Ecosystems (2014) 17: 228–241 DOI: 10.1007/s10021-013-9719-3 " 2013 Springer Science+Business Media New York

Local Climate Forces Instability in Long-Term Productivity of a Mediterranean Oak Along Climatic Gradients Guillermo Gea-Izquierdo1,2* and Isabel Can˜ellas1 1

INIA-CIFOR, Ctra. La Corun˜a km 7.5, 28040 Madrid, Spain; 2Present address: CEREGE UMR 7330, CNRS/Aix-Marseille Universite´, Europole de l’Arbois, BP 8013545, Aix-en-Provence cedex 4, France

ABSTRACT growth trees were located at the coldest sites and exhibited a long-term increase in productivity starting 150 years ago which could express a dominant positive effect of warming temperatures since the mid 1800 s at cold-humid sites. Conversely, trees at dry sites exhibited negative growth trends. Particularly low elevation stands located at latitudes below 40! displayed enhanced growth constraints with the increase in water stress around 1970, which suggests vulnerability of Quercus pyrenaica at the sampled altitudinal dry edge. The response of trees to future changes in climate should be monitored, particularly in threatened transitional zones.

Forests modify their productivity, composition, and distribution in response to global change. We studied the radial growth trends of the Western Mediterranean oak Quercus pyrenaica over the last two centuries to analyze whether trees exhibited instability in productivity in response to climatic changes. Trees were sampled to build annual growth chronologies following climatic gradients of increasing moisture availability and decreasing temperature with altitude and latitude. The species’ response to climate showed high variability linked to local climatic conditions. The strength in the positive response of trees to moisture availability was inversely related to precipitation (that is, enhanced by higher water stress) whereas high temperature in the growing season was positive for tree-growth only at cold sites. The oldest ages of trees expanded back to the late 1500 s. These old-

Key words: Quercus pyrenaica; water stress; growth enhancement; vulnerability; global change; forest decline.

INTRODUCTION The climate got warmer from the end of the Little Ice Age (LIA) in the mid 1800 s but temperatures have risen sharply over recent decades (Luterbacher and others 2006; IPCC 2007). Climate change will modify ecosystem productivity and species distribution in different ways depending on local conditions (Boisvenue and Running 2006; Bertrand and others 2011; Thuiller and others 2005). Warming can enhance productivity and lead to the

Received 28 January 2013; accepted 16 September 2013; published online 11 October 2013 Electronic supplementary material: The online version of this article (doi:10.1007/s10021-013-9719-3) contains supplementary material, which is available to authorized users. Author Contribution: G.G.I. conceived and designed the study, performed research, analyzed data and wrote the paper; I.C. contributed to the study design and paper writing. *Corresponding author; e-mail: [email protected], [email protected]

228

Growth Instability of Mediterranean Oaks expansion of forests which are mainly limited by temperature (Myneni and others 1997; Lloyd and others 2011). Conversely, global change has augmented the vulnerability of forests limited by moisture availability and several species show symptoms of decline and death episodes related to an increase in local water stress (Breshears and others 2009; van Mantgem and others 2009; Allen and others 2010). Forest decline is a consequence of multiple stress factors, including drought, which concurrently act at different temporal scales (Bigler and others 2006; Niinemets 2010). Drought increase can lead to ecophysiological imbalances and eventual death of trees (Rice and others 2004; Wiley and Helliker 2012). The analysis of inter-annual growth variability can be used to study changes in productivity and forest vitality to detect vulnerability to stress (Fritts 1976; Dobbertin 2005). The inverse relationship between radial growth [ring-width (RW)] and age disappears when analyzed as basal area increments (BAI). Theoretically, under constant average climatic conditions and in the absence of major disturbances, after several years of increasing BAI at young ages this would reach a plateau and stabilize for a long period before starting a final decrease prior to death at old ages (Biondi and Qeadan 2008; Johnson and Abrams 2009). Negative trends in growth using BAI (for example, Pen˜uelas and others 2008; Piovesan and others 2008; Linares and others 2009; Di Filippo and others 2012) but also RW (for example, Gea-Izquierdo and others 2011; Sarris and others 2011) have been considered to express a decline in productivity and a consequent increase in forest vulnerability whereas an increase in recent growth was explained by a positive effect of climate or atmospheric fertilization in productivity (for example, Salzer and others 2009; Carrer and others 2010). However, disentangling the direct effects of climate on stand productivity can be complex. The sensitivity of trees to climate is generally non-stationary because an ensemble of nonlinear confounding factors including age, reproduction, climate, atmospheric fertilization, patterns in nonstructural carbon allocation or inter-tree competition simultaneously modify growth (D’Arrigo and others 2008; Salzer and others 2009; Niinemets 2010; Wiley and Helliker 2012). Crowding and long-term increase in water stress are the main predisposing factors affecting sizedependent individual mortality (Bigler and others 2006; Luo and Chen 2011). However, high competition reduces growth and enhances mortality (Pretzsch and Biber 2005; Luo and Chen 2011) even in the absence of long-term climatic con-

229

straints. Climatic events can ultimately trigger death pulses (Sua´rez and others 2004; Bigler and others 2006, 2007) but distinguishing growth decline and increased mortality rates which threaten ecosystem sustainability from baseline mortality as part of natural sustainable dynamics is crucial (Voelker and others 2008; Dietze and Moorcroft 2011). Quercus pyrenaica Willd. (QUPY) is a winter deciduous oak endemic to siliceous soils from submediterranean areas [that is, cooler and more humid than typical Mediterranean areas dominated by evergreen taxa, but still with water summer deficit; see Sa´nchez-de-Dios and others (2009) for further explanation] of southwestern France, the Iberian Peninsula, and northern Morocco (Costa and others 2005). The species is often present today in almost pure monospecific stands partly because of impoverishment of the original mixed forest composition by humans (Manuel and Gil 1999; Costa and others 2005). Likewise other Mediterranean oaks, Q. pyrenaica woodlands were traditionally managed as coppice for firewood or open silvopastoral systems. Increased human populations and two big socioeconomic changes shaped Spanish forests over the last two centuries: disentailment processes in the 1800 s and the substitution of firewood by other fuels in the mid 1900 s (Manuel and Gil 1999; Valbuena-Caraban˜a and others 2010). The latter resulted in the expansion of oak coppice and forest densification and encroachment, yet there is great concern for the persistence of Q. pyrenaica under global change (Sa´nchez de Dios and others 2009). Its drought-tolerance is lower compared to that of cooccurring Mediterranean species such as Q. ilex L. (Corcuera and others 2006; Montserrat-Martı´ and others 2009) and future forest simulations suggest that submediterranean species are particularly vulnerable to the expected regional increase in water stress (Sa´nchez de Dios and others 2009). However, the potential use of long dendrochronological series to analyze vulnerability of submediterranean oaks to climatic changes has not been widely explored (Tessier and others 1994; Rathgeber and others 1999; Rozas and others 2009; Di Filippo and others 2010). In this study, we analyzed the instability in longterm BAI trends of Q. pyrenaica along climatic gradients in relation to global change. Although a BAI trend could express changes in stand competition in the mid-term, we hypothesize that a long-term negative trend in BAI of non-senescent adult dominant trees will express vulnerability to water stress. Conversely, a positive long-term trend will express a beneficial effect for trees of changing

