Trees

https://doi.org/10.1007/s00468-021-02158-z

ORIGINAL ARTICLE

Oxygen isotopes in tree rings of Cedrela odorata L. as an indicator

of hydroclimate variations in a seasonally dry tropical forest
in northeastern Brazil
Mariana Alves Pagotto1 · Itallo Romany Nunes Menezes1,2 · Clayane Matos Costa1 · Claudio Sergio Lisi1,2 ·
Achim Bräuning3

Received: 24 November 2020 / Accepted: 4 June 2021

© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2021

Abstract
Key message We present the first δ18O chronology of a tropical dry forest tree species in Brazil. Our results showed
a strong relationship of δ18O with inter-annual variation of moisture conditions during the growing season.
Abstract In seasonally dry tropical environments, natural archives of isotopes present in cellulose of tree rings can be used to
study changes in the hydrological cycle. We investigated the stable oxygen isotope ratios in tree-ring cellulose (δ18OTRC​) of
Cedrela odorata L. from Caatinga forest, northeastern Brazil, and tested if δ18OTRC​ is correlated to climate parameters. We
extracted the α-cellulose from the most recent 50 tree rings (1968–2017). The inter-annual variation of δ18OTRC​ was strongly
synchronized (r = 0.65) among all five studied individuals and closely related to changes in humidity during the growing
season. The δ18OTRC​ was significantly correlated (p < 0.01) with precipitation (r = − 0.59), relative humidity (r = − 0.37),
air temperature (r = 0.55), soil water deficit (r = 0.62) and PDSI drought index (r = − 0.62), with most significant correla-
tion during wet season and in the first month of the dry season. The relationship with local climate variables reflects the
effect of evaporative enrichment of leaf water on δ18OTRC​, whereby dry soil conditions, the low relative humidity and high
air temperature reduce stomatal conductance and lead to increased evaporative enrichment of δ18OTRC​. Spatial correlations
revealed a strong impact of PDSI in δ18OTRC​. Our results also show an increase in δ18OTRC​ values from the 1990s onwards
due to extreme and recurring droughts in northeastern Brazil. We conclude that δ18OTRC​ variations are related to the climatic
conditions, especially during the wet period due to the greater water availability for plant physiological processes.

Keywords Tree-ring isotope · Tropical dendroecology · δ18O chronology · Drought · Caatinga vegetation · Semi-arid
climates

Introduction

In most terrestrial biomes, climate seasonality strongly

Communicated by S. Leavitt . influences the physiological processes in plants, controlling
growth and thereby affecting the global carbon and water
* Mariana Alves Pagotto
cycles (Frank et al. 2015; Wagner et al. 2016). In temperate
maripagotto07@gmail.com
environments, tree growth mainly occurs during the warm
1
Laboratory of Plant Anatomy and Dendrochronology, season, while in dry or seasonally dry regions, such as sea-
Federal University of Sergipe, Av. Marechal Rondon s/n, sonally dry tropical forests (SDTF), tree growth is mainly
Rosa Elze, Tr. José Alves dos Santos, n. 75, casa 04, Atalaia, restricted to the rainy season (Castagneri et al. 2018).
Aracaju,, São Cristóvão, Sergipe 49100‑000, Brazil
2
The largest and most diversified SDTF in South America
Department of Biology, Federal University of Sergipe, is known as the Caatinga, located in northeastern Brazil,
Av. Marechal Rondon s/n, Rosa Elze, São Cristóvão,
Sergipe 49100‑000, Brazil with an area of approximately 900,000 k­ m2 (Pennington
3 et al. 2009; Silva et al. 2017). In this environment, the rainy
Institute of Geography, Friedrich-Alexander University
Erlangen-Nuremberg, Wetterkreuz 15, 91058 Erlangen, season is normally restricted to a period of three to four
Germany months, while the prolonged dry season lasts eight to nine

13
Vol.:(0123456789)
 Trees

months, receiving less than 60 mm ­month−1 of rainfall theoretical evidence supporting seasonal cycles of δ18OTRC​
(Sampaio 1995; Corrêa et al. 2019). This can induce a period in tropical tree species. Subsequently, Brienen et al. (2013)
of cambium dormancy in woody plants, which may trigger analyzed δ18OTRC​ of Mimosa acantholoba (Willd.) Poir. in a
the formation of anatomically distinct tree-ring boundaries dry tropical forest of Mexico and identified a negative corre-
in the xylem (Worbes 1995). lation with the amount and intensity of rainfall. The authors
Recent dendrochronological studies found that sev- also showed that large-scale events, such as cyclones and
eral species of Caatinga trees form annual tree rings as a variations in sea surface temperatures, affect the δ18OTRC​
response to hydroclimate seasonality, and ring-width vari- signal. More recently, Olson et al. (2020) presented oxy-
ations are strongly correlated with annual precipitation gen and carbon isotope data from Prosopis sp. trees from
amounts (Pagotto et al. 2015, 2017; Nogueira et al. 2018, the Atacama in South America and showed evidence for
2019; Aragão et al. 2019; Aragão and Lisi 2019; Lisi et al. decreasing water availability and increasing climate vari-
2020). These studies demonstrated that the tree rings of ability in the Atacama from the early- to late- Holocene.
Caatinga species are valuable archives of hydroclimate To date, few studies on δ18OTRC​ of South American SDTF
variability and store information about past environmental species have been conducted, despite their great potential to
conditions. improve the understanding of tree functioning and climate
Environmental variations, however, do not only influence reconstructions (van der Sleen et al. 2017).
ring widths, but also affect the chemical composition of In semi-arid regions as the Caatinga, global and regional
wood (van der Sleen et al. 2017). The stable oxygen isotope climate change scenarios suggest that droughts in the sec-
ratio in tree-ring cellulose (δ18OTRC​) has been reported as a ond half of the twenty-first century will negatively affect
tool that can be used as a paleoclimate proxy (McCarroll and the region through an increase in aridity (Santos et al. 2014;
Loader 2004, 2006; Sternberg 2009) and also for the analysis Marengo et al. 2016; Brito et al. 2017). Moreover, global
of hydroclimate seasonality in tropical forests (Cullen and temperatures and VPD have increased over recent decades
Grierson 2007; Brienen et al. 2013; Schollaen et al. 2013; and are expected to continue to rise in the future (Gros-
Ohashi et al. 2015; Cintra et al. 2019; Islam et al. 2021). siordi et al. 2020). The Caatinga vegetation is adapted and
In most tropical regions, the δ18O composition of pre- tolerant to drought and high temperatures, but the effects
cipitation correlates negatively with the rainfall amount, a caused by changes in large-scale atmospheric circulation,
phenomenon known as “amount effect” (Dansgaard 1964; rising temperature and water deficit over extended periods
Rozanski et al. 1993). Therefore, if tropical trees incorporate can substantially increase the evaporative demand of the
the δ18O signal of rainwater in cellulose in a systematic way, atmosphere, aggravating water stress in plants (Souza et al.
it is possible to correlate the δ18OTRC​ with past changes in 2017; Grossiordi et al. 2020). Within this context, stud-
precipitation amount (Evans and Schrag 2004; Schollaen ies on how increasing climate seasonality is affecting tree
et al. 2013). In addition, the δ18O signals in tree rings may physiology, particularly on native tree species, are crucial
be related to evaporative enrichment of leaf water, which in to understand ecosystems’ responses and regional impacts
turn is controlled by the vapor pressure deficit (VPD) of the of climate change.
surrounding air and related to temperature, precipitation, and Our study presents the first analysis of δ 18OTRC​ of a
relative humidity (Kahmen et al. 2011). Therefore, δ18OTRC​ tree species occurring in a Brazilian SDTF (Caatinga).
time series can be used to identify past variability of differ- We chose Cedrela odorata L. (Meliaceae), since it shows
ent climatic factors in tropical environments. clearly visible annual tree rings, which are well delimited
A considerable number of publications relating the by bands of marginal parenchyma and vessel arrangement
δ18OTRC​ of tropical tree species to hydroclimatic factors in in semi-porous rings (Tomazello Filho et al. 2000; Lisi
tropical environments were recently summarized in a review et al. 2020). The growth periodicity of C. odorata has been
by van der Sleen et al. (2017). However, most of these inves- studied in different environments (Worbes 1995, 1999;
tigations were performed in tropical rainforests (Anchukaitis Tomazello-Filho et al. 2000; Dünisch and Morais 2002;
et al. 2008; Pons and Helle 2011; Volland et al. 2016), par- Dünisch and Puls 2003; Dünisch et al. 2003; Brienen and
ticularly in South America, where a lot of research activities Zuidema 2005, 2006; Brienen et al. 2010; Costa et al.
focus on the Amazon forest (Ballantyne et al. 2011; Brienen 2013; Lisi et al. 2020). Moreover, several researches
et al. 2012; Baker et al. 2015; Ohashi et al. 2015; Cintra et al. have explored stable isotopes in C. odorata as a proxy of
2019). So far, only few studies have investigated the isotope- tropical climate variability (Tomazello Filho et al. 2000;
climate relationships in neotropical semi-arid environments Fichtler et al. 2010; Ballantyne et al. 2011; Baker et al.
in South and Central America. For example, Evans and 2015, 2016, 2017; Cintra et al. 2019). In the Amazon for-
Schrag (2004) showed a strong δ18O signal related to large- est, studies have reported that isotopic variation in the tree-
scale extreme climate events such as El Niño in Prosopis sp. ring cellulose of C. odorata trees is an excellent recorder
trees in a dry forest of Peru. They presented empirical and of the isotopic composition of precipitation during the wet

