of grapevine (Kuhn et al., 2014; Martínez-Lüscher et al., 2016) and (c) a quite different water content can derive from the same precipitation amount in function of the time distribution of the events (precipitation concentrated in a few events results in more relevant infiltration and runoff losses). The present work focuses on meteorological data and geophysical or biological proxies in order to highlight climatic cycles whose existence will be corroborated by the detection of their presence in the three different segments of the MCTP causal chain. This exercise is referred to Europe, an ideal area because such historical records are among the longest of the world. In this general context, a preliminary comparison between our approach and that adopted by other authors (Tourre et al., 2011; Berger, 2008) was performed in order to verify the pros and cons with respect to an analysis based on the Multi Taper Method (hereafter MTM; Ghil et al., 2002). An attempt to highlight the signature of the causal chain MCPP in the selected time series was also made. 1.1. Some insights on the MCTP causal chain The time variability of macroscale circulation for the Euro-Atlantic area is often analyzed by means of NAO, AMO and ENSO. AMO describes the temperature of the North Atlantic surface (Kilbourne, 2014) and its presence, documented since 1857 by direct measurements is also testified for the last 8000 years by means of a multiproxy approach (Knudsen et al., 2011). As implied by the term “Oscillation”, AMO is subject to characteristic cycles that have a relevant effect on the European thermal and pluviometric regime (Sutton and Dong, 2012). The NAO index is based on the difference of the normalized sea level pressure between two stations in the North Atlantic, one in the far North (typically in Iceland) and another more to the south (typically in the Azores islands or in the Iberian Peninsula). The winter NAO (NAOI, from December to March) is the most effective on European climate because the thermal contrast between the Atlantic Ocean and the Eurasia is stronger than in summer. More specifically a positive NAOI gives advection towards Europe of mild and moist maritime air masses from the Atlantic while advection of polar continental air from the Siberia, very cold and dry, is observed with a negative NAO (Hurrell, 1995). A possible link between AMO and NAOI was proposed by McCarthy et al. (2015) which hypothesized that the long persistence of NAOI on positive values triggers the transition of AMO from negative to positive,

as for example happened with the 1994 transition, triggered by a long positive phase of NAOI that begun in 1988. In this context the positive NAOI determined the phase change of European temperatures from negative to positive anomaly whose persistence over time was then guaranteed by the positive AMO. ENSO is the most important coupled ocean–atmosphere phenomenon that causes global climate variability on seasonal to interannual time scales (Wolter and Timlin, 2011). Europe offers the longest world instrumental records of surface weather variables because first meteorological instruments (thermometer, pluviometer, barometer and evaporimeter) were invented by the Galilean school, an highly original scientific forum that conceived the idea of first regular meteorological measurements within the Tuscany network, active from 1655 (Camuffo and Bartolin, 2012) and at the same time spread meteorological instruments at the European level to promote an observing network that would operate in a coordinated manner (Camuffo and Jones, 2002). The availability of very long time series is quite interesting for historical and climatological purposes although the homogeneity of such series is negatively affected by problems like changes in urban heat island effect, measurement units, observational standards (e.g.: time of observation) and location of instruments. Climatic reconstructions before the beginning of the instrumental period are carried out with proxy data, such as the time series of 18O in marine sediments that enabled Cesare Emiliani to show the occurrence of a sequence of ice ages during the Pleistocene (Berger, 2013), gases and dust embedded in the Greenland and Antarctic ice sheets useful to rebuild global temperatures, wind and other atmospheric features (Jouzel, 2013) and the tree growth rings of Pinus aristata Engelm, in USA, useful to rebuild temperature and precipitation or a very long period (Carrara and McGeehin, 2015). Agriculture is a relevant source of secular proxy data because this revolutionary technology was discovered at the end of the last ice age and was initially based on the domestication of herbaceous species (wheat, rice, corn, sorghum, etc.) while the domestication of woody plants like grapevine (Vitis vinifera L.), apple (Malus communis L.), peach (Prunus persica L.), cherry (Prunus avium L.) and black cherry (Prunus cerasus L.) happened later. About grapevine it can be considered that the first wine was produced about 8000 years ago in Georgia (McGovern, 2003) and the vine has been cultivated for at least 6000 years between the Caucasus and the Zagros (Zohary et al., 2012). After the domestication, grapevine migrated first to Europe and then to other continents. From this long history follows that grapevine phenological time series are an important source of data for paleoclimate reconstructions. While they are not the only type of available phenological data for grapevine (Parisi et al., 2014), grapevine harvest dates (GHD) are by far the predominant phenological information on this crop which in plains or low hills of the Euro-Mediterranean areas with Koeppen climate Csa, Cfa or Cfb is harvested from midsummer to late autumn (July for earliest varieties, November for latest ones). GHD are registered since the Middle age by the municipalities of a wide area involving France, Switzerland, Austria and Northern Italy that established the official date of the beginning of the harvest in order to protect the wine quality and prevent the grapes theft. The gathering of these time series was started by the physicist Louis Dufour (1870) for Swiss and the climatologist Alfred Angot (1885) for France while more recent contributions came for example from the historians Le Roy Ladurie (1976) and Labbé and Gaveau (2013). GHD are mainly determined by temperatures (more daily maximum than minimum) of the spring period before grapevine flowering (April, May and June). More in detail, mild temperatures during spring give an early flowering that is generally followed by an early harvest. On this causal scheme is founded the reconstruction of spring temperatures of the past on the base of GHD dates carried out for example in Switzerland (Meier et al., 2007), Italy (Mariani et al., 2009b), Austria (Maurer et al., 2011) and France (Chuine et al., 2004).

