Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
INSTITUTO DE QUÍMICA
PROGRAMA DE PÓS-GRADUAÇÃO EM GEOCIÊNCIAS – GEOQUÍMICA
ANGELA AMPUERO GRÁNDEZ
THE FOREST EFFECTS ON THE ISOTOPIC COMPOSITION OF RAINFALL IN THE NORTHWESTERN AMAZON BASIN
NITERÓI
2019
ANGELA AMPUERO GRÁNDEZ
THE FOREST EFFECTS IN THE ISOTOPIC COMPOSITION OF RAIN IN THE NORTHWESTERN AMAZON BASIN
Dissertação apresentada ao Curso de Pós-
Graduaçãoem Geociências da Universidade
Federal Fluminense, como requisito parcialpara a
obtenção do Grau de Mestre. Área de
Concentração: Geoquímica Ambiental.
Orientador:
Prof. Dr. Nicolás MisaílidisStrikis
Co-Orientador:
Prof. Dr. James Emiliano Apaéstegui Campos
NITERÓI
2019
Ficha catalográfica automática - SDC/BGQGerada com informações fornecidas pelo autor
Bibliotecária responsável: Catia Vasconcellos Marques - CRB7/5116
G751t Grández, Angela Ampuero THE FOREST EFFECTS ON THE ISOTOPIC COMPOSITION OF RAINFALLIN THE NORTHWESTERN AMAZON BASIN / Angela Ampuero Grández ;Nicolás Misaílidis Strikis, orientador ; James EmilianoApaéstegui Campos, coorientador. Niterói, 2019. 70 f. : il.
Dissertação (mestrado)-Universidade Federal Fluminense,Niterói, 2019.
DOI: http://dx.doi.org/10.22409/PPG-Geo.2019.m.06419394775
1. Amazon. 2. Stable isotopic composition of water. 3.Deuterium excess. 4. Speleothem. 5. Produção intelectual. I.Strikis, Nicolás Misaílidis, orientador. II. Campos, JamesEmiliano Apaéstegui, coorientador. III. Universidade FederalFluminense. Instituto de Química. IV. Título.
CDD -
ACKNOWLEDGMENTS
I thank my parents for all their love and for always encouraging me to do better.
I thank my family and friends for being a great support team, even though we were
far away.
I thank Rod for the time we spent together before I came here. I sure learnt some
survival skills that came out useful in this process. What a ride!
I thank my peruvian friends from the Instituto Geofísico del Perú for all the things you
thought me and the good times too!
I thank the class of 2017, you were with no doubt, the best class of my life.
I thank the professors, staff and fellow students from the Geoquimica Department for
making such an amazing and welcoming environment for learning and have fun.
I thank the climbers from Rio, specially the guys from Niteroi. I will never forget the
night climbs at Pracinha. I specially thank Kristy, my Colombian sister, climbing
partner and Python instructor.
I thank my roommates who were like a family to me, in the most complete sense.
I thank Abdel and Chico Bill, because it was you who set the base for the projects we
are currently working on.
I thank the community of the Palestina Cave in Peru, specially Graciela for collecting
the water samples, doing a wonderful job. Also, the Speleo Club Andino (ECA), the
Institute de Recherche pour le development (IRD) in Peru, for their support during the
field trips.
I thank Mathias Vuille, Renato Campello and Heitor Evangelista, for all your
comments and recommendations to improve this study. I also thank Delphine Zemp
and Jelena Maksic, for their disposition to solve my doubts and share their data.
I thank James for all the things you taught me, from driving to scientific writing, which
helped me along the way. Also, for showing me Peru and introducing me to
speleology, which is way cooler than I initially thought.
I thank Nicolás, in first place, for accepting me as his student. I know the whole
process was challenging but, in the end, it turned out ok. Who would have guessed! I
thank you for all the things you patiently taught me and for our long discussions.
Thank you for keeping me motivated, for making everything a little cooler and fun.
I thank CAPES for granting the scholarship over the period of this master's degree.
“I had nothing to offer anybody except my
own confusion.”