13
Trees

season, with weaker influences of temperature and VPD Material and methods
(Brienen et al. 2012), also suggesting large-scale rainfall
influences the δ18OTRC​ (Baker et al. 2015; Volland et al. Study site and climate conditions
2016; Cintra et al. 2019).
Our objective in this study was to examine how Caat- The study area is located in a fragment of forest in Sergipe
inga climatic factors influence the δ18OTRC​ variability in C. state, northeastern region of Brazil (10° 08′01.90’’S and 37°
odorata trees. Given the previous oxygen isotopic studies 03′23.58’’W; Fig. 1A). The shrub-tree vegetation is known
on C. odorata in tropical forests of South America, we as Caatinga and consists mainly of a great diversity of cactus
hypothesized that δ18OTRC​ in C. odorata from Caatinga species and deciduous angiosperms, whose leaves fall in the
reflect variations in source water. We also expect that dry season and start to sprout at the beginning of the rainy
δ18OTRC​ is most strongly correlated with climatic condi- season (Silva et al. 2017).
tions during the wet season, considering the greater water The literature classifies the vegetation as SDTF (Pen-
availability during this period, which corresponds to the nington et al. 2009, 2018; Banda-R 2016) subjected to a
period of tree-ring formation. semi-arid tropical climate (Alvares et al. 2014). We calcu-
lated regional precipitation amounts for the common period

Fig. 1  A Location of the study site (dark star) in Sergipe State (SE), symbolizes the entire rainy season, while dark blue shows the wet-
Northeast of Brazil. The gray area represents the Caatinga for- test months. Precipitation data were obtained from a mean between
est boundary and the numbers show different climatic station that the five local weather stations (Silva et al. 2012). The air tempera-
were used to create the regional precipitation. (i) Aquidabã Semarh/ ture data set are from the NOAA NCEP Reanalysis Dataset (https://​
INPE, 217msl; (ii) Aquidabã Agritempo/TRMM 5237, 13msl; (iii) psl.​noaa.​gov/​cgi-​bin/​data/​times​eries/​times​eries1.​pl). Data cover
Itabi Semarh/CEMADEN, 187msl; (iv) Nossa Senhora de Lourdes the period from 1955–2017. C Climatological water balance using
Semarh, 200msl and (v) Nossa Senhora de Lourdes Agritempo/ 100 mm soil water storage capacity (Thornthwaite and Mather, 1955).
TRMM 5368, 453msl; BA Bahia State, AL Alagoas State, AO Atlantic Soil water deficit and water usage occur between January to April
Ocean. B Climate diagram of the study site. The monthly variations and August to December. Note that as the soil loses water, the deficit
in temperature and precipitation are represented by red and blue lines, increases. The soil water recharge is present from May to July. There
respectively. The yellow area represents the dry season, light blue is no soil water surplus in the study region

13
 Trees

1955–2017 from five meteorological stations (see “Data This demonstrates a negative water balance and shows that
analysis” Section and Fig. 1A). The regional climate series water availability for vegetation is restricted to the rainy sea-
(1955–2018) shows 781 mm of annual rainfall and an annual son (Fig. 1C).
mean air temperature of 25.5 °C. Annual potential evapo-
transpiration amounts fall between 1500 and 2000 mm, in
which the precipitation/evapotranspiration ratio is < 0.65, Tree‑ring data treatment and oxygen isotope
which is characterized as semi-arid (Sampaio 1995, 2010). analysis
The well-defined dry season lasts from September-Feb-
ruary, with a mean precipitation of 32.98 mm ­month−1 We collected 5-mm-diameter cores from 30 C. odorata trees
(1955–2018), while the wet season lasts from March-August with an increment borer. We extracted two cores per tree
(97.16 mm ­month−1), with the months of April-July being at breast height (approximately 1.30 m above the ground).
the wettest (111.97 mm ­month−1; Fig. 1B). Consequently, Each core was mounted on wooden sample holders and pol-
the rainy season has a unimodal distribution, with 50% or ished manually with progressively finer sandpaper (up to
more of the total annual rainfall recorded in four months (De 4000 grit) to make all cellular wood structures clearly vis-
Andrade et al. 2017). ible. Subsequently, the wood surfaces were digitized with a
However, the seasonal precipitation pattern is not con- high-resolution scanner (EPSON® Expression 10,000 XL).
stant, with prolonged droughts in some dry years, and a pro- The digital images with 1200 dpi resolution were transferred
longed rainy season in some rainy years (see Supplementary to the Image Pro Plus® program (version 4.5.0.29) for the
Material S1). For example, in certain years the April-July measurement of tree-ring widths to an accuracy of 0.01 mm
rains may fall well below average, increasing the length of and for cross-correlation. We dated a total of 20 trees and 30
the dry season from six to eight months or even more (see radii, synchronizing the ring widths between trees using the
1955, 1959, 2012–2018 in Supplementary Material S2). On Cofecha software (Holmes 1983) for quality control of this
the other hand, in humid years, the sequence of rainy months procedure and for identification of missing rings.
may last for six or seven months (see 1964, 1984, 1988 and For stable isotopes analysis, the five oldest trees exhib-
1989 in Supplementary Material S2). In a third case typical iting high correlations with the master chronology were
of Caatinga region, the dry years receive heavy downpours selected. The mean age was 61 years (maximum 63 and
at any time during the dry season (see 1963 and 1993 in minimum 56 years), and the outer 50 rings were used for
Supplementary Material S2). The causes of all these vari- the analysis of δ18O.
ations are due to large-scale atmospheric systems and oce- The wood of each tree ring was carefully separated with
anic processes over the tropical Atlantic Ocean, such as the a thin blade under a binocular microscope, and transferred
position of the intertropical convergence zone (ITCZ) and to labeled and sterile Eppendorf tubes prior to the extraction
anomalous north–south gradients in ocean surface tempera- of α-cellulose. Some very narrow rings did not have enough
tures near the Equator (Chiang and Koutavas 2004). These isolated wood for analysis, resulting in some gaps in the
large-scale climate factors can influence the displacement of isotope time series (for example, 1969, 1983, 2003, 2006
humid air masses and determine wetter or drier conditions and 2013). The cellulose was extracted from the whole wood
in the region of northeastern Brazil. following the method of Wieloch et al. (2011): (i) solvent
The Brazilian semi-arid region is characterized by the extraction (5% NaOH) to remove components such as resins;
prevalence of a crystalline geological basement, forming (ii) delignification with acidified sodium chloride ­NaCLO2
shallow litholic neosols with a depth of less than one meter, (7%); (iii) alkaline hydrolysis with sodium hydroxide solu-
usually associated with rocky outcrops and reduced water tion (17% NaOH) to eliminate short-chain cellulose products
storage capacity (EMBRAPA 2018). These shallow soils and other possible mixtures. Next, cellulose underwent ultra-
have low water storage capacity. The typical period for soil sonic homogenization (Laumer et al. 2009) and then lyo-
water recharge occurs exclusively during the rainy season philization. Between 300 and 500 µg of cellulose from each
(Fig. 1C). Therefore, the humid months constitute the only sample were placed in silver capsules for isotopic analysis.
period in which there is sufficient water availability for trees. Isotope analysis was done at the Friedrich-Alexander
The long dry period initiates a high soil water deficit and University Erlangen-Nuremberg (Germany) using an Iso-
creates a pronounced seasonal contrast in soil water condi- tope Ratio Mass Spectrometer (IRMS, Delta V Advantage,
tions. The high-water deficit in the soil and the high arid- Thermo Fisher) coupled to a high temperature (1450 °C)
ity (Jesus et al. 2019) contribute to the intermittent flow pyrolysis reactor (HT Oxygen Analyzer), with analytical
regime of rivers, which may remain dry for many months. precision ± 0.25%. Stable isotope composition is reported
Of the total rainfall occurring in the Caatinga, 88% is turned as δ-values in parts per thousand (denoted as %) against
into evapotranspiration, 9% becomes runoff, and only 3% the international VSMOW standard using the following
becomes groundwater recharge (De Andrade et al. 2017). equation:

13
Trees

� � chronology and PDSI were assessed with maps generated

⎡ 18 O ⎤
⎢ 16 O
sample ⎥ using the KNMI Climate Explorer (http://​clime​xp.​knmi.​nl/​
𝛿 18 O = ⎢ � 18 � − 1⎥ × 1000%
start.​cgi).” All analyses were performed with current-year
⎢ 16 O
⎥
⎣ O standard ⎦ climate data. Tree growth in the study site starts in March
and lasts until the end of July (Pagotto et al. 2015; Nogueira
et al. 2018; Lisi et al. 2020), which represents the wet sea-
Data analysis son (Fig. 1B). The cambial growth dynamics of C. odorata
shows that active and dormant phases of the cambium are
To detect the climatic signal in the δ18OTRC​, we correlated determined by the change from the wet to the dry season
the isotope series with the local climatic data and also with (Dünisch et al. 2002).
sea surface temperatures (SST) around the tropics and in El
Niño 3.4 region to explore large-scale controls. Addition-
ally, we compared our δ18OTRC​ chronology with the self- Results
calibrating Palmer Drought Severity Index (PDSI).
We considered the local climate variables annual pre- During the crossdating process, we identified missing rings
cipitation, air temperature, relative humidity (RH), and in the analyzed tree population. The years 2012 and 2013,
soil water deficit (SWD). To better represent the regional which were especially dry, showed narrow tree rings and
moisture conditions, we calculated the regional rainfall missing rings, respectively, in more than 70% of the studied
(Silva et al. 2012) since the precipitation data from a unique trees. Among the five oldest trees selected for isotopic analy-
meteorological station presented missing record in some sis, four had a missing ring in 2013. Only one tree formed a
months or years. The regional rainfall was obtained from very narrow ring, which did not supply enough material for
five weather stations close to the study region: (i) Aquid- the extraction of cellulose for isotope analysis in this year.
abã Semarh/INPE, 217 m a.s.l. (data from 1955–2009), (ii) The mean δ18OTRC​ value for the 50-year analysis period
Aquidabã Agritempo/TRMM 5237, 13 m a.s.l. (2000–2017), 1968–2017 was 29.71 ± 0.24%, ranging from 27.13% (1979)
(iii) Itabi Semarh/CEMADEN, 187 m a.s.l. (1963–2005), to 31.59% (2012). Inter-annual variations of δ18OTRC​ fol-
(iv) Nossa Senhora de Lourdes Semarh, 200 m a.s.l. lowed a similar trend in all trees, with a mean correlation
(1985–2005), (v) Nossa Senhora de Lourdes Agritempo/ of r = 0.65 (p < 0.01; see Supplementary Material S3) and
TRMM 5368, 453 m a.s.l. (2000–2017). Air temperature and an Expressed Population Signal (EPS) of 0.90, which is
RH indices were obtained from the NOAA NCEP Reanalysis above the recommended threshold of 0.85 for sufficient sig-
Dataset (https://​psl.​noaa.​gov/​cgi-​bin/​data/​times​eries/​times​ nal strength of a chronology (Wigley et al. 1984; Table 1).
eries1.​pl) between − 10 N to − 10.5S and − 35E to − 40 W. However, correlations between individual trees were higher
The monthly and annual values of SWD were produced from in the last 33 years (1985–2017), with a mean correlation
the statistical spreadsheet provided by Rolim et al. (1998), value of r = 0.70 (p < 0.01; see Supplementary Material S3)
based on calculations of the Normal Climatic Water Balance and an EPS of 0.93. These significant correlations demon-
developed by Thornthwaite 1948; Thornthwaite and Mather strate a coherent pattern among the δ18OTRC​ series of indi-
(1955). We used the monthly tropical South Atlantic index vidual trees (Fig. 2), and indicate a strong common forcing
(TSA) defined as anomalies from the average of the SSTs by environmental factors. The mean δ18OTRC​ chronology
between Equator-20S and 10E-30 W (http://​www.​esrl.​noaa.​ shows high inter-annual variability and an increasing trend
gov/​psd/​data/​corre​lation/​tsa.​data). Monthly values of the El since 1990 (Fig. 2).
Niño 3.4 are from the US National Oceanic and Atmospheric The ring width and δ18OTRC​ series of the individual trees
Administration-Physical Sciences Laboratory (http://​www.​ did not show significant correlations. However, the mean
esrl.​noaa.​gov/​psd/​data/​clima​teind​ices/​list/​index.​html). The ring width and δ18OTRC​ chronologies comparison showed
PDSI data were obtained from the Royal Netherlands Mete-
orological Institute (KNMI), Climate Explorer (http://​clime​
xp.​knmi.​nl/​start.​cgi). Table 1  Summary of tree ring width and δ18OTRC​ data of Cedrela
We computed Pearson’s correlation coefficients between odorata, Caatinga, Brazil
the C. odorata δ18OTRC​ chronology and climatic variables. Ring width δ18O
We also calculated partial correlations between δ18OTRC​ and
temperature and RH, with precipitation as the control vari- Total number of samples (radii) 30 5
able (Brienen et al. 2013). The monthly correlations were Time span (years) 1957–2017 (61) 1968–2017 (50)
calculated between δ18OTRC​ and rainfall, RH, temperature, Mean interseries correlation (r 0.518 0.648
bar)
SWD and PDSI for the 12 months from January to Decem-
EPS 0.97 0.90
ber. Additionally, spatial relationships between the δ18OTRC​

13
 Trees

variables. We found highly negative correlations between

δ18OTRC​ and precipitation during March-September, while
highest negative correlations occurred with the months of
May–July (Fig. 5A). δ18OTRC​ was also significantly nega-
tively correlated with relative humidity in the wet months
of March–May and the dry month of September (Fig. 5B).
In contrast, we found positive correlations with air temper-
ature during January-July and November–December, with
strongest correlations during the wet season (March-July;
Fig. 5D). In parallel, positive correlations between SWD
and δ18OTRC​ prevailed during the wet season of March-Sep-
tember (Fig. 5E).
We didn’t find any significative correlation with the large-
scale variables, SST and El Niño. However, δ18OTRC​ in C.
Fig. 2  50-years inter-annual variations of δ18OTRC​ extracted from five odorata was strongly negatively correlated with drought
trees of Cedrela odorata, Caatinga, Brazil. The mean inter-tree cor- index PDSI (r = − 0.62, p < 0.01; Fig. 4) of the period
relation is 0.648 (0.702) and EPS is 0.90 (0.93) for the period 1968– 1968–2017 (Table 2; Fig. 5C). We also found significant
2017 (1985–2017)
spatial correlations between the δ18OTRC​ and PDSI at the
study site and along a large part of the Caatinga region
(Fig. 6).

Discussion

The δ18OTRC​ values of the studied C. odorata trees were

significantly correlated among individuals showing a strong
common signal. Nevertheless, greater variations in δ18OTRC​
among trees were observed before 1985 (Fig. 2). A possible
explanation for that is the juvenile effect, since the studied
trees were on average 61 years old. Studies of stable isotopes
in tree rings reported anomalous values in the inner rings
closest to the pith (Duffy et al. 2017). Some of the causes,
Fig. 3  The mean tree-ring width (dotted line) and δ18O mean (solid especially for stable carbon isotopes, likely reflect changes
line) chronologies of Cedrela odorata, Caatinga, Brazil. The correla- in atmospheric C ­ O2-levels close to the soil surface, due to
tion coefficient (r) is r =  − 0.40 (− 0.60) for the period 1968–2017 intense soil respiration, lower light levels or hydraulic con-
(1985–2017). The δ18O chronology represents the mean of five indi-
ductivity (Loader et al. 2007). A similar effect, but more
viduals of cedar, while the tree-ring width chronology comprises 30
radii from 20 trees. Note that the scale of the tree-ring width chronol- attenuated, is expected in oxygen stable isotopes, once the
ogy is reversed for better visual comparison fractionation of water isotopes in the leaf is controlled to
some extent by stomatal conductance (McCarroll and Loader
2006). For C. odorata that grows in humid forests, juvenile
similar annual patterns, but of opposite sign, and hence a effects were observed in stable carbon isotopes by Fichtler
correlation coefficient of r = − 0.40 (p < 0.01; Fig. 3), which et al. (2010) and Brienen et al. (2017). However, there is no
indicates greater 18O enrichment in narrow rings than wide. available information on the nature and duration of juve-
The δ18OTRC​ chronology was significantly negatively cor- nile effects of δ18OTRC​ in C. odorata. Therefore, we can-
related with total annual precipitation (Fig. 4), as well as not exclude a possible ontogenetic effect in the years before
with precipitation of the wet season (March-August) or of 1985. The competition for soil water may explain some of
the months with more intense rain (April-July; Table 2). In this variability. C. odorata has most of the root biomass con-
particular, the correlations were higher when precipitation centrated in the upper soil horizons, especially during the
was compared with the period of strongest synchronization early growth years (Cintron 1990). Therefore, it strongly
between trees (1985–2017), both for total annual and wet relies on upper soil layer water sources (Schwendenmann
season precipitation (Table 2). et al. 2014). Since the C. odorata population in the study site
Figure 5 contains results of the monthly correlation usually aggregates in small forest fragments, the individual
analysis between the δ 18OTRC​ chronology and climatic trees may compete for water resources. In this regard, some

13
Trees

Fig. 4  Relationship between

δ18O tree-ring cellulose of
Cedrela odorata (black line)
and A total annual precipitation
and B PDSI. The correlation
coefficient (r) with precipita-
tion is r =  − 0.59 (− 0.70) and
PDSI is r =  − 0.62 (− 0.60),
both for the period 1968–2017
(1985–2017). The scale of the
total annual precipitation and
PDSI is reversed for better
visual comparison

young trees may have easier access to water than others, and actually more pronounced in the last 33 years (r = − 0.70).
a slightly inferior water supply could lead to a higher leaf-to- The strong inverse correlation with precipitation amount
air water vapor pressure gradient with higher 18O enrichment reveals the importance of rainfall in the oxygen isotopic sig-
(Gessler et al. 2014). For example, the δ18OTRC​ value in C. nature of C. odorata from Caatinga and the ability of trees
odorata tree 5 in 1976 is higher compared to the other trees to record the δ18O variability of rainwater, which is strongly
(Fig. 2). Despite that, the 50 years analyzed showed a coher- determined by the “amount effect” in the tropics (Dansgaard
ent δ18OTRC​ pattern among the trees, demonstrating that the 1964; Kurita et al. 2009). This is observed in the majority
fluctuations in stable isotope ratios were synchronized by of δ18OTRC​ studies from tropical environments (Cullen and
external forcing. Grierson, 2007; Schollaen et al. 2013; van der Sleen et al.
The δ18OTRC​ in C. odorata at our study site can be related 2015; Locosselli et al. 2020), as well as in studies with C.
to rainfall conditions. The correlation between δ18OTRC​ and odorata from South America humid forests (Ballantyne et al.
total annual precipitation of the region showed a strong 2011; Brienen et al. 2012; Cintra et al. 2019). It is important
negative relationship in the last 50 years (r = − 0.59) and to note that the δ18OTRC​ of C. odorata from Caatinga shows