Lat = latitude; Long = longitude; D = estimated mean stand density; P = precipitation for the hydrological year; PET = potential evapotranspiration for hydrological year; SPEI = mean SPEI for hydrological year. Dbh, age and height are site means. Standard deviations are shown in brackets. Climatic data refer to the period 1960–2008. If two values are shown in D it means that trees were sampled at two stands with different densities.

18.0 17.3 14.6 19.7 14.4 11.6 15.9 12.5 11.5 11.6 111 502 235 201 339 281 118 156 101 197 (19.3) (148.3) (48.3) (48.9) (82.9) (50.6) (20.3) (32.2) (18.1) (52.8) 65.5 270.2 75.3 149.2 142.7 165.3 81.2 90.6 68.0 120.5 (19.4) (22.1) (14.5) (7.2) (19.9) (8.7) (6.8) (5.2) (6.7) (12.0) 100 175 70–125 20–275 325 200 150 275 65 90 0.15 0.17 -0.03 -0.00 -0.16 -0.13 -0.21 -0.31 -0.39 -0.32 614.3 568.8 704.6 708.7 684.5 623.3 784.0 835.6 738.4 731.0 9.3 7.6 12.2 12.7 11.8 9.7 14.3 15.6 14.1 13.0 857.8 965.4 478.0 783.2 692.8 791.7 464.3 507.6 446.1 578.8 1030 1310 760 900 1300 1650 900 890 1140–1370 1525 -6.5 -6.7 -6.1 -6.8 -3.9 -3.9 -4.1 -3.9 -3.4 -3.4 42.1 42.1 41.9 40.3 40.9 40.9 39.4 38.4 37.1 37.0 Sanabria Sanabria Ta´bara Navasfrı´as Rascafrı´a Rascafrı´a Quintos de Mora Andu´jar Sierra Nevada Sierra Nevada

P (mm) Tmean (!C) PET (mm) SPEI

QUPY1 QUPY2 QUPY3 QUPY4 QUPY5 QUPY6 QUPY7 QUPY8 QUPY9 QUPY10

Figure 1. Map showing the distribution of the ten chronologies. Overlapping points mean different chronologies coming from a single site because of sampling along altitudinal gradients.

Table 1.

Characteristics of Sampled Plots

Dominant Q. pyrenaica trees were sampled at ten sites (Figure 1; Table 1) following climatic gradients of increasing moisture availability and decreasing temperature with increasing latitude (from 37! to 42! North) and altitude (from 760 to 1650 m asl) in Spain. The sampling sites included transitional zones in altitude for the species local distribution (Sa´nchez de Dios and others 2009). Stand density was estimated from variable radius plots including ten trees. The species occurred within stands of variable density (Table 1) generally with only one canopy layer of dominant–codominant trees, hence where competition for light was low. Except in QUPY5 where oaks were scattered on a dominant close canopy of Pinus sylvestris L., all other stands presented a dominant canopy of Q. pyrenaica. At the sampled sites we did not observe widespread oak mortality but seed regeneration seemed to be often absent particularly in southern stands. In the south the species is only

46.8 72.7 40.1 53.9 58.8 55.2 39.2 30.9 33.9 43.6

D (trees/ha) dbh (cm)

Study Sites

Mean

Age

METHODS

Lat (!) Long (!) Altitude (m) Climate

AND

Site

MATERIALS

Max

atmospheric conditions (that is, climate and CO2). In particular, we addressed the following questions: (1) how do long-term trends in oak growth relate to major climatic changes over the last 200 years? (2) do oak trees at opposite ends of their distribution (cold-humid limit and warm-dry limit) exhibit contrasting growth trends in response to climate change?

(3.0) (3.4) (3.9) (3.5) (2.0) (1.9) (3.7) (1.8) (2.1) (1.6)

Height (m)

G. Gea-Izquierdo and I. Can˜ellas

Name

230

Growth Instability of Mediterranean Oaks marginally present at humid sites in the mountains and managers are greatly concerned by the low vitality of oaks.

Climatic and CO2 data Climatic gridded CRU TS 3.10 data (Mitchell and Jones 2005) including monthly precipitation and mean, minimum, and maximum temperature were obtained from the KNMI explorer (http://climexp. knmi.nl/get_index.cgi) for the period 1901–2008. Gridded data were linearly corrected using local data from AEMET (http://www.aemet.es/es/ portada) when available and along altitudinal gradients temperatures were corrected considering a mean lapse rate of 0.6!C for every 100 m in elevation range (Nobel 2009). In addition, we calculated the monthly standardized precipitation– evapotranspiration index (SPEI) using the SPEI package in R (http://sac.csic.es/spei/) with a scale of 6 months. This drought index is calculated using water balances from precipitation and potential evapotranspiration (PET), which we estimated using the Thornthwaite equation (Vicente-Serrano and others 2010). Water stress increases inversely to SPEI. For our analyses we calculated climatic values for periods of maximum growth response for the studied Mediterranean species (Tessier and others 1994; Montserrat-Marti and others 2009; Rozas and others 2009): hydrological year (from October t - 1 to September t), winter (from January to February), spring (from April to June) and June–July. On a complementary analysis and to compare with long-term growth trends of oldgrowth trees we used two historical precipitation and temperature series (1850–2005) for Madrid and Ca´diz (Central and Southern Spain, respectively) and CO2 records from Keeling and others (2001) which combined ice core data before 1958 from Etheridge and others (1996) and in situ data from the Mauna Loa and South Pole observatories.

