Agricultural and Forest Meteorology 201 (2015) 153–164

Contents lists available at ScienceDirect

Agricultural and Forest Meteorology journal homepage: www.elsevier.com/locate/agrformet

Summer drought and ENSO-related cloudiness distinctly drive Fagus sylvatica growth near the species rear-edge in northern Spain Vicente Rozas a,∗ , J. Julio Camarero b , Gabriel Sangüesa-Barreda b , Manuel Souto c , Ignacio García-González c a

Misión Biológica de Galicia, Consejo Superior de Investigaciones Científicas (MBG-CSIC), Apdo. 28, E-36080 Pontevedra, Spain Instituto Pirenaico de Ecología, Consejo Superior de Investigaciones Científicas (IPE-CSIC), Avda. Monta˜ nana 1005, Apdo. 202, E-50192 Zaragoza, Spain c Departamento de Botánica, Escola Politécnica Superior, Campus de Lugo, Universidade de Santiago de Compostela, E-27002 Lugo, Spain b

a r t i c l e

i n f o

Article history: Received 30 May 2014 Received in revised form 4 November 2014 Accepted 11 November 2014 Keywords: Dendrochronology Drought Latitudinal gradient Atlantic climate Mediterranean climate Teleconnection

a b s t r a c t The ample distribution of common beech (Fagus sylvatica) across Europe implies that this key tree species occurs under a broad variety of climatic conditions despite its sensitivity to drought stress. Iberian beech rear-edge (southernmost) forests are located along the boundary between the Eurosiberian and Mediterranean biogeographical regions. Therefore, those forests are considered to be sensitive monitors of the effects of warming-induced drought stress on marginal tree populations. We evaluate if the radial growth of Iberian beech populations is mainly constrained by drought. Since previous findings indicated that El ˜ Nino-Southern Oscillation (ENSO) teleconnections may influence the rainfall regime in northern Spain, we also assessed if beech response to drought and water availability is modulated by this large-scale climatic pattern. We compared the recent growth patterns and responses to climate across a network of 30 tree-ring site chronologies established throughout northern Spain where beech forests are subjected to contrasting climatic conditions. Iberian beech populations located near or within the Mediterranean biogeographical region were the most sensitive to June water deficit. However, the dependency of beech growth on drought stress near the rear-edge of the species was mitigated where cloudy conditions prevail in summer, namely in mesic stands located in the Eurosiberian region. Drought stress in the latter populations was alleviated by cloudiness, which in turn depended on ENSO, and this effect on growth has been intensifying for the last decades. We prove that the sensitivity of rear-edge populations to drought, in terms of growth reduction, is greatly modulated by local or regional environmental gradients, but also by the influence of large-scale climatic variation. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Common beech (Fagus sylvatica L.) is one of the most widely distributed tree species over European forests. This highlycompetitive, late-successional, deciduous hardwood tree species shows a wide tolerance to climatic conditions, growing under continental, sub-boreal, oceanic and sub-Mediterranean climates, but usually requires a humid atmosphere and a well-drained soil (Fang and Lechowicz, 2006). F. sylvatica forms pure or mixed stands over large areas, and therefore it is considered a key species since its

∗ Corresponding author. Tel.: +34 986 854800; fax: +34 986 841362. E-mail addresses: [email protected], [email protected] (V. Rozas), [email protected] (J.J. Camarero), [email protected] (G. Sangüesa-Barreda), [email protected] (M. Souto), [email protected] (I. García-González). http://dx.doi.org/10.1016/j.agrformet.2014.11.012 0168-1923/© 2014 Elsevier B.V. All rights reserved.

forests contain a high biodiversity and they encompass a wide range of habitat types at elevations ranging from sea level to the upper tree line (von Wuehlisch, 2008; Packham et al., 2012). Beech is also known to be sensitive to water deficit mainly due to its vulnerability to xylem cavitation leading to a loss in hydraulic conductivity and enhanced tree mortality under severe drought conditions (Herbette et al., 2010; Barigah et al., 2013). In fact, the geographical distribution of this species depends on its low tolerance to summer drought, being therefore abundant at sites with sub-oceanic and temperate climates, i.e. subjected to cool and mesic conditions (Giesecke et al., 2007; Bradshaw et al., 2010). In the Mediterranean Basin, F. sylvatica grows in mountain areas where rainfall is high enough, just barely within the limit of its requirements (Fotelli et al., 2009). On the other hand, low temperatures that induce late frosts and shorten the length of the growing season determine the upper altitudinal, northern latitudinal and eastern longitudinal limits of distribution of the species (Bolte et al., 2007; Maxime and Hendrik, 2011).

154

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

Much research has recently focused on the mechanisms of F. sylvatica drought tolerance either by studying saplings in provenance trials or by sampling adults living in rear-edge populations, i.e. stands forming the southernmost or xeric limit of distribution (Aranda et al., 2000; Nahm et al., 2006; Jump et al., 2007; Meier and Leuschner, 2008; Rose et al., 2009; Pluess and Weber, 2012; Robson et al., 2013). This research has stimulated an active debate on the future performance of rear-edge F. sylvatica forests under the prospect of a drier and warmer climate (e.g., Gessler et al., 2007; Kramer et al., 2010; Meier et al., 2011). From a dendrochronological perspective, adult F. sylvatica trees have been relatively well studied with networks of tree-ring chronologies developed in central, southern, western and northern European forests (Dittmar et al., 2003; Lebourgeois et al., 2005; Piovesan et al., 2005; Di Filippo et al., 2007; Friedrichs et al., 2009; Drobyshev et al., 2010; Babst et al., 2013; Tegel et al., 2014). Several studies showed coherent geographic patterns in the tree-ring growth response to climate, but these also depend upon local environmental gradients modulated by altitude or latitude (Piovesan et al., 2005; Di Filippo et al., 2007). For instance, analyses of F. sylvatica tree-ring width chronologies from southwestern Europe have shown that May temperature and water availability in summer, together with the amount of winter precipitation, are the major drivers of radial growth (Biondi, 1993; Piovesan et al., 2005; Lebourgeois et al., 2005). On the other side, drought is the major constraint of growth near the F. sylvatica southwestern range boundary in northern Spain (Gutiérrez, 1988; Rozas, 2001; Jump et al., 2007). Growth dynamics of western European F. sylvatica forests may also reflect large-scale atmospheric circulation patterns, as was the case for the Northern Atlantic Oscillation (NAO) in Italy (Piovesan and Schirone, 2000). In the Iberian Peninsula, climate and hydrological conditions are transitional between Atlantic influences westwards and Mediterranean influences eastwards, being therefore not only influenced by the NAO but also by other atmospheric dynamics operating at large distances (teleconnection) like the ˜ El Nino-Southern Oscillation (ENSO) (Rodó et al., 1997; Knippertz et al., 2003; Pozo-Vázquez et al., 2005). In fact, ENSO represents the strongest interannual variation of Earth’s climate affecting a wide range of geographic areas (Stenseth et al., 2003), but we still do not know how ENSO affects the long-term growth of rear-edge Iberian F. sylvatica forests. We hypothesize that F. sylvatica growth in the Iberian Peninsula is (i) mainly constrained by water scarcity across regional and local geographical gradients, and (ii) such drought response is modulated by the ENSO variability, as it has been recently found for other tree species in Spain (Rozas and García-González, 2012a, Rozas and García-González, 2012b). In this work, we tested both hypotheses by comparing growth responses to climate across a network of tree-ring chronologies established throughout northern Spain where F. sylvatica forests are subjected to diverse climatic conditions. These rear-edge populations represent the southwestern distribution limit of the species, being located along the boundary between the Eurosiberian and Mediterranean biogeographical regions. Therefore, they are considered sensitive forests potentially recording the effects of ongoing warming and drying trends on tree populations.

