Wet season precipitation during the past century reconstructed from tree-rings of a tropical dry forest in Southern Ecuador

Global and Planetary Change 133 (2015) 65–78

Contents lists available at ScienceDirect

Global and Planetary Change journal homepage: www.elsevier.com/locate/gloplacha

Wet season precipitation during the past century reconstructed from tree-rings of a tropical dry forest in Southern Ecuador Darwin Pucha-Cofrep a,b,⁎, Thorsten Peters a, Achim Bräuning a a b

Institute of Geography, University of Erlangen-Nuremberg, Wetterkreuz 15, D-91058 Erlangen, Germany Universidad Nacional de Loja, Carrera de Ingeniería Forestal, Ciudadela Universitaria Guillermo Falconí Espinosa “La Argelia”, EC-110101 Loja, Ecuador

a r t i c l e

i n f o

Article history: Received 20 February 2015 Received in revised form 31 July 2015 Accepted 4 August 2015 Available online 5 August 2015 Keywords: Tree-rings Tropical dry forest Reconstructed precipitation Wet season Bursera graveolens Maclura tinctoria

a b s t r a c t This study investigates the dendroclimatic potential of tree species in a tropical dry forest in southern Ecuador. From 10 selected tree species, Bursera graveolens and Maclura tinctoria exhibited distinct annual and crossdatable tree-rings. It was possible to synchronize individual tree-ring series and to establish two tree-ring chronologies of 203 and 87 years length, respectively. The characteristic ENSO frequency band is reflected in wavelet power spectra of both chronologies. Both species show a strong correlation between ring width and precipitation of the wet season (January–May). Strong El Niño events (1972, 1983 and 1998) lead to strong growth responses in the tree-ring chronologies, whereas ‘normal’ ENSO events do not trigger long-lasting growth responses. The first ring-width based wet-season precipitation reconstruction for the past 103 years was developed. Statistical and spatial correlation analysis verified the skills of the reconstructed precipitation which captures a great part of the Rainfall Index over the land area of Ecuador and the equatorial Pacific. Furthermore, teleconnections with central Pacific precipitation and SST patterns were found. © 2015 Elsevier B.V. All rights reserved.

1. Introduction After their initial development almost one century ago (Douglass, 1919), tree-ring studies were conducted worldwide. Currently, more than 3000 tree-ring data sets are stored in the International Tree-Ring Data Bank (ITRDB, NOAA http://www.ncdc.noaa.gov/data-access/ paleoclimatology-data/datasets/tree-ring). In comparison to the temperate climate zones, the tropical regions are still strongly underrepresented and require much further research (Zhou et al., 2013). A major challenge for tropical dendrochronology is the indistinct or missing formation of tree-rings in most species due to the lack of climatic seasonality (Rozendaal and Zuidema, 2011). However, as early as 1927, tropical tree species with distinct rings have first been described (Coster, 1927). In the last decades, tropical dendrochronology has been developing rapidly through identifying a considerable number of tree species with annual growth rings. The annual nature of these rings was proven by studying cambial phenology (Krepkowski et al., 2011; Volland-Voigt et al., 2011), wood anatomy (Harley, 2013; Rozendaal and Zuidema, 2011; Worbes, 1995), and stable isotopes (Anchukaitis et al., 2008; Evans and Schrag, 2004). Tree-ring formation in humid tropical forests can be an effect of the occurrence of short dry periods or of leaf phenology (Brienen and Zuidema, 2005). While tree-

⁎ Corresponding author at: Institute of Geography, University of Erlangen-Nuremberg, Wetterkreuz 15, D-91058 Erlangen, Germany. E-mail address: [email protected] (D. Pucha-Cofrep).

http://dx.doi.org/10.1016/j.gloplacha.2015.08.003 0921-8181/© 2015 Elsevier B.V. All rights reserved.

ring formation in inundation forests is triggered by the annual flooding cycle (Callado et al., 2001; Schöngart et al., 2005), in dry forests it is controlled by a marked annual rainfall seasonality regulating the water balance of the ecosystem (Fichtler et al., 2004; Reich, 1995; Rodríguez et al., 2005; Worbes, 1999). The lack of a comprehensive network of tree-ring chronologies in tropical ecosystems poses a deficit for a better understanding of large-scale tropical forest ecology and global carbon balances. Additional tree-ring chronologies providing species-specific information on physiological responses to the local and regional climate variability are urgently needed to get a broader picture of long-term changes and adaptations of tropical forest ecosystems (Rozendaal and Zuidema, 2011). While humid tropical forests received considerable attention by ecologists and dendrochronologists, tropical dry forests are almost understudied (Brienen and Zuidema, 2005; Chazdon et al., 2007; Mooney et al., 1995; Sanchez-Azofeifa et al., 2005). The El Niño/Southern Oscillation (ENSO) tropical circulation system influences large parts of tropical South America and other parts of the earth via long-distance teleconnections. Existing tree-ring based climate reconstructions for the South American tropics or ENSO have often been based on moisture-sensitive chronologies from subtropical or semiarid temperate regions in Central and North America (D'Arrigo et al., 2005; Stahle et al., 1988). However, teleconnections between local moisturesensitive proxies and ENSO are often not stable, and so correlations between proxy-derived ENSO reconstructions vary over time (D'Arrigo et al., 2005; Wilson et al., 2010). Hence, more reconstructions of local precipitation and variations of ENSO-related climate parameters from tropical areas are needed.

66

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

Fig. 1. Study area, climate stations (triangles) and sample sites (circles). For climate stations labels refer to Table 1.

For that purpose the first dendrochronological study in the semideciduous dry forests of southern Ecuador was conducted to generate regional knowledge on historical climate changes in the equatorial tropics, and to study the influence of climate factors on these ecosystems. The potential of new tree species for dendrochronology was evaluated in order to achieve the following aims: (1) to select and identify potential tree species for dendrochronology based on the identification of distinct annual growth rings through analyzing wood anatomy; (2) to construct local climate-sensitive tree-ring width (TRW) chronologies in order to analyze and understand the impact of the local and regional

climate on the annual growth of this forest ecosystem; and (3) to develop a ring-width based precipitation reconstruction. 2. Materials and methods 2.1. Study site This study was carried out in the dry forest Laipuna Nature Reserve (4°22′S, −79°90′W; 1.600 ha) located in the core of UNESCO “Bosque Seco” Biosphere Reserve in the Catamayo River canyon in southern

Table 1 Parameters of the climate stations used to build the regional precipitation and temperature series. No

Label

Station name

Correlation coefficient r [Prec. Laipuna]

Distance from Laipuna [km]

Altitude [m a.s.l.]

Available records [time period]/[no. years]

Annual mean precipitation [mm]

1 2 3 4 5 6 7 8 9 10 11 12 13 14

LA LG EL SB CA SG ME SO CO UT AL SC MA ZA

Laipuna Lauro Guerrero El Limo Sabanilla Catacocha Sabiango Mercadillo Sozoranga Colaisaca Utuana Alamor Saucillo Macaraa Zapotilloa

1.00 0.97 0.95 0.94 0.93 0.92 0.92 0.92 0.91 0.91 0.89 0.84 0.83 0.82

0 30 34 24 32 20 22 18 26 27 25 35 20 40

590 1910 1150 733 1808 734 1125 1510 2410 2410 1250 328 427 223

2007–2012/6 1975–2013/22 1982–2013/10 1975–2013/24 1963–2013/50 1972–2013/41 1975–2013/21 1979–2013/39 1963–2013/49 1982–1986/5 1963–2013/48 1967–2013/44 1958–1986/26 1963–2013/46

648 1522 2140 730 907 1234 1238 1208 1165 1573 1357 788 630 688

a

Climate stations with available temperature data. All correlations are significant at the 0.01 level.

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

67

Fig. 2. Monthly mean precipitation and temperature of Laipuna Nature Reserve (2007–2012) and a calculated regional climate series (1960–2013).

Ecuador (Fig. 1). The area covers an altitudinal gradient from 590 to 1480 m a.s.l. and is located in a semihumid region between the Pacific coastline and the lower western slopes of the Andes mountain range. This ecosystem is classified as a premontane semi-deciduous dry forest (Aguirre et al., 2006; MAE, 2013; Sierra, 1999) and characterized by different climate types along altitudinal gradients. In this study climate data of five years of local instrumental measurements (2007–2012) were used from the climate station called “Laipuna (LA)” located at 590 m a.s.l. Climate data were recorded by an automatic local climate station (THIES Clima, Germany; Table 1) since 2007. Air temperature was measured 2 m above ground level while precipitation was measured 1 m above ground. Both climate variables were recorded in 10 minute intervals and hourly arithmetic means were stored by a data logger. The local mean annual precipitation is 625 mm, the rainy season lasts from January to May, and the dry season last from middle of May until December. Temperature shows little variability over the course of the year, with an annual mean of 23.4 °C (Fig. 2).