Dendroecological Data Processing Between 16 and 22 dominant trees at each site were cored 2–3 times at 1.3 m from 2009 to 2010 (Appendix 1 in supplementary material). Two-perpendicular diameters at 1.3 m (dbh) and the total height of trees were measured. We sampled trees representative of stand conditions at each site including the oldest non-pollarded individuals. Increment cores were dried, glued onto wooden mounts, sanded, and annual radial growth (RW) measured using a measuring device connected to a computer. Individual series were visually crossdated

231

then verified using COFECHA (Holmes 1983). In the case of trees where the pith was not reached, age was estimated for the missing radius in proportion to growth of the last 20 years (Rozas 2003). The growth trends in relation to climate were analyzed at different time scales. To analyze the highfrequency (short-term, annual) response of growth to climate we used residual chronologies (GIres) calculated from biweight means of ratios between raw measurements and individual cubic splines with a 50% frequency cutoff at 30 years to remove the agetrend (Cook and Kairiukstis 1990). The low-frequency (long-term, multidecadal) and mid-frequency (mid-term, decadal) response of growth were studied using BAI calculated from RW and measured dbh (for example, Piovesan and others 2008). For the long-term analysis, BAI data were smoothed (BAIspl) using LOESS (Crawley 2007) with a = 0.8 when we were interested in calculating confidence intervals and cubic splines with a 50% frequency cutoff at 40 years when we analyzed just the growth trends. Climate data were smoothed likewise. Because the influence of competition may differ with tree age or size and to avoid a bias in the estimation of recent growth trends produced by in-growth of young trees we excluded from the mean site BAI chronology those series with an estimated age below 80 years (Figure 2). In addition, we considered ‘Old-growth’ trees those with ages over 250 years, which were used on an independent analysis to compare with the historical climatic records. We aimed to include the longest possible period previous to the LIA and minimize the influence of decadal, mid-frequency nonclimatic events in our analysis of the growth trends.

Assessing Woodland History from Growth Patterns: Disturbance and Climate Events Because, we lacked historical management data we assessed stand-history indirectly from growth. We built site disturbance chronologies using RW (Nowacki and Abrams 1997) to identify abrupt positive (releases) or negative (suppressions) changes in growth. Growth change (GC) was calculated annually using 10-year windows either as positive GCs (PGC) or negative GCs (NGC): 8 h i Þ > < PGCi ¼ ðG2"G1 % 100; if G2[G1 G1 h i GCi ¼ ; with Þ > : NGCi ¼ ðG2"G1 % 100; if G1[G2 G2 G1 ¼

ðX i"1Þ

ði"10Þ

RWi ; G2 ¼

ðX iþ9Þ i

RWi

232

G. Gea-Izquierdo and I. Can˜ellas

Figure 2. BAI trends and LOESS confidence intervals from year 1700 of the ten site chronologies divided according to the two age cohorts. Cumulative PET (black lines, in mm) and mean SPEI (gray lines) for the period 1901–2008 are depicted in small subplots. PET and SPEI are referred to the hydrological year (from October t - 1 to September t). Vertical dashed lines in 1850 and 1970 show, respectively, the end of the LIA and an abrupt change toward more arid conditions. A horizontal dashed line of low growth is represented at 10-cm2/year as reference baseline growth to ease the comparison between sites. n is the number of trees in each age cohort (‘‘old’’ or ‘‘young’’, see text for details).

and RWi being RW from year i of an individual tree-ring series. Site disturbance chronologies were constructed by averaging GCi annually. We were interested in separating growth peaks produced by disturbance events from those by climate. The GC threshold considered as a non-climatic disturbance event needs to be carefully selected (Black and Abrams 2003). To analyze whether there was a climatic signal in the original threshold of GC equal to 25% considered to reflect a disturbance event in Nowacki and Abrams (1997) and to check whether recent negative peaks were responding to particularly dry conditions we used bivariate event analyses (BEA), which is a temporal modification of the Ripleys’ K function applied for one-dimensional, bivariate data (Gavin and others

2006). In BEA PGC and NGC events were compared with precipitation events (Ppt10) calculated from year i to year i + 9, the same period as the growth events (PGC or NGC). Growth events were considered as those years with GC over ± 25 or 50% of the trees with GC greater than 25%. Precipitation events were considered as those dates for which Ppt10 was greater than 1SD (for PGC, SD is standard deviation) or -1 SD (for NGC [Sua´rez and others 2004]). The number of growth events (G) following precipitation events (P) at different lags t was expressed as nG ! nP P P KPG ðt Þ ¼ nGTnP I ðGj "Pi Þ ' tjGj ( Pi ). T is the i¼1 j¼1

length of the record; nG and nP are the number of G and P events; and Gj and Pj are the dates of G and P events. Results were expressed as LPG(t) = KPG(t) - t

Growth Instability of Mediterranean Oaks

233

Figure 3. BAI growth trends of trees with ages over 250-years old (‘old-growth’, 12 trees, black solid line) and atmospheric CO2 records (dashed solid line, see text for details on original source). In a smaller sub-graph we highlight the last 150 years of growth (note that the units of BAI are not in the y axis of the subgraph but can be seen in the main graph) to compare with the mean temperature (solid gray lines) and precipitation (dashed gray lines) long climatic series from Ca´diz (upper lines, southern Spain), and Madrid (lower lines, central Spain). All mean curves were smoothed using splines with a 50% frequency at a 40-year period.

Figure 4. Dendrograms of cluster analyses (A, C) and mean normalized BAI trends (B, D) of groups as suggested by HCA. Above, in (A) and (B), are results from 1864 to 2008 for the six sites where data were available back to 1864 (HCA1864– 2008). Below, in (C) and (D) we show results for all sites (except QUPY1, where records did not reach 1920) for the period 1920–2008 (HCA1920–2008). In (B) and (D) we show standard deviations of the annual means with shadowed polygons. Data used in these analyses exclude the ‘‘young’’ trees as from Figure 2 (see main text for more details).

to remove the time dependence of KPG (Gavin and others 2006; Bigler and others 2007). We only evaluated positive associations between years where PGC was greater than 25% and Ppt10 was greater than +1SD, and years where NGC was less than -25% and Ppt10 was less than -1 SD because we hypothesized that, whether PGC (NGC) events had been linked to climate they should have followed wet (dry) P events (Bigler and others 2007).