2. Materials and methods 2.1. Study area We sampled 30 forest stands located along a wide area in northern Spain, at the southwestern boundary of the species range (Fig. 1a), ranging from 42.10◦ N to 43.43◦ N, and from 0.37◦ W to 7.07◦ W. Sites were distributed along a broad elevation range

between 140 and 1400 m a.s.l. (Table 1), which covers the complete elevation range of F. sylvatica in the study region, even if the upper forest limit is mainly due to historical deforestation of mountain areas for pasture. Sampled stands were more abundant in the central Cantabrian area, including lowland mixed deciduous stands (Supplementary material, Fig. S1a), as well as monospecific stands on littoral mountains and on the Cantabrian Range (Fig. S1b). In addition, pure or mixed stands were also sampled on the western Cantabrian Range, Galician mountains, mountains of Álava and Navarra, Iberian Range, pre-Pyrenean mountains, and the western Pyrenees including sites located in the Eurosiberian and Mediterranean regions, and also in transitional areas (Fig. 1b). Associated tree species in the Cantabrian area were mainly deciduous oaks (Quercus robur, Q. petraea, Q. pyrenaica), and other species such as Fraxinus excelsior, Acer campestre, Acer pseudoplatanus or Taxus baccata (Table 1). In the pre-Pyrenees, however, the main co-dominant species was Pinus sylvestris (Fig. S1c), and Abies alba in the Pyrenees (Fig. S1d). A considerable diversity of habitats occupied by F. sylvatica in northern Spain were included in our sampling, with abundant representation of western Cantabrian acidofilous, sub-humid neutrophilous oro-Cantabrian, and sub-Mediterranean calcicolous F. sylvatica forests (Table 1). F. sylvatica forests in the Cantabrian area are mainly associated with foggy conditions, under the influence of wet winds from the Cantabrian Sea. These stands occur in lowland hilly areas (Fig. S1e), and on northern-faced mountain slopes (Fig. S1f). There, soils were mainly deep brown on slightly acidic, neutral or basic bedrock, even if some study stands were located on calcium-rich karst plateaus (e.g., sites CUE and URB). The climate in the study area is diverse, varying from temperate humid Atlantic conditions without a dry period westwards or northwards, to sub-humid Mediterranean conditions with a marked summer drought eastwards or southwards (Fig. 1). The majority of stands were located within the Eurosiberian biogeographical region, but eleven stands were near the Eurosiberian-Mediterranean boundary (FON, LIN, SIS, ZAN, VAL, HIJ, IZK, URB, LUE, PEI, MON), and one stand from the Iberian Range (DIU) was located well within the Mediterranean Region (Fig. 1b). This referred boundary between Eurosiberian and Mediterranean biogeographical regions in the northern Iberian Peninsula is based on both bioclimatic and botanical criteria (Rivas-Martínez et al., 2002). The climate in the Cantabrian and Pyrenean Ranges is humid temperate, with a mean annual precipitation ranging from 930 to 1945 mm, and a mean annual temperature of 8.3 to 10.2 ◦ C. In the mountains of Álava and Navarra, there are transitional conditions between Atlantic and Mediterranean climates, with mean annual precipitation of over 900 mm and around 9.8 ◦ C in mean annual temperature. A dry period in summer, mainly in July and August, was evident in the Galician mountains and the western Cantabrian Range (Fig. 1). Dry summer was more remarkable under Mediterranean climate conditions in the pre-Pyrenees, but also in the Iberian Range (site DIU) where the climate is continental Mediterranean, with a mean annual precipitation of over 660–670 mm, and a mean annual temperature of 10.6 to 11.9 ◦ C. 2.2. Field sampling, sample processing and tree-ring measuring Most of the sampled stands were selected in protected areas (e.g., Regional and National Parks) preferentially without signs of recent human disturbance. Latitude and longitude, elevation, dominant tree species, and European Nature Information System (EUNIS) habitat type (Davies et al., 2004; http://eunis. eea.europa.eu/habitats.jsp) were recorded for every sampled stand. We randomly chose six to 43 dominant or co-dominant trees at each stand, separated by at least 10 m, for sampling between 1998 and 2012. We collected two wood cores per tree, along

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

155

Fig. 1. Study area located at the southwestern distribution limit of the Fagus sylvatica European range ((a) modified from von Wuehlisch, 2008) showing the location of the study sites in the northern Iberian Peninsula (b). The grey line in (b) shows the limit for the Eurosiberian (nortwards) and Mediterranean (southwards) biogeographical regions according to Rivas-Martínez et al. (2002). Blue dots indicate Eurosiberian sites, green dots indicate transitional sites between the Eurosiberian and Mediterranean regions, and the red dot indicates the Mediterranean site. Site codes are as in Table 1. Climatic diagrams of O Courel- western Cantabrian Range (A), central Cantabrian Range (B), western Pyrenees (C), mountains of Álava-Navarra (D), Iberian Range (E), and pre-Pyrenean mountains (F), for the period 1940–2000, are shown (abbreviations: T, mean monthly temperature; Prec , monthly total precipitation).

opposing radii oriented parallel to the slope contour line when possible, using increment borers at 1.3 m. Bole diameter at 1.3 m or breast height (DBH) was also registered for each sampled tree. The wood cores were air dried, glued onto wooden mounts, surfaced and finely polished until the xylem cellular structure was clearly visible in transversal section. Tree-ring series along the cores were dated by assigning calendar years to the rings through the identification of characteristic ring sequences. Total ring widths were measured to the nearest 0.001 mm using either Velmex or LINTAB sliding-stage micrometers interfaced with a computer. The computer program COFECHA (Grissino-Mayer, 2001) served to quantitatively check for potential absent rings and cross-dating errors. We individually estimated tree age at coring height based on the oldest core per tree, and no correction for the number of missing rings due to coring height was performed. 2.3. Chronologies calculation and assessment To obtain detrended tree-ring width indices, every raw ring-width series was standardized with the ARSTAN computer program (Cook and Holmes, 1996). The series were fitted to a spline function with a 50% frequency response of 32 years, which was flexible enough to maximize the high-frequency climatic information, and minimize the non-climatic variance related to ontogenetic trends and/or local disturbances (Helama et al., 2004). The indices were obtained as ratios, and calculated by dividing each measured ring width by its expected value, according to fitted spline function. The obtained indices were pre-whitened by

autoregressive modelling. Each series was modelled as an autoregressive process where the order was selected by first minimum Akaike Information Criterion. Autoregressive modelling gives dimensionless indices that represent independent, normalized, and homogenized records of annual growth for each measured series. All growth indices from a site were averaged on an annual basis into a whole population chronology using a biweight robust mean. The reliable portion of each chronology to be used was the period with at least five trees included in the chronology. The similitude among all site chronologies was assessed by calculating a correlation matrix among all chronologies. Temporal variation of the similitude among chronologies and their common signal was assessed by running correlations and by the percentage of tree-ring growth variation accounted for by the first component (PC1) of a Principal Component Analysis (PCA), respectively. Both statistics were computed in 35-yr intervals lagged by five years. The statistical quality of the chronologies was assessed for 1942–1998, the common period of all chronologies, using standard basic statistics to measure the common signal (Fritts, 2001), such as the first order autocorrelation (AC1), mean sensitivity (MS), mean between-trees correlation (Rbt), signal-to-noise ratio (SNR), and expressed population signal (EPS). Common patterns of growth behaviour within our tree-ring network were explored using twotailed Pearson correlations and PCA calculated on the 30 chronologies for the common period 1942–1998. PCA transformed our collection of chronologies into a set of principal components (PCs), calculated on the covariance matrix of the original data. We assessed the number of PCs to be retained in two steps. We initially selected those PCs potentially accounting for the complete variation in the

156

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

Table 1 Stand and tree characteristics for the sampled Fagus sylvatica stands in northern Iberian Peninsula. The site name and code, geographical latitude and longitude, elevation, mean DBH and age values, maximum age, dominant tree species, and EUNIS habitat type, are shown. Site name (Province)

Code

North latitude (◦ )

West longitude (◦ )

Elevation (m)

DBH (cm)a

Age (yrs.)a

˜ (Cantabria) Hayal de Alonos Alto Asón (Cantabria) ˜ de Bértiz (Navarra) Senorío Caviedes (Cantabria) Sierra de Cuera (Asturias) Hayedo de Diustes (Soria) Fonteformosa (Lugo) Fuente De (Cantabria) Gamueta (Huesca) Monte Hijedo (Cantabria) Izki (Álava) ˜ (Lugo) Linares Luesia (Zaragoza) Puerto de Monrepós (Huesca) Monte Muniacos (Asturias) Monte Peiró (Huesca) Hayedo de Pome (Asturias) ˜ (Asturias) Puerto de Ranadoiro ˜ Sagra (Cantabria) Pena Hayedo del Saja (Cantabria) O Sisto (Lugo) Soto de Sajambre (León) Sierra del Sueve (Asturias) Monte Tejas (Cantabria)

ALO ASO BER CAV CUE DIU FON FUE GAM HIJ IZK LIN LUE MON MUN PEI POM RAN SAG SAJ SIS SOT SUE TEJ

43.23 43.18 43.17 43.33 43.35 42.10 42.63 43.13 42.88 42.90 42.67 42.68 42.40 42.37 43.25 42.37 43.27 42.98 43.13 43.10 42.65 43.17 43.43 43.23

3.88 3.62 1.62 4.28 4.77 2.42 7.03 4.80 0.78 3.95 2.43 7.07 1.08 0.37 5.30 0.55 5.02 6.62 4.52 4.25 7.07 5.02 5.22 4.02

720 1260 300 140 1150 1320 1330 1210 1400 915 800 1200 1250 1290 775 1360 1100 1140 875 830 860 1350 530 520

60.8 51.5 46.5 61.8 57.1 54.1 40.9 56.9 62.8 50.2 58.4 37.9 51.3 26.3 66.9 46.0 66.7 52.9 60.9 69.5 29.3 56.6 92.7 71.3

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

6.6 8.0 4.6 10.3 22.3 9.6 7.2 11.5 12.4 11.3 6.2 14.5 10.9 4.8 12.1 8.7 10.5 7.4 13.4 14.4 11.9 9.8 23.3 11.2

114 180 125 191 139 90 83 130 152 109 93 70 70 90 117 97 166 140 88 96 57 97 97 70

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

Monte Texedo (Asturias) Ucieda (Cantabria) Sierra de Urbasa (Navarra) Monte Valgrande (Asturias) Puerto de Ventana (Asturias) Zanfoga (Lugo)

TEX UCI URB VAL VEN ZAN

43.28 43.20 42.83 42.98 43.08 42.65

5.18 4.22 2.15 5.78 6.05 7.05

680 590 920 890 1175 1030

62.7 75.3 86.8 65.3 52.5 37.2

± ± ± ± ± ±

14.4 15.5 18.5 18.0 14.9 9.7

99 95 144 134 118 65

± ± ± ± ± ±

Maximum age (yrs.)