2.2. Regional precipitation, temperature and SST data Regional instrumental climate records of precipitation (42 climate stations) and temperature (7 climate stations) were provided from the National Institute of Meteorology and Hydrology of Ecuador (INAMHI). Monthly means of all climate variables were correlated with local climate data series, and the climate stations with the highest positive Pearson's correlation values (r N 0.8) were selected. The retained climate stations (13 for precipitation and 2 for temperature, see Table 1) were used to calculate a regional time series for precipitation (1960–2013) and temperature (1970–2013) applying the technique described by Jones and Hulme (1996) (Fig. 2). Sea surface temperature (SST) anomalies from the Niño 3.4 region were accessed from the National Center for Atmospheric Research (NCAR; period 1900–2007 with smoothed and normalized anomalies of a 5-month running mean), and from the Climate Prediction Center (CPC) of the

National Oceanic Atmospheric Administration (NOAA) to complete the period 2008–2012. 2.3. Tree sampling, ring-width measurement and chronology development A total of 10 tree species (Bursera graveolens, Capparis scabrida, Ceiba trichistandra, Celpis loxensis, Cynothalla mollis, Ipomoea calodendrum, Maclura tinctoria, Tabebuia chrysantha, Terminalia valverdae and Viguiera sp.) were used in this study. The selection of trees in the field was focused on collect samples from larger trees to develop a long chronology. Samples were collected with a Pressler borer of 5 mm at breast height of the trunk. In a first field campaign, three to four individual trees of each species were sampled. Four cores per tree were collected in the four cross directions of the trunk (total 34 trees, 136 radii). The increment cores were mounted in wooden grooved holders with a vertical direction of the vessels, air-dried and brought to the laboratory. Then, the surface of the cores was smoothed with consecutively finer sanding pads up to 4000 grit for tree-ring measurements. To confirm the presence of true annual rings in the selected tree species, thin sections of 10 to 20-μm thicknesses were cut with a microtome to take microscopic photographs and to determine the ring boundaries (Fig. 3). The sections were stained with solutions of safranin and astra blue, and washed with ethanol, following standard procedures (Gärtner and Schweingruber, 2013). The description of the wood anatomical features was done following the classification in (Carlquist, 1988). After inspection of the wood anatomy of all collected tree species for anatomically visible treering boundaries, a selection of the tree species with anatomically clear tree-ring characteristics for developing tree-ring chronologies was made. Accordingly, B. graveolens and M. tinctoria were selected for further study. After selection of the target tree species, two additional field campaigns were made to collect additional samples of the selected species, completing at the end of this study a total of 59 individual trees and 224 radii. At this time, stem discs from fallen trees only from B. graveolens

Fig. 3. Wood cross-sections of a) B. graveolens and b) M. tinctoria showing the wood anatomy elements. White arrows indicate the distinct boundaries in narrow tree-rings.

68

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

Table 2 El Niño/La Niña events classification. El Niño ‘normal’

El Niño ‘strong’

La Niña ‘normal’

La Niña ‘long’

1902, 1905, 1912, 1914, 1919, 1923, 1926, 1931, 1941, 1951, 1958, 1963, 1965, 1969, 1977, 1987, 1992, 2002, 2004, 2006, 2009

1972, 1982, 1997

1904, 1916, 1924, 1933, 1942, 1964, 1988, 2007, 2010, 2012

1909, 1949, 1955, 1971, 1973, 1985, 2000

Table 3 Statistical characteristics of raw tree-ring width chronologies (Rbar and mean EPS statistics were calculated from the detrended chronology). Parameter

Bursera graveolens

Maclura tinctoria

No. of trees/radii Time period Mean length of series [years] Mean ring width [mm]/SD Mean sensitivity 1st order autocorrelation Mean glka Interval analysis Rbar.wta/Rbar.bta Mean EPSa

18/50 1809–2011 (203 years) 92 1.44 ± 1.18 0.66 0.27 0.56 1908–2011 (103 years) 0.53/0.22 0.85

12/47 1926–2012 (87 years) 36 3.01 ± 2.13 0.68 0.22 0.66 1964–2012 (48 years) 0.77/0.62 0.93

a glk: Gleichläufigkeit (sign test), Rbar.wt: mean interseries correlation withintrees, Rbar.bt: mean interseries correlation between-trees, EPS: Expressed Population Signal.

were found (three discs), and these discs were used as references for a better precision in the crossdating procedure. Tree-ring width (TRW) measurements were made using a Lintab 6 measuring system (Rinntech, Heidelberg, Germany) with a precision of 0.01 mm using the TSAP-Win program (Rinn, 2012). Cross-dating of ring-width series was accomplished by visual inspection of treering patterns (Stokes and Smiley, 1968) and statistical tests (Cook et al., 1990). In the first step, ring-width series from the stem discs were synchronized. In the second step, the mean curves of these discs were used as reference for synchronizing the ring-width series of the increment cores, where the occurrence of missing rings could not a priori be excluded. Every ring-width series was checked visually and

statistically for cross-dating quality, and series with poor visibility of ring boundaries, the occurrence of several missing rings or crossdating problems after several measurement attempts were discarded. From B. graveolens 18 trees/50 cores (90% trees/91% cores from a total sampled size of 20 trees/55 cores) were used to build a site chronology. For M. tinctoria 12 trees/47 cores (92% trees/96% cores from a total of 13 trees and 49 cores) were selected to construct another species-specific site chronology. The statistical quality of the cross-dated TRW series was checked by using the open-source packages “dendrochronology program library in R (dplR)” (Bunn, 2008, 2010) and “detrendR” (Campelo et al., 2012) within the R statistical programming environment (R Development Core Team, 2009) (Table 3). To remove age related growth trends and natural competitions effects, the individual series were standardized using a one-step interactive detrending procedure through the “detrenderR” package. Due to the shortness of the tree-ring series (92 years on average), a 32-year cubic smoothing spline length with 60% frequency response (p = 0.6) and a spline ratio of 0.6 was used for age trend elimination to retain multi-decadal decadal variability in the final chronology. The chronologies were built with the chron (dplR-package) function using the detrended ring-width index (RWI) series with a robust mean to reduce the influence of outliers (Briffa and Jones, 1990). Due to the low first order autocorrelation of the raw series (Table 3), autoregressive modeling was not applied and the standard chronology was used for this study. 2.4. Data analysis and climate response To evaluate the effect of climate on TRW, bootstrapped correlation and response functions analysis based on principal components

Fig. 4. Ring-width index chronologies of Bursera graveolens and Maclura tinctoria (black lines). The red lines show a 10-year smoothing spline. The green lines indicate the EPS, the green dotted lines represent the recommended threshold of 0.85. The filled gray areas indicate the number of samples included in the chronologies. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

69

Fig. 5. RWI series and wavelet spectrum with cone of influence (shaded area). Black contours show frequencies significant on the 0.01 confidence level, black dashed lines show signals with decreasing periodicity trends. a) Bursera graveolens, and b) Maclura tinctoria.

regression with monthly precipitation, temperature and ENSO3.4 SSTs data were calculated using the dendroclimatic calibration bootRes package (Zang and Biondi, 2012) in R. Changing frequency domains over the length of the RWI chronologies were studied by a Morlet-Wavelet analysis (Torrence and Compo, 1998) through the “dplR” morlet and wavelet.plot functions. The considered periods cover 1876–2011 and 1938–2012 for B. graveolens and M. tinctoria, respectively. A Superposed Epoch Analysis (SEA) (Samson and Yeung, 1986) was applied to test the influence of El Niño/La Niña events on the local precipitation and temperature and on tree growth. This analysis tests the significance of a time series response to certain events within a window including three years before and after the event. Bootstrapped confidence intervals were calculated with 1000 random resampling with the “dplR” sea function (Bunn, 2008). To compute the SEA, El Niño years were classified as ‘normal’ and ‘strong’ events, and La Niña years as ‘normal’ and ‘long’ events. ‘Normal’ events were defined as years in which SST anomalies passed a threshold ±0.4 °C for six months or more (Trenberth, 1997). ‘Strong’ events were defined as events with anomalies exceeding ± 2.5 °C. La Niña ‘long’ events were defined as events longer than 15 consecutive months exceeding a threshold − 0.4 °C (Table 2). These data were compared with event lists from the literature (Brönnimann et al., 2007; Hoell et al., 2013; Yu and Kim, 2013). SEA was calculated for the period 1900–2011 for B. graveolens, and for 1950–2012 for M. tinctoria based on the Expressed Population Signal (EPS), which indicates the chronology quality as a fraction of the common variance of a theoretical infinite tree population (Wigley et al., 1984). To quantify the regional climatic forcing on the RWI chronologies, the RWI with surface level precipitation and temperature datasets