Analysis of Climate–Growth Relationships at Different Temporal Scales Firstly, to group sites with similar long-term growth trends we used hierarchical cluster analysis using the complete linkage agglomerative clustering method (HCA) on the Euclidean distances matrix of normalized mean annual BAIspl (ZBAI Legendre and

234

G. Gea-Izquierdo and I. Can˜ellas

Figure 5. Mean GC following Nowacki and Abrams (1997). We only show periods replicated more than five times. Periods 1840–1860 and 1940–1960 are shadowed in light gray. Small (big) dark gray rectangles at the upper (lower) part of graphs indicate the first year of 10-year period (same as GC) during which the mean hydrological year precipitation was 1SD (2SD) over (below) the average.

Legendre 1998). To test the influence of different time frames on the analyses of long-term growth trends we carried out HCA for two periods, 1920– 2008 (HCA1920–2008) to compare the trends since 1920 (this analysis included 9 chronologies) and 1864–2008 (HCA1864–2008) to compare the growth trends starting right after the LIA (6 chronologies were long enough to be included). We excluded ‘‘young’’ trees from the HCA for the reasons previously explained. Secondly, the relationship between climate and growth in the low-frequency was studied comparing smoothed climatic and growth data (BAIspl). Finally, we used bootstrap correlations to study the relationship between climate and the highfrequency of growth (GIres). Unless otherwise indicated, all significance levels are at a = 0.05.

RESULTS A warmer climate without a concurrent increase in rainfall from the early 1900 s increased PET (Figure 2) and led to an abrupt increase in water stress (as

reflected by SPEI) after the 1970 s. The water stress enhancement was greater at southern sites whereas the mean BAI of trees was larger at more humid sites (Table 1; Figure 2; Appendix 2 in supplementary material).

Trends of Old-Growth Trees in Relation to Climate (Years 1650–2008) The oldest ages were found at the coldest locations among those sampled: seven of the 12 old-growth Q. pyrenaica trees came from a mountain site bordering the temperate ecosystems in the north (QUPY2) and five from the mountains in central Spain (QUPY5-QUPY6). The long-term productivity of old-growth trees increased from the second half of the nineteenth century simultaneously to the increase in temperature observed in historical climatic records but also to rising CO2 atmospheric levels. However, the long-term growth trend seemed to match better that of temperature than that of CO2, particularly for the last 50–75 years (Figure 3).

Growth Instability of Mediterranean Oaks

235

in C + presented higher densities (233.3 ± 80.4 trees/ha) than stands in C - (95.0 ± 60.1 trees/ha, Table 1). At low altitudes and below 40! latitude the species exhibited minimum growth and negative growth trends since the 1970 s (Figures 2, 4D). Had we analyzed the growth trends for a shorter period as in HCA1920–2008 the clustering would have been fuzzier than that for HCA1864–2008 (Figure 4D). However, HCA1920–2008 showed that growth either leveled off or started to decrease after the 1970 s at most sites (see also Figure 2). In addition, the ‘‘young’’ trees exhibited negative growth trends in recent years and these trends were often different and more abrupt than those in older age-cohorts just discussed in HCA (Figure 2).

Growth Events: Management and Response to Moisture Availability

Figure 6. BEA between growth and precipitation events. Only significant relationships are shown. We term ‘‘positive’’ the PGC events and ‘‘negative’’ the NGC events. 99% confidence envelopes are in gray dashed lines and were calculated based on 1000 Monte-Carlo simulations. Significant year lags indicating connection between growth and precipitation events are represented with stars.

Oak Growth Trends Along Climatic Gradients (Years 1864–2008) When we analyzed the similarities of the long-term growth trends for the period 1864–2008 the HCA1864–2008 split the data into two groups. The first group included those sites with overall increasing long-term growth trends (C +). The second group (C -) included stands with negative growth trends from 1864 to 1940 and thereafter fairly constant lower BAI than that of trees from C + (Figures 2, 4A, B). Trees within the group with negative trends (C -) suffered harsher water stress conditions than those in the group with positive trends (C +). Mean annual PET (646.3 ± 63.0 mm) and temperature (10.5 ± 2.3!C) were lower and precipitation higher (808.3 ± 113.9 mm) in sites included in C + compared to those in C - (717.8 ± 114.7 mm, 12.6 ± 3.8!C and 528.4 ± 273.4 mm, respectively). Stands

There were some disturbance events between 1820 and 1860 in the three chronologies expanding earlier than 1800. These were particularly evident in QUPY5 and QUPY6, two stands coming from the same slope. Big early peaks in disturbance chronologies of the two sites included in C - , QUPY3 (around 1920) and QUPY10 (around 1860), suggested that the present low density at those sites was presumably the result of thinning (Figure 5). A GC threshold of 25% was significantly related to precipitation events only at three sites (QUPY5, QUPY8, and QUPY9) and PGC peaks were driven by humid periods (events) only at the two driest sites (Table 1), namely QUPY8 and QUPY9 (Figures 5, 6). These two sites presented the largest late NGC events (> 50%) after 1970 among the ten sites studied and their negative growth events were forced by dry conditions (Figures 5, 6). This together with the slow growth and negative recent growth trends at those sites suggest poor tree performance and maximum dependence on moisture availability.

Variability in the Climate–Growth Relationship with Different Local Climatic Conditions Moisture availability was the main climatic factor driving growth (Ppthydrol, SPEIhydrol) and the strength of the positive relationship increased inversely with the amount of precipitation (Figures 7A, 8A). Correlations with PET were very similar to those with temperature, hence we just discuss the latter. Warm springs were beneficial for trees only at colder sites and the intensity of the positive response to spring minimum temperatures

236

G. Gea-Izquierdo and I. Can˜ellas

Figure 7. Bootstrap correlations between GIres and site climatic covariates for the ten sites (period 1950–2008): A precipitation of the hydrological year. B minimum spring temperature. C maximum temperature in June–July; D minimum temperature in January–February. Significant values are shadowed in gray. Dotted thick lines correspond to a linear fit. We show coefficients of determination (R2) values to ease the comparison of trends.

was linearly related to mean site temperature (Figures 7B, 8A). Conversely, the positive effect of warm winters on growth increased with site temperature although this relationship was only significant at the warmest site QUPY8 (Figure 7D). High early summer temperature was negative for growth at some sites (Figures 7C, D, 8A). The group of chronologies with a negative long-term trend (C -) were more sensitive to moisture availability and less to spring temperature than the group with positive trends (C +) in the short-term (Figure 8A) and more sensitive to variability in SPEI in the long-term (Figure 8B, C).