Dominant speciesb

EUNIS habitatc

19 50 13 32 48 5 9 19 29 25 18 14 16 17 29 23 57 20 23 33 9 14 19 28

146 263 152 267 250 95 95 159 192 177 138 95 95 113 156 132 274 178 126 206 73 115 139 130

G1.624 G1.643 G1.8621 G1.8622 G1.643 G1.627 G1.626 G1.643 G1.675 G1.8623 G1.675 G1.626 G1.675 G1.675 G1.8623 G1.675 G1.643 G1.625 G1.625 G1.625 G1.626 G1.643 G1.643 G1.A19

23 20 27 44 23 11

164 138 193 239 155 89

FS FS QR, FS, QP FS, QR FS FS FS FS FS, AA QP, FS, TB FS FS FS, PS FS, PS FS, QP FS, TB, PS FS FS FS FS, QP FS, AP FS, QP FS QR, FE, FS, AC, TB FS, QR QR, FS FS FS FS FS, QY

G1.643 G1.8622 G1.675 G1.625 G1.625 G1.626

Mean ± SD. FS: Fagus sylvatica, QR: Quercus robur, QP: Quercus petraea, QY: Quercus pyrenaica, FE: Fraxinus excelsior, AA: Abies alba, AC: Acer campestre, AP: Acer pseudoplatanus, PS: Pinus sylvestris, TB: Taxus baccata. c European Nature Information System (EUNIS) habitat types classification [Davies et al. (2004), http://eunis.eea.europa.eu/habitats.jsp]. G1.A19: Pyreneo-Cantabrian oak-ash forest. G1.624: Pyreneo-Cantabrian acidophilous beech forest. G1.625: Western Cantabrian acidophilous beech forest. G1.626: Galician acidophilous beech forest. G1.627: Humid Iberian acidophilous beech forest. G1.643: Sub-humid neutrophilous oro-Cantabrian beech forest. G1.675: Sub-Mediterranean calcicolous beech forest. G1.8621: Eastern Cantabrian acidophilous oak forest. G1.8622: Western Cantabrian acidophilous oak forest. G1.8623: Oro-Cantabrian acidophilous oak forest. a

b

chronologies network using the PVP criterion; this is an a priori method to retain a relatively large number of PCs, by determining the point where the cumulative product of the eigenvalues falls just below 1 (Guiot, 1985). Afterwards, the final number of meaningful PCs was obtained with the restrictive Kaiser–Guttman criterion; for this, we computed the mean of the eigenvalues of the retained PCs, and interpreted only the axes whose eigenvalues were larger than that mean (Legendre and Legendre, 1998). Orthogonal rotation of the PC axes was performed according to the Varimax criterion, which maximizes the spread of original PC loadings. Principal component analysis was performed with the statistical package SPSS ver. 15.0 (SPSS Inc., Chicago IL, USA). 2.4. Computing tree-ring growth responses to climate The climatic data used in this study were monthly gridded time series for maximum mean temperatures (abbreviated Tmax ), total precipitation (Prec ), and cloud fraction (Cloud) obtained from the CRU TS 3 data set for the period 1901–2009 (Mitchell and Jones, 2005). CRU climate time series were taken from the Climate Explorer of the Royal Netherlands Meteorological Institute (http://climexp.knmi.nl/) for the 0.5◦ longitude × 0.5◦ latitude sector in which the study stands were located. Several indices describing ENSO dynamics (Southern Oscillation Index, Sea Surface Tem˜ 1.2, 3, 4, and 3.4 in the tropperature indices from the regions Nino ical Pacific Ocean) were obtained from the National Oceanic and Atmospheric Administration website (http://www.cdc.noaa.gov/). Among these, only the Sea Surface Temperature (SST) index from

˜ 3.4 region (hereafter SST 3.4) showed significant relathe Nino tionships with chronologies, and was further used to evaluate the effects of ENSO on F. sylvatica growth and regional climate. The SST 3.4 time series covers the period 1872–2007, and it represents SST anomalies relative to the 1950–1979 base period in the central equatorial Pacific area bounded by latitude 5◦ S–5◦ N and longitude ˜ 3.4 region captures a 120–170◦ W (Stenseth et al., 2003). The Nino ˜ time scales, and it is close to the region large variability on El Nino where changes in local sea-surface temperature are important for shifting rainfall patterns across the equatorial Pacific (Trenberth, 1997). The temporal window for climatic predictors was taken from April of the year prior to ring formation (Apr(−1)) to September of the year of ring formation (Sep). Pearson’s correlations were used to estimate the response of the selected PCs of the whole network to the monthly climate time series, considering the common period 1942–1998. The geographical structure and consistency of the most significant tree-ring growth-to-climate correlations at local level were, respectively, assessed by kriging interpolation of the Pearson’s correlations of each local chronology to climate, and by calculating Moran’s I coefficients of spatial autocorrelation. Kriging is one of the most flexible methods for interpolated mapping, and Moran’s I coefficient provides statistical testing for the presence of spatial autocorrelation (Legendre and Legendre, 1998). For more details on kriging and Moran’s I coefficient see the Supplementary material (Appendix S1). The dependency of the growth-to-climate correlations on geographical (north latitude, west longitude, elevation) and stand (mean DBH, mean and maximum tree ages)

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

predictors was quantified by using linear regressions. We also calculated moving Pearson correlations, in intervals of 35-yr width shifted year-by-year, to assess the temporal consistency of treering growth-to-climate relationships, and of ENSO-to-local climate relationships. Main temporal trends of temperature and precipitation were assessed by calculating mean values, in intervals of 35-yr width shifted year-by-year, over the study area (latitude 42–43.5◦ N, longitude 0–7.5◦ W). Lastly, the geographical variation of the relationships of the PC scores of the whole network and ENSO as related to climate in northern Spain was assessed by calculating spatial field correlations using Climate Explorer (http://climexp.knmi.nl/). 3. Results 3.1. Stand and tree-ring network characteristics Mean DBHs of the sampled trees ranged between 26.3 and 92.7 cm, and mean tree ages varied between 57 and 191 years, with maximum ages ranging between 73 and 274 years (Table 1). Larger trees were mainly found in northernmost stands, located in the central Cantabrian area and at a low elevation. Oldest trees were also mainly located in northernmost stands, irrespective of their elevation, while youngest trees were found in southernmost stands, mainly on the Iberian Range and on mountains near the Eurosiberian-Mediterranean boundary. The overall mean ring width of the whole network was 2.09 mm. Mean ring width at each stand was significantly and negatively related to mean tree age (r = −0.58, P < 0.001), with older stands showing lower growth rates. The indexed tree-ring chronologies showed a high statistical quality, with absent rings at only eight localities, though with a low frequency. MS values ranged 0.123–0.301, Rbt values ranged

157

0.260–0.575, and SNR values ranged 3.87–24.35. Elevated EPS values, greater than 0.85, were observed throughout the tree-ring network, with the unique exception of site SIS (Table 2). These values suggest an appropriate replication and a very robust common signal, probably of climatic origin, among radial growth of trees within every studied stand. As regards chronology descriptive statistics, only Rbt was significantly and positively related to elevation (r = 0.39, P = 0.032). The other descriptive statistics were independent of any geographical and stand variables considered. The correspondence between radial growth variability was geographically consistent and shared by most of the studied localities, which fitted well to a pattern of common variation (Fig. 2a). Mean correlation among all chronologies was high over the 20th century, with a great proportion of tree-ring growth variation explained by the first principal component ranging between 32 and 50% since the mid-19th century (Fig. 2b). We observed a lower similarity among chronologies from the 1950s to the 1970s, when extreme growth values were rare and summers were relatively wet, which suggest that annual growth dispersion from the global average was low. The observed discrepancy in the initial period between PC1 variation and mean correlation could be due to a lower sample size or a weaker climatic response of younger individuals, with a resulting lower similarity among chronologies. Correlations between site chronologies were high and significant mostly for close stands and stands located at a similar elevation, but even for distant sites located more that 500 km away. In fact, elevation difference was more important than intersite distance in explaining the similitude within tree-ring growth variation. Inter-site distance explained 7.5% of the correlation variation between tree-ring growth (Supplementary material, Fig. S2a), while elevation difference explained 19.9% of this variation (Fig. S2b).