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

The date of cherry flowering (CFD) is another phenological information widely used for climatological purposes. In plains or low hills of the Euro-Mediterranean areas with Koeppen climate Csa, Cfa or Cfb, cherry flowering happens during spring and is mainly determined by temperatures of the late winter and early spring. The most ancient phenological time series of CFD is undoubtedly the record of the beginning of flowering of Prunus jamasakura Sieber ex Koidz. (Japanese cherry) at the royal court of Kyoto, Japan, which dates back to 705 CE and is probably the longest written phenological record of the world (Aono and Kazui, 2008). The study of the periodic cycles in the above-mentioned time series can be approached by means of the technique of spectral analysis that is an interesting tool to retrieve information on the periodicities existing in time series of physical and biological variables (Ghil et al., 2002). This approach was for example adopted by Schlesinger and Ramankutty (1994) that defined the 65–70 year AMO spectral peak and Humlum et al. (2011) that compared the characteristic frequencies of Svalbard's temperature with AMO spectrum. Tularam and Ilahee (2010) studied by means of FFT (Fast Fourier Transform) the interaction between precipitation and temperature in Queensland (Australia) working on very short time-scales (from hours to 10 days). Lüdecke et al. (2013) compared the spectral features (maxima above the 95% confidence level) of an average series, derived from 6 temperature series from 1757 CE with that of the speleothems of the Spannagel Cave (Austrian Alps). Moreover Olafsdottir (2010) in a thesis for the MS degree at the University of Iceland showed the significant correlation, based on spectral analysis, between varve thickness of 2-lakes sediments and NAO-AMO summer indexes while Thiéblemont et al. (2015) proposed a 1–2 years lagged solar/AMO relationship, highly misrepresented in climate model simulations and Zanzi et al. (2007) suggested that Pinus

255

montana growth in the Central Italian Alps, if not disturbed by external geomorphological factors, is controlled by environmental and/or climatic conditions that in the last ~150 years oscillated on scales ranging from interdecadal (~20 years) to decadal.

2. Data The main features and the sources of the time series used for this work are listed in Table 1. The winter NAO index of the Climate Research Unit of the East Anglia University is the difference of the normalized sea level pressure between Ponta Delgada, Azores, and Gibraltar. The AMO time series of the NOAA is the area weighted average over the North Atlantic (0 to 70°N) of the sea surface temperatures from the Kaplan dataset and the resulting time series was detrended to eliminate the positive trend existing after the end of the little ice age (Enfield et al., 2001). The spectral features of ENSO were analyzed for the NOAA weekly series (http://www.cpc.ncep.noaa.gov/data/indices/wksst8110.for visited 7 December, 2016) of the four El Niño regions (Nino1 + 2, Nino3, Nino3.4 and Nino4). The ETA dataset for the period from 1655 to 2015 was obtained working on yearly data of 34 weather stations as described in the Supplementary material (Appendix S1). Despite ETA was selected as the reference dataset for European temperatures, the two time series of Paris Montsouris and Geneva Cointrin were also considered, the first because analyzed also in the seminal work of Tourre et al. (2011) on the causal link between AMO and GHD in Burgundy (France) while the latter is representative of the two main sources of phenological data (the Swiss Plateau and the Burgundy).

Table 1 List of datasets analyzed in this paper. The web sites have been visited in October 2016. N.

Data

Country

Unit

1

AMO

–

2

NAO

–

AMO 1857–2015 http://www.esrl.noaa.gov/psd/data/correlation/amon.us.long.data (NOAA, 2016) Index NAO Index 1825–2015 http://www.cru.uea.ac.uk/cru/data/nao/nao.dat (CRU, 2016)

3

European Temperature Anomaly (ETA)

–

°C

1660–2010 http://www.climatemonitor.it/?page_id=40210&lang=en

France

°C

1900–2013 http://www.rimfrost.no

4

Paris Montsouris Temperature 5 Geneve-Cointrin Temperature 6 CFD of Swiss Plateau 7 GHD of Beaune Cote d'or – Burgundy 8 GHD of 16 vineyards of Burgundy 9 GHD of 16 vineyards of Burgundy. Dataset used by Tourre et al. (2011) 10 GHD of Swiss plateau 11 GHD of Haut Medoc/Bordeaux

12 GHD of Tirano

13 Lower Bavaria precipitation

Switzerland °C

Time range Link and reference

Notes Unsmoothed data Basic series and Osborne update Sources of data are listed in the Supplementary material (Appendix S1)

1864–2016 http://www.meteoswiss.admin.ch/product/output/climate-data/homogenous-monthlydata-processing/data/homog_mo_GVE.txt 1721–2012 ftp://ftp.ncdc.noaa.gov/pub/data/paleo/historical/europe/switzerland/swiss-cherryLocal phenology2003.txt (Rutishauser, 2008) 1370–2010 Labbé and Gaveau (2013) Local

Switzerland Day of the year France Days from 1 September France Days from 1354–2006 http://www.clim-past.net/8/1403/2012/cp-8-1403-2012-supplement.zip (Daux et al., 1 2012) September France Days from 1676–2006 Tourre et al. (2011) 1 September

Switzerland Days from 1 September Days from France 1 September Italy Days from 1 September Germany mm

1624–2006 Meier et al. (2007)

1752–2006 Chevet and Soyer (2006)

1624–1930 Mariani et al. (2009b)

1480–1978 https://www.ncdc.noaa.gov/paleo-search/study/6334 (Wilson et al., 2005)