- Jack Kerouac,
On the road
ABSTRACT
Moisture recycling is a major process of the hydrologic cycle in the Amazon. Recent studies based on remote sensing tools and numerical modeling have shed light on the importance of this process, and even quantified it on a climatology perspective. Parallel efforts suggested the effects of recycled moisture in the isotopic composition of rainfall, although this subject is still controversial. Even more uncertain are the implications for the paleoclimate registers based on water stable isotopes (δ18O, δD). To assess temporal variations of water isotopic composition, we take an empirical approach and present four-year monitoring of isotopic composition of rainfall in the northwestern Amazon basin. We explore the effects of climate and vegetation on isotopic composition by establishing air mass history based on atmospheric back-trajectory analyses, satellite observations of precipitation, leaf area index and modeled moisture recycling along the transport pathway. The results suggest that all variables exert control on isotope variability and that control varies with prevailing atmospheric transport pathways. Observations agree that precipitation upwind is the main control on δ18Oin the northwestern Amazon basin. Furthermore, the results suggest that the forest exerts a significant control on the isotopic composition of precipitation, evidenced in dxs variations. In the light of these findings, we interpret Holocene dxs record based on fluid inclusions from well dated stalagmites from the Tigre Perdido cave. Results show that dxs points variations in the gradient of speleothem δ18Ocalcite from the east and west ends of the Amazon basin. Moreover, dxs follows shifts in rainforest cover inferred from lake pollen registers and global vegetation modelling. Finally, we suggest that the continental moisture contribution to precipitation in the northwestern Amazon involves the forest activity and has a significant imprint in the isotopic composition precipitation.
Keywords: Amazon. Stable isotopic composition of water. Deuterium excess. Speleothem. Fluid inclusions. Holocene.
RESUMO
A ciclagem de umidade é um processo importante do ciclo hidrológico na Amazônia. Estudos recentes baseados em ferramentas de sensoriamento remoto e modelagem numérica esclareceram a importância desse processo e até o quantificaram em uma perspectiva climatológica. Esforços paralelos sugeriram efeitos da umidade reciclada na composição isotópica da chuva, embora esse assunto ainda seja controverso. Ainda mais incertas são as implicações para os registros paleoclimáticos baseados em isótopos estáveis da água (δ18O, δD). Para avaliar as variações temporais dos isótopos da água, adotamos uma abordagem empírica e apresentamos um monitoramento de quatro anos da composição isotópica da precipitação no noroeste da bacia Amazônica. Nós exploramos os efeitos do clima e da vegetação na composição dos isótopos estabelecendo a historia das parcelas de ar com base na análise reretro-trajetórias, observações de precipitação por satélite, índice de área foliar e com ciclagem de umidade modelada no caminho das massas de ar. Os resultados sugerem que todas as variáveis exercem controle sobre a variabilidade isotópica e que o controle varia com as vias de transporte atmosféricas predominantes. Observações concordam que a precipitação a caminho dos fluxos de umidade é o principal controle no δ18O do noroeste da baciaAmazônica. Além disso, os resultados sugerem que a floresta exerce um controle significativo sobre a composição isotópica da precipitação, evidenciada nas variações de dxs. À luz dessas descobertas, nós interpretamos o registro do Holocene de dxs baseado em inclusões fluidas de estalagmites bem datadas da caverna do Tigre Perdido. Os resultados mostram que dxs aponta variações no gradiente δ18O de espeleotemas na borda leste e oeste da bacia Amazônica. Além disso, dxs segue mudanças na cobertura de floresta tropical inferidas a partir de registros de pólen de lagos e modelagem global de vegetação. Finalmente, sugerimos que a contribuição da umidade continental para a precipitação no noroeste da Amazônia envolve a atividade florestal e tem uma impressão significativa na composição isotópica das precipitações.
Palavras-chave: Amazônia. Composição isotopica estável da água. Excesso de deutério. Espeleotema. Inclusões fluidas. Holoceno.