DISCUSSION Enhanced Long-Term Oak Productivity at Species-specific Cold Locations The positive growth trend observed for the last 150 years matched the increase in temperature since the end of the LIA (Luterbacher and others 2006) and was similar to that reported for other species at cold locations in different climates (Spiecker 1999; Salzer and others 2009; Lloyd and others 2011; Di Filippo and others 2012). It also

matched rising CO2 levels until the 1950 s but seemed to diverge thereafter. Some of the many studies analyzing the net effect of the CO2 increase on plants reported a growth enhancement (for example, Huang and others 2007). Nevertheless the positive effect that the CO2 increase can have on trees (for example, improved water use efficiency) often seems to be counteracted by other factors including moisture constraints in Mediterranean ecosystems (Pen˜uelas and others 2008, 2011; Andreu-Hayles and others 2011). Trees present certain variability in their capacity of acclimation to climatic variability and the response of trees to specific stress factors augments when the intensity of the stress factor increases (Cook and Kairiukstis 1990; Tessier and others 1994). The positive effect of spring temperatures for Q. pyrenaica increased inversely to site temperature. Tree growth did not respond negatively to winter temperature, contrary to growth results from Rozas and others (2009) in a maritime location and studies on earlywood hydraulic traits of ring-porous species (Garcia-Gonzalez and Eckstein 2003; Gea-Izquierdo and others 2012). Therefore an increase in productivity at the old-growth cold sites with reduced water constraints could have been

Growth Instability of Mediterranean Oaks

237

Figure 8. Relationships between growth and selected climatic variables for the two groups selected in Figure 4A (that is, C + and C - in the HCA1864–2008). In A we show correlations between climate and the detrended growth index GIres for the period 1950–2008 for all sites (QUPY), C + and C - ; error bars correspond to 1 SD; bootstrap significant levels are as in Figure 7. In B and C we show mean normalized BAI for the two groups (black lines) against normalized mean SPEI of the hydrological year (solid gray lines with circles) smoothed with 40-years splines for the period where we had available climatic data per site (1901–2008).

driven by the rise in temperatures since the LIA. However because water stress was reduced at these humid sites existence of a synergistic positive nonlinear effect of CO2 fertilization should not be discarded (Huang and others 2007). Other factors such as atmospheric nitrogen fertilization could enhance growth but to our knowledge there is no published evidence of a nitrogen-related growth enhancement in the studied area probably also because nitrogen deposition has been rather low (De Vries and Posch 2011).

Land History Influences the Analysis of Growth Dynamics in Managed Woodlands In highly transformed Mediterranean landscapes it can be difficult to disentangle the factors forcing growth trends because changes in historical land use can be more influential than climate to shape present forest structure and composition (Franco Mu´gica and others 1998; Chauchard and others 2007; Ame´ztegui and others 2010). The positive release events observed at the beginning of the two sites with long-term negative trends (C -) likely reflected conversion from a closed forest to an open

silvopastoral system, a common management applied in the past to many stands in the region (Manuel and Gil 1999). However, management was likely not responsible for the decreasing trends observed in Q. pyrenaica at these two sites with open stands (C -). For instance, the overall growth trends of the more drought-tolerant species Q. ilex at the same site as QUPY3 (ZAM in Appendix C in supplementary material) and other similar open stands were positive for the same period (Appendix C in supplementary material; Gea-Izquierdo and others 2011). The climate–growth relationship can be agedependent and is generally not stable over time (Carrer and Urbinati 2004; D’Arrigo and others 2008; Lloyd and others 2011; Porter and Pisaric 2011). It can be misleading to analyze changes in long-term productivity in relation to climate in highly stocked stands if available data have a short time span because it is complex to filter out any mid-frequency non-climatic disturbance in young and crowded stands. We specifically separated the young trees in the analysis and observed contrasting growth trends of different age classes at some sites, which could be explained by non-climatic factors (Johnson and Abrams 2009). For instance,

238

G. Gea-Izquierdo and I. Can˜ellas

they could reflect differential size-dependent competition, be explained by rooting-depth (Rice and others 2004; Niinemets 2010) or whether young resprouts had different growth dynamics than older cohorts of seedling origin (Clarke and others 2010). Old-growth trees were present only at the coldest sites (Di Filippo and others 2012) which were also those less accessible in the mountains. Slow growth of old-growth trees up to 1850 could partly explain their older age (Bigler and Veblen 2009; Di Filippo and others 2012) but the location of old stands was also likely influenced by land use (Manuel and Gil 1999).

Recent Instability in Long-Term Growth Suggests Vulnerability at the Dry Edge The growth trends of shorter chronologies analyzed since 1864 also suggested the existence of an increase in productivity since 150 years ago at coldhumid sites but only up to the 1970 s. That decade marks the start of an abrupt increase in temperature triggering an advance in plant phenology in Iberia (Gordo and Sanz 2009) and with the exception of the coldest site (QUPY2) the BAI of all sites leveled off or decreased after the 1970 s. The sensitivity of trees to moisture availability exhibited an inverse linear relationship with site precipitation and contrary to trees at cold-humid sites the trees at dry sites did not exhibit a long-term growth enhancement. Q. pyrenaica occupies small residual areas in the southern mountains of the Iberian Peninsula (Costa and others 2005) and the recent minimum growth rates with negative trends were observed at latitudes below 40!. Slow growth was also evidenced at cold sites but not a simultaneous age-independent negative growth trend characteristic of declining stands (Bigler and Veblen 2009). Increasing belowground competition could have been responsible for the observed recent negative trends after the 1970 s but then this should have been expressed similarly at most sites, which was not the case. The growth trends discussed were neither an age effect because young trees were excluded from the analysis and the oldest trees exhibited positive growth trends, that is, did not show senescence. Different provenances can have different adaptations to climate (Schenk 1996) and along gradients growth will be determined by the interaction between the climatic variables studied and factors such as nutrient availability, incident radiation or the photoperiod (Niinemets 2010). Nevertheless, the negative trends over recent years and the greater response to moisture availability found at warmer and drier low elevations in the