Proxy from tree rings

256

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

The spring-summer precipitation for the period 1480–1978 derived from Norway spruces (Picea abies L.) of a Lower Bavaria forest (Wilson et al., 2005) was also considered in our analysis. About the use of GHD time series, some preliminary aspects that should be considered are hereafter listed: • The choice of the GHD is mainly dictated by sugar content and acidity of grapes. In the past these variables were subjectively verified by tasting while in the last 50 years they are increasingly evaluated by means of quantitative analytical methods and the thresholds for harvest are often changed in function of the oenological objectives. • To obtain long and continuous time series the data are in many cases gathered from a lot of sites which poses a problem of homogeneity. Exceptions are for example the time series from Labbé and Gaveau (2013) from Beaune (Côte d'or) and Valtellina series from Tirano (Italy). • In some cases researchers validated data from harvest edicts with ancillary data, such as data from registers that reported the wages or the food purchased for the harvesters by the owners of the vineyards (Labbé and Gaveau, 2013). • There is a varietal problem (each variety has a specific harvest period genetically determined) that is overcome in the selected series because the dominant variety is traditionally Pinot noir (aka Burgundry) in Burgundy, the Bordeaux blend (Cabernet Sauvignon and Merlot) in Bordeaux area and the Nebbiolo variety (locally named Chiavennasca) in Valtellina. • When grapes are ripe, harvest can be postponed in case of rainy periods, but this delay is generally not taken into account by the harvest edicts. • In the second half of the XIX century the European viticulture was stricken by severe and widespread pests and diseases coming from the America (the aphid Phyilloxera vastatrix and the two fungal

diseases powdery mildew caused by Erysiphe necator and downy mildew caused by Plasmopara viticola). These maladies influenced the dates of harvest and for example harvest was anticipated in Valtellina to prevent the losses due to powdery mildew while the harvest was not performed in case of strong attacks of downy mildew o phyilloxera. The GHD time series 1480–2006 for Switzerland come from 15 locations of the Swiss Plateau region and north-western Switzerland and were homogenized and quality checked as described by Meier et al. (2007). The GHD time series 1752–2006 for Bordeaux were collected from the archives of the castles of Medoc (Chevet and Soyer, 2006) while GHD time series 1624–1930 for Tirano (Valtellina – Italy) were gathered from local archives (Mariani et al., 2009b). Also useful for this work was the composite time series 1721–2000 of cherry flowering dates (CFD) for the extended Swiss Plateau region from Alps to the Basel area (Rutishauser, 2008), recently updated to 2012 on the NOAA paleoclimatology web site. All sites used for our analysis (Fig. 1) are directly affected by the westerlies with the only exception of Tirano, shielded by the Alpine divide which interacts with the westerlies giving rise to mesoscale phenomena like foehn - Stau systems or enhanced thunderstorm activity in summer (Barry, 1992).

3. Methods The spectral analysis was based on the two main analytical methods Maximum Entropy Method (hereafter MEM; Childers, 1978) and LombScargle Periodogram (hereafter LOMB; Lomb, 1975; Scargle, 1982).

Fig. 1. Locations of the 12 European time series used for the spectral analysis (solid dots) and the 34 time series (empty dots) used to produce the time series of the European Temperature Anomaly. Please refer to Tables 1 and 2 of the Supporting material for a more detailed information.

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

The MEM method was applied on evenly-spaced data. We followed what published in Press et al. (2009) and used the highest allowed pole order, which is half of the length of the record. Such technique can resolve distinct sinusoids (i.e. sharp spectral features) and, on the other hand, gives rise to spurious peaks in the flat noise background. The LOMB method was applied when the series are unevenly spaced and in general when they have missing data. Again we followed what Press et al. (2009) made available, with ofac = 5 and hifac = 4 as normal setting, hifac = fhi/fc being the ratio between “how high in frequency to go, or fhi” and the Nyquist frequency fc and ofac as an oversampling parameter which allows a better sampling at finer intervals than 1/T, T being the span of the input data or the frequency such that the data can include a complete cycle. When and if it would be necessary, ofac and hifac were both lowered by 1 (4 and 3 respectively). Lomb method required a preliminary data detrending carried out computing linear trends by a least square fit and subtracting them to the data.

257

We also used the MTM method (Ghil et al., 2002) with 2 tapers (default value) and resolution 2 and 3, applied to a couple of our datasets (Paris temperature and Swiss CFD), as a comparison with our main procedures. Within the MTM approach, we also used the confidence levels of 50, 90, 95, 99%. The public procedure (SSA-MTM toolkit) available at http://research.atmos.ucla.edu/tcd//ssa/ has been applied. Finally a Student's t-test has been used to evaluate the influence of El Niño on the spectral peaks detected in the selected time series. 4. Results and discussion 4.1. Spectral analysis Some preliminary tests of the selected approach were carried out in order to evaluate its reliability and are reported in the Supplementary material - Appendix S2. More specifically our results are coherent with

Fig. 2. Data and LOMB spectrum for the GHD in Beaune-Burgundy-France (Labbé and Gaveau, 2013).

258

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

that obtained in two other works based on the MTM approach (Tourre et al., 2011; Berger, 2008). For the aim of this work, depending on the evenly or unevenlyspaced nature of the available data, the MEM spectrum or the LOMB periodogram were computed for the time series listed in Table 1 and a suitable diagram was produced for any of them, with data and spectrum plotted at two (or three) different degrees of resolution. All datasets and spectral plots are collected in the figures listed in Supplementary material - Appendix S3 while in Fig. 2 are represented the time series and the LOMB spectrum for the GHD of Beaune (Labbé and Gaveau, 2013) and in Fig. 3 are represented the AMO series (1857–2015) and its MEM spectrum.