Table 2 Summary of descriptive statistics for tree-ring width chronologies of Fagus sylvatica at the study stands and considering the common period 1942–1998. Code

No. trees/cores

Time spana

MRW (mm)b

ALO ASO BER CAV CUE DIU FON FUE GAM HIJ IZK LIN LUE MON MUN PEI POM RAN SAG SAJ SIS SOT SUE TEJ TEX UCI URB VAL VEN ZAN

16/26 20/38 20/38 43/64 33/48 7/14 34/56 18/31 19/20 20/40 20/39 37/57 12/24 11/22 16/27 6/12 17/32 18/35 18/29 19/34 12/24 20/34 18/32 20/35 24/35 23/41 21/40 19/33 18/36 16/29

1890–2008 1793–2008 1874–2008 1775–2007 1777–1998 1924–2012 1915–2012 1863–2007 1831–2000 1878–2007 1906–2008 1916–2012 1938–2012 1919–2009 1879–2008 1927–2011 1760–2008 1854–2009 1909–2007 1908–2007 1942–2001 1901–2008 1892–2008 1922–2007 1881–1999 1894–2007 1847–2008 1868–2008 1858–2008 1927–2001

1.27 1.24 1.54 1.25 1.39 1.72 2.05 1.72 1.38 1.82 2.78 2.28 1.92 1.39 2.19 2.08 1.41 1.72 2.43 2.87 2.20 2.24 2.55 4.92 2.80 3.20 2.13 1.92 1.60 2.59

a

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.69 0.34 1.27 0.80 0.68 1.04 0.84 0.79 0.92 0.97 1.49 0.93 1.03 0.78 1.22 1.09 0.75 0.82 1.24 1.37 1.16 1.12 1.19 1.70 1.34 1.49 1.08 0.81 0.85 1.05

LAR (%)c

AC1c

MSc

Rbtc

SNRc

EPSc

0 0 0 0.019 0.047 0 0.193 0 0 0.074 0 0.108 0 0 0 0 0 0 0 0 0.385 0 0 0 0 0 0 0 0.024 0.121

0.415 0.291 0.474 0.349 0.356 0.291 0.161 0.160 0.199 0.258 0.095 0.285 0.329 0.218 0.393 0.545 0.361 0.280 0.328 0.516 0.266 0.383 0.467 0.155 0.411 0.336 0.296 0.468 0.296 0.305

0.207 0.270 0.236 0.221 0.219 0.219 0.206 0.258 0.123 0.241 0.231 0.205 0.190 0.203 0.215 0.301 0.235 0.239 0.242 0.264 0.211 0.207 0.199 0.226 0.228 0.225 0.267 0.188 0.265 0.215

0.334 0.575 0.395 0.373 0.414 0.556 0.479 0.510 0.296 0.388 0.468 0.380 0.514 0.434 0.338 0.542 0.425 0.445 0.315 0.409 0.260 0.440 0.306 0.475 0.438 0.451 0.483 0.357 0.463 0.393

7.53 24.35 11.12 16.67 17.65 8.78 23.92 17.71 6.72 12.02 16.72 15.95 10.58 7.68 7.13 7.10 11.07 14.46 6.92 11.76 3.87 14.14 7.51 18.07 14.01 15.59 15.88 10.01 14.66 8.40

0.883 0.961 0.917 0.943 0.946 0.898 0.960 0.947 0.870 0.923 0.944 0.941 0.914 0.885 0.877 0.877 0.917 0.935 0.874 0.922 0.795 0.934 0.883 0.948 0.933 0.940 0.941 0.909 0.936 0.894

Period with at least five trees. Mean ring width ± SD. LAR: locally absent rings, AC1: first-order autocorrelation, MS: mean sensitivity of the unstandardized series, Rbt: mean between-trees correlation, SNR: signal-to-noise ratio; EPS: expressed population signal. b c

158

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

1.9

(a)

1998

1942

1.0 0.7

30 20

0.4

10 0.1 1830

1850

1870

1890

1910

1930

1950

1970

1990

0 2010

Year

Mean correlation

0.5

Mean correlation Variation in PC1

(b)

0.4

60 50 40 30

0.3

20 0.2 10 0.1 1830

1850

1870

1890

1910

1930

1950

1970

1990

0 2010

No. of chronologies

1.3

Variation in PC1 (%)

Tree-ring index

1.6

Central year of a 35-yr interval Fig. 2. Fagus sylvatica site chronologies (orange lines), mean chronology for all sites (black line), and temporal variation of the number of site chronologies with at least five trees (grey line, right y axis) for the period 1830–2008 (a). The common period with at least five trees for all site chronologies (1942–1998) is highlighted as grey background. Temporal variation of mean correlation among all chronologies, and the percentage of tree-ring growth variation explained by the PC1, calculated in 35-year intervals lagged by 5 years (b). Dots represent the central point of 35-year intervals.

The Kaiser–Guttman criterion revealed that the three first principal components were meaningful when interpreting the chronologies network (Supplementary material, Fig. S3a). These three principal components accounted for 63.13% of the total tree-ring network variance. No conspicuously separated groups of stands were detected in the component loadings scatter plots (Figs. S3b and S3c), which supports a common signal shared among chronologies in the tree-ring network. In general, low- and high-elevation stands scored lower and higher on the PC1 axis, respectively. In fact, the first principal component was mainly related to stand elevation, which explained 56.6% of PC1 loadings variation (Supplementary material, Fig. S4a). Chronologies from all stands located at an elevation of 800 m or higher, and also the BER site located at 300 m, were significantly positively correlated with PC1 scores (Fig. S4b). However, stands located in the central Cantabrian Mountains scored higher on the PC2 axis, while easternmost and westernmost stands from Galicia, pre-Pyrenees and the Iberian Range scored lower on the PC2 axis (Figs. S3b and S3c), suggesting a shared growth variation. The second principal component was actually dependent on north latitude, which explained 75.6% of PC2 loadings variation (Fig. S4c). Indeed, the chronologies from stands located at latitudes greater than 42.8◦ N, were significantly positively correlated with PC2 scores variation (Fig. S4d), excepting the Pyrenean forests (BER, GAM). Only the chronology located on the Iberian Range (DIU), well within the Mediterranean biogeographical region, was significantly negatively correlated with PC2 scores variation. PC3 loadings did not show any significant correlation with the geographical and stand variables considered. 3.2. Correlations between growth and climate The two first principal components (PC1, PC2) calculated using the chronologies network revealed significant and different responses to the considered climatic variables, while PC3 scores

did not show any significant correlation with the climatic variables. Tmax was the main temperature variable driving F. sylvatica growth, with a negative correlation of PC1 scores with June Tmax , and a positive correlation of PC2 scores with previous-December Tmax , i.e. during the winter before tree-ring formation (Fig. 3a). With regard to precipitation, the observed relationships with PC scores were positive in all cases, in April and mainly June for PC1, and in previous October for PC2 (Fig. 3b). Cloud fraction showed only positive correlations with PC2 scores, in previous June, July and October, and in current April to July, but was unrelated to PC1 scores variation (Fig. 3c). The response to ENSO greatly differed for the two first PC scores, with no significant correlations for PC1, and significant positive correlations with SST 3.4 since previous May to current April for PC2 (Fig. 3d). The highest correlations between PC1 and PC2 scores and the evaluated climatic variables were observed for June precipitation and previous November to January SST 3.4, respectively. 3.3. Geographic and stand drivers of climate-growth associations The observed correlations between June precipitation and PC1 were geographically structured, with three main areas of significant correlation. The most important was located in the Galician mountains and the western-central Cantabrian Range, one small nucleus was observed just over the Iberian Range, and one third area covered the western Pyrenees (Fig. 4a). In addition, a positive relationship was found between April-to-July cloudiness and November-to-January SST 3.4 (r = 0.376, P = 0.004 over the complete study area), which was most significant in the northwestern Cantabrian area (Fig. 4b), where the highest correlation was observed (r = 0.451, P < 0.001). Some of the correlations of local chronologies with the identified main climatic drivers for F. sylvatica growth were linearly related to geographical or stand traits, as shown by linear regression models

0.4

Tmax

PC1 PC2

(a)

0.4 0.2

0.0

0.0

-0.2

-0.2

-0.4

-0.4

Cloud fraction

(c)

0.4

(b)

0.2

0.0

0.0

-0.2

-0.2

-0.4

-0.4

SST 3.4

(d)