south suggest vulnerability to warming at the local, low elevation dry edge of the species’ range (for example, QUPY8, QUPY9). Growth is negatively affected and mortality enhanced by drought episodes (Cook and others 1987; Leblanc and Foster 1992; Voelker and others 2008) and other studies in the Mediterranean report a similar decrease in productivity with a recent increase in vulnerability to water stress at species-specific dry sites (Pen˜uelas and others 2011; Sarris and others 2011; Gea-Izquierdo and others 2011; Di Filippo and others 2012). Currently we seem to be within one of the driest periods in the last centuries in the studied region (for example, Luterbacher and others 2006) and a recent overall increase in moisture constraints is evident (Garcı´a-Ruiz and others 2011). It seems thus reasonable to assume that the observed instability in growth trends reflects long-term changes in Q. pyrenaica performance with climate change. This assumption agrees with the expected contraction of the submediterranean range with climate change (Sa´nchez de Dios and others 2009) and shift of species at the limit of their distribution, including the advance of more drought-tolerant taxa at dry sites but also forest expansion at cold sites (Thuiller and others 2005; Galiano and others 2010; Ruiz-Labourdette and others 2012). Trees can eventually recover from sharp growth decreases (Biondi 1999; Johnson and Abrams 2009) and resilience from recurrent drought can be patchy and related to soil conditions (Jenkins and Pallardy 1995; Lloret and others 2004; Rice and others 2004). Therefore, it should be monitored whether the increase in resource limitation suggested by the growth trends since the 1970 s leads to forest decline and a shortterm increase in mortality or self-thinning in the case of highly stocked even-aged young stands (Dietze and Moorcroft 2011).

CONCLUSIONS Growth of Q. pyrenaica was mainly driven by cumulative precipitation and spring temperatures but the importance of the two climatic factors dependent on local site conditions and was not constant in time. After a period of reduced productivity since the 1600 s, growth of Q. pyrenaica trees from cold and moist sites increased since the mid 1800 s simultaneous to a warming climate. The positive increase in the sensitivity of trees to spring temperatures at colder sites and the agreement of the long-term growth trends of old-growth trees with those of temperature suggest the exis-

Growth Instability of Mediterranean Oaks tence of an increase in productivity at cold-humid sites driven mainly by warming temperatures. Conversely changes in climate were detrimental for long-term growth at drier sites expressing more water limitation. The intensity of the positive growth response to moisture availability decreased linearly with rainfall, hence increased with overall water stress, and after an abrupt increase in temperature since the 1970 s the growth rate slowed down at most sites, except the coldest. This may reflect that rising moisture constraints threaten particularly the sustainability of low elevation stands below 40! latitude. It should be monitored whether this negative response to climate marks the beginning of Q. pyrenaica decline at the local dry edge of its distribution.

ACKNOWLEDGMENTS G.G.I. thanks the Dendro group and Paolo Cherubini at WSL for hosting and the Spanish Ministry of Science (MICINN) for funding through a postdoctoral contract. We are grateful to all foresters who made possible the sampling at the different sites and to Enrique Garriga for his help during field sampling and processing of wooden cores. This research was supported by projects AGL2010-2115301, funded by MICINN, and S2009/AMB-1668. It is also a contribution to the Labex OT-Med (no. ANR11-LABX-0061) funded by the «Investissements d’Avenir» program of the French National Research Agency through the A*MIDEX project (no. ANR-11-IDEX-0001-02).

REFERENCES Allen CD, Macalady AK, Chenchouni H et al. 2010. A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. For Ecol Manag 259:660–84. Ame´ztegui A, Brotons L, Coll L. 2010. Land-use changes as major drivers of mountain pine (Pinus uncinata ram.) expansion in the pyrenees. Glob Ecol Biogeogr 19:632–41.

239

Bigler C, Veblen TT. 2009. Increased early growth rates decrease longevities of conifers in subalpine forests. Oikos 118:1130–8. Biondi F. 1999. Comparing tree-ring chronologies and repeated timber inventories as forest monitoring tools. Ecol Appl 9:216–27. Biondi F, Qeadan F. 2008. A theory-driven approach to tree-ring standardization: defining the biological trend from expected basal area increment. Tree-Ring Res 64:81–96. Black BA, Abrams MD. 2003. Use of boundary-line growth patterns as a basis for dendroecological release criteria. Ecol Appl 13:1733–49. Boisvenue C, Running SW. 2006. Impacts of climate change on natural forest productivity—evidence since the middle of the 20th century. Glob Chang Biol 12:862–82. Breshears DD, Myers OB, Meyer CW et al. 2009. Tree die-off in response to global change-type drought: mortality insights from a decade of plant water potential measurements. Front Ecol Environ 7:185–9. Carrer M, Urbinati C. 2004. Age-dependent tree-ring growth responses to climate in Larix decidua and Pinus cembra. Ecology 85:730–40. Carrer M, Nola P, Motta R, Urbinati C. 2010. Contrasting treering growth to climate responses of Abies alba toward the southern limit of its distribution area. Oikos 119:1515–25. Chauchard S, Carcaillet C, Guibal F. 2007. Patterns of land-use abandonment control tree-recruitment and forest dynamics in Mediterranean mountains. Ecosystems 10:936–48. Clarke PJ, Lawes MJ, Midgley JJ. 2010. Resprouting as a key functional trait in woody plants—challenges to developing new organizing principles. New Phytol 188:651–4. Cook ER, Johnson AH, Blasing TJ. 1987. Forest decline: modeling the effect of climate in tree rings. Tree Physiol 3:27–40. Cook ER, Kairiukstis LA, Eds. 1990. Methods of dendrochronology. Applications in the environmental sciences. Dordrecht: Kluwer. Corcuera L, Camarero JJ, Siso S, Gil-Pelegrı´n E. 2006. Radialgrowth and wood-anatomical changes in overaged Quercus pyrenaica coppice stands: functional responses in a new Mediterranean landscape. Trees-Struct Funct 20:91–8. Costa M, Morla C, Sa´inz H, Eds. 2005. Los bosques ibe´ricos. Una interpretacio´n geobota´nica. Barcelona: Editorial Planeta. Crawley MJ. 2007. The R Book. West Sussex: Wiley. 942 pp D’Arrigo R, Wilson R, Liepert B, Cherubini P. 2008. On the ‘divergence problem’ in northern forests: a review of the treering evidence and possible causes. Glob Planet Chang 60:289– 305.