The best-visible spectral maxima are resumed in Fig. 4 and in Table 2. Fig. 4 illustrates as the spectral peaks aggregate around some frequencies which include either NAO or AMO periods and, in a couple of cases, both. The gray bands were drawn to outline the groupings that include at least half of the 12 monthly/yearly datasets actually analyzed. Note that the 6–7 years band is the only spectral range that does not include the atmospheric and oceanic oscillation peaks. These latter are present in the largest number of datasets (7) with the remarkable exception of the temperature of Paris Montsouris and the CFD of the Swiss plateau. In the specific case of the Paris Montsouris this decoupling from the macroscale oceanic signal could be due to the fact that it is representative of a quite peculiar urban microclimate. Something similar could

Fig. 3. Spectra of AMO time series, whose data are the un-smoothed ones. Please note the 63.5 years maximum: not a 60 years solar peak, but, given its width, probably the characteristic 65–70 years AMO cycle.

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

259

Fig. 4. Spectral maxima of all series: the bottom frame is the enlargement through 30 years periods of the above plot. The gray bands are the 65, 24, 20.5, 13.5, 11.5, 7.7, 5.5, 4.1, 3, 2.4 yrs. periods ±5%.

happen for the CFD of the Swiss plateau because it seems strange that a signal picked from Valtellina grapevine is not caught by another species in an area of the Alpine massif that is better exposed to the Atlantic macroscale signal. An alternative explanation may be sought in a peculiar biological response of the phenology of the flowering cherry to the macroscale thermal forcing although we realize that this assumption clashes with the extreme sensitivity of the CFD to the abrupt climate change of the 80's of the XXth century (Reid et al., 2016). The groupings highlighted in Fig. 4 may be considered the fingerprint of a causal relation along the MCTP causal chain which establishes a link between macro, meso and microscale phenomena. In this context we can suppose for example that the 3 and 65 years bands are influenced by AMO; the 20.5 and the 4.1 years bands by NAO and the

13.5 years band by both. The 6–7 years band (not plotted in gray in Fig. 4) appears in such a scheme as a strong modifier of the ocean imprinting in favor of other causal factors that are not considered in our approach, like solar activity or meso and microscale phenomena (soil water resources, cold air drainages from mountains to the bottom of valleys and so on). Some periods larger than the 160 years limit are reported in Table 2 shows and are also grouped in Fig. 5. From Table 3 we can draw the conclusion that the Atlantic oscillations reflect their influence - at different degrees - on European temperatures and by consequence on phenological time series for both sea-faced (Medoc/Bordeaux) and continental places (Swiss plateau). In particular, we note that the 83% of our datasets are signed by the circa 7.7 years period which seems to be

Table 2 Best-visible spectral maxima for the selected series. N.

Series

Time range Period of spectral maxima (years)

1 2 3

AMO NAO European Temperature Anomaly (ETA)

1857–2015 63.6, 33.5, 23.6, 17.2, 13.3, 10.1, 8.96, 7.4, 5.5, 2.87 1825–2015 382, 152, 59, 36.4, 20.6, 17, 13.4, 11.2, 7.8, 4.93, 4, 2.3 1660–2010 253.6, 126.8, 88.8, 68.3, 55.5, 46.7, 34.8, 26.5, 23.1, 19.1, 15.3, 13.1, 12.4, 11.2, 7.8, 7.4, 6.5, 5.7, 5.0, 4.7, 4.0, 3.6, 3.5, 3.1, 2.9, 2.3, 2.2, 2.1, 2.0, 1.98, 1.91, 1.86 1900–2013 65, 22.7, 11.66, 7.7, 5.54, 4.54, 3.82, 3.09, 2.38, 1.99, 1.51 1864–2016 305.5, 67.88, 23.50, 13.28, 8.86, 7.64, 6.64, 5.50, 3.94, 3.82, 3.07, 2.66, 2.39, 2.06, 1.86, 1.82 1721–2012 145.5, 91, 60.6, 46.9, 35.5, 27.5, 18, 13.3, 11.2, 10.5, 7.7, 2.1, 1.9 1370–2010 399, 152, 118, 64, 48.4, 44.4, 24.2, 20.2, 15.1, 12.5, 11.5, 6.7, 6.0, 4.1, 3.9, 2.397, 2.23 1354–2006 326, 155.2, 75.8, 65.2, 57.2, 44.1, 36.2, 24.1, 20.2, 12.5,10.6, 7.5, 6.8, 5.9, 5.2, 4.1, 3.9, 3.1, 2.1 1676–2006 330, 150, 75, 53.2, 45.8, 35.1, 26.6, 24.3, 20.4, 4.7, 3.8, 2.4

4 5 6 7 8 9

Paris Montsouris Temperature Geneve-Cointrin Temperature CFD of Swiss Plateau GHD of Beaune Cote d'or – Burgundy GHD of 16 vineyards of Burgundy GHD of 16 vineyards of Burgundy. Dataset used by Tourre et al. (2011) 10 GHD of Swiss plateau 11 GHD of Haut Medoc/Bordeaux 12 GHD of Tirano 13 Lower Bavaria precipitation

1624–2006 382, 90.95, 70.74, 39.79, 22.73, 20.32, 17.52, 14.15, 11.51, 10.49, 8.27, 7.82, 6.77, 5.86, 4.28, 3.93, 3.56, 2.51, 2.41, 1.71, 1.66 1752–2006 48, 103.3, 68.89, 41.33, 32.63, 26.96, 23.40, 20.67, 12.40, 7.80,5.66, 3.90, 3.72, 3.10, 2.60, 2.32, 1.76, 1.63 1624–1930 170, 63.75, 38.25, 25.5, 19.87, 11.0, 6.77, 5.63, 4.97, 4.1, 3.99, 3.94, 3.87, 3.12, 2.69, 2.52, 2.42, 2.40, 1.71, 1.66 1480–1978 166,75.5, 39.5, 16, 10.6, 7.1, 5.8, 4.3, 1.3

260

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

Fig. 5. Spectral peaks with periods in the range 160 through 400 years. The gray band refers to 385 ± 2% years. It does not appear that (multi) century-class periods have been influenced by AMO or NAO but the uncertainty here is quite high, the peaks being derived by only one cycle (time range is about 400 years).