LIST OF FIGURES
Figure 1. (a) Mean annual precipitation in mm for the period 1989-1995, based on
the average of the products CRU, GPCC, GPCP and CPC and(b) number of months
with less than 100 mm rainfall. Deforested (red dots) and non-forest (gray dots) areas
are indicated. ........................................................................................................... 19
Figure 2. Long term mean precipitation (shades) and wind circulation (vectors and
stream lines). Left panels show low level (950hPa) wind circulation for (a) January
and (b) July and right panels show high level (300hPa) wind circulation for (c)
January and (d) July. ................................................................................................ 20
Figure 3. (a) Mean annual evapotranspiration in mm for the period from 1989 to
1995 from forty products. including field observations, reanalysis and model outputs.
(b) Precipitation recycling defined as the fraction of rainfall that comes from
continental evapotranspiration and vertically integrated moisture fluxes (black
arrows). Deforested (red dots) and non-forest (gray dots) areas are indicated ......... 22
Figure 4. a) Schematic representation of tree transpiration recycling. On average
20% of rainfall in the Amazon has been transpired at least once. About half of this
transpiration (51%) is recycled in the first cycle, the remainder occurs after multiple
re-evaporation cycles (moisture cascading). b) Transpiration recycling ratio in the
Amazon against average rainfall from oceanic origin. 2005 and 2010 are labeled
because vast drought was registered in vast areas across the Amazon. .................. 23
Figure 5. a) Mean annual transpiration by trees that precipitate over land. b) Fraction
of mean annual rainfall that has been transpired by trees in the Amazon basin. The
Amazon basin is shown by black outline. ................................................................. 24
Figure 6. Schematic representation of cascading recycling effects in the vegetation-
rainfall system. a) System in equilibrium. b) Initial forest loss triggered by decreasing
oceanic moisture influx. c) Altered rainfall regime in another location, leading to
further forest loss and reduced moisture transport. .................................................. 25
Figure 7. Vegetation distribution map for different periods of the Holocene, from
numerical simulation. The model used from the Center for Weather Prediction and
Climate Studies (CPTEC) is the Potential Vegetation Model version 2 (CPTEC-
PVM2). Two experiments were performed for the 6K period, one with present SST
(6k-pvm veg type) and the other with mid Holocene SST (6k-pvm-MH-sst veg type).
................................................................................................................................. 29
Figure 8. Simplified Rayleigh distillation scheme in the water cycle. ........................ 30
Figure 9. Images of typical fluid inclusions in the speleothems analyzed in VAN
BREUKELEN et al., 2008. ........................................................................................ 32
Figure 10. Location of the Palestina station and precipitation climatology. Dark blue
shadows represent the Andes mountain range and the green line, the limit of the
Amazon basin taken from the HyBAM data base. Precipitation at the Palestina
station was obtained from TRMM 3B42 from 1998 to 2018. ..................................... 33
Figure 11. Records of δ18O (upper panel) and dxs (lower panel) from the stations
Palestina (black) and Pomacochas (red) on biweekly timestep. ............................... 34
Figure 12. Scheme of the rain collector used in this study. ...................................... 35
Figure 13. Schematic representation of one back-trajectory (black arrow). The back-
trajectory is divided in seven segments, each one covering one day. The beginning of
the back-trajectory is marked with a star at the Palestina station on the most recent
day (0), and then progresses back in time until completing 7 days (-7). Climate data
on precipitation, LAI and PR was selected from the tiles spanned by the back-
trajectory on the corresponding day. In this example, on the sixth day (between -5
and -6), the trajectory goes over the ocean (blue shadow), so information on those
tiles is excluded. The dashed line represents the coast line. .................................... 39
Figure 14. Clusters of seven-day back trajectories at 1500 m.a.g.l. from June 2012 to
May 2018.The bars under each map show the frequency of back-trajectory clusters
per month. ................................................................................................................ 42
Figure 15. Annual cycle of the degree of rainout upstream (DRU) and local
precipitation from 2012 to 2018. DRU is the accumulated precipitation along the
back-trajectories that initiate on precipitation days at the Palestina station. Local