Apr(-1) May(-1) Jun(-1) Jul(-1) Aug(-1) Sep(-1) Oct(-1) Nov(-1) Dec(-1) Jan Feb Mar Apr May Jun Jul Aug Sep

0.2

Apr(-1) May(-1) Jun(-1) Jul(-1) Aug(-1) Sep(-1) Oct(-1) Nov(-1) Dec(-1) Jan Feb Mar Apr May Jun Jul Aug Sep

Correlation coefficient

0.4

Prec

159

Apr(-1) May(-1) Jun(-1) Jul(-1) Aug(-1) Sep(-1) Oct(-1) Nov(-1) Dec(-1) Jan Feb Mar Apr May Jun Jul Aug Sep

0.2

Apr(-1) May(-1) Jun(-1) Jul(-1) Aug(-1) Sep(-1) Oct(-1) Nov(-1) Dec(-1) Jan Feb Mar Apr May Jun Jul Aug Sep

Correlation coefficient

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

Fig. 3. Correlation coefficients obtained by relating the PC1 (white bars) and PC2 (black bars) scores of the Spanish Fagus sylvatica chronologies and several climatic variables: ˜ 3.4 region (SST 3.4) located maximum mean temperatures (Tmax , a), total precipitation (Prec , b), cloud fraction (c), and the Sea Surface Temperature indices from the Nino in the tropical Pacific Ocean (d). Correlations were calculated monthly since April of the previous year (Apr(–1)) up to September of the current growing year (Sep) for the common period 1942–1998. No significant correlations with climate were found for PC3. Correlations outside the shaded area are statistically significant (P < 0.05).

(Supplementary material, Table S1), with the only exception of the response to cloudiness, which occurred irrespective of stand elevation or geographical position. For instance, the positive response to previous December Tmax was a direct function of the maximum tree age, with the oldest stands showing a more pronounced beneficial effect of maximum winter temperatures on growth. The positive response to previous October precipitation was an inverse function of elevation, being growth enhanced at the lowest stands under wet autumn conditions. By contrast, the positive effect of April precipitation on growth was directly influenced by elevation, with a beneficial effect on growth at the highest stands. Northern latitude was the most influential geographical trait on tree-ring growth response to climate, with a positive influence on the responses to June Tmax and November–January SST 3.4, and a negative influence on the response to June precipitation (Table S1). Almost 48.2% of the response to June precipitation was explained by latitude (Fig. 5a), with most intense and significant positive correlations for the western area (e.g., sites LIN, SIS, ZAN, VEN), the southern Cantabrian Range (e.g., sites FUE, SAG, SOT), and especially at southern stands of the eastern area of the network more frequently suffering from summer drought (e.g., sites IZK, DIU, LUE, MON, PEI). As regards November-January ENSO, a 45.9% of the local response was explained by northern latitude, with a strong and significant response throughout the whole Cantabrian area (Fig. 5b). In fact, all stands located at a longitude between 1.6 and 6.1◦ W showed a significant positive response to November-January SST 3.4, with the only exceptions of two stands located at hilly Cantabrian lowlands (CAV, TEJ), and the single stand on the Iberian Range (DIU). The responses to both climatic variables showed significant positive spatial autocorrelation at inter-site distances of up to 100 km, as displayed by the Moran spatial correlograms, while errors derived from kriging interpolation did not show any significant spatial structure (Fig. S5).

3.4. Temporal trends of climate-growth associations In addition to be geographically structured, responses to the main climatic factors controlling F. sylvatica tree-ring growth in northern Spain, i.e. June precipitation and November-January SST 3.4, were unstable through time. Both time-dependent responses increased during the last decades, as moving correlations of climatic time series with local chronologies revealed. In the case of the eight most drought-sensitive stands, its sensitivity to June precipitation became significant from the 1920s to the 1960s, with a maximum correlation found in the 1970s (Fig. 6a). We detected a significant and negative trend of mean June precipitation over the study area (latitude 42–43.5◦ N, longitude 0–7.5◦ W) for the period 1920–1990, considering the CRU climate data set (r = −0.352, P = 0.0026). Such trend towards drier June conditions was overlapped with increasing June maximum temperatures, particularly in the 1920–1930s and since the 1970s, whose negative effects on radial growth would caused a decline in the rising positive response to June precipitation (Fig. 6b). The sensitivity of growth to November-January SST 3.4, in the case of the ten most ENSO-sensitive stands, was significant since the 1960s, with maximum correlations being found in the 1980s (Fig. 6c). The relationship between April-July cloudiness and November-January SST 3.4 was higher for the stands located in the Eurosiberian area, which showed significant correlations for 1960–1985, while correlations were sporadic in the transitional and Mediterranean areas within this period (Fig. 6d). 4. Discussion Our findings revealed that growth of Iberian F. sylvatica forests mostly responded positively to early-summer water availability in areas suffering from seasonal drought, but to spring-summer cloudiness in the central Cantabrian area under more mesic

160

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

44

North latitude (º)

PC1 vs. Jun Prec

N

(a)

43

42

8

7

6

5

4

3

2

1

0

West longitude (º)

P 0.0001 0.0005 0.001

0.005

0.01

0.05

0.1

44

North latitude (º)

Nov-Jan SST 3.4 vs. Apr-Jul Cloudiness

N

(b)

43

42

8

7

6

5

4

3

2

1

0

West longitude (º) Fig. 4. Geographical associations found between the PC1 scores of the chronologies network and June precipitation (a), and between the Sea Surface Temperature indices ˜ 3.4 region (SST 3.4) in November-January (Nov-Jan SST 3.4) and April-July cloud fraction (b). Field correlations were calculated for the period 1942–1998 and from the Nino their statistical significance levels (P) are shown.

conditions. The most drought-sensitive populations were usually rear-edge southern stands, which are located in mountain areas near or within the Mediterranean biogeographical region. Drought in April and June negatively influenced growth of F. sylvatica, mainly because cambial activity starts in April, and maximum radial growth rate occurs in June, rapidly decreasing in July and August; this is evidenced by previous works on cambial activity ˇ et al., 2008; Michelot et al., 2012b; Prislan and xylogenesis (Cufar et al., 2013; Vavrˇcík et al., 2013). This pattern confirms the droughtsensitive growth dynamics widely supported by the literature throughout the natural F. sylvatica range in northern, central, and southern Europe (Piovesan et al., 2005; Jump et al., 2007; Friedrichs et al., 2009; Drobyshev et al., 2010; Michelot et al., 2012a). A higher sensitivity upon drought, in comparison to other hardwood deciduous trees, has been found for F. sylvatica particularly for stem increment reduction and embolism in its conducting system in dry summers, pre-senescent leaf shedding after droughts, and a reduced fine root biomass in dry soil (Cochard et al., 2005; Granier et al., 2007; Meier and Leuschner, 2008). The observed findings support that this species is strongly limited by relatively high temperatures plus a moisture deficit in rear-edge Iberian populations, but these limitations are buffered when cloudy conditions prevail. According to our results, cloudy conditions in the Cantabrian area mitigated the drought-sensitivity of F. sylvatica particularly

during the last decades. Cloud immersion and fogginess occurs at northern mesic stands in the Eurosiberian region, irrespective of their elevation, in both littoral and higher-elevation inner stands. Cloud immersion is known to affect the forest water budget directly via the capture of cloud water by the canopy (e.g., Burgess and Dawson, 2004; Eugster et al., 2006). There are also indirect hydrological effects of foggy conditions that produce a reduction of evaporative demand, through attenuation of incoming solar radiation plus increased atmospheric humidity (Reinhardt and Smith, 2008). These indirect effects result in strong reductions of leaf transpiration rates compared to those under sunny conditions and, to a lesser extent, foliar absorption of cloud water accumulated on the canopy (Ritter et al., 2009; Alvarado-Barrientos et al., 2014). Such effects seem to be more remarkable for the northernmost studied F. sylvatica populations in late spring and early summer, but also during previous summer (June and July) with a pure effect of cloudiness, and in autumn (October) with a combined effect of both rainfall and cloudiness. Cloud immersion increases mean relative humidity, and greatly reduces crown transpiration and mean leaf-to-air temperature difference (Berry and Smith, 2012; García-Santos, 2012), which has been recorded as a positive growth reaction in cloud-sensitive F. sylvatica populations. These evidences suggest that the northernmost F. sylvatica stands studied should behave similarly to cloud forests, and probably rely on fogginess for improved carbon gain and water use.

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

161

Fig. 5. Spatial variation of correlations between chronologies of Fagus sylvatica and the climatic variables interpolated by kriging, and relationships between northern latitude and the correlations of chronologies with June precipitation (a) and previous November-current January SST 3.4 (b), respectively. Colour code for sites is as in Fig. 1. Bold dots indicate statistically significant (P < 0.05) correlations for the compared common period 1942–1998. The amount of variance explained by latitude (r2 ) and the linear fits are shown. Both linear fits are highly significant (P < 0.001).