Andreu-Hayles L, Planells O, Gutierrez E et al. 2011. Long treering chronologies reveal 20th century increases in water-use efficiency but no enhancement of tree growth at five Iberian pine forests. Glob Chang Biol 17:2095–112.

De Vries W, Posch M. 2011. Modelling the impact of nitrogen deposition, climate change and nutrient limitations on tree carbon sequestration in Europe for the period 1900–2050. Environ Pollut 159(10):2289–99.

Bertrand R, Lenoir J, Piedallu C, Riofrio-Dillon G, de Ruffray P, Vidal C, Pierrat JC, Ge´gout JC. 2011. Changes in plant community composition lag behind climate warming in lowland forests. Nature 479:517–20.

Dietze MC, Moorcroft PR. 2011. Tree mortality in the eastern and central United States: patterns and drivers. Glob Chang Biol 17:3312–26.

Bigler C, Braker OU, Bugmann H, Dobbertin M, Rigling A. 2006. Drought as an inciting mortality factor in Scots pine stands of the Valais, Switzerland. Ecosystems 9:330–43.

Di Filippo A, Alessandrini A, Biondi F et al. 2010. Climate change and oak growth decline: dendroecology and stand productivity of a Turkey oak (Quercus cerris L.) old stored coppice in Central Italy. Ann For Sci 67:706.

Bigler C, Gavin DG, Gunning C, Veblen TT. 2007. Drought induces lagged tree mortality in a subalpine forest in the Rocky Mountains. Oikos 116:1983–94.

Di Filippo A, Biondi F, Maugeri M et al. 2012. Bioclimate and growth history affect beech lifespan in the Italian Alps and Apennines. Glob Chang Biol 18:960–72.

240

G. Gea-Izquierdo and I. Can˜ellas

Dobbertin M. 2005. Tree growth as indicator of tree vitality and of tree reaction to environmental stress: a review. Eur J For Res 124:319–33.

Lloret F, Pen˜uelas J, Ogaya R. 2004. Establishment of co-existing Mediterranean tree species under a varying soil moisture regime. J Veg Sci 15:237–44.

Etheridge DM, Steele LP, Langenfelds RL et al. 1996. Natural and anthropogenic changes in atmospheric CO2 over the last 1000 years from air in Antarctic ice and firn. J Geophys Res 101:4115–28.

Lloyd AH, Bunn AG, Berner L. 2011. A latitudinal gradient in tree growth response to climate warming in the Siberian taiga. Glob Chang Biol 17:1935–45. Luo Y, Chen HYH. 2011. Competition, species interaction and ageing control tree mortality in boreal forests. J Ecol 99:1470– 80.

Franco Mu´gica F, Anton MG, Ollero HS. 1998. Vegetation dynamics and human impact in the Sierra de Guadarrama, Central System, Spain. Holocene 8:69–82. Fritts HC. 1976. Tree rings and climate. Caldwell: Blackburn Press. 567 pp Galiano L, Martı´nez-Vilalta J, Lloret F. 2010. Drought-induced multifactor decline of Scots pine in the pyrenees and potential vegetation change by the expansion of co-occurring oak species. Ecosystems 13:978–91. Garcia-Gonzalez I, Eckstein D. 2003. Climatic signal of earlywood vessels of oak on a maritime site. Tree Physiol 23:497–504. Garcia-Ruiz JM, Ignacio Lopez-Moreno J, Vicente-Serrano SM, Lasanta-Martinez T, Begueria S. 2011. Mediterranean water resources in a global change scenario. Earth-Sci Rev 105:121–39. Gavin DG, Hu FS, Lertzman K, Corbett P. 2006. Weak climatic control of stand-scale fire history during the late Holocene. Ecology 87:1722–32. Gea-Izquierdo G, Cherubini P, Can˜ellas I. 2011. Tree-rings reflect the impact of climate change along a temperature gradient in Spain over the last 100 years. For Ecol Manag 262:1807–16. Gea-Izquierdo G, Fonti P, Cherubini P, Martı´n-Benito D, Chaar H, Can˜ellas I. 2012. Xylem hydraulic adjustment and growth response of Quercus canariensis Willd. to climatic variability. Tree Physiol 32:401–13. Gordo O, Sanz JJ. 2009. Long-term temporal changes of plant phenology in the Western Mediterranean. Glob Chang Biol 15:1930–48. Holmes RL. 1983. Computer-assisted quality control in tree-ring dating and measurement. Tree-Ring Bull 43:69–78. Huang JG, Bergeron Y, Denneler B, Berninger F, Tardif J. 2007. Response of forest trees to increased atmospheric CO2. Crit Rev Plant Sci 26:265–83. IPCC. 2007. R.K. Pachauri and A. Reisinger, Eds. Climate change 2007: synthesis report. IPCC, Geneva, Switzerland, 104 p. Jenkins MA, Pallardy SG. 1995. The influence of drought on red oak group species growth and mortality in the Missouri Ozarks. Can J For Res 25:1119–27. Johnson SE, Abrams MD. 2009. Age class, longevity and growth rate relationships: protracted growth increases in old trees in the eastern United States. Tree Physiol 29:1317–28. Keeling CD, Piper SC, Bacastow RB, et al. 2001 Exchanges of atmospheric CO2 and 13CO2 with the terrestrial biosphere and oceans from 1978 to 2000. Global aspects, SIO Ref Ser 01–06. San Diego: Scripps Institution of Oceanography, 88 pp. Legendre P, Legendre L. 1998. Numerical ecology. Amsterdam: Elsevier. Leblanc DC, Foster JR. 1992. Predicting effects of global warming on growth and mortality of upland oak species in the Midwestern United States—a physiologically based dendroecological approach. Can J For Res 22:1739–52. Linares JC, Camarero JJ, Carreira JA. 2009. Interacting effects of changes in climate and forest cover on mortality and growth of the southernmost European fir forests. Glob Ecol Biogeogr 18:485–97.