(NCEP; National Centers for Environmental Prediction) of the Climate Forecast System Reanalysis period 1979–2009 (CFSR; Saha et al., 2010) were correlated. The CFRS data set was selected because it has a high spatial resolution and wide coverage including land and Ocean areas. Spatial correlation analyses (p b 0.05) were carried out with the KNMI Climate Explorer (Royal Netherlands Meteorological Institute, http://climexp.knmi.nl) for the Pacific Ocean and American region from 60°S to 50°N latitude and from 20°W to 100°E longitude. Considering that this study site has clear rainfall seasonality, the correlations were performed for the rainy (January–May) and the dry season (June–December) independently. To test for regional representativeness, spatial correlation fields of the regional precipitation and temperature instrumental records were computed, too. 2.5. Precipitation reconstruction Due to the length and quality of their RWI series, the chronology of B. graveolens was selected to derive a local precipitation reconstruction,. The transfer function used for the reconstruction was based on the regional precipitation as the predictand variable, and the B. graveolens tree-ring width data as the predictor or regressor. The split sample method (Gordon, 1982) was applied, with periods of calibration and verification from 1959–1984 and 1985–2011. After a first calculation, verification and calibration periods were exchanged. The final precipitation reconstruction was derived from a linear regression model calculated over the whole period of climate data. To test the reconstruction quality, several statistics were calculated for split calibration and verification periods. The F-test (F) was applied to test the significance of the final regression model for reconstruction. The Sign-test (ST) proofed the similarity between the reconstructed and the instrumental climate data by counting the number of similarities

Fig. 6. Annual precipitation at Laipuna climate station (2007–2012) and regional precipitation (1959–2012) compared with the Ring Width Index curves of B. graveolens and M. tinctoria. In the top left Pearson correlations (r) and Gleichläufigkeit (glk) between regional precipitation and other variables are given. All correlations are significant at p b 0.01.

70

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

Fig. 7. Correlation and response coefficients of B. graveolens and M. tinctoria tree-ring chronologies with monthly temperature (1970–2012), precipitation (1960–2012), and NINO3.4 SST anomalies (1950–2012). The darker bars indicate a coefficient significant at p b 0.1 (*), p b 0.05 (**) and p b 0.01 (***). The lines represent the 95%-confidence interval.

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

and dissimilarities between this two series. The Product Mean Test (PMT) takes into account both the sign and magnitude of the departure from the calibration average, if the departure sign is correctly estimated, the product is positive (Fritts, 1976). The Durbin–Watson (DW) statistic was used to test the autocorrelation in the model residuals (Durbin and Watson, 1950). Finally, the Reduction of Error (RE) and Coefficient of Efficiency (CE) were used to measures the model skill between the reconstructed and observed precipitation, in both cases any positive value was considered as successful verification (Cook et al., 1990, 1994). Furthermore, to proof the reliability of the reconstructed precipitation, a spatial correlation for the rainy and dry season was computed using the Netherlands Meteorological Institute (KNMI) Climate Explorer (http://climexp.knmi.nl), with the Climatic Research Unit (CRU) TS (time-series) 3.22 dataset (Mitchell and Jones, 2005) with a grid 0.5° × 0.5° covering only the land area during the period 1901–2011. In addition, the NCEP/NCAR reanalysis data set for the surface level (Kalnay et al., 1996) was used during the period 1948–2011 to cover the tropical Pacific Ocean area. 3. Results 3.1. Tree-ring chronologies According to wood anatomical analysis, the species C. scabrida, C. loxensis, C. mollis, T. chrysantha and T. valverdae were discarded because the indistinct rings were confused by the occurrence of frequent parenchyma bands. C. trichistandra and I. calodendrum exhibited distinct rings but demonstrated a poor and insignificant crossdating between trees. Viguiera sp. being one of the species with clearer and distinct rings was discarded because the brittle wood and hardness do not permitted collection of complete increment cores with the borer, and also the absence of large and old trees in this forest was not promising to construct long ring-width chronologies. Distinct growth rings were detected in B. graveolens and M. tinctoria cross-sections (Fig. 3). B. graveolens has a diffuse porous distribution of solitary vessels and groups of 2–3 clustered vessels in radial to diagonal arrangement. The tree-ring boundaries in B. graveolens are marked by 1–2 very well defined rows of radially flattened latewood cells. Vessels in M. tinctoria are diffuse porous and occur solitary or in groups of 2–3 clustered vessels in radial arrangement. The tree ring boundaries in M. tinctoria are determined by a transition of radially flattened latewood fibers, often accompanied by marginal parenchyma. Hence, these two species were selected for construction of tree-ring chronologies. Chronology statistics are summarized in Table 3. Most of the sampled trees were rather young; the mean length is 92 years for B. graveolens and 36 years for M. tinctoria, respectively. A chronology of 203 years (1809–2011) created from 50 radii of 18 individual trees was established for B. graveolens, which is the longest tree-ring chronology from tropical dry forests of South America developed so far. In the case of M. tinctoria, a chronology of 87 years (1926–2012) was established from 47 radii of 12 individual trees (Fig. 4). B. graveolens has a mean ring width of 1.44 mm, and a mean interseries correlation (Rbar) of 0.55. M. tinctoria shows a wider mean ring width of 3.01 mm and higher interseries correlation (0.77). Both species show a high mean sensitivity (B. graveolens = 0.66, M. tinctoria = 0.68), and very low first-order autocorrelation (B. graveolens = 0.27, M. tinctoria = 0.22), indicating that tree-ring widths in both species width are hardly influenced by growth of the previous year, but mostly respond to environmental factors during the current growth season. The two final ring-width chronologies of the two species showed a high similarity in radial growth patterns (r = 0.61, p b 0.01, Gleichlaeufigkeit or sign test; glk = 0.73; Fig. 4). Concerning chronology quality, the time period with EPS values close to or above the recommended threshold of 0.85 (Wigley et al., 1984) in the B. graveolens chronology covers 103 years (1908–2011)

71

with a mean EPS value of 0.83 (Fig. 4). For M. tinctoria, this period is restricted to 48 years (1964–2012) with a higher mean EPS value of 0.93. The B. graveolens chronology shows a significant periodicity of ~2–4 years, corresponding to the ENSO frequency band (Fig. 5). During 1970– 1990, before and after the strong El Niño event of 1983, this frequency band is very prominent in both trees species. Multiple frequencies (Mallat, 2009) were spotted in a decreasing signal with direction from higher to lower periods, that is from older to younger rings. One signal is observed more clearly in B. graveolens in the time ~1905–2000, with a period starting at ~ 16 and finishing at 8 years (Fig. 5a), and also in M. tinctoria during the time ~1955 to ~2005 (Fig. 5b). The second signal is less prominent and occurs in the frequency band of 8 to 4 years in B. graveolens. 3.2. Climate interactions 3.2.1. Climatic influence on ring width A comparison between the RWI chronologies and regional precipitation shows a good positive relationship (Fig. 6) during the last decades (1959–2012), with slightly higher values for M. tinctoria (r = 0.78, p b 0.01, glk = 0.77) than for B. graveolens (r = 0.60, p b 0.01; glk = 0.63). The trees show very high growth in wet years which are evident after strong El Niño events (e.g., 1983, 1998). A significant correlation between RWI and monthly precipitation sums was found for M. tinctoria during and after the end of the rainy season (January–August). B. graveolens shows a significant correlation one month after the beginning of the rainy season (February to June and September). According to the response function analysis, M. tinctoria shows two months with significant values in the rainy season and two in the dry season (February–March, June–July), whereas B. graveolens does not show any significant responses at all (Fig. 7). The monthly temperature series (1970–2012) show few significant correlations with the RWI series. A positive correlation with B. graveolens is present only in October. However, temperatures show a negative correlation with M. tinctoria from February to April and in September (Fig. 7), revealing species-specific differences between B. graveolens located at the lower parts and M. tinctoria growing in higher areas of the Laipuna Nature Reserve. No significant response coefficients occur between RWI of both species and temperature in any month. SST anomalies reveal a significant positive correlation with B. graveolens RWI during the whole rainy season and the first months of the dry season (January–July) (Fig. 7). M. tinctoria RWI shows a significant negative correlation in December, but a negative response in January and from October to December and a positive growth response to the dry season during May to July (p b 0.01). A direct impact of ENSO activity on tree growth can only be found after strong El Niño events (Fig. 8), when precipitation and tree growth are increased. Normal El Niño events do not affect the annual growth in both tree species. At the same time, neither ‘normal’ nor ‘long’ La Niña events have a significant influence on RWI. 3.2.2. Precipitation and sea surface temperatures (SST from El Niño region 3.4) The monthly NINO3.4 SST anomalies show low correlations with the regional precipitation (r = 0.08, p b 0.05; 1960–2013). Only during the beginning of strong El Niño events (e.g., 1982 and 1997) a notable and continuous increase on precipitation is observed (Fig. 9), which is also visible in the Superposed Epoch Analysis (SEA) (Fig. 8). ‘Normal’ intensity El Niño events which are associated with warmer Pacific SSTs do not have a significant influence on local precipitation, but ‘strong’ events (1972, 1982, 1997) show a strong precipitation increase in the following rainy season during January to April. ‘Normal’ La Niña events which are