13
 Trees

Table 2  Pearson’s correlation and partial correlations (Italic) coefficients between the Cedrela odorata δ18OTRC​ chronology and climate vari-
ables, Caatinga, Brazil
Annual Rainy months Wet season Dry season
1968–2017 1985–2017 1968–2017 1985–2017 1968–2017 1985–2017 1968–2017 1985–2017

Rainfall − 0.59** − 0.70** − 0.52** − 0.60** − 0.57** − 0.64** − 0.23 − 0.31*

Temperature 0.55** 0.36* 0.50** 0.33* 0.52** 0.36* 0.51** 0.22
Controlled for rainfall 0.32* 0.39* 0.37** 0.48** 0.34* 0.49** 0.26 0.22
RH − 0.37** − 0.36** − 0.38** − 0.39** − 0.45** − 0.45** − 0.23 − 0.23
Controlled for rainfall − 0.31* − 0.31* − 0.34* − 0.38** − 0.38** − 0.43** − 0.20 − 0.16
SWD 0.62** 0.66** 0.50** 0.52** 0.59** 0.61** 0.46** 0.53**
PDSI − 0.62** − 0.60** − 0.60** − 0.57** − 0.63** − 0.60** − 0.58** − 0.58**

Annual: Jan to Dec; Rainy months: Apr to Jul (> 100 mm ­month−1); Wet season: Mar to Aug (> 50 mm ­month−1); Dry season: Sep to Feb
(< 50 mm ­month−1); n = 50; Significant correlations: *p < 0.05; **p < 0.01

a strong correlation with local precipitation, whereas in C. photosynthesizing at least in the beginning of the dry season,
odorata from the Amazon the main effect driving δ18OTRC​ while reducing the overall loss of evapotranspiration (Jolly
is the large-scale rainout upstream, rather than local rainfall and Running 2004).
(Brienen et al. 2012; Cintra et al. 2019). The isotopic enrichment in the leaf water can be seen
The months between March to September constitute the in the significant correlation between δ18OTRC​ and RH
period of rainfall that significantly influences the variability (Table 2), especially at the beginning of the rainy season
δ18OTRC​ in C. odorata from Caatinga (Fig. 5A). However, (March, April and May; Fig. 5B), when temperatures are
we observed a slight decrease in strength of correlation still high and RH is low. At the leaf site of evaporation,
(Table 2) when we related the δ18OTRC​ with precipitation transpiration causes enrichment of leaf water due to prefer-
in the wet season (March-August), or with the rainy months ential evaporation of lighter H­ 216O (Dongmann et al. 1974;
(April-July). Generally, the δ18OTRC​ series correlates more Farquhar et al. 2007). The amount of transpiration depends
strongly with precipitation during the wet season that cor- on stomatal conductance and VPD, which are both linked
responds to the main growth period for this species, which to RH (McCarroll and Loader 2004). Marques et al. (2020)
is deciduous during the dry season (Brienen et al. 2012). showed a sharp linear increase in evapotranspiration and
Nevertheless, monthly correlations (Fig. 5A) show that pre- surface conductance in Caatinga vegetation at the beginning
cipitation or humidity conditions outside of the wet season of the wet season, when leaf flush occurred as soon as soil
(September) can also affect δ18OTRC​ C. odorata trees from moisture was re-established. Also, Cintra et al. (2019) found
Caatinga forest. This can be explained by the fact that Sep- a leaf enrichment contribution to δ18OTRC​of C. odorata from
tember corresponds to the beginning of the dry season, and the Amazon in initial ring sections, caused by relatively dry
historically, has a higher amount of rain (44 mm ­month−1) conditions during the start of the growing season. Consist-
and lower SWD (Fig. 1C) when compared to the other ent with this, the relationship between RH and δ18OTRC​ of
months of the dry period (October-February). This result C. odorata from Caatinga reflects the effect of atmospheric
is consistent with findings from a semi-deciduous forest in humidity on stomatal conductance and rates of transpiration,
northern Brazil, where the growth period of C. odorata con- in a way that lower (higher) humidity reduces (increases)
tinues until the first month of the dry season (Costa et al. stomatal conductance, leading to increased (decreased)
2013). Indeed, phenological studies with Caatinga trees evaporative enrichment of 18O. The partial correlations
show that the dry period induces leaf fall in deciduous spe- also reveal that, after removing the rainfall effect on the RH
cies gradually, with trees being completely leafless at the variable, the correlation with RH remains significant, espe-
peak of the dry season (Sampaio 1995; Machado et al. 1997; cially during the rainy season (Table 2), and therefore, is not
Amorim et al. 2009; Queiroz et al. 2017). In other words, only an effect of covariation between the climatic variables.
many deciduous species start rapid leaf flush as soon as the These results from Caatinga contrast to findings from the
first major precipitation events occur at the beginning of the Amazon forests, where correlations of δ18OTRC​ variations in
rainy season, while leaf senescence responds more slowly C. odorata were not significant and showed high covariation
to decreases in the soil water availability (Borchert 1994). between different climate variables (Brienen et al. 2013).
These adaptations allow the trees to make full use of the The strong negative correlations with both rainfall and RH
period of favorable water relations and optimize photosyn- during the wet season and the beginning of dry season (Sep-
thesis. Dropping part of their foliage allows trees to continue tember) supports our hypothesis that δ18OTRC​ in C. odorata

13
Trees

Fig. 5  Correlations between δ18OTRC​ of Cedrela odorata and A pre- and black bars indicate significant correlations at the p < 0.05 and
cipitation, B relative air humidity, C PDSI, D temperature and E soil p < 0.01 levels, respectively. The light blue box shows the entire wet
water deficit, from January until December of the current year. Cor- season, while the dark blue box represents months with precipita-
relations were calculated for the period 1968–2017 (50 years). Gray tion > 100 mm

trees from Caatinga is most strongly related to climatic con- stomatal conductance dominates and the environmental
ditions in the months that receive the most rainfall, due to controls are RH and soil moisture status (McCarroll and
the availability of water for physiological processes and tree Loader 2004). Tsuchya (1995) was the first who related soil
growth. water storage and tree growth in Caatinga and found that
The significant relationship of δ18OTRC​ and SWD and also wood volume growth depends on the length of the period of
with temperature (Table 2) reinforces our interpretation that soil moisture retention. Recent studies in Caatinga indicated
δ18OTRC​ reflects variations in the evaporative enrichment that the pronounced seasonality of rainfall and soil moisture
of leaf water more than in source water, especially in the affects the seasonality of stomatal conductance and photo-
wet season (Fig. 5D, E). Where moisture stress is limiting, synthesis (Mendes et al. 2017; Souza et al. 2017). Although

13
 Trees

non-significant correlations with climate during the dry

months August to October can probably be explained by
the mechanism of leaf shedding. However, the significant
correlation with temperatures during November to Febru-
ary probably results from the high covariation with rain-
fall events during the dry season (Table 2). Intense rainfall
events (> 100 mm ­month−1) in dry months (see supplemen-
tary material II) can arise because of the occurrence of an
upper tropospheric cyclonic vortex, resulting in a concentra-
tion of atmospheric moisture in the region (Oliveira 2013).
This may induce formation of new leaves, cambial activity
and plant growth, as has been documented in other Caatinga
species (Pagotto et al. 2015; Nogueira et al. 2018). However,
these heavy rains do not occur every year and should not
have a systematic influence on the δ18OTRC​ of C. odorata.
Furthermore, C. odorata is considered conservative concern-
ing the maintenance of growth seasonality, even when there
is no water deficit during the dry season (Costa et al. 2013).
Evaporative enrichment of the leaf water would explain
Fig. 6  Spatial correlation between δ18OTRC​ and Palmer drought sever- the high values of δ18OTRC​ during dry years, such as in 1987,
ity index (PDSI) obtained from the Royal Netherlands Meteorological 1990, 1998, 1999, 2001, and 2006, as well as in years with
Institute (KNMI), Climate Explorer (http://​clime​xp.​knmi.​nl/​start.​cgi)
over the months of January to December of the period 1968–2017.
extreme droughts, such as in 1993, 2003, 2011–2017. High
Red dot in the northeast Brazil indicates the study area. Correlation δ18OTRC​ values often correspond to narrow rings in the
coefficients significant at p < 0.05 are shown at the right tree-ring width chronology, for example in 1993 and 2012
(Fig. 3). Years with less rainfall experience higher radia-
tion and air temperature, which contributes to higher VPD
we cannot completely rule out that soil water isotope enrich- and reduced surface conductance (Marques et al. 2020). On
ment in the upper soil horizons might have increased during the other hand, lower values of δ18OTRC​ occur in wide tree
the dry season, thus influencing the source water δ18O value, rings that are formed in years with high precipitation, such
we suggest that increasing leaf water enrichment is the main as in 1988, 1989, 1994–1997. Air temperatures tend to be
factor behind the observed δ18OTRC​ values due to the sig- lower in wet years as a result of increased cloud cover and
nificant positive δ18OTRC​-air temperature and air humidity decreased irradiation. Under such conditions, RH is higher,
relationships. so that leaf transpiration will be reduced and evaporative
Temperature has an indirect effect on δ18OTRC​ varia- enrichment of δ18O in leaf water will be lower, resulting in
tions in C. odorata by increasing vapor pressure deficit and lower δ18OTRC​ (Roden and Ehleringer 1999; Treydte et al.
hence the evaporative enrichment during the growth season, 2014), although the δ18O value of leaf water may be altered
which has also been observed in other tree-ring stable iso- by Péclet-effect (Barbour et al. 2004; Farquhar et al. 2007;
tope studies on semi-arid forests (Cullen and Grierson 2007; Sternberg 2009). Therefore, our findings do not support the
Gebrekirstos et al. 2011; Foroozan et al. 2020). The evapora- hypothesis that variability in δ18OTRC​ in C. odorata from
tive enrichment at the leaf contributes to carbohydrates with Caatinga reflect variations in source water isotope composi-
a higher δ18O signature, which will be transported through tion. Even though our results show a strong correlation with
the phloem and be incorporated into the wood (Gessler et al. total annual precipitation, the statistical relationships dem-
2014). This may explain why C. odorata from Caatinga onstrate that δ18OTRC​ is also driven by local climate influ-
shows higher mean δ18OTRC​ (29.71 ± 0.24%) than the same ences on leaf water enrichment, via evaporative enrichment
species growing in humid forests, where δ18OTRC​ generally and stomatal restrictions to transpiration.
ranges between 18 and 28% (Ballantyne et al. 2011; Brienen Our δ18OTRC​ chronology shows a significant increasing
et al. 2012; Baker et al. 2015; Cintra et al. 2019). This is trend during recent decades, which can reflect the concur-
consistent with McCarroll and Loader (2004), who affirmed rent decrease in regional rainfall (Marengo et al. 2018) and
that trees from dry sites show different isotope ratios and are also an increase in VPD (Grossiord et al. 2020). A signifi-
more sensitive to variations in moisture regime than trees cant long-term increase of δ18OTRC​ was found by van der
from moist sites. Sleen (2015) from 1970 to 1990 in tropical Africa and also
It is important to highlight that in our study site the air related it to lowered precipitation. Although the short series
temperature exhibits little annual variability (Fig. 1B). The in the present study does not allow an analysis of long-term