Growth of the northernmost stands responding positively to cloud cover, were also strongly related to previous fall-winter sea surface temperatures in the central equatorial Pacific, a surrogate of ENSO atmospheric dynamics. Our analyses suggest that such teleconnection can be explained as an increasing influence of ENSO on summer cloud conditions in the Cantabrian area, which was maximal in the 1960s to 1980s. ENSO influence has been demonstrated in southwest Europe and northern Africa, with a significant impact on winter–spring rainfall, river discharge, and lake level variation in the Iberian Peninsula (Rodó et al., 1997; Knippertz et al., 2003; Pozo-Vázquez et al., 2005). However, ENSO does not modulate June precipitation in northern Spain, the most influencing factor on beech growth in our study area. In fact, correlation between Nov-Jan SST 3.4 and June precipitation was not significant (r = 0.259, P = 0.052, period 1942–1998). We can expect negative sea-level pressure anomalies in northern Spain and over southern Europe ˜ events (Pozo-Vázquez et al., 2005; Brönnimann during El Nino et al., 2007), which, in combination with conditions in other tropical regions, can significantly influence hydrological regimes in the Mediterranean area (Losada et al., 2012). Both observational and modelling studies revealed that the strength of ENSO teleconnections to extratropical areas, and particularly to southern Europe and the Mediterranean basin, recently shifted since the 1970s owing to interactions with other large-scale atmospheric patterns, heat flux anomalies, stochastic forcing, and recent warming trends (Sterl et al., 2007; Losada et al., 2012). The climatic conditions in the study area illustrate their transitional nature, since Spain is influenced by Atlantic (westward), Mediterranean (eastward, southward) and continental (northward, central plateaus) climatic regimes, further modulated by the mountain character of the Iberian Peninsula. Rodó et al. (1997) established a close link between ENSO events and spring precipitation in eastern Spain, while the North Atlantic Oscillation mainly

controls winter climate in western and southwestern Iberia (Zorita et al., 1992). SSTs over the equatorial Pacific drive spring climatic conditions in northwestern Spain mainly through negative phases ˜ events characterized by low SST 3.4 valof the ENSO (‘La Nina’ ues) which precede dry springs in that area (Lorenzo et al., 2011). We argue that those conditions are also linked to reduced cloud cover in the Cantabrian area, thus decreasing F. sylvatica growth at mesic sites located in that area. In addition, the relationship between ENSO and increased cloud fraction in northern Spain could be a consequence of the link between tropical Atlantic-Pacific ENSO events since the 1960–1970s (Rodríguez-Fonseca et al., 2009). The geographical consistency of F. sylvatica growth responses to June water availability and previous autumn-winter ENSO was modulated by northern latitude, probably because of separation between the Cantabrian-Eurosiberian region, i.e. the area where Atlantic cyclogenesis is most active (Zorita et al., 1992), and the Mediterranean region. The positive response to winter temperature (previous December) may be related to the fact that this species develop a significant amount of winter embolism, which can be recovered during spring. Active refilling of the embolized vessels and the formation of new functional vessels leads to an increase in conductive area and largely increase xylem conductivity after the onset of cambial activity, but these processes are costly in terms of carbohydrate investment (Cochard et al., 2001). This process seems to be modulated by maximum ages of the sampled trees, since older stands showed a more beneficial effect of winter temperatures on growth. However, the importance of temperature on F. sylvatica growth described in other European regions was not evidenced in this work. For instance, growth of F. sylvatica in southern Sweden is strongly negatively correlated with previous summer temperature, but positively with previous October temperature (Drobyshev et al., 2010). In this regard, Dittmar et al.

162

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

0.8

(a) Jun Prec

0.6

Correlation coefficient

Correlation coefficient

0.8

0.4 0.2 0.0

Mean correlation SD interval

-0.2 -0.4 1920

1930

1940

1950

1960

1970

1980

(c) Nov-Jan SST 3.4

0.6 0.4 0.2 0.0

Mean correlation SD interval

-0.2 -0.4 1920

1990

Central year of a 35-yr interval 23

(b) Mean Jun Prec and Jun Tmax variation

65

r 2 = 0.124

22

60 21

55 50 45 1920

Mean Jun Prec Mean Jun Tmax 1930

1940

1950

1960

1970

1980

1950

1960

1970

1980

1990

20 1990

Central year of a 35-yr interval

0.8

Correlation coefficient

70

1940

Central year of a 35-yr interval

Mean June Tmax (ºC)

Mean June Prec (mm)

75

1930

0.6

(d) Nov-Jan SST 3.4 vs. Apr-Jul Cloudiness

0.4 0.2 0.0

Eurosiberian Transitional Mediterranean

-0.2 -0.4 1920

1930

1940

1950

1960

1970

1980

1990

Central year of a 35-yr interval

Fig. 6. Temporal variation of correlation coefficients calculated in 35-year intervals, lagged by 1 year, between the variation of drought-sensitive Fagus sylvatica chronologies (DIU, IZK, LUE, MON, SAG, SIS, VEN, ZAN) and June precipitation (a), and between ENSO-sensitive chronologies (BER, CUE, POM, RAN, SAG, SOT, SUE, UCI, VAL, VEN) and previous November–January SST 3.4 (c). Colour code for sites in (a) and (c) is as in Fig. 1. Smoothed mean correlations (±SD) are also shown. Temporal variation of mean June precipitation and maximum temperature (b), and correlation coefficients calculated in 35-year intervals, lagged by 1 year, between Nov–Jan SST 3.4 and April–July cloud fraction (d). The linear fit of mean June precipitation in (b), and its coefficient of determination (P = 0.0026), are shown. Correlations outside the shaded area are statistically significant (P < 0.05).

(2003) suggested a reduced ecological fitness with increasing elevation, as low-elevation F. sylvatica growth mainly suffers from below-average precipitation, whereas high elevation F. sylvatica growth is generally limited by low temperatures. Elevation has also been suggested as a factor that greatly influences the length of the growing season in F. sylvatica (Prislan et al., 2013). For instance, in Southern Germany and Slovenia a significant dependence of leaf unfolding on elevation was observed, with a main positive effect of elevation on spring temperatures and the length of the vegetation ˇ period (Dittmar and Elling, 2006; Cufar et al., 2012). In our study, however, only April rainfall showed to be beneficial on growth at high-elevation stands. This result could be related to an earlier start of the growing period in mountain sub-Mediterranean areas favoured by improved spring water availability. A reduced sensitivity to April water availability in low-elevation stands is also an alternative reliable explanation. Lower stands with F. sylvatica in the Cantabrian area are mixed forests, usually coexisting with other deciduous, mainly oak species. F. sylvatica has been suggested to be significantly more resistant to drought stress in mixed than pure stands, and especially when mixed with oak, growth of F. sylvatica is facilitated (Pretzsch et al., 2013). On the contrary, October precipitation showed the opposite effect, being beneficial for growth mainly in low-elevation Cantabrian stands. Wet late autumn conditions could enhance carbohydrate synthesis in those mesic sites and that stored carbon could be used to form wood in spring. However, at high-elevation stands leaf fall could occur before, and therefore the beneficial effect of previous wet autumns on growth would be negligible.

In conclusion, this work revealed that the expected dependency of F. sylvatica growth on water stress near the rear-edge of the species distribution area can be greatly mitigated when cloudy conditions prevail. Iberian F. sylvatica populations located near or within the Mediterranean biogeographical region were the most sensitive to water deficit, particularly to June drought. However, the widely demonstrated sensitivity of this species to drought, as related to reduced radial growth, can be greatly alleviated by cloudiness and fog immersion. This finding is especially relevant in climate change impact assessment on rear-edge F. sylvatica forests under a drier and warmer climate (Gessler et al., 2007; Kramer et al., 2010), and supports that drought stress would not be as relevant as expected at its Mediterranean distribution limit (Tegel et al., 2014). In fact, F. sylvatica populations in the mesic Cantabrian area did not significantly respond to water shortage, but were strongly influenced by cloudy conditions, particularly in spring-summer of the growth year. These evidences suggest that the Cantabrian F. sylvatica stands may behave as cloud forests, and may rely on cloud immersion for improved growth and carbon gain. In our case, local cloudiness in the Cantabrian area is greatly modulated by an ENSO teleconnection, whose effects on F. sylvatica radial growth intensified during the last decades. Acknowledgements We thank M. A. Álvarez, P. Álvarez-Uría, S. Corés, C. Díaz, A. González, S. Lamas, I. Outeda, M. Rodríguez, and B. RodríguezMorales for field and laboratory tasks. We thank the Services for