72

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

Fig. 8. Superposed Epoch Analysis (SEA) of ENSO events from 1900 to 2012 (see the El Niño/La Niña years in Table 2) on the regional temperature (1965–2013), precipitation (1960–2013), and Ring Width Indexes of B. graveolens and M. tinctoria. The diagram presents the reactions of three years prior to and after the El Niño/La Niña events. The dark bars indicate values significant at the p b 0.01 level.

associated with colder SSTs do not lead to any significant responses in climate and tree growth, but during longer events (N 15 consecutive months, years: 1909, 1949, 1955, 1971, 1973, 1985, 2000), a significant impact is evident two years before the event. In most cases, this coincides with an El Niño event predating the respective La Niña. Mean monthly regional temperatures are correlated with SST anomalies (r = 0.31, p = b 0.01; 1970–2013). However, the SEA does not reveal any significant effect on temperatures. Only during longer La Niña events, a slight increase of the temperatures three years before the event is evident (Fig. 8). 3.2.3. Spatial correlations with the global precipitation and temperature The spatial correlation maps reveal strong interactions between precipitation as well as temperatures in the tropical Pacific Ocean and RWI of both studied tree species (Fig. 10). The northern Inter-tropical convergence zone (ITCZ) (Liu, 2002) acts as a boundary line between positive correlations further south and negative correlations further

north. In the rainy season (January–May), strong positive correlations (r N0.6, p b 0.05) are prominent in the zone of the calms within the double ITCZ, between the Ecuadorian coast (5°N–10°S latitude) and the waters east of Indonesia and Papua New Guinea. These correlations have a longitudinal maximum influence zone, reaching 20°S in the central Pacific at 140°W. The correlation decreases towards Indonesia and fades out at 160°E longitude. In contrast, a zone of significant negative correlations (r N −0.6, p b 0.05) is evident over the northern ITCZ around 10°N latitude, including the seas around the Philippines and the whole Pacific Ocean until the Central American coast. During the dry season (June–December), only small areas maintain positive and negative correlations with RWI, but most correlations disappear, indicating little influence of the ocean climate on tree growth in the dry forests of Ecuador during the dry season. Both tree species react very similar to tropical Pacific precipitation. Similar to precipitation, a correlation between temperature and RWI is evident for both species. Over the Pacific Ocean, the extension of the

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

73

Fig. 9. Monthly time series of NINO3.4 SSTs and precipitation, and annual time series of ring with (RWI). The left axis shows the normalized SST anomalies from NOAA 1871–2012, with values exceeding a threshold of ±0.4 °C indicating El Niño events (positive values in red) or La Niña events (negative values in blue). The green curves in the top panel show the Ring Width Index (RWI) of B. graveolens (1809–2011) and M. tinctoria (1926–2012), respectively. Blue bars in the bottom panel show the annual regional precipitation (1960–2012 in blue) for the dry forest area and local precipitation (2007–2012 in dark blue) at Laipuna.

significant correlation fields is smaller in longitude, but wider in latitude compared with the correlation fields for precipitation. However, for B. graveolens the correlation with temperature during the wet season extends over parts of the Amazon Basin (Fig. 10). 3.3. Precipitation reconstruction The calibration models passed all verification tests (Table 4, Fig. 11). The R2Adj accounted for 40%, 34%, and 35% of the climatic variance in the two subperiods, and the full calibration period, respectively. The t-values are significant at p b 0.001 for the coefficient β1, but not significant for the constant or β0. The Durbin‐Watson‐statistic is not significant, revealing that the residuals from the regression are statistically random and the model is appropriate. The verification (Table 5, Fig. 11) shows a Pearson correlation of 0.60 (p b 0.001) and a Signtest at p b 0.01 for the full calibration period. Reduction of error and coefficient of efficiency show positive values, indicating the skill of the reconstruction (Cook et al., 1994; Fritts, 1976). The final wet season (Jan–May) precipitation reconstruction covers the period 1908–2011 (103 years) (Fig. 12). 3.3.1. Spatial verification of the reconstructed precipitation The significant correlations of the reconstructed Jan–May precipitation with two gridded data sets (CRU TS3.22 precipitation period 1901–2011, and NCEP/NCAR reanalyzed precipitation period 1948–2011) confirm the validity of the model (Fig. 13). During the last 110 years, the CRU TS3.22 precipitation correlates only in Ecuador, with an evident difference between the rainy (January to May) and dry season (June to December). The NCEP/NCAR precipitation shows higher correlations in the equatorial Pacific from 160°W to 80°W, with highest correlations along the Ecuadorian and Nord Peru coasts. 4. Discussion and conclusions 4.1. Dendrochronology In this study the first 203-year long chronology for B. graveolens and the first 87-year long chronology for M. tinctoria from a tropical dry forest in South America was developed. The high correlations between the individual TRW series within each species and the correlation of the two species-specific chronologies prove strong synchronicity between inter-annual growth variations. While this has been shown

before for B. graveolens growing in northern Peru (Rodríguez et al., 2005), M. tinctoria is constituted as a new tree species suitable for tropical dendrochronology. The very high mean sensitivity and very low firstorder autocorrelation of the RWI chronologies in both species are typical for highly drought-sensitive trees (Fritts, 1976; Grissino-Mayer, 2008), indicating that growth of both species is strongly controlled by climatic factors. On the other hand, for the species C. scabrida, C. trichistandra, C. loxensis, C. mollis, I. calodendrum, T. chrysantha, T. valverdae and Viguiera sp., it was not feasible to develop chronologies, although their growth behavior is probably also influenced by climatic factors. More detailed wood and tree physiological anatomical studies are necessary to identify their growth rates and their rings borders, as intra-annual stable isotope analysis (Anchukaitis et al., 2008; Evans and Schrag, 2004; Krepkowski et al., 2013). So far, annual rings of these species have only been reported for T. chrysantha (Volland-Voigt et al., 2011). In the chronology of B. graveolens (Fig. 5) a periodicity of ~2–4 years was found which corresponds to the frequency band of ENSO events (D'Arrigo et al., 2005; Jevrejeva et al., 2003; Maruyama et al., 2011; Wilson et al., 2010). During the El Niño 1900 and 1983 a strong and significant range of influence is evident on RWI. Additional periodicity signals occur in two longer frequency band of ~ 8 and ~ 16 years. Although the causes of these longer signals are difficult to determine, they may be related to low-frequency climate modes like the Pacific Decadal Oscillation (PDO) as also found in tree-ring chronologies close to the North Pacific (Gray et al., 2003), or in the influence of solar variations (Rigozo et al., 2002, 2003).