13
Trees

variations in precipitation, it is important to note that from reduced precipitation (Grossiord et al. 2020), as described
the 1990s onwards, the annual rainfall amounts in our study for South America (Barkhordarian et al. 2019). Our results
site declined (Fig. 4). Marengo et al. (2018) affirmed that the highlight the importance of region’s water availability and
rainfall inter-annual variability in northeastern Brazil has the drought impact on δ18OTRC​ in C. odorata from Caat-
decreased, while at the same time the presence of sequences inga. Prospects for the reconstruction of past droughts with
of multiple dry years has been observed, especially in the δ18OTRC​ from older trees will allow a better understanding of
early years of 2000s and 2010s. This was also observed by long-term natural hydroclimate variability. Although stud-
Brito et al. (2017), who reported that the period 2011–2016 ies in this region have found trees between 40 to 80 years,
showed more intense droughts in terms of duration, sever- and a unique location that we discovered with live trees that
ity and recurrence throughout Brazil’s northwest region. reached up to 108 years, it is possible to extend the chronol-
Our results are in agreement with these findings, since the ogy through construction timbers, because C. odorata was
highest values of δ18OTRC​ occurred in the years 2010–2017 widely used to build windows, doors and fences from farms
(Fig. 4). The explanations for recurrent droughts in north- in the Caatinga rural communities (Lisi et al. 2020).
eastern Brazil are the occurrence of large-scale El Niño
events in tropical Pacific Ocean, positive anomalies of SSTs
in the northern tropical Atlantic, and a combination of both
(Marengo et al. 2018). This leads to a more northerly posi- Conclusions
tion of the ITCZ, leaving the northeast region of Brazil drier
(Servain et al. 2014). Marengo et al. (2018) report that from We present the first investigation of climatic influence on
the recent droughts of 1992, 1998, 2002, and 2010–16, only δ18O fractionation patterns in Caatinga trees, northeastern
those of 1998, 2002 and 2015–16 occurred during El Niño Brazil. Inter-annual variations in δ18OTRC​ were highly corre-
events. The extreme drought in 2012, however, occurred dur- lated among trees, showing a common environmental signal
ing a La Niña event, which is often associated with enhanced for the C. odorata population. The δ18OTRC​ record correlated
precipitation. Other factors, such distribution of SST anoma- with moisture conditions during the wet season on a regional
lies in the tropical Atlantic, the position and intensity of the scale. Although changes in rainfall amount coincide with
ITCZ, cold fronts and upper-level subtropical cyclonic vorti- inter-annual variations of δ18OTRC​, we suggest that increas-
ces are also important and affect rainfall variability in north- ing leaf water enrichment during dry conditions is the main
eastern Brazil (Marengo et al. 2018). Although we measured factor controlling stable isotope variations of C. odorata
higher δ18OTRC​ values (Fig. 4) in years with intense El Niño trees from Caatinga, which was not found in any other C.
events (1982; 1993; 1998; 2002 and 2015–16), we did not odorata population so far.
find a direct relationship between δ18OTRC​ and the oceanic Hence, we reject the hypothesis that δ18OTRC​ variations
variables SST and El Niño. The lack of a significant correla- in C. odorata from Caatinga reflect changes in source water
tion between oceanic variables and δ18OTRC​ may be related isotope composition, but depend on regional climate vari-
to local rainfall events, RH, soil water availability and air ables, including local rainfall, RH, air temperature, SWD
temperature, which are important factors influencing the and PDSI, in a way that dry conditions lead to an enrichment
isotopic composition of tree rings. of δ18OTRC​. On the other hand, we confirmed the hypoth-
The present study showed a strong negative correlation esis that δ18OTRC​ is strongly correlated with climatic condi-
with δ18OTRC​ values and PDSI in the study site (Fig. 4) and tions during the wet season, which corresponds to the wood
along a large part of the Caatinga (Fig. 6). Similar negative formation period. A probable enrichment of isotopic oxy-
relations with PDSI were also observed in a humid tropical gen occurs during evaporative processes, especially along
forest (Pumijumnong et al. 2020; Islam et al. 2021), but to earlywood formation at the beginning of the rainy season
our knowledge, it was not reported for C. odorata in South (March, April and May), when RH is low and local climatic
America. Low PDSI indicates dry soil conditions and, when conditions are relatively dry, as well as the first month of the
it is associated with low relative humidity, can causes the dry season (September) because leaf senescence responds
oxygen isotopes in leaf water to be enriched (Roden et al. slowly to decreases in the soil water availability. Under these
2000). The co-occurrence of low precipitation in north- conditions, the relationship between δ18OTRC​ and relative
eastern Brazil with the global increase in temperatures will humidity reflects a direct response of stomatal conductance
intensify the arid conditions in the Caatinga (Marengo et al. to atmospheric humidity. The strong relation with PDSI rein-
2016). High VPD reduces stomatal conductance and pho- forces that hydrological drought influences δ18OTRC​ variabil-
tosynthesis, while increasing plant water losses through ity. We also observed a trend of higher δ18OTRC​ values over
transpiration (Grossiord et al. 2020). This situation leads the past few years, which may be reflecting the increasing
to a faster depletion of soil moisture, increasing the risk of aridity in northeastern Brazil due to the decrease in precipi-
plant water stress, particularly if high VPD is combined with tation and the increase in global temperature and VPD.