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

Forest and Wildlife Conservation of Aragón, Asturias, Cantabria, Castilla y León and Galicia, and the staffs of Picos de Europa National Park, and Collados del Asón, Guara, Izki, Oyambre, Saja-Besaya, ˜ de Bértiz, Valles Occidentales and Urbasa-Andía Regional Senorío Parks, for sampling permissions and guidance during fieldwork. This work was partially funded by Fundación para el Fomento de la Investigación Científica Aplicada y la Tecnología, Asturias Regional Government (PB-REC98-04), Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria, Spanish Government (RTA200600117), Spanish Economy and Competitiveness (CGL2011-26654) and Agriculture and Environment (OAPN, 387/2011) Ministries projects. JJC acknowledges the support of ARAID. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.agrformet.2014. 11.012. References Alvarado-Barrientos, M.S., Holwerda, F., Asbjornsen, H., Dawson, T.E., Bruijnzeel, L.A., 2014. Suppression of transpiration due to cloud immersion in a seasonally dry Mexican weeping pine plantation. Agric. For. Meteorol. 186, 12–25. Aranda, I., Gil, L., Pardos, J., 2000. Water relations and gas exchanges in Fagus sylvatica L. and Quercus petraea (Matuschka) Liebl. in a mixed stand at their southern limit of distribution in Europe. Trees 14, 344–352. Babst, F., Poulter, B., Trouet, V., Tan, K., Neuwirth, B., Wilson, R., Carrer, M., Grabner, M., Tegel, W., Levanic, T., Panayotov, M., Urbinati, C., Bouriaud, O., Ciais, P., Frank, D., 2013. Site- and species-specific responses of forest growth to climate across the European continent. Global Ecol. Biogeogr. 22, 706–717. Barigah, T.S., Charrier, O., Douris, M., Bonhomme, M., Herbette, S., Améglio, T., Fichot, R., Brignolas, F., Cochard, H., 2013. Water stress-induced xylem hydraulic failure is a causal factor of tree mortality in beech and poplar. Ann. Bot. 112, 1431–1437. Berry, Z.C., Smith, W.K., 2012. Cloud pattern and water relations in Picea rubens and Abies fraseri, southern Appalachian Mountains, USA. Agric. For. Meteorol. 162–163, 27–34. Biondi, F., 1993. Climatic signals in tree-rings of Fagus sylvatica L. from the central Apennines, Italy. Acta Oecol. 14, 57–71. Bolte, A., Czajkowski, T., Kompa, T., 2007. The north-eastern distribution range of European beech—a review. Forestry 80, 413–429. Bradshaw, R.H.W., Kito, N., Giesecke, T., 2010. Factors influencing the Holocene history of Fagus. For. Ecol. Manage. 259, 2204–2212. Brönnimann, S., Xoplaki, E., Casty, C., Pauling, A., Luterbacher, J., 2007. ENSO influence on Europe during the last centuries. Clim. Dyn. 28, 181–197. Burgess, S.S.O., Dawson, T.E., 2004. The contribution of fog to the water relations of Sequoia sempervirens (D. Don): foliar uptake and prevention of dehydration. Plant Cell Environ. 27, 1023–1034. Cochard, H., Lemoine, D., Ameglio, T., Granier, A., 2001. Mechanisms of xylem recovery from winter embolism in Fagus sylvatica L. Tree Physiol. 21, 27–33. Cochard, H., Coste, S., Chanson, B., Guehl, J.M., Nicolini, N., 2005. Hydraulic architecture correlates with bud organogenesis and primary shoot growth in Beech (Fagus sylvatica). Tree Physiol. 25, 1545–1552. Cook, E.R., Holmes, R.L., 1996. Guide for computer program ARSTAN. In: GrissinoMayer, H.D., Holmes, R.L., Fritts, H.C. (Eds.), The International Tree-Ring Data Bank Program Library Version 2. 0 User’s Manual. Laboratory of Tree-Ring Research. University of Arizona, Arizona, pp. 75–87. ˇ ˇ Cufar, K., De Luis, M., Saz, M.A., Crepinˇ sek, Z., Kajfeˇz-Bogataj, L., 2012. Temporal shifts in leaf phenology of beech (Fagus sylvatica) depend on elevation. Trees 26, 1091–1100. ˇ Cufar, K., Prislan, P., Griˇcar, J., 2008. Cambial activity and wood formation in beech (Fagus sylvatica) during the 2006 growth season. Wood Res. 53, 1–12. Davies, C.E., Moss, D., Hill, M.O., 2004. EUNIS Habitat Classification Revised 2004. Report to the European Topic Centre on Nature Protection and Biodiversity. European Environment Agency, Paris, pp. 307. ˇ Di Filippo, A., Biondi, F., Cufar, K., De Luis, M., Grabner, M., Maugeri, M., Presutti Saba, E., Schirone, B., Piovesan, G., 2007. Bioclimatology of beech (Fagus sylvatica L.) in the Eastern Alps: spatial and altitudinal climatic signals identified through a tree-ring network. J. Biogeogr. 34, 1873–1892. Dittmar, C., Elling, W., 2006. Phenological phases of common beech (Fagus sylvatica L.) and their dependence on region and altitude in Southern Germany. Eur. J. For. Res. 125, 181–188. Dittmar, C., Zech, W., Elling, W., 2003. Growth variations of common beech (Fagus sylvatica L.) under different climatic and environmental conditions in Europe: a dendroecological study. For. Ecol. Manage. 173, 63–78. Drobyshev, I., Övergaard, R., Saygin, I., Niklasson, M., Hickler, T., Karlsson, M., Sykes, M.T., 2010. Masting behaviour and dendrochronology of European beech (Fagus sylvatica L.) in southern Sweden. For. Ecol. Manage. 259, 2160–2171.