4.2. Climate interactions The influence of the El Niño 3.4 SSTs is evidenced with high precipitation increases in the study area only during climatic strong events (1983, 1998; Fig. 9). A strong correlation of the El Niño 3.4 SSTs with precipitation along the equatorial Pacific coast is well documented (Bendix et al., 2011; Espinoza et al., 2009; Rodbell, 1999; Vuille et al., 2000). (Bazo et al., 2013) found lagged correlations with 1–6 months delay between tropical Pacific SSTs and local rainfall in northwest Peru, which is close to the study site. This time lag is confirmed by SEA (Fig. 8) that indicates a significant impact of El Niño strong events on tree growth in the year following the event, whereas normal El Niño events do not have a strong impact on precipitation in the study area. La Niña long events show an increase in precipitation occurring two years before, which can be interpreted in a way that two years

74

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

Fig. 10. Spatial correlations (p b 0.05) between the ring width index chronologies (RWI) and CFRS monthly mean precipitation and temperature from 60°S to 50°N latitude and from 20°W to 100°E longitude. The analyzed period is 1979–2009.

after higher precipitation following an El Niño, long droughts are frequent in these ecosystems. 4.2.1. Climatic influence on tree ring width Significant correlations between local and regional precipitation and the RWI of both studied species during the last five decades (1959–2012, Fig. 6) were found. M. tinctoria shows higher

correlations (r = 0.78) with annual precipitation than B. graveolens (r = 0.60). Similarly, a study in northern Peru found high correlation (r = 0.83) between a chronology of B. graveolens and local precipitation (Rodríguez et al., 2005). These correlation values are even higher than at the site of this study since the climatic conditions in northern Peru are more extreme, with annual precipitation of less than 50 mm per year. This indicates that the main climate

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

75

Table 4 Calibration statistics for the reconstructed wet season (Jan–May) precipitation. F: F‐value, R2: explained variance, R2Adj: explained variance adjusted for loss of degrees of freedom, DW: Durbin–Watson‐statistic. Period

F

R2

R2Adj

Coefficient (mm) β0

β1

β0

β1

β0

β1

1959–2011 1959–1984 1985–2011

28.98⁎⁎⁎ 17.79⁎⁎⁎ 13,13⁎

0.36 0.43 0.34

0.35 0.40 0.32

170.40 51.97 225.10

853.40 868.27 903.60

157.28 213.70 235.9

158.53 205.88 249.40

1.08 0.24 0.95

5.38⁎⁎⁎ 4.22⁎⁎⁎ 3.62⁎⁎⁎

Standard error (mm)

t-Statistic (H0: β = 0)

DW

1.89 1.97 2.01

Significance levels: ⁎⁎⁎ p b 0.001. ⁎ p b 0.05.

factor limiting radial growth in these forest ecosystems is available soil moisture, which is usually controlled by precipitation (Lugo et al., 1978). Correlation and response functions between the climate data and RWI indicated the strong impact of rainy season precipitation on tree growth (Fig. 7). M. tinctoria evidenced significant response to the wettest months February to March, and also June to July, when the trees may still profit from soil water accumulated during the rainy season. In contrast, B. graveolens evidenced no significant values in this response analysis, denoting other mechanisms to uptake the water stored in the soil during and after the rainy season. This may be related with a deeper root system adapted to site conditions on slopes. In contrast to precipitation, the impact of temperature on radial growth is weaker and less consistent. M. tinctoria RWI shows negative correlations to temperatures of February to April and September, indicating that high temperatures affect growth of M. tinctoria more strongly, probably by increasing potential evapotranspiration. In contrast to correlation analysis, response analysis did not produce significant correlations with temperatures in any species. Since in tropical environments, temperature is rather constant throughout the year, and this is not a limiting factor for tree growth, these results are consistent with expectation and confirm the larger role of moisture availability for growth conditions of tropical trees (Enquist and Leffler, 2001; Reich, 1995; Way and Oren, 2010; Wils et al., 2011). With respect to monthly SST anomalies, B. graveolens RWI shows a significant and positive correlation during the wet season January to July, whereas M. tinctoria revealed a weak and negative correlation (p b 0.1) only in December. Response analysis for B. graveolens shows a positive and significant effect in the rainiest month (March). M. tinctoria shows negative response from October to January, and a positive response from May to July. This response effect could be linked to species-specific factors as rooting depth or foliage cover and needs further study by tree physiological measurements.

In summary, these results indicate that tree growth of several species in the dry tropical forest ecosystem is strongly influenced by precipitation and only slightly and indirectly by temperatures. The limiting impact of precipitation on tree growth has been demonstrated for several other tropical forest ecosystems in seasonally dry climates (Enquist and Leffler, 2001; Krepkowski et al., 2011; Rodríguez et al., 2005; Therrell et al., 2006; Wils et al., 2011). While in some seasonally dry forest ecosystems, a highly individualistic and species-specific reaction to precipitation and soil water use was observed (Enquist and Leffler, 2001), the two studied species in this forest ecosystem responded highly synchronous to moisture availability. 4.2.2. Spatial correlations with the global precipitation and temperature A good relationship between precipitation and radial tree growth in southern Ecuador was demonstrated in an analysis with climate datasets from the CSFR reanalysis (Fig. 10). As (Wise, 2015) state, tree-rings can reflect ocean–atmosphere oscillations and a more varied set of climate controls. In this study higher and positive correlation values were found in windless areas in the central Pacific, especially in the equatorial region within the Doldrums (Gentilli, 2005). In these areas, the higher correlations of precipitation shift parallel with the displacement of the ITCZ and trade wind strength over the year. These spatial correlations make evident a clear contrast between the rainy (January to May) and dry season (June to December). In the rainy season, the positive correlations with precipitation mark a well-defined area of positive correlations with dry forest species' RWIs along the equator near latitudes from the Ecuadorian coasts (−80°E) stretching until Papua New Guinea (160°E). Conversely, in the rainy season the northern part of the ITCZ with the highest rainfall over the Pacific Ocean (at 10°N) specially in the Western Pacific Warm Pool (WPWP) shows inverted, negative correlations with RWIs (Fig. 10, prec. in rainy season). Interestingly, this pattern is similar to teleconnections

Fig. 11. Observed and reconstructed wet season (Jan–May) precipitation on the calibrated full 53-year period (1959–2011) showing the correlation in the embedded scatter graph. The vertical line shows the split of the full period for two calibration and validation subperiods 1959–1984 and 1985–2011. See statistical analysis in Tables 4 and 5.

76

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

Table 5 Verification statistics for the reconstructed wet season (Jan–May) precipitation. R: explained variance, ST: sign‐test, PMT: product‐means‐test, RE: reduction of error, CE: coefficient of efficiency. Period

R (Pearson corr.)

ST (hit/miss)

PMT

RE

CE

1959–2011 1959–1984 1985–2011

0.60⁎⁎⁎ 0.65⁎⁎⁎ 0.59⁎⁎⁎

33/19⁎⁎

1.67⁎ 1.14⁎⁎⁎⁎ 1.30⁎

0.36 0.43 0.34

0.26 0.09

16/9 16/10

Significance levels: ⁎⁎⁎⁎ p b 0.1. ⁎⁎⁎ p b 0.001. ⁎⁎ p b 0.01. ⁎ p b 0.05.

between tree-ring based precipitation reconstructions and global SST in southeastern China (Shi et al., 2015). Likewise, dipole-like spatial patterns were found between stable isotope ratios in precipitation across the USA and the North Pacific (Liu et al., 2014). On the other hand, during the dry season the positive correlations over the central Pacific vanish almost completely, and only negative correlations with a small area around 15°N between the western Pacific from 170°W to the Philippine waters are retained. M. tinctoria shows stronger negative large-scale correlations than B. graveolens. Spatial correlation fields with surface temperatures show similar patterns as for precipitation in the rainy season, but within a wider area in latitude along the eastern coast of the Central Pacific, and a smaller area around 160°W. Identical patterns occur in a spatial correlation between tree-ring cellulose δ18O-based precipitation reconstruction from northwest Thailand and tropical SSTs (Xu et al., 2015), revealing the influence of temperatures over the central Pacific on the tropical lands at both flanks of the Pacific. Additionally, strong teleconnections were also reported between tree-ring records and the SST and PDO index over the North Pacific (Biondi et al., 1999; D'Arrigo et al., 1999; Douglass, 1980). These climate patterns are currently not well understood, and dendroclimatic studies about teleconnections between tree growth and climate in the Central Pacific are still scant. This study demonstrates that the radial growth of tree species growing in dry tropical forest ecosystems close to the Pacific coast are highly linked to the central Pacific precipitation and SST patterns. These results confirm former studies (Enquist and Leffler, 2001) reporting that different tree species growing in the same tropical ecosystem may show individual sensitivity to climate on a local and global scale. 4.3. Reconstructed precipitation Although longer reconstructions of past ENSO activities have been established (Wilson et al., 2010), the present reconstruction of wet-