Luterbacher J, Xoplaki E, Casty C, et al. 2006. Mediterranean climate variability over the last centuries: a review. In: Lionello P, Malanotte-Rizzoli P, Boscolo R, Eds. The Mediterranean climate: an overview of the main characteristics and issues. Amsterdam: Elsevier. pp 27–148. Manuel C, Gil L. 1999. La transformacio´n histo´rica del paisaje forestal en Espan˜a. En ‘Segundo IFN 1986-1996’. Madrid: Ministerio de Medio Ambiente. pp 15–104. Montserrat-Marti G, Camarero JJ, Palacio S et al. 2009. Summer-drought constrains the phenology and growth of two coexisting Mediterranean oaks with contrasting leaf habit: implications for their persistence and reproduction. TreesStruct Function 23:787–99. Mitchell TD, Jones PD. 2005. An improved method of constructing a database of monthly climate observations and associated high-resolution grids. Int J Climatol 25:693–712. Myneni RB, Keeling CD, Tucker CJ, Asrar G, Nemani RR. 1997. Increased plant growth in the northern latitudes from 1981 to 1991. Nature 386:698–702. Niinemets U. 2010. Responses of forest trees to single and multiple environmental stresses from seedlings to mature plants: past stress history, stress interactions, tolerance and acclimation. For Ecol Manag 260:1623–39. Nobel PS. 2009. Physicochemical and environmental plant physiology. 4th edn. Elsevier: Academic Press. Nowacki GJ, Abrams MD. 1997. Radial-growth averaging criteria for reconstructing disturbance histories from presettlement-origin oaks. Ecol Monogr 67:225–49. Pen˜uelas J, Hunt JM, Ogaya R, Jump AS. 2008. Twentieth century changes of tree-ring delta C-13 at the southern rangeedge of Fagus sylvatica: increasing water-use efficiency does not avoid the growth decline induced by warming at low altitudes. Glob Chang Biol 14:1076–88. Pen˜uelas J, Canadell JG, Ogaya R. 2011. Increased water-use efficiency during the 20th century did not translate into enhanced tree growth. Glob Ecol Biogeogr 20:597–608. Piovesan G, Biondi F, Di Filippo A, Alessandrini A, Maugeri M. 2008. Drought-driven growth reduction in old beech (Fagus sylvatica L.) forests of the central Apennines, Italy. Glob Chang Biol 14:1265–81. Porter TJ, Pisaric MFJ. 2011. Temperature–growth divergence in white spruce forests of Old Crow Flats, Yukon Territory, and adjacent regions of northwestern North America. Glob Chang Biol 17:3418–30. Pretzsch H, Biber P. 2005. A re-evaluation of Reineke’s rule and stand density index. For Sci 51:304–20. Rathgeber C, Guiot J, Roche P, Tessier L. 1999. Quercus humilis increase of productivity in the mediterranean area. Ann For Sci 56:211–19. Rice KJ, Matzner SL, Byer W, Brown JR. 2004. Patterns of tree dieback in Queensland, Australia: the importance of drought stress and the role of resistance to cavitation. Oecologia 139:190–8.

Growth Instability of Mediterranean Oaks Rozas V. 2003. Tree age estimates in Fagus sylvatica and Quercus robur: testing previous and improved methods. Plant Ecol 167:193–212. Rozas V, Lamas S, Garcı´a-Gonza´lez I. 2009. Differential treegrowth responses to local and large-scale climatic variation in two Pinus and two Quercus species in northwest Spain. Ecoscience 16:299–310. Ruiz-Labourdette D, Nogues-Bravo D, Sainz Ollero H, Schmitz MF, Pineda FD. 2012. Forest composition in Mediterranean mountains is projected to shift along the entire elevational gradient under climate change. J Biogeogr 39:162–76. Salzer MG, Hughes MK, Bunn AG, Kipfmueller KF. 2009. Recent unprecedented tree-ring growth in bristlecone pine at the highest elevations and possible causes. PNAS 106:20348–53. Sa´nchez de Dios R, Benito-Garzo´n M, Sainz-Ollero H. 2009. Present and future extension of the Iberian submediterranean territories as determined from the distribution of marcescent oaks. Plant Ecol 204:189–205. Sarris D, Christodoulakis D, Korner C. 2011. Impact of recent climatic change on growth of low elevation eastern Mediterranean forest trees. Clim Chang 106:203–23. Schenk HJ. 1996. Modeling the effects of temperature on growth and persistence of tree species: a critical review of tree population models. Ecol Model 92:1–32. Spiecker H. 1999. Overview of recent growth trends in European forests. Water Air Soil Pollut 116:33–46.

241

Sua´rez ML, Ghermandi L, Kitzberger T. 2004. Factors predisposing episodic drought-induced tree mortality in Nothofagus—site, climatic sensitivity and growth trends. J Ecol 92:954–66. Tessier L, Nola P, Serrebachet F. 1994. Deciduous Quercus in the Mediterranean Region—tree-ring/climate relationships. New Phytol 26:355–67. Thuiller W, Lavorel S, Araujo MB, Sykes MT, Prentice IC. 2005. Climate change threats to plant diversity in Europe. PNAS 102:8245–50. Valbuena-Caraban˜a M, de Heredia UL, Fuentes-Utrilla P, Gonzalez-Doncel I, Gil L. 2010. Historical and recent changes in the Spanish forests: a socio-economic process. Rev Palaeobot Palynol 162:492–506. Van Mantgem PJ, Stephenson NL, Byrne JC et al. 2009. Widespread increase of tree mortality rates in the Western United States. Science 323:521–4. Vicente-Serrano SM, Beguerı´a S, Lo´pez-Moreno JI. 2010. A multi-scalar drought index sensitive to global warming: the standardized precipitation evapotranspiration index—SPEI. J Clim 23(7):1696–718. Voelker SL, Muzika RM, Guyette RP. 2008. Individual tree and stand level influences on the growth, vigor, and decline of red oaks in the Ozarks. For Sci 54:8–20. Wiley E, Helliker B. 2012. A re-evaluation of carbon storage in trees lends greater support for carbon limitation to growth. New Phytol 195:285–9.