“driven” by some combination of NAO and AMO. As far as the periods lesser than 7.7 years are concerned, they appear a mixing of AMO and NAO also if any period refers to only one oscillation. The longer periods are mostly attributable to AMO while NAO seems to play a role only below 24 years. A feature of summer (AMJJAS) and winter (ONDJFM) AMO spectra stated for example by Jiang et al. (2015) and confirmed by our analysis is the existence of periods respectively of 50–60 and 70–80 years which are coherent with the three peaks of cold of the LIA attained in 1695, 1740 and 1816 (Le Roy Ladurie, 2004) and that at least partially correspond to the peaks of lateness in GHD (1740 and 1816) and CFD (1816). Fig. 17sum of Supplementary materials shows that spectral maxima of the Bavarian precipitation proxy are near (but not included in) the respective gray band in a couple of cases (4.3 and 5.8 year). Moreover the 1.3 and 16 year maxima cannot be found in any other dataset and the 77.5-year is comparable only to the Tourre dataset in both its ranges. The almost total absence of spectral maxima shared between Lower Bavaria precipitation proxy and the other 12 time series analyzed in this work is in our view the result of two main phenomena: 1. Precipitation in the Euro-Mediterranean area is mainly the result of Atlantic troughs, their cut-off lows and Mediterranean lows triggered by outbreaks of cold air masses (Mariani and Parisi, 2013). This makes crucial the role in precipitation genesis of mesoscale phenomena not strictly related to the macroscale circulation indexes AMO, NAO and ENSO. 2. Precipitation effect on plant phenology is largely mediated by the soil water budget which in its turn is strongly influenced by the amount and time distribution of precipitation. The abovementioned aspects make extremely elusive the precipitation effect on plant phenology and justify the absence, as far as we know, of phenological models that adopt precipitation as driving variable (Wilczek et al., 2010; Cola et al., 2016). . 4.2. Teleconnections The global peculiarity of the 65 year cycle at Euro-Atlantic scale was stated by Barcikowska et al. (2016) that highlighted that the center of action of this cycle is located over the North Atlantic but it manifests also over the Pacific and Indian Oceans, suggesting a peculiar interbasin teleconnection maintained through an atmospheric bridge. Teleconnections of the spectral peaks with solar activity and El Niño Southern Oscillation (ENSO) are hereafter analyzed and discussed. Jiang et al. (2015) looks for solar influence on North Atlantic summer SST and claims a close link between solar activity and SST in this area during the past 4000 yr. He also suggests a more notable solar influence during cold periods with less vigorous ocean circulation. The paper of Jiang et al.

(2015) includes in its Supplementary material the detailed summer SST derived from diatom data in the northern coast of Iceland (table DR1) and its Lomb-Scargle Fourier spectral analysis (figure DR3). The comparison of figure DR3 with Fig. 3 shows that the 61.1 years (the last maximum on the right side of figure DR3) is not the 63.6 years powerful and large peak in the average AMO spectrum. Also, summer and winter AMO spectra are different in such a way that the main maxima spans between 63 and 70 years periods, as it appears in Fig. 10sum of Supplementary materials. So, the (weak) solar signature present in Jiang et al., 2015 (the 61.1 years of figure DR3) cannot be confirmed by the AMO spectra, unless we admit that the inertia of the sea temperature cyclicity gives rise to a delay with respect to the solar forcing. El Niño Southern Oscillation impact on the Euro-Atlantic area is still under discussion. To approach this question the spectral features of the four El Niño regions (Nino1+2, Nino3, Nino3.4 and Nino4) were analyzed and reported in Fig. 11sum of the Supplementary materials. The influence of El Niño on air temperature and plant phenology in Europe can be analyzed by a comparison among the main ENSO spectral features and the maxima reported in Fig. 4, Fig. 5 and Table 2. The main ENSO maxima are at about 6.0, 4.0–4.2, 3.8–3.9, 3.0–3.1, 2.3–2.4, 1.8– 1.9, 1.5–1.6 years. Table 4 presents the short-period spectral maxima of the twelve series used for this work, compared with the abovementioned El Niño peaks. Ten out of the twelve series show periods around 2.3–2.4 years, which made apparent that the European temperature and phenology have a “feeling” with the 2.3–2.4 years periodicity of equatorial/tropical Pacific without forgetting the 50–67% frequencies of the periods 4.0–4.2, 3.8–3.9, 3.0–3.1 yr. It should also be noted (see Fig. 11sum of the Supplementary materials) that the period 3.8–3.9 years is referred only to the Nino1+2 region, the easternmost one of the equatorial Pacific, strictly linked to subtropical Atlantic Ocean (Alexander et al., 2002) whose influence on the European climate was stated for example by Seager et al. (2002).