season precipitation at Laipuna in southern Ecuador constitutes the first 103-year local tree-ring based climate reconstruction in this tropical region. The calibration period of 53 years explains 36% of precipitation variability in the calibration period 1959–2011. Non-linear effects in very wet years, when a surplus of precipitation after soil saturation with water might not lead to additional growth performance (Fig. 11) might limit the explanatory power of a linear regression model, but the rarity of such strong events justifies the application of a linear climate-growth model. In a historical analysis, the well-known 1968 drought in Loja which caused massive migration (Gondard, 1983; Ospina et al., 2011; Ramón, 2012) is reflected in this precipitation reconstruction (Fig. 12). Other drought periods are evident during 1953–1957, 1939–1941, 1922–1925, and 1895–1898. Likewise, the strong El-Niño events of 1998, 1983 and 1972 are recorded in the periods with higher values in this reconstructed precipitation, as well as the very wet periods that occurred in 1958, 1949, 1927, 1917, 1899–1901 (Fig. 12). The spatial correlation analysis verifies that the reconstructed precipitation captures a great part of the Rainfall Index over the land area of Ecuador and the equatorial Pacific in their rainy season (Fig. 13). Spatial correlation fields of the original tree-ring chronologies and the precipitation reconstruction are in agreement (Fig. 10). After this first study proved the climatic sensitivity of selected dry tropical forest tree species, it is crucial to establish a network of tree-ring chronologies in the tropical dry forest ecosystem, including additional tree species to develop more and longer historical records which can help to better understand the climatic impact on carbon sequestration and resilience of these forest ecosystems on economy of local communities. This study demonstrates the applicability of dendrochronological methods on locally growing tropical tree species for climate reconstruction in the dry tropical forests. Future studies shall try to extend the reconstruction further into the past and include additional tree-ring parameters like stable isotopes that have been proven to be useful indicators of hydroclimate variability in tropical ecosystems (e.g., Brienen et al., 2012; Schollaen et al., 2013). Acknowledgments This study was funded by the German Research Foundation (DFG) in the framework of the research unit FOR816 (BR 1895/19; BR 1895/ 23). We acknowledge support by the German Academic Exchange Service (DAAD) (No. A/08/96621) and Secretaría Nacional de Educación Superior, Ciencia, Tecnología e Innovación (SENESCYT) (No. SENESCYT-DMPF-2014-0779-CO) for the grant of DPC. Naturaleza y Cultura Internacional (NCI, Loja, Ecuador) for their help in accessing the Laipuna Nature Reserve. Besides, we thank the Dendro-Group at the University of Erlangen-Nuremberg for the technical assistance.

Fig. 12. Reconstructed precipitation for the period 1908–2011 in southern Ecuador at Laipuna Nature Reserve. The 10 year smoothing spline highlights decadal precipitation variations.

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

77

Fig. 13. Spatial correlations of the reconstructed wet season (January–May) precipitation. The left graph indicates the correlations with the NCEP/NCAR precipitation during the period 1948–2011, covering land and sea areas from 60°S to 50°N latitude and from 20°W to 100°E longitude. The smaller graph at the right indicates the correlations with the CRU TS3.22 observed precipitation during the longer period 1901–2011 in the land area from +/−15° latitude and from 50°W to 95°W longitude.

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.gloplacha.2015.08.003. These data include the Google map of the most important areas described in this article. References Aguirre, Z., Kvist, P.L., Sánchez, T.O., 2006. Bosques secos en Ecuador y su diversidad. In: Moraes R., M., Øllgaard, B., Kvist, P.L., Borchsenius, F., Balslev, H. (Eds.), Botánica Económica de los Andes Centrales, La Paz, pp. 162–187. Anchukaitis, K.J., Evans, M.N., Wheelwright, N.T., Schrag, D.P., 2008. Stable isotope chronology and climate signal calibration in neotropical montane cloud forest trees. J. Geophys. Res. 113 (G3). http://dx.doi.org/10.1029/2007JG000613. Bazo, J., Lorenzo, María de las Nieves, Porfirio da Rocha, Rosmeri, 2013. Relationship between monthly rainfall in NW Peru and tropical sea surface temperature. Adv. Meteorol. 2013 (2), 1–9. http://dx.doi.org/10.1155/2013/152875. Bendix, J., Trache, K., Palacios, E., Rollenbeck, R., Goettlicher, D., Nauss, T., Bendix, A., 2011. El Niño meets La Niña — anomalous rainfall patterns in the “traditional” El Niño region of southern Ecuador. ERDKUNDE 65 (2), 151–167. http://dx.doi.org/10.3112/ erdkunde.2011.02.04. Biondi, F., Cayan, D.R., Berger, W.H., 1999. Decadal scale changes in Southern California tree-ring records. In: American Meteorological Society (Ed.), Proceedings of the 10th Symposium on Global Change Studies. 79th Annual Meeting of the American Meteorological Society, pp. 303–306 (10–15 January, Dallas, TX). Brienen, Roel J.W., Zuidema, P.A., 2005. Relating tree growth to rainfall in Bolivian rain forests: a test for six species using tree ring analysis. Oecologia 146 (1), 1–12. http://dx.doi.org/10.1007/s00442-005-0160-y. Brienen, Roel J.W., Helle, G., Pons, T.L., Guyot, J.-L., Gloor, M., 2012. Oxygen isotopes in tree rings are a good proxy for Amazon precipitation and El Nino-Southern Oscillation variability. Proc. Natl. Acad. Sci. U. S. A. 109 (42), 16957–16962. http://dx.doi.org/ 10.1073/pnas.1205977109. Briffa, K.R., Jones, P.D., 1990. Basic chronology statistics and assessment. In: Cook, E.R., Kairiukstis, L.A. (Eds.), Methods of Dendrochronology: Applications in the Environmental Sciences. Kluwer Academic Publishers, pp. 137–152. Brönnimann, S., Xoplaki, E., Casty, C., Pauling, A., Luterbacher, J., 2007. ENSO influence on Europe during the last centuries. Clim. Dyn. 28 (2–3), 181–197. http://dx.doi.org/10. 1007/s00382-006-0175-z. Bunn, A.G., 2008. A dendrochronology program library in R (dplR). Dendrochronologia 26 (2), 115–124. http://dx.doi.org/10.1016/j.dendro.2008.01.002. Bunn, A.G., 2010. Statistical and visual crossdating in R using the dplR library. Dendrochronologia 28 (4), 251–258. http://dx.doi.org/10.1016/j.dendro.2009.12.001. Callado, C., da Silva Neto, Sebastião, Scarano, F., Costa, C., 2001. Periodicity of growth rings in some flood-prone trees of the Atlantic Rain Forest in Rio de Janeiro, Brazil. Trees 15 (8), 492–497. http://dx.doi.org/10.1007/s00468-001-0128-4. Campelo, F., García-González, I., Nabais, C., 2012. detrendeR — a graphical user interface to process and visualize tree-ring data using R. Dendrochronologia 30 (1), 57–60. http:// dx.doi.org/10.1016/j.dendro.2011.01.010. Carlquist, S., 1988. Comparative Wood Anatomy: Systematic, Ecological, and Evolutionary Aspects of Dicotyledon Wood. Springer (ISBN: 3540188274, 9783540188278). Chazdon, R.L., Letcher, S.G., van Breugel, M., Martinez-Ramos, M., Bongers, F., Finegan, B., 2007. Rates of change in tree communities of secondary Neotropical forests following