13
 Trees

In conclusion, we show that δ18OTRC​ of C. odorata can be Baker JCA, Hunt SF, Clerici SJ et al (2015) Oxygen isotopes in tree
used as a proxy for inter-annual variation in regional precipi- rings show good coherence between species and sites in Bolivia.
Glob Planet Change 133:298–308. https://d​ oi.o​ rg/1​ 0.1​ 016/j.g​ lopl​
tation and moisture conditions of Caatinga during the wet acha.​2015.​09.​008
period (growth season). The potential to reconstruct humid- Baker JCA, Gloor M, Spracklen DV, Arnold SR, Tindall JC, Clerici
ity conditions from δ18OTRC​ offers opportunities for complex SJ, Leng MJ, Brienen RJW (2016) What drives interannual vari-
investigations of past dynamics of the Caatinga climate and ation in tree ring oxygen isotopes in the Amazon? Geophys Res
Lett 43:11831–11840. https://​doi.​org/​10.​1002/​2016G​L0715​07
can improve our understanding of changes in rainfall pat- Baker JCA, Santos GM, Gloor M, Brienen RJW (2017) Does Cedrela
terns in northeastern Brazil. always form annual rings? Testing ring periodicity across
South America using radiocarbon dating. Trees 31(6):1999–
Supplementary Information The online version contains supplemen- 2009. https://​doi.​org/​10.​1007/​s00468-​017-​1604-9
tary material available at https://​doi.​org/​10.​1007/​s00468-​021-​02158-z. Ballantyne AP, Baker PA, Chambers JQ, Villalba R, Argollo J (2011)
Regional differences in South American monsoon precipitation
Acknowledgements We are particularly grateful to Ms Iris Burchardt inferred from the growth and isotopic composition of tropi-
for technical assistance during sample preparation and Ms Roswitha cal trees. Earth Interact 15(5):1–35. https://​doi.​org/​10.​1175/​
Hoefner-Stich for laboratory analyses. We would like to thank the anon- 2010E​I277.1
ymous reviewers for their helpful comments. The first author expresses Banda-R K et al (2016) Plant diversity patterns in neotropical dry
a deep appreciation to the tree-ring research group of the Institute of forests and their conservation implications. Science 353:1383–
Geography, University of Erlangen-Nuremberg for their assistance dur- 1387. https://​doi.​org/​10.​1126/​scien​ce.​aaf50​80
ing the Post-Doctoral internship. Barbour MM, Roden JS, Farquhar GD, Ehleringer JR (2004)
Expressing leaf water and cellulose oxygen isotope ratios as
enrichment above source water reveals evidence of a Péclet
Author Contribution Statement MAP, conducted the fieldwork, estab- effect. Oecologia 138:426–435. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 007/​
lished the chronology, performed the isotope analysis and wrote the s00442-​003-​1449-3
manuscript. IRNM, contributed to the interpretation of the results Barkhordarian A, Saatchi SS, Behrangi A, Loikith PC, Mechoso CR
and helped in writing the manuscript. CMC, participated in the field- (2019) A recent systematic increase in vapor pressure deficit over
work and contributed to the chronology. CSL, supervised the research tropical South America. Sci Rep 9(1):1–12. https://​doi.​org/​10.​
and the tree ring analyses. AB, supervised the research, the tree ring 1038/​s41598-​019-​51857-8
and isotope analyses, provide critical feedback and helped shape the Borchert R (1994) Soil and stem water storage determine phenology
research. All authors read and approved the final manuscript. and distribution of tropical dry forest trees. Ecology 75(5):1437–
1449. https://​doi.​org/​10.​2307/​19374​67
Funding This study was funded by Coordenação de Aperfeiçoamento Brienen RJW, Zuidema PA (2005) Relating tree growth to rainfall in
de Pessoal de Nível Superior—CAPES, Post-Doctoral Research Bolivian rain forests: a test for six species using tree ring analysis.
Abroad (88881.170289/2018-01) and was also by Conselho Nacional Oecologia 146:1–12. https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 00442-0​ 05-0​ 160-y
de Desenvolvimento Científico e Tecnológico—CNPq (MCTI/CNPQ/ Brienen RJW, Zuidema PA (2006) Lifetime growth patterns and ages
UNIVERSAL 14/2011-449668/2014-2). of Bolivian rain forest trees obtained by tree ring analysis. J Ecol
94:481–493. https://​doi.​org/​10.​1111/j.​1365-​2745.​2005.​01080.x
Declarations Brienen RJW, Zuidema PA, Martínez-Ramos M (2010) Attaining the
canopy in dry and moist tropical forests: strong differences in tree
growth trajectories reflect variation in growing conditions. Oeco-
Conflict of interest The authors declare that there is no conflict of in- logia 163:485–496. https://​doi.​org/​10.​1007/​s00442-​009-​1540-5
terest. Brienen RJW, Helle G, Pons TL, Guyot JL, Gloor M (2012) Oxygen
isotopes in tree rings are a good proxy for Amazon precipitation
and El Nino-Southern Oscillation variability. PNAS 109:16957–
16962. https://​doi.​org/​10.​1073/​pnas.​12059​77109
References Brienen RJW, Hietz P, Wanek W, Gloor M (2013) Oxygen isotopes
in tree rings record variation in precipitation δ18O and amount
Alvares CA, Stape JL, Sentelhas PC, Gonçalves JLM, Sparovek G effects in the south of Mexico. J Geophys Res Biogeosci
(2014) Koppen’s climate classification map for Brazil. Meteorol 118:1604–1615. https://​doi.​org/​10.​1002/​2013J​G0023​04
Z 22(6):711–728. https://​doi.​org/​10.​1127/​0941-​2948/​2013/​0507 Brienen RJW, Gloor E, Clerici S, Newton R, Arppe L, Boom A, Bot-
Amorim IL, Sampaio EVSB, Araújo EL (2009) Fenologia de espécies trell S, Callaghan M, Heaton T, Helama S, Helle G, Leng MJ,
lenhosas da Caatinga do Seridó. RN Rev Árvore 33(3):491–499. Mielikäinen K, Oinonen M, Timonen M (2017) Tree height
https://​doi.​org/​10.​1590/​S0100-​67622​00900​03000​11 strongly affects estimates of water-use efficiency responses to
Anchukaitis KJ, Evans MN, Wheelwright NT, Schrag DP (2008) Stable climate and ­CO2 using isotopes. Nat Commun. https://​doi.o​ rg/​
isotope chronology and climate signal calibration in neotropical 10.​1038/​s41467-​017-​00225-z
montane cloud forest trees. J Geophys Res 113:G03030. https://​ Brito SSB, Cunha APMA, Cunningham CC, Alvalá RC, Marengo
doi.​org/​10.​1029/​2007J​G0006​13 JA, Carvalho MA (2017) Frequency, duration and severity of
Aragão JRV, Lisi CS (2019) Caatinga tree wood anatomy: perspectives drought in the semiarid Northeast Brazil region. Int J Climatol
on use and conservation. Floresta e Ambient 26(2):3–14. https://​ 38(2):517–529. https://​doi.​org/​10.​1002/​joc.​5225
doi.​org/​10.​1590/​2179-​8087.​099717 Castagneri D, Battipaglia G, von Arx G, Pacheco A, Carrer M (2018)
Aragão JRV, Groenendijkl P, Lisi CS (2019) Dendrochronological Tree-ring anatomy and carbon isotope ratio show both direct and
potential of four neotropical dry-forest tree species: climate- legacy effects of climate on bimodal xylem formation in Pinus
growth correlations in northeast Brazil. Dendrochronologia pinea. Tree Physiol 38:1098–1109. https://​doi.​org/​10.​1093/​treep​
53:5–16. https://​doi.​org/​10.​1016/j.​dendro.​2018.​10.​011 hys/​tpy036