Table 3 Spectral periods, band width, no. of peaks in the band, percent of the time series affected and presence of NAO/AMO signal for the 12 datasets used. Central period (years)

Band width (±%)

N

%

NAO/AMO

65 24 20.5 13.5 11.5 7.7 5.5 4.1 3 2.4

5 5 5 5 5 5 5 5 5 5

7 7 7 6 7 9 7 7 6 8

58 58 58 50 58 75 58 58 50 67

AMO AMO NAO NAO + AMO NAO NAO + AMO AMO NAO AMO NAO

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

261

Table 4 El Nino and the spectral maxima of the European series. N

Series

1 2 3 4 5 6 7 8 9 10 11 12

AMO NAO European Temperature Anomaly (ETA) Paris Montsouris Temperature Geneve-Cointrin Temperature CFD of Swiss Plateau GHD of Beaune Cote d'or – Burgundy GHD of 16 vineyards of Burgundy GHD of 16 vineyards of Burgundy. Dataset used by Tourre et al. (2011) GHD of Swiss plateau GHD of Haut Medoc/Bordeaux GHD of Tirano Total

Period yr 6

A simple Student's test has been computed via | t | = [(Xm − Xn)√N] / Sx to test the null hypothesis Ho:Xm − Xn = 0 (against H1:Xm − Xn ≠ 0) of a compatibility between the mean value Xm of the periods of all the series relative to any El Niño range Xn (see e.g. Table 4), Sx being the reduced (relative to N-1) standard deviation of Xm and N the dimension of any sample. Complete data and results (excel file STUDENT.xlsx of the Supplementary materials) show in summary that the null hypothesis must be rejected for the periods 3.8, 3, 3.05 years at 99% confidence and, in addition, for 4, 4.2, 3.9, 2.4, 1.5 years at 95% confidence. 5. Future developments and conclusions Climatic cycles relevant for the Euro-Atlantic area were studied in the three different segments of the MCTP causal chain. Results showed that the three levels of the causal chain are marked by the presence of cycles with central period ranging from 65 to 2.4 years and that can be related to AMO, NAO and ENSO. The analysis highlights the links among physical and biological variables of the climate system at scales that range from macro to microscale. The knowledge of such links is crucial to reach a suitable understanding of the eco-systemic response to climate forcing. We think that in the future this work could be improved in order to observe the time evolution of the Pacific-Atlantic-Europe links and analyzing how the macroscale signals from NAO, AMO and ENSO propagate towards the centre of the Eurasian continent. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2017.03.182. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Acknowledgments We thank the two reviewers for their valuable suggestions. References Alexander, M., Blade, I., Newman, M., Lanzante, J.R., Ngar, Cheung L., Scott, J., 2002. The atmospheric bridge: the influence of ENSO teleconnections on air–sea interaction over the global oceans. J. Clim. 15, 2205–2231. Angot, A., 1885. Etude sur les vendanges en France. Annales du Bureau Central Météorologique de France, a. I883, t. 1: B.29-120. Aono, Y., Kazui, K., 2008. Phenological data series of cherry tree flowering in Kyoto, Japan, and its application to reconstruction of springtime temperatures since the 9th century. Int. J. Climatol. 28, 905–914. Barcikowska, M., Knutson, T., Zhang, R., 2016. Observed and simulated fingerprints of multidecadal climate variability, and their contributions to periods of global SST stagnation. J. Clim. http://dx.doi.org/10.1175/JCLI-D-16-0443.1 (in press).