163

Eugster, W., Burkard, R., Holwerda, F., Scatena, F.N., Bruijnzeel, L.A., 2006. Characteristics of fog and fogwater fluxes in a Puerto Rican elfin cloud forest. Agric. For. Meteorol. 139, 288–306. Fang, J., Lechowicz, M.L., 2006. Climatic limits for the present distribution of beech (Fagus L.) species in the world. J. Biogeogr. 33, 1804–1819. Fotelli, M.N., Nahm, M., Radoglou, K., Rennenberg, H., Halyvopoulos, G., Matzarakis, A., 2009. Seasonal and interannual ecophysiological responses of beech (Fagus sylvatica) at its south-eastern distribution limit in Europe. For. Ecol. Manage. 257, 1157–1164. Friedrichs, D., Trouet, V., Büntgen, U., Frank, D.C., Esper, J., Neuwirth, B., Löffler, J., 2009. Species-specific climate sensitivity of tree growth in Central-West Germany. Trees 23, 729–739. Fritts, H.C., 2001. Tree Rings and Climate. Blackburn Press, Caldwell, ID. García-Santos, G., 2012. Transpiration in a sub-tropical ridge-top cloud forest. J. Hydrol. 462–463, 42–52. Gessler, A., Keitel, C., Kreuzwieser, J., Matyssek, R., Seiler, W., Rennenberg, H., 2007. Potential risks for European beech (Fagus sylvatica L.) in a changing climate. Trees 21, 1–11. Giesecke, T., Hickler, T., Kunkel, T., Sykes, M.T., Bradshaw, R.H.W., 2007. Towards an understanding of the Holocene distribution of Fagus sylvatica L. J. Biogeogr. 34, 118–131. Granier, A., Reichstein, M., Bréda, N., Janssens, I.A., et al., 2007. Evidence for soil water control on carbon and water dynamics in European forests during the extremely dry year: 2003. Agric. For. Meteorol. 143, 123–145. Grissino-Mayer, H.D., 2001. Evaluating crossdating accuracy: a manual and tutorial for the computer program COFECHA. Tree-Ring Res. 57, 205–221. Guiot, J., 1985. The extrapolation of recent climatological series with spectral canonical regression. J. Climatol. 5, 325–335. Gutiérrez, E., 1988. Dendroecological study of Fagus sylvatica L. in the Montseny mountains (Spain). Acta Oecol. 9, 301–309. Helama, S., Lindholm, M., Timonen, M., Eronen, M., 2004. Detection of climate signal in dendrochronological data analysis: a comparison of tree-ring standardization methods. Theor. Appl. Climatol. 79, 239–254. Herbette, S., Wortemann, R., Awad, H., Huc, R., Cochard, H., Barigah, T.S., 2010. Insights into xylem vulnerability to cavitation in Fagus sylvatica L.: phenotypic and environmental sources of variability. Tree Physiol. 30, 1448–1455. ˜ Jump, A., Hunt, J.M., Penuelas, J., 2007. Climate relationships of growth and establishment across the altitudinal range of Fagus sylvatica in the Montseny Mountains, northeast Spain. Écoscience 14, 507–518. Knippertz, P., Ulbrich, U., Marques, F., Corte-Real, J., 2003. Decadal changes in the link ˜ and springtime North Atlantic Oscillation and European-north between El Nino African rainfall. Int. J. Climatol. 23, 1293–1311. Kramer, K., Degen, B., Buschbom, J., Hickler, T., WThuiller, W., Sykes, M.T., de Winter, W., 2010. Modelling exploration of the future of European beech (Fagus sylvatica L.) under climate change—range, abundance, genetic diversity and adaptive response. For. Ecol. Manage. 259, 2213–2222. Lebourgeois, F., Bréda, N., Ulrich, E., Granier, A., 2005. Climate–tree-growth relationships of European beech (Fagus sylvatica L.) in the French Permanent Plot Network (RENECOFOR). Trees 19, 385–401. Legendre, P., Legendre, L., 1998. Numerical ecology. In: Developments in Environmental Modelling, 20, second English ed. Elsevier Science B. V., Amsterdam. Lorenzo, M.N., Taboada, J.J., Iglesias, I., Gómez-Gesteira, M., 2011. Predictability of the spring rainfall in Northwestern Iberian Peninsula from sea surfaces temperature of ENSO areas. Clim. Change 107, 329–341. Losada, T., Rodríguez-Fonseca, B., Kucharski, F., 2012. Tropical influence on the summer Mediterranean climate. Atmos. Sci. Lett. 13, 36–42. Maxime, C., Hendrik, D., 2011. Effects of climate on diameter growth of cooccurring Fagus sylvatica and Abies alba along an altitudinal gradient. Trees 25, 265–276. Meier, E.S., Edwards, T.C., Kienast, F., Dobbertin, M., Zimmermann, N.E., 2011. Cooccurrence patterns of trees along macro-climatic gradients and their potential influence on the present and future distribution of Fagus sylvatica L. J. Biogeogr. 38, 371–382. Meier, I.C., Leuschner, C., 2008. Genotypic variation and phenotypic plasticity in the drought response of fine roots of European beech. Tree Physiol. 28, 297–309. Michelot, A., Bréda, N., Damesin, C., Dufrêne, E., 2012a. Differing growth responses to climatic variations and soil water deficits of Fagus sylvatica, Quercus petraea and Pinus sylvestris in a temperate forest. For. Ecol. Manage. 265, 161–171. Michelot, A., Simard, S., Rathgeber, C., Dufrêne, E., Damesin, C., 2012b. Comparing the intra-annual wood formation of three European species (Fagus sylvatica, Quercus petraea and Pinus sylvestris) as related to leaf phenology and non-structural carbohydrate dynamics. Tree Physiol. 32, 1033–1045. Mitchell, T.D., Jones, P.D., 2005. An improved method of constructing a database of monthly climate observations and associated high resolution grids. Int. J. Climatol. 25, 693–712. Nahm, M., Radoglou, K., Halyvopoulos, G., Gessler, A., Rennenberg, H., Fotelli, M.N., 2006. Physiological performance of beech (Fagus sylvatica L.) at its Southeastern distribution limit in Europe: seasonal changes in nitrogen, carbon and water balance. Plant Biol. 8, 52–63. Packham, J.R., Thomas, P.A., Atkinson, M.D., Degen, T., 2012. Biological Flora of the British Isles: Fagus sylvatica. J. Ecol. 100, 1557–1608. Piovesan, G., Schirone, B., 2000. Winter North Atlantic oscillation effects on the tree rings of the Italian beech (Fagus sylvatica L.). Int. J. Biometeorol. 44, 121–127. Piovesan, G., Biondi, F., Bernabei, M., Di Filippo, A., Schirone, B., 2005. Spatial and altitudinal bioclimatic zones of the Italian peninsula identified from a beech (Fagus sylvatica L.) tree-ring network. Acta Oecol. 27, 197–210.

164

V. Rozas et al. / Agricultural and Forest Meteorology 201 (2015) 153–164

Pluess, A.R., Weber, P., 2012. Drought-adaptation potential in Fagus sylvatica: Linking moisture availability with genetic diversity and dendrochronology. PLoS One 7, e33636, http://dx.doi.org/10.1371/journal.pone.0033636. Pozo-Vázquez, D., Gámiz-Fortis, S.R., Tovar-Pescador, J., Esteban-Parra, M.J., Castro˜ Oscillation events and associated European Díez, Y., 2005. El Nino-Southern winter precipitation anomalies. Int. J. Climatol. 25, 17–31. Pretzsch, H., Schütze, G., Uhl, E., 2013. Resistance of European tree species to drought stress in mixed versus pure forests: evidence of stress release by inter-specific facilitation. Plant Biol. 15, 483–495. ˇ Prislan, P., Griˇcar, J., de Luis, M., Smith, K.T., Cufar, K., 2013. Phenological variation in xylem and phloem formation in Fagus sylvatica from two contrasting sites. Agric. For. Meteorol. 180, 142–151. Reinhardt, K., Smith, W.K., 2008. Impacts of cloud immersion on microclimate, photosynthesis and water relations of Abies fraseri (Pursh.) Poiret in a temperate mountain cloud forest. Oecologia 158, 229–238. Ritter, A., Regalado, C.M., Aschan, G., 2009. Fog reduces transpiration in tree species of the Canarian relict heath-laurel cloud forest (Garajonay National Park, Spain). Tree Physiol. 29, 517–528. Rivas-Martínez, S., Díaz, T.E., Fernández González, F., Izco, J., Loidi, J., Lousa, M., Penas, A., 2002. Vascular plant communities of Spain and Portugal. Addenda to the syntaxonomical checklist of 2001. Itinera Geobotanica 15, 5–432. Robson, T.M., Rasztovits, E., Aphalo, P.J., Alia, R., Aranda, I., 2013. Flushing phenology and fitness of European beech (Fagus sylvatica L.) provenances from a trial in La Rioja, Spain, segregate according to their climate of origin. Agric. For. Meteorol. 180, 76–85. Rodó, X., Baert, E., Comin, F.A., 1997. Variations in seasonal rainfall in Southern Europe during the present century: relationships with the North Atlantic Oscil˜ lation and the El Nino-Southern Oscillation. Clim. Dyn. 13, 275–284. Rodríguez-Fonseca, B., Polo, I., García-Serrano, J., Losada, T., Mohino, E., Mechoso, ˜ enhancing Pacific ENSO C.R., Kucharski, F., 2009. Are Atlantic Ninos events in recent decades? Geophys. Res. Lett. 36, L20705, http://dx.doi.org/ 10.1029/2009GL040048.

Rose, L., Leuchner, C., Köckemann, B., Buschmann, H., 2009. Are marginal beech (Fagus sylvatica L.) provenances a source for drought tolerant ecotypes? Eur. J. For. Res. 128, 335–343. Rozas, V., 2001. Detecting the impact of climate and disturbances on tree-rings of Fagus sylvatica L. and Quercus robur L. in a lowland forest in Cantabria, Northern Spain. Ann. For. Sci. 58, 237–251. ˜ Rozas, V., García-González., I., 2012a. Non-stationary influence of El Nino-Southern Oscillation and winter temperature on oak latewood growth in NW Iberian Peninsula. Int. J. Biometeorol. 56, 787–800. Rozas, V., García-González, I., 2012b. Too wet for oaks? Inter-tree competition and recent persistent wetness predispose oaks to rainfall-induced dieback in Atlantic rainy forest. Global Planet. Change 94–95, 62–71. Stenseth, N.C., Ottersen, G., Hurrell, J.W., Mysterud, A., Lima, M., Chan, K.S., Yoccoz, N.G., Ådlandsvik, B., 2003. Studying climate effects on ecology through the use ˜ Southern Oscillation of climate indices: the North Atlantic Oscillation, El Nino and beyond. Proc. R. Soc. London, B 270, 2087–2096. Sterl, A., van Oldenborgh, G.J., Hazeleger, W., Burgers, G., 2007. On the robustness of ENSO teleconnections. Clim. Dyn. 29, 469–485. Tegel, W., Seim, A., Hakelberg, A., Hoffmann, S., Panev, M., Westphal, T., Büntgen, U., 2014. A recent growth increase of European beech (Fagus sylvatica L.) at its Mediterranean distribution limit contradicts drought stress. Eur. J. For. Res. 133, 61–71. ˜ Trenberth, K.E., 1997. The definition of El Nino. Bull. Am. Meteorol. Soc. 78, 2771–2777. Vavrˇcík, H., Gryc, V., Menˇsík, L., Baar, J., 2013. Xylem formation in Fagus sylvatica during one growing season. Dendrobiology 69, 69–75. von Wuehlisch, G., 2008. EUFORGEN Technical Guidelines for Genetic Conservation and Use for European Beech (Fagus sylvatica). Bioversity International, Rome, Italy, pp. 6. Zorita, E., Kharin, V., von Storch, H., 1992. The atmospheric circulation and sea surface temperature in the North Atlantic area in winter: their interaction and relevance for Iberian precipitation. J. Climatol. 5, 1097–1108.