13
Trees

Chiang J, Koutavas A (2004) Tropical flip-flop connections. Nature derived from tree-ring δ18O data. Atmosphere 11:889. https://​
432:684–685. https://​doi.​org/​10.​1038/​43268​4a doi.​org/​10.​3390/​atmos​11090​889
Cintra BBL, Gloor M, Boom A, Schöngart J, Locosselli GM, Brienen Frank D, Reichstein M, Bahn M, Thonicke K, Frank D, Mahecha MD,
R (2019) Contrasting controls on tree ring isotope variation for Smith P, van der Velde M et al (2015) Effects of climate extremes
Amazon floodplain and terra firme trees. Tree Physiol 39(5):845– on the terrestrial carbon cycle: concepts, processes and potential
860. https://​doi.​org/​10.​1093/​treep​hys/​tpz009 future impacts. Glob Change Biol 21:2861–2880. https://d​ oi.o​ rg/​
Cintron BB (1990) Cedrela odorata L. Cedro hembra. Spanish Cedar 10.​1111/​gcb.​12916
2(654):250–257 Gebrekirstos A, Bräuning A, Van Noordwijk M, Mitlöhner R (2011)
Corrêa ACB, Tavares BAC, Lira DR, Mutzenberg DS, Cavalcanti LCS Keynote Paper: Understanding past, present and future climate
(2019) The semi-arid domain of the Northeast of Brazil. In: Sal- changes from east to west Africa. CTA and FARA. Agricultural
gado AAR, Santos L, Paisani J (eds) The physical geography of Innovations for Sustainable Development. Contributions from
Brazil: Geography of the Physical Environment. Springer, Cham, the Finalists of the 2009/2010 Africa-wide Women and Young
pp 119–150 (10.1007/978-3-030-04333-9_7) Professionals in Science Competitions. Accra, Ghana, pp 77–86
Costa MS, Vasconcellos TJ, Barros CF, Callado CH (2013) Does Gessler A, Ferrio JP, Hommel R, Treydte K, Werner RA, Monson RK
growth rhythm of a widespread species change in distinct growth (2014) Stable isotopes in tree rings: towards a mechanistic under-
sites? IAWA J 34(4):498–509. https://d​ oi.o​ rg/1​ 0.1​ 163/2​ 29419​ 32-​ standing of isotope fractionation and mixing processes from the
00000​040 leaves to the wood. Tree Physiol 34:796–818. https://​doi.​org/​10.​
Cullen LE, Grierson PF (2007) A stable oxygen, but not carbon, iso- 1093/​treep​hys/​tpu040
tope chronology of Callitris columellaris reflects recent climate Grossiord C, Buckley TN, Cernusak LA, Novick KA, Poulter B, Sieg-
change in north-western Australia. Clim Change 85:213–229. wolf RTW, Sperry JS, McDowell NG (2020) Plant responses to
https://​doi.​org/​10.​1007/​s10584-​006-​9206-3 rising vapor pressure deficit. New Phytol 226:1550–1566. https://​
Dansgaard W (1964) Stable isotopes in precipitation. Tellus 16(4):436– doi.​org/​10.​1111/​nph.​16485
468. https://​doi.​org/​10.​3402/​tellu​sa.​v16i4.​8993 Holmes RL (1983) Computer-assisted quality control in tree-ring dat-
De Andrade EM, Aquino DN, Chaves LCG, Lopes FB (2017) Water ing and measurement. Tree-Ring Bull 43:69–78
as capital and its uses in the Caatinga. In: Silva JMC, Leal Islam M, Rahman M, Gebrekirstos A, Bräuning A (2021) Tree-ring
IR, Tabarelli M (eds) Caatinga: the largest tropical dry for- δ18O climate signals vary among tree functional types in South
est region in South America. Springer, Cham, pp 281–302 Asian tropical moist forests. Sci Total Environ 756:143939.
(10.1007/978-3-319-68339-3_10) https://​doi.​org/​10.​1016/j.​scito​tenv.​2020.​143939
Dongmann G, Nürnberg HW, Förstel H, Wagener K (1974) On the Jesus JB, Souza BB, Oliveira AMS, Gama DC (2019) Aridity index
enrichment of H ­ 218O in the leaves of transpiring plants. Radiat and climatic risk of desertification in the semi-arid state of Ser-
Environ Biophys 11:41–52. https://d​ oi.o​ rg/1​ 0.1​ 007/B​ F0132​ 3099 gipe. Revis Bras De Climatol 15(24):214–227. https://​doi.​org/​
Duffy JE, McCarroll D, Barnes A, Ramsey CB, Davies D, Loader 10.​5380/​abcli​ma.​v24i0.​62847
NJ, Miles D, Young GHF (2017) Short-lived juvenile effects Jolly W, Running S (2004) Effects of precipitation and soil water
observed in stable carbon and oxygen isotopes of UK oak trees potential on drought deciduous phenology in the Kalahari. Glob
and historic building timbers. Chem Geol 472:1–7. https://​doi.​ Change Biol 10:303–308. https://​doi.​org/​10.​1046/j.​1365-​2486.​
org/​10.​1016/j.​chemg​eo.​2017.​09.​007 2003.​00701.x
Dünisch O, Morais RR (2002) Regulation of xylem sap flow in an Kahmen A, Sachse D, Arndt SK, Tu KP, Farrington H, Vitousek
evergreen, a semi-deciduous, and a deciduous Meliaceae species PM, Dawson TE (2011) Cellulose δ18O is an index of leaf-to-
from the Amazon. Trees 16:404–416. https://​doi.​org/​10.​1007/​ air vapour pressure difference (VPD) in tropical plants. PNAS
s00468-​002-​0182-6 108:1981–1986. https://​doi.​org/​10.​1073/​pnas.​10189​06108
Dünisch O, Puls J (2003) Changes in content of reserve materials in an Kurita N, Ichiyanagi K, Matsumoto J, Yamanaka MD, Ohata T (2009)
evergreen, a semi-deciduous, and a deciduous Meliaceae species The relationship between the isotopic content of precipitation and
from the Amazon. J Appl Bot 77:10–17 the precipitation amount in tropical regions. J Geochem Explor
Dünisch O, Bauch J, Gasparotto L (2002) Formation of increment 102:113–122. https://​doi.​org/​10.​1016/j.​gexplo.​2009.​03.​002
zones and intra-annual growth dynamics in the xylem of Swi- Laumer W, Andreu L, Helle G, Schleser GH, Wieloch T, Wissel H
etenia macrophylla, Carapa guianensis, and Cedrela odorata (2009) A novel approach for the homogenization of cellulose to
(Meliaceae). IAWA J 23:101–119. https://d​ oi.o​ rg/1​ 0.1​ 163/2​ 2941​ use micro-amounts for stable isotope analyses. Rapid Commun
932-​90000​292 Mass Spectrom 23(13):1934–1940. https://​doi.​org/​10.​1002/​rcm.​
Dünisch O, Montóia VR, Bauch J (2003) Dendroecological investiga- 4105
tions on Swietenia macrophylla King and Cedrela odorata L. Lisi CS, Pagotto MA, Anholetto CR, Nogueira FC Jr, Santos HL, Costa
(Meliaceae) in the central Amazon. Trees 17:244–250. https://​ CM, Menezes IRN, Roig Juñet FA, Tommasello Filho M (2020)
doi.​org/​10.​1007/​s00468-​002-​0230-2 Dendroecological studies with Cedrela odorata L., Northeastern
EMBRAPA—Empresa Brasileira de Pesquisa Agropecuária (2018) Brazil. In: Pompa-Garcia M, Camarero JJ (eds) Latin American
Sistema Brasileiro de Classificação de Solos, 5th edn. Embrapa Dendroecology. Springer, Cham, pp 37–59
Solos, Rio de Janeiro Loader NJ, McCarroll D, Gagen M, Robertson I, Jalkanen R (2007)
Evans MN, Schrag DP (2004) A stable isotope-based approach to tropi- Extracting climatic information from stable isotopes in tree rings.
cal dendroclimatology. Geochim Cosmochim Acta 68(16):3295– In: Dawson TE, Siegwolf RTW (eds) Stable Isotopes as Indica-
3305. https://​doi.​org/​10.​1016/j.​gca.​2004.​01.​006 tors of Ecological Change. Elsevier, Oxford, pp 25–48
Farquhar GD, Cernusak LA, Barnes B (2007) Heavy water fractiona- Locosselli GM, Brienen R, Martins VTS, Gloor E, Boom A, Souza
tion during transpiration. Plant Physiol 143:11–18. https://​doi.​ EP, Saldiva PHN, Buckeridge MS (2020) Intra-annual oxygen
org/​10.​1104/​pp.​106.​093278 isotopes in the tree rings record precipitation extremes and water
Fichtler E, Helle G, Worbes M (2010) Stable-carbon isotope time series reservoir levels in the Metropolitan Area of São Paulo. Brazil
from tropical tree rings indicate a precipitation signal. Tree Ring Sci Total Environ 743:140798–141000. https://d​ oi.o​ rg/1​ 0.1​ 016/j.​
Res 66(1):35–49. https://​doi.​org/​10.​3959/​2008-​20.1 scito​tenv.​2020.​140798
Foroozan Z, Grießinger J, Pourtahmasi K, Bräuning A (2020) 501 years Machado ICS, Barros LM, Sampaio EVSB (1997) Phenology of Caat-
of spring precipitation history for the semi-srid Northern Iran inga species at Serra Talhada, PE. Northeastern Brazil Biotropica

13
 Trees

29(1):57–68. https://​doi.​org/​10.​1111/j.​1744-​7429.​1997.​tb000​ Pons TL, Helle G (2011) Identification of anatomically non-distinct