4–4.2

3.8–3.9

3–3.1

x x

x x x

x x

x x

x x x

x

3

x 6

x x x x x x 8

2.3–2.4 x x x x x x

1.8–1.9

1.5–1.6

x x x x

x

x x 6

x x x x 10

3

x x x 4

Barry, R.G., 1992. Mountain Weather and Climate. second ed. Routledge, London (404 pp.). Barry, R.G., Carleton, A.M., 2001. Synoptic and Dynamic Climatology. Routledge, New York, and London 620 pp. Berger, W.H., 2008. Solar modulation of the North Atlantic oscillation: assisted by the tides? Quat. Int. 188, 24–30. Berger, W.H., 2013. On the Milankovitch sensitivity of the Quaternary deep-sea record. Clim. Past 9, 2003–2011. Camuffo, D., Bartolin, C., 2012. The earliest temperature observations in the world: The Medici Network (1654–1670). Clim. Chang. 111 (2), 335–363 (July 2012). Camuffo, D., Jones, P.D., 2002. Improved Understanding of Past Climatic Variability From Early Daily European Instrumental Sources. Kluwer Academic Publisher http:// www.wkap.nl/prod/b/1-4020-0556-3 (sito visitato il 7 gennaio 2016). Carrara, P.E., McGeehin, J.P., 2015. Evidence of a higher late-Holocene treeline along the Continental Divide in central Colorado. The Holocene 25, 1829–1837 November 2015. Chevet, J.M., Soyer, J.P., 2006. Phénologie et climat dans le Haut-Médoc,1800-2005 (Vine phenology and climate in Bordeaux, since the beginning of the XIXth century). VIe Congrès International des Terroirs Viticoles 2006 (in French). Childers, D.G. (Ed.), 1978. Modern Spectrum Analysis. IEEE Press, New York (chapter II). Chuine, I., Yiou, P., Viovy, N., Seguin, B., Daux, V., Ladurie, E.L., 2004. Historical phenology: grape ripening as a past climate indicator. Nature 432:289–290 Data available at. https://www.ncdc.noaa.gov/cdo/f?p=519:1:::::P1_STUDY_ID:5414. Cola, G., Failla, O., Maghradze, D., Megrelidze, L., Mariani, L., 2016. Grapevine phenology and climate change in Georgia. Int. J. Biometeorol. http://dx.doi.org/10.1007/ s00484-016-1241-9. CRU, 2016. North Atlantic Oscillation (NAO). https://crudata.uea.ac.uk/cru/data/nao/ (web site visited at 15 November 2016). Daux, V., Garcia de Cortazar-Atauri, I., Yiou, P., Chuine, I., Garnier, E., Le Roy Ladurie, E., Mestre, O., Tardaguila, J., 2012. An open-access database of grape harvest dates for climate research: data description and quality assessment. Clim. Past 8:1403–1418 (data available at: http://www.clim-past.net/8/1403/2012/cp-8-1403-2012supplement.zip). Dufour, L., 1870. Notes sur le problème de la variation du climat. Bull. Soc. Vaud. Sci. Nat. 10, 359–556 (in French). Enfield, D.B., Mestas-Nunez, A.M., Trimble, P.J., 2001. The Atlantic Multidecadal Oscillation and its relationship to rainfall and river flows in the continental U.S. Geophys. Res. Lett. 28, 2077–2080. Ghil, M., Allen, R.M., Dettinger, M.D., Ide, K., Kondrashov, D., Mann, M.E., Robertson, A., Saunders, A., Tian, Y., Varadi, F., Yiou, P., 2002. Advanced spectral methods for climatic time series. Rev. Geophys. 40 (1):3.1–3.41. http://dx.doi.org/10.1029/2000RG000092. Humlum, O., Stordahl, K., Solheim, J.E., 2011. Spectral analysis of the Svalbard temperature record 1912–2010. Adv. Meteorol. 2011. http://dx.doi.org/10.1155/2011/ 175296. Hurrell, J.W., 1995. Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation. Science 269, 676–679. Jiang, H., Muscheler, R., Björck, S., Seidenkrantz, M., Olsen, J., Sha, L., Sjolte, J., Eiríksson, J., Ran, L., Knudsen, K., Knudsen, M.F., 2015. Solar forcing of Holocene summer sea-surface temperatures in the northern North Atlantic. Geology February 2015. 10.1130/ G36377.1. Jouzel, J., 2013. A brief history of ice core science over the last 50 yr. Clim. Past 9, 2525–2547. Kilbourne, K.H., 2014. A paleoclimate perspective on Atlantic multidecadal variability. J. Mar. Syst. 133 (2014), 4–13. Knudsen, M.F., Seidenkrantz, M., Jacobsen, B.H., Kuijpers, A., 2011. Tracking the Atlantic Multidecadal Oscillation through the last 8,000 years. Nat. Commun. 2, 178. Kuhn, N., Guan, L., Dai, Z.W., Wu, B.H., Lauvergeat, V., Gomès, E., et al., 2014. Berry ripening: recently heard through the grapevine. J. Exp. Bot. 65:4543–4559. http://dx.doi. org/10.1093/jxb/ert395. Labbé, T., Gaveau, F., 2013. Les dates de vendange à Beaune (1371–2010). Analyse et données d'une nouvelle série vendémiologique. Revue historique 666 (2013/2), 333–367. Le Roy Ladurie, E., 1976. Tempo di festa, tempo di carestia, storia del clima dall'anno 1000 Torino. (in Italian).

262

L. Mariani, F. ZavattiScience of the Total Environment 593–594 (2017) 253–262

Le Roy Ladurie, E., 2004. Histoire humaine et comparée du climat. I. Canicules et glaciers (XIIIe-XVIIIe siècles). Fayard, Paris (740 pp). Lieth, H., 1974. Phenology and Seasonality Modeling. Ecological Studies 8. SpringerVerlag, Berlin (444 pp.). Lomb, N.R., 1975. Least-squares frequency analysis of unequally spaced data. Astrophys. Space Sci. 39, 447–462. Lüdecke, H.J., Hempelmann, A., Weiss, C.O., 2013. Multi-periodic climate dynamics: spectral analysis of long-term instrumental and proxy temperature records. Clim. Past 9: 447–452. http://dx.doi.org/10.5194/cp-9-447-2013. Mariani, L., Parisi, S., 2013. Extreme rainfalls in the Mediterranean area. In: Diodato, Bellocchi (Eds.), Storminess and Environmental Changes: Climate Forcing and Responses in Mediterranean Region. Springer. Mariani, L., Parisi, S.G., Cola, G., 2009a. Space and time behavior of climatic hazard of low temperature for single rice crop in the mid latitude. Int. J. Climatol. 29:1862–1871. http://dx.doi.org/10.1002/joc.1830. Mariani, L., Parisi, S., Failla, O., Cola, G., Zoia, G., Bonardi, L., 2009b. Tirano (1624–1930): a long time series of harvest dates for grapevine. Ital. J. Agrometeorol. 1, 7–16. Martínez-Lüscher, J., Kizildeniz, T., Vucetic, V., Dai, Z., Luedeling, E., van Leeuwen, C., Gomès, E., Pascual, I., Irigoyen, J.J., Morales, F., 2016. Sensitivity of Grapevine Phenology to Water Availability, Temperature and CO2 Concentration. Front. Environ. Sci. 4 (article 48). Maurer, C., Hammerl, C., Koch, E., Hammerl, T., 2011. Extreme grape harvest data of Austria, Switzerland and France from A.D. 1523 to 2007 compared to corresponding instrumental/reconstructed temperature data and various documentary sources. Theor. Appl. Climatol. 106, 55–68. McCarthy, G.D., Haigh, I.D., Hirschi, J.J.M., Grsit, J.P., Smeed, D.A., 2015. Ocean impact on decadal Atlantic climate variability revealed by sea-level observations. Nature 521 508Ð510. 10.1038/nature14491. McGovern, P., 2003. Ancient Wine: The Search for the Origins of Viniculture. Princeton University Press (365 pp). Meier, N., Rutishauser, T., Pfister, C., Wanner, H., Luterbacher, J., 2007. Grape harvest dates as a proxy for Swiss April to August temperature reconstructions back to AD 1480. Geophys. Res. Lett. 34 (L20705):2007. http://dx.doi.org/10.1029/2007GL031381. NOAA, 2016. Linear Correlations in Atmospheric Seasonal/Monthly Averages. http:// www.esrl.noaa.gov/psd/data/correlation/ (web site visited at 15 November 2016). Olafsdottir, M.B., 2010. Time Series Analysis Based on Paleoclimatic Proxies From Lake Sediments in Iceland. MS Thesis. University of Iceland, Reykjavik (February 2010). Parisi, S.G., Antoniazzi, M.M., Cola, G., Lovat, L., Mariani, L., Moreale, G., Zoltan, K., Calò, A., 2014. Spring thermal resources for grapevine in Koszeg (Hungary) deduced from a very long pictorial time series (1740–2009). Clim. Chang. http://dx.doi.org/10.1007/ s10584-014-1220-2. Press, W.H., Teukolsky, S.A., Vetterling, W.T., Flannery, B.P., 2009. Numerical Recipes in Fortran. Cambridge University Press India Pvt. Ltd., New Delhi. Reid, P.C., Hari, R.E., Beaugrand, G., Livingstone, D.M., Marty, C., Straile, D., Barichivich, J., Goberville, E., Adrian, R., Aono, Y., Brown, R., Foster, J., Groisman, P., Hélaouët, P., Hsu, H.H., Kirby, R., Knight, J., Kraberg, A., Li, J., Lo, T.T., Myneni, R.B., North, R.P., Pounds, J.A., Sparks, T., Stübi, R., Tian, Y., Wiltshire, K.H., 2016. Global impacts of the