major disturbances. Philos. Trans. R. Soc. B 362 (1478), 273–289. http://dx.doi.org/10. 1098/rstb.2006.1990. Cook, E.R., Briffa, K.R., Shiyatov, S., Mazepa, A., Jones, P.D., 1990. Data analysis. In: Cook, E.R., Kairiukstis, L.A. (Eds.), Methods of Dendrochronology: Applications in the Environmental Sciences. Kluwer Academic Publishers, pp. 97–162. Cook, E.R., Briffa, K.R., Jones, P.D., 1994. Spatial regression methods in dendroclimatology: a review and comparison of two techniques. Int. J. Climatol. 14 (4), 379–402. http:// dx.doi.org/10.1002/joc.3370140404. Coster, C., 1927. Zur Anatomie und Physiologie der Zuwachszonen und Jahresbildung in den Tropen. Jardin Botanique de Buitenzorg 37 pp. 49–160. D'Arrigo, R.D., Wiles, G., Jacoby, G., Villalba, R., 1999. North Pacific sea surface temperatures: past variations inferred from tree rings. Geophys. Res. Lett. 26 (17), 2757–2760. http://dx.doi.org/10.1029/1999GL900504. D'Arrigo, R., Cook, E.R., Wilson, R., Allan, R., Michael, E., 2005. On the variability of ENSO over the past six centuries. Geophys. Res. Lett. 32 (3). http://dx.doi.org/10.1029/ 2004GL022055. Douglass, A.E., 1919. Climatic Cycles and Tree-growth: A Study of the Annual Rings of Trees in Relation to Climate and Solar Activity. Carnegie Institution of Washington. Douglass, A.V., 1980. Geophysical estimates of sea-surface temperatures off western North America since 1671. CalCOFI Report XXI, pp. 102–112. Durbin, J., Watson, G.S., 1950. Testing for serial correlation in least squares regression: I. Biometrika 37 (3/4), 409. http://dx.doi.org/10.2307/2332391. Enquist, B.J., Leffler, A.J., 2001. Long-term tree ring chronologies from sympatric tropical dry-forest trees: individualistic responses to climatic variation. J. Trop. Ecol. 17 (1), 41–60. http://dx.doi.org/10.1017/S0266467401001031. Espinoza, J.C., Ronchail, J., Guyot, J.L., Cochonneau, G., Naziano, F., Lavado, W., De Oliveira, E., Pombosa, R., Vauchel, P., 2009. Spatio-temporal rainfall variability in the Amazon basin countries (Brazil, Peru, Bolivia, Colombia, and Ecuador). Int. J. Climatol. 29 (11), 1574–1594. http://dx.doi.org/10.1002/joc.1791. Evans, M.N., Schrag, D.P., 2004. A stable isotope-based approach to tropical dendroclimatology. Geochim. Cosmochim. Acta 68 (16), 3295–3305. http://dx. doi.org/10.1016/j.gca.2004.01.006. Fichtler, E., Trouet, V., Beeckman, H., Coppin, P., Worbes, M., 2004. Climatic signals in tree rings of Burkea africana and Pterocarpus angolensis from semiarid forests in Namibia. Trees 18 (4). http://dx.doi.org/10.1007/s00468-004-0324-0. Fritts, H.C., 1976. Tree Rings and Climate. Academic Press, London, New York 9780122684500, p. 567 (xii). Gärtner, H., Schweingruber, F.H., 2013. Microscopic Preparation Techniques for Plant Stem Analysis. Kessel Publishing House, Birmendorf, Switzerland (78 pp. ISBN: 3783-941300-76-7). Gentilli, J., 2005. Doldrums. In: Oliver, J.E. (Ed.), Encyclopedia of World Climatology. Springer, Dordrecht, Netherlands, New York, p. 338. Gondard, P., 1983. Ritmos pluviométricos y contrastes climáticos en la provincia de Loja. Cultura: Revista del Banco Central del Ecuador 5 (15) pp. 38–57. Gordon, G.A., 1982. Verification of dendroclimatic reconstructions. In: Hughes, M.K., Kelly, P.M., Pilcher, J.R., Lamarche JR., V.C. (Eds.), Climate From Tree Rings. Cambridge University Press, Cambridge [Cambridgeshire], New York, pp. 58–62. Gray, S.T., Betancourt, J.L., Fastie, C.L., 2003. Patterns and sources of multidecadal oscillations in drought-sensitive tree-ring records from the central and southern Rocky Mountains. Geophys. Res. Lett. 30 (6). http://dx.doi.org/10.1029/2002GL016154. Grissino-Mayer, H., 2008. User Guide to COFECHA Output Files. Harley, G.L., 2013. Tropical tree rings and environmental change. Southeast. Geogr. 53 (1), 1–3. http://dx.doi.org/10.1353/sgo.2013.0000. Hoell, A., Funk, C., Barlow, M., 2013. The regional forcing of Northern hemisphere drought during recent warm tropical west Pacific Ocean La Niña events. Clim. Dyn. 1–23. http://dx.doi.org/10.1007/s00382-013-1799-4.

78

D. Pucha-Cofrep et al. / Global and Planetary Change 133 (2015) 65–78

Jevrejeva, S., Moore, J.C., Grinsted, A., 2003. Influence of the Arctic Oscillation and El NiñoSouthern Oscillation (ENSO) on ice conditions in the Baltic Sea: the wavelet approach. J. Geophys. Res. 108 (D21). http://dx.doi.org/10.1029/2003JD003417. Jones, P.D., Hulme, M., 1996. Calculating regional climatic time series for temperature and precipitation: methods and illustrations. Int. J. Climatol. 16 (4), 361–377. Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Leetmaa, A., Reynolds, R., Chelliah, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K.C., Ropelewski, C., Wang, J., Jenne, R., Joseph, D., 1996. The NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteorol. Soc. 77 (3), 437–471. http://dx.doi.org/10.1175/1520-0477(1996)077b0437:TNYRPN2.0.CO;2. Krepkowski, J., Bräuning, A., Gebrekirstos, A., Strobl, S., 2011. Cambial growth dynamics and climatic control of different tree life forms in tropical mountain forest in Ethiopia. Trees 25 (1), 59–70. http://dx.doi.org/10.1007/s00468-010-0460-7. Krepkowski, J., Gebrekirstos, A., Shibistova, O., Bräuning, A., 2013. Stable carbon isotope labeling reveals different carry-over effects between functional types of tropical trees in an Ethiopian mountain forest. New Phytol. 199 (2), 431–440. http://dx.doi. org/10.1111/nph.12266. Liu, W.T., 2002. Double intertropical convergence zones — a new look using scatterometer. Geophys. Res. Lett. 29 (22). http://dx.doi.org/10.1029/2002GL015431. Liu, Z., Yoshmura, K., Bowen, G.J., Welker, J.M., 2014. Pacific–North American teleconnection controls on precipitation isotopes (δ 18 O) across the contiguous United States and adjacent regions: a GCM-based analysis. J. Clim. 27 (3), 1046–1061. http://dx.doi.org/10.1175/JCLI-D-13-00334.1. Lugo, A.E., Gonzalez-Liboy, J.A., Cintron, B., Dugger, K., 1978. Structure, productivity, and transpiration of a subtropical dry forest in Puerto Rico. Biotropica 10 (4), 278. http://dx.doi.org/10.2307/2387680. MAE, 2013. Mapa de Ecosistemas del Ecuador Continental. 1:1.2500.000. Ministerio del Ambiente del Ecuador (MAE). Mallat, S., 2009. Time meets frequency. In: Mallat, S.G. (Ed.), A Wavelet Tour of Signal Processing. Elsevier, pp. 89–153. Maruyama, F., Kai, K., Morimoto, H., 2011. Wavelet-based multifractal analysis of the El Niño/Southern Oscillation, the Indian Ocean Dipole and the North Atlantic Oscillation. SOLA 7, 65–68. http://dx.doi.org/10.2151/sola.2011-017. 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 (6), 693–712. http://dx.doi.org/10.1002/joc.1181. Mooney, H.A., Bullock, S.H., Medina, E., 1995. Introduction. In: Bullock, S.H., Mooney, H.A., Medina, E. (Eds.), Seasonally Dry Tropical Forests. Cambridge University Press, Cambridge, New York, NY, USA, pp. 1–8. Ospina, P., Andrade, D., Castro, S., Chiriboga, M., Hollenstein, P., Larrea, C., Larrea, A., Poma Loja, J., Portillo, B., Rodríguez, L., 2011. “Dinámicas económicas territoriales en Loja, Ecuador: ¿crecimiento sustentable o pasajero?”. Programa Dinámicas Territoriales Rurales 76. Rimisp, Santiago-Chile (www.rimisp.org/dtr). R Development Core Team, 2009. R: R Foundation for Statistical Computing, Vienna, Austria. Ramón, M., 2012. Dinámica poblacional comparativa de la provincia de Loja. Informe de Coyuntura Económica 9. Universidad Técnica Particular de Loja. Instituto de Investigaciones Económicas. Reich, P.B., 1995. Phenology of tropical forests: patterns, causes, and consequences. Can. J. Bot. 73 (2), 164–174. http://dx.doi.org/10.1139/b95-020. Rigozo, N.R., Nordemann, D., Echer, E., Zanandrea, A., Gonzalez, W.D., 2002. Solar variability effects studied by tree-ring data wavelet analysis. Adv. Space Res. 29 (12), 1985–1988. http://dx.doi.org/10.1016/S0273-1177(02)00245-4. Rigozo, N.R., Vieira, L.E.A., Echer, E., Nordemann, D.J.R., 2003. Wavelet analysis of solarENSO imprints in tree ring data from Southern Brazil in the last century. Clim. Chang. 60 (3), 329–340. http://dx.doi.org/10.1023/A:1026048124353. Rinn, F., 2012. TSAPWin Scientific: Time Series Analysis and Presentation for Dendrochronology and Related Applications. RINNTECH, D-69124 Heidelberg, Germany. Rodbell, D.T., 1999. An 15,000-year record of El Niño-driven alluviation in Southwestern Ecuador. Science 283 (5401), 516–520. http://dx.doi.org/10.1126/ science.283.5401.516. Rodríguez, R., Mabres, A., Luckman, B., Evans, M., Masiokas, M., Ektvedt, T.M., 2005. “El Niño” events recorded in dry-forest species of the lowlands of northwest Peru. Dendrochronologia 22 (3), 181–186. http://dx.doi.org/10.1016/j.dendro.2005.05.002. Rozendaal, Danaë M.A., Zuidema, P.A., 2011. Dendroecology in the tropics: a review. Trees 25 (1), 3–16. http://dx.doi.org/10.1007/s00468-010-0480-3. Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., Tripp, P., Kistler, R., Woollen, J., Behringer, D., 2010. The NCEP climate forecast system reanalysis. Bull. Am. Meteorol. Soc. 91 (8), 1015–1057. http://dx.doi.org/10.1175/2010BAMS3001.1. Samson, J.C., Yeung, K.L., 1986. Some generalizations on the method of superposed epoch analysis. Planet. Space Sci. 34 (11), 1133–1142. http://dx.doi.org/10.1016/00320633(86)90025-5. Sanchez-Azofeifa, G.A., Quesada, M., Rodriguez, J.P., Nassar, J.M., Stoner, K.E., Castillo, A., Garvin, T., Zent, E.L., Calvo-Alvarado, J.C., Kalacska, M.E., Fajardo, L., Gamon, J.A.,