06.x annual rings in tropical trees using stable isotopes. Trees 25:83–
Marengo JA, Torres RR, Alves LM (2016) Drought in Northeast Bra- 93. https://​doi.​org/​10.​1007/​s00468-​010-​0527-5
zil—past, present, and future. Theor Appl Climatol 129:1189– Pumijumnong N, Bräuning A, Sano M, Nakatsuka T, Muangsong C,
1200. https://​doi.​org/​10.​1007/​s00704-​016-​1840-8 Buajan S (2020) A 338-year tree-ring oxygen isotope record
Marengo JA, Alves LM, Alvala RCS, Cunha AP, Brito S, Moraes from Thai teak captures the variations in the Asian summer
OLL (2018) Climatic characteristics of the 2010–2016 drought monsoon system. Sci Rep 10:8966. https://​doi.​org/​10.​1038/​
in the semiarid Northeast Brazil region. An Acad Bras Ciênc s41598-​020-​66001-0
90(2 Suppl. 1):1973–1985. https://​doi.​org/​10.​1590/​0001-​37652​ Queiroz LP, Cardoso D, Fernandes MF, Moro MF (2017) Diversity and
01720​170206 evolution of flowering plants of the Caatinga domain. In: Silva
Marques TV, Mendes K, Mutti P, Medeiros S, Silva L, Perez-Marin JMC, Leal IR, Tabarelli M (eds) Caatinga: the largest tropical
A et al (2020) Environmental and biophysical controls of evap- dry forest region in South America. Springer, Cham, pp 23–63
otranspiration from Seasonally Dry Tropical Forests (Caatinga) (10.1007/978-3-319-68339-3_2)
in the Brazilian Semiarid. Agric for Meteorol 287:107957. Roden JS, Ehleringer JR (1999) Hydrogen and oxygen isotope ratios
https://​doi.​org/​10.​1016/j.​agrfo​r met.​2020.​107957 of tree ring cellulose for field-grown riparian trees. Oecologia
McCarroll D, Loader NJ (2004) Stable isotopes in tree rings. Quat 123:481–489. https://​doi.​org/​10.​1007/​s0044​20000​349
Sci Rev 23:771–801. https://​doi.​org/​10.​1016/j.​quasc​irev.​2003.​ Roden JS, Lin G, Ehleringer JR (2000) A mechanistic model for inter-
06.​017 pretation of hydrogen and oxygen isotope ratios in tree-ring cel-
McCarroll D, Loader NJ (2006) Isotopes in tree rings. In: Leng MJ (ed) lulose. Geochim Cosmochim Acta 64(1):21–35. https://​doi.​org/​
Isotopes in paleoenvironmental research. Springer, Dordrecht, 10.​1016/​S0016-​7037(99)​00195-7
pp 67–106 Rolim GS, Sentelhas PC, Barbieri V (1998) Planilhas no ambiente
Mendes KR, Granja JAA, Ometto JP, Antonino ACD, Menezes RSC, ExcelTM para os cálculos de balanços hídricos: normal, sequen-
Pereira EC, Pompelli MF (2017) Croton blanchetianus modu- cial, de cultura e de produtividade real e potencial. Rev Bras
lates its morphophysiological responses to tolerate drought in a Agrometeorol 6:133–137
tropical dry forest. Funct Plant Biol 44(10):1039–1051. https://​ Rozanski K, Araguas L, Gonfiantini R (1993) Isotopic patterns in mod-
doi.​org/​10.​1071/​FP170​98 ern global precipitation. Geophys Monogr 78:1–36. https://​doi.​
Nogueira FC Jr, Pagotto MA, Roig FA, Lisi CS, Ribeiro AS (2018) org/​10.​1029/​GM078​p0001
Responses of tree-ring growth in Schinopsis brasiliensis to Sampaio EVSB (1995) Overview of the Brazilian Caatinga. In: Bullock
climate factors in the dry forests of northeastern Brazil. Trees SM, Mooney HA, Medina E (eds) Seasonally dry tropical forests.
32(2):453–464. https://​doi.​org/​10.​1007/​s00468-​017-​1642-3 Cambridge University Press, Cambridge, pp 35–63 (10.1017/
Nogueira FC Jr, Pagotto MA, Aragão JRV, Roig FA, Ribeiro AS, Lisi CBO9780511753398.003)
CS (2019) The hydrological performance of Prosopis juliflora Sampaio EVSB (2010) Caracterização do Bioma Caatinga—caracterís-
(Sw.) growth as an invasive alien tree species in the semiarid ticas e potencialidades. In: Gariglio et al. (orgs) Uso Sustentável
tropics of northeastern Brazil. Biol Invasions 21:2561–2575. e Conservação dos Recursos Florestais da Caatinga, Serviço Flo-
https://​doi.​org/​10.​1007/​s10530-​019-​01994-y restal Brasileiro, Brasília, pp 27–48. ISBN: 978-85-63269-04-1
Ohashi S, Durgante FM, Kagawa A et al (2015) Seasonal variations in Santos MG, Oliveira MT, Figueiredo KV et al (2014) Caatinga, the
the stable oxygen isotope ratio of wood cellulose reveal annual Brazilian dry tropical forest: can it tolerate climate changes?
rings of trees in a Central Amazon terra firme forest. Oecologia Theor Exp Plant Physiol 26(1):83–99. https://​doi.​org/​10.​1007/​
180:685–696. https://​doi.​org/​10.​1007/​s00442-​015-​3509-x s40626-​014-​0008-0
Oliveira PT, Santos e Silva CM, Lima KC (2013) Synoptic environ- Schollaen K, Heinrich I, Neuwirth B, Krusic PJ, D’Arrigo RD, Kary-
ment associated with heavy rainfall events on the coastland of anto O, Helle G (2013) Multiple tree-ring chronologies (ring
Northeast Brazil. Adv Geosci 35:73–78. https://d​ oi.o​ rg/1​ 0.5​ 194/​ width, δ13C and δ18O) reveal dry and rainy season signals of
adgeo-​35-​73-​2013 rainfall in Indonesia. Quat Sci Rev 73:170–181. https://​doi.​org/​
Olson EJ, Dodd JP, Rivera MA (2020) Prosopis sp. tree-ring oxygen 10.​1016/j.​quasc​irev.​2013.​05.​018
and carbon isotope record of regional-scale hydroclimate vari- Schwendenmann L, Pendall E, Sanchez-Bragado R, Kunert N, Höls-
ability during the last 9500 years in the Atacama Desert. Palaeo- cher D (2014) Tree water uptake in a tropical plantation vary-
geogr Palaeoclimatol Palaeoecol 538:109408. https://​doi.​org/​10.​ ing in tree diversity: interspecific differences, seasonal shifts
1016/j.​palaeo.​2019.​109408 and complementarity. Ecohydrol 8(1):1–12. https://​doi.​org/​10.​
Pagotto MA, Roig FA, Ribeiro AS, Lisi CS (2015) Influence of 1002/​eco.​1479
regional rainfall and Atlantic sea surface temperature on tree- Servain J, Caniaux G, Kouadio YK, McPhaden MJ, Araujo M (2014)
ring growth of Poincianella pyramidalis, semiarid forest from Recent climatic trends in the tropical Atlantic. Clim Dyn
Brazil. Dendrochronologia 35:14–23. https://​doi.​org/​10.​1016/j.​ 43:3071–3089. https://​doi.​org/​10.​1007/​s00382-​014-​2168-7
dendro.​2015.​05.​007 Silva VPR, Pereira ERR, Almeida RSR (2012) Estudo da variabilidade
Pagotto MA, DeSoto L, Carvalho A, Nabais C, Tomazello Filho M, anual e intra-anual da precipitação na região nordeste do Bra-
Ribeiro A, Lisi CS (2017) Evaluation of X-ray densitometry to sil. Rev Bras Meteorol 27(2):163–172. https://​doi.​org/​10.​1590/​
identify tree-ring boundaries of two deciduous species from S0102-​77862​01200​02000​05
semi-arid forests in Brazil. Dendrochronologia 42:94–103. Silva JMC, Barbosa LCF, Leal IR, Tabarelli M (2017) The Caat-
https://​doi.​org/​10.​1016/j.​dendro.​2017.​01.​007 inga: understanding the challenges. In: Silva JMC, Leal
Pennington RT, Lavin M, Oliveira-Filho A (2009) Woody plant diver- IR, Tabarelli M (eds) Caatinga: the largest tropical dry for-
sity, evolution, and ecology in the tropics: perspectives from sea- est region in South America. Springer, Cham, pp 3–22
sonally dry tropical forests. Annu Rev Ecol Evol Syst 40:437– (10.1007/978-3-319-68339-3_1)
457. https://​doi.​org/​10.​1146/​annur​ev.​ecols​ys.​110308.​120327 Souza LSB, Moura MSB, Sediyama GC, Silva TGF (2017) Carbon
Pennington RT, Lehmann CER, Rowland LM (2018) Tropical savannas exchange in a caatinga area during an unusually drought year.
and dry forests. Curr Biol 28:541–545. https://d​ oi.o​ rg/​10.​1016/j.​ Agrometeoros 25(1):37–44 (ISSN 2526-7043)
cub.​2018.​03.​014

13
Trees

Sternberg LSLOR (2009) Oxygen stable isotope ratios of tree-ring Erdkunde 70:69–82. https://​doi.​org/​10.​3112/​erdku​nde.​2016.​01.​
cellulose: the next phase of understanding. New Phytol 181:553– 05
562. https://​doi.​org/​10.​1111/j.​1469-​8137.​2008.​02661.x Wagner FH, Hérault B, Bonal D et al (2016) Climate seasonality lim-
Thornthwaite CW (1948) An approach toward a rational classification its leaf carbon assimilation and wood productivity in tropical
of climate. Geogr Rev 38:55–94. https://d​ oi.o​ rg/1​ 0.2​ 307/2​ 10739 forests. Biogeosciences 13:2537–2562. https://​doi.​org/​10.​5194/​
Thornthwaite CW, Mather JR (1955) The water balance. Publ Climatol bg-​13-​2537-​2016
8(1):1–104 Wieloch T, Helle G, Heinrich I, Voigt M, Schyma P (2011) A novel
Tomazello Filho M, Botosso PC, Lisi CS (2000) Potencialidade da device for batch-wise isolation of α-cellulose from small-amount
família Meliaceae para dendrocronologia em regiões tropicais wholewood samples. Dendrochronologia 29(2):115–117. https://​
e subtropicais. In: Roig FA (ed) (comp) Dendrocronología en doi.​org/​10.​1016/j.​dendro.​2010.​08.​008
América Latina, Editorial de la Universidad Nacional de Cuyo, Wigley T, Briffa KR, Jones PD (1984) On the average value of cor-
Mendoza, pp 381–431. ISBN: 950-39-0122-7 related time series, with applications in dendroclimatology
Treydte K, Boda S, Graf Pannatier E et al (2014) Seasonal trans- and hydrometeorology. J Appl Meteorol Climatol 23:201–213.
fer of oxygen isotopes from precipitation and soil to the tree https://​doi.​org/​10.​1175/​1520-​0450(1984)​023%​3c0201:​OTA-
ring: source water versus needle water enrichment. New Phytol VOC%​3e2.0.​CO;2
202:772–783. https://​doi.​org/​10.​1111/​nph.​12741 Worbes M (1995) How to measure growth dynamics in tropical trees—
Tsuchiya A (1995) Preliminary study on the relationship between ves- a review. IAWA J 16:337–351. https://​doi.​org/​10.​1163/​22941​
sel growth of thorny shrubs and water balance in the semi-arid 932-​90001​424
region, northeastern Brazil. Geogr Sci 50:123–131 Worbes M (1999) Annual growth rings, rainfall-dependent growth and
van der Sleen P, Groenendijk P, Zuidema PA (2015) Tree-ring δ18O long-term growth patterns of tropical trees from the Caparo For-
in African mahogany (Entandrophragma utile) records regional est Reserve in Venezuela. J Ecol 87:391–403. https://​doi.​org/​10.​
precipitation and can be used for climate reconstructions. Glob 1046/j.​1365-​2745.​1999.​00361.x
Planet Change 127:58–66. https://​doi.​org/​10.​1016/j.​glopl​acha.​
2015.​01.​014 Publisher’s Note Springer Nature remains neutral with regard to
van der Sleen P, Zuidema PA, Pons TL (2017) Stable isotopes in jurisdictional claims in published maps and institutional affiliations.
tropical tree rings: theory, methods and applications. Funct Ecol
31:1674–1689. https://​doi.​org/​10.​1111/​1365-​2435.​12889
Volland F, Pucha D, Bräuning A (2016) Hydro-climatic variability
in southern Ecuador reflected by tree-ring oxygen isotopes.

13