1980s regime shift. Glob. Chang. Biol. 22 (682–703):2016. http://dx.doi.org/10. 1111/gcb.13106. Rutishauser, T., 2008. Swiss Cherry Tree Phenology 1721-2000 AD.IGBP PAGES/World Data Center for Paleoclimatology Data Contribution Series # 2008-080. NOAA/NCDC Paleoclimatology Program, Boulder CO, USA ftp://ftp.ncdc.noaa.gov/pub/data/paleo/ historical/europe/switzerland/swiss-cherry-phenology2003.txt. Scargle, J.D., 1982. Studies in astronomical time series analysis. II. Statistical aspects of spectral analysis of unevenly spaced data. Astrophys. J. 263, 835–853. Schlesinger, M.E., Ramankutty, N., 1994. An oscillation in the global climate system of period 65–70 years. Nature 367, 723–726. Seager, R., Battisti, D.S., Yin, J., Gordon, N., Naik, N., Clement, A.C., Cane, M.A., 2002. Is the Gulf Stream responsible for Europe's mild winters? Q. J. R. Meteorol. Soc. 128, 2563–2586 2002. Sutton, R.T., Dong, B., 2012. Atlantic Ocean influence on a shift in European climate in the 1990s. Nat. Geosci. 5, 788–792 November 2012. Thiéblemont, R., Matthes, K., Omrani, N.E., Kodera, K., Hansen, F., 2015. Solar forcing synchronizes decadal climate variability North Atlantic. Nat. Commun. http://dx.doi.org/ 10.1038/ncomms9268. Tourre, Y.M., Rousseau, D., Jarlan, L., Le Roy Ladurie, E., Daux, V., 2011. Western European climate, and Pinot noir grape harvest dates in Burgundy, France, since the 17th century. Clim. Res. 46:243–253. http://dx.doi.org/10.3354/cr00991d. Tularam, G.A., Ilahee, M., 2010. Time series analysis of rainfall and temperature interactions in coastal catchments. J. Math. Stat. 6 (3), 372–380. Wilczek, A.M., Burghardt, L.T., Cobb, A.R., Cooper, M.D., Welch, S.M., Schmitt, J., 2010. Genetic and physiological bases for phenological responses to current and predicted climates. Philos. Trans. R. Soc. B 365:3129–3147 2010. 10.1098/rstb.2010.0128. Wilson, R.J.S., Luckman, B.H., Esper, J., 2005. A 500 year dendroclimatic reconstruction of spring-summer precipitation from the lower Bavarian forest region, Germany. Int. J. Climatol. 25:611–630. http://dx.doi.org/10.1002/joc.1150 (data at. https://www. ncdc.noaa.gov/paleo-search/study/6334. Wolter, K., Timlin, M.S., 2011. El Niño/Southern Oscillation behaviour since 1871 as diagnosed in an extended multivariate ENSO index (MEI.ext). Int. J. Climatol. 31 (7): 1074–1087. http://dx.doi.org/10.1002/joc.2336. Zanzi, A., Pelfini, M., Muttoni, G., Santilli, M., Leonelli, G., 2007. Spectral analysis on mountain pine tree-ring chronologies. Dendrochronologia 24:145–154. http://dx.doi.org/ 10.1016/j.dendro.2006.10.002. Zeng, N., Zhao, F., Collatz, G.J., Kalnay, E., Salawitch, R.J., West, T.O., Guanter, L., 2014. Agricultural Green Revolution as a driver of increasing atmospheric CO2 seasonal amplitude. Nature 5015:394–397. http://dx.doi.org/10.1038/nature13893. Zhang, R., Zuo, Z., 2011. Impact of spring soil moisture on surface energy balance and summer monsoon circulation over East Asia and precipitation in East China. J. Clim.:3309–3322 July 2011. 10.1175/2011JCLI4084.1. Zohary, D., Hopf, M., Weiss, E., 2012. Domestication of Plants in the Old World: The Origin and Spread of Domesticated Plants in Southwest Asia, Europe, and the Mediterranean Basin. Oxford University Press.