Cuevas-Reyes, P., 2005. Research priorities for Neotropical dry forests 1. Biotropica 37 (4), 477–485. http://dx.doi.org/10.1111/j.1744-7429.2005.00066.x. Schollaen, K., Heinrich, I., Neuwirth, B., Krusic, P.J., D'Arrigo, R.D., Karyanto, O., Helle, G., 2013. Multiple tree-ring chronologies (ring width, δ13C and δ18O) reveal dry and rainy season signals of rainfall in Indonesia. Quat. Sci. Rev. 73, 170–181. http://dx. doi.org/10.1016/j.quascirev.2013.05.018. Schöngart, J., Maria Teresa F, Piedade, Wittmann, F., Junk, W.J., Worbes, M., 2005. Wood growth patterns of Macrolobium acaciifolium (Benth.) Benth. (Fabaceae) in Amazonian black-water and white-water floodplain forests. Oecologia 145 (3), 454–461. http://dx.doi.org/10.1007/s00442-005-0147-8. Shi, J., Lu, H., Li, J., Shi, S., Wu, S., Hou, X., Li, L., 2015. Tree-ring based February–April precipitation reconstruction for the lower reaches of the Yangtze River, southeastern China. Glob. Planet. Chang. 131, 82–88. http://dx.doi.org/10.1016/j.gloplacha.2015. 05.006. Sierra, R., 1999. Propuesta preliminar de un sistema de clasificación de vegetación para el Ecuador continental. EcoCiencia, INEFAN/GEF-BIRF and, Quito (174 pp.). Stahle, D.W., Cleaveland, M.K., Hehr, J.G., 1988. North Carolina climate changes reconstructed from tree rings: A.D. 372 to 1985. Science (New York, N.Y.) 240 (4858), 1517–1519. http://dx.doi.org/10.1126/science.240.4858.1517. Stokes, M.A., Smiley, T.L., 1968. An Introduction to Tree-Ring Dating. University of Chicago Press, Chicago, IL. Therrell, M.D., Stahle, D.W., Ries, L.P., Shugart, H.H., 2006. Tree-ring reconstructed rainfall variability in Zimbabwe. Clim. Dyn. 26 (7–8), 677–685. http://dx.doi.org/10.1007/ s00382-005-0108-2. Torrence, C., Compo, G.P., 1998. A practical guide to wavelet analysis. Bull. Am. Meteorol. Soc. 79 (1), 61–78. http://dx.doi.org/10.1175/1520-0477(1998)079b0061: APGTWAN2.0.CO;2. Trenberth, K.E., 1997. The definition of el Niño. Bull. Am. Meteorol. Soc. 78 (12), 2771–2777. http://dx.doi.org/10.1175/1520-0477(1997)078b2771:TDOENON2. 0.CO;2. Volland-Voigt, F., Bräuning, A., Ganzhi, O., Peters, T., Maza, H., 2011. Radial stem variations of Tabebuia chrysantha (Bignoniaceae) in different tropical forest ecosystems of southern Ecuador. Trees 25 (1), 39–48. http://dx.doi.org/10.1007/s00468-0100461-6. Vuille, M., Bradley, R.S., Keimig, F., 2000. Climate variability in the Andes of Ecuador and its relation to tropical Pacific and Atlantic sea surface temperature anomalies. J. Clim. 13 (14), 2520–2535. http://dx.doi.org/10.1175/1520-0442(2000)013b2520: CVITAON2.0.CO;2. Way, D.A., Oren, R., 2010. Differential responses to changes in growth temperature between trees from different functional groups and biomes: a review and synthesis of data. Tree Physiol. 30 (6), 669–688. http://dx.doi.org/10.1093/treephys/tpq015. Wigley, T.M.L., Briffa, K.R., Jones, P.D., 1984. On the average value of correlated time series, with applications in dendroclimatology and hydrometeorology. J. Clim. Appl. Meteorol. 23 (2), 201–213. http://dx.doi.org/10.1175/1520-0450(1984)023b0201: OTAVOCN2.0.CO;2. Wils, T.H.G., Sass-Klaassen, U.G.W., Eshetu, Z., Bräuning, A., Gebrekirstos, A., Couralet, C., Robertson, I., Touchan, R., Koprowski, M., Conway, D., Briffa, K.R., Beeckman, H., 2011. Dendrochronology in the dry tropics: the Ethiopian case. Trees 25 (3), 345–354. http://dx.doi.org/10.1007/s00468-010-0521-y. Wilson, R., Cook, E., D'Arrigo, R.D., Riedwyl, N., Evans, M.N., Tudhope, A., Allan, R., 2010. Reconstructing ENSO: the influence of method, proxy data, climate forcing and teleconnections. J. Quat. Sci. 25 (1), 62–78. http://dx.doi.org/10.1002/jqs.1297. Wise, E.K., 2015. Tropical Pacific and Northern Hemisphere influences on the coherence of Pacific Decadal Oscillation reconstructions. Int. J. Climatol. 35 (1), 154–160. http://dx. doi.org/10.1002/joc.3966. Worbes, M., 1995. How to measure growth dynamics in tropical trees. A review. IAWA J. 16 (4), 337–351. http://dx.doi.org/10.1163/22941932-90001424. Worbes, M., 1999. Annual growth rings, rainfall-dependent growth and long-term growth patterns of tropical trees from the Caparo Forest Reserve in Venezuela. J. Ecol. 87 (3), 391–403. http://dx.doi.org/10.1046/j.1365-2745.1999.00361.x. Xu, C., Pumijumnong, N., Nakatsuka, T., Sano, M., Li, Z., 2015. A tree-ring cellulose δ18O-based July–October precipitation reconstruction since AD 1828, northwest Thailand. J. Hydrol. http://dx.doi.org/10.1016/j.jhydrol.2015.02.037. Yu, J., Kim, S.T., 2013. Identifying the types of major El Niño events since 1870. Int. J. Climatol. http://dx.doi.org/10.1002/joc.3575. Zang, C., Biondi, F., 2012. Dendroclimatic calibration in R: the bootRes package for response and correlation function analysis. Dendrochronologia 31, 68–74. http://dx. doi.org/10.1016/j.dendro.2012.08.001. Zhou, X., Fu, Y., Zhou, L., Li, B., Luo, Y., 2013. An imperative need for global change research in tropical forests. Tree Physiol. 33 (9), 903–912. http://dx.doi.org/10.1093/treephys/ tpt064.