precipitation is calculated from TRMM3B42 and GPM by averaging the nearest tiles
to the Palestina station. ............................................................................................ 44
Figure 16. Annual cycle of the degree of rainout upstream (DRU), Leaf area index on
the back-trajectories (LAI) and Precipitation Recycling on the back-trajectories (PR)
for the period from June 2012 to May 2018. ............................................................. 45
Figure 17. Precipitation recycling (PR) computed with WAM2lyr and data from ERA
Interim. a) Mean annual PR. b-e) Seasonal anomalies. ........................................... 46
Figure 18. The maps show mean moisture recycling (PR) per season in percentage
(%). Only values in the dominant back-trajectory areas are shown. Panels under the
maps show the meridional average PR. Clusters 1, 5 and 11 are frequent during
austral summer; clusters 2 and 3, during austral autumn and winter; and clusters 6
and 7, during austral winter. . .................................................................................. 46
Figure 19. Biweekly δ18O andδD of precipitation at the Palestina station. The GMWL
is represented by a solid red line and deviations of ±5‰ are represented with dashed
lines. ........................................................................................................................ 47
Figure 20. Correlation between δ18O and dxs of precipitation at the Palestina station
on biweekly timestep. ............................................................................................... 48
Figure 21. Correlation between of δ18O and dxs of precipitation at the Palestina
station on biweekly timestep for the seasons (a)JJA, (b)SON, (c)DJF and (d)MAM. 48
Figure 22. Panels on the left show the original records at the Paletina station and the
ones in the right, the records without seasonality (anomalies). (a) and (b) show δ18O
of precipitation and local precipitation accumulated along the water sampling period.
(c) and (d) show δ18O of precipitation and calculated degree of rainout upstream
(DRU), weighted by local precipitation for the water sampling period. (e) and (f) show
deuterium excess (dxs) of precipitation and Leaf area index accumulated along the
back-trajectories (LAI). LAI was weighted by local precipitation for the water sampling
period. ...................................................................................................................... 51
Figure 23. Time series of dxs, PR in the upper panels and LAI in the lower panels.
Lines represent the series in biweekly time step dots, the seasonal averages for (a,c)
wet season and (b,d) extended dry season. The gray shadow indicates the period
filled with dxs data from Pomacochas station. .......................................................... 52
Figure 24. Spatial correlation between LAI and VIDMF on monthly time step for the
period from June 2012 to May 2018. Seasonality was removed from both datasets.
Shadows show only significant grid cells (p<0.05).................................................... 53
Figure 25. Geographic location of the geochemical and palynological records
referenced in this study. ........................................................................................... 56
Figure 26. Comparison between stable isotope record from speleothems from the
western and eastern Amazon edges: a) dxs reconstruction from Tigre Perdido
speleothem record (VAN BREUKELEN et al., 2008); b) Tigre Perdido - Paraíso
δ18O; c) 20-yr interpolated δ18O record from Tigre Perdido cave speleothem (VAN
BREUKELEN et al., 2008). The δ18O values from Tigre Perdido are corrected by 1.4
‰ to account for temperature variations between the caves following the procedure
of Wang et al. (2017); d) as in (c) 20-yr interpolated δ18O record from Paraíso cave
speleothems (WANG et al., 2017). ........................................................................... 57
Figure 27. Comparison between calculated dxs from Tigre Perdido stalagmite (VAN
BREUKELEN et al., 2008) with the integrated record of rainforest pollen frequency
from five lakes: Iriri, Altamira - PA (SANTOS, 2019); Ilha Arapujá, Altamira - PA
(SANTOS, 2019);Laguna Chochos, Peru (BUSH et al. 2005);Saci, central-south
Amazon (FONTES et al., 2017) and Serra Sul Carajás Lake, CSS2 (ABSY et al.,
1991; SIFEDDINE et a., 2001). ................................................................................ 59
Figure 28. Normalized interpolated frequency of arboreal pollen from lakes used in
this study and average value. ................................................................................... 59
LIST OF TABLES
Table 1. Values of the quality layer for LAI/FPAR (8bit) in MODIS. .......................... 37
Table 2. Linear correlation coefficient (r) and p-value between isotope records and
potential climatic controls based on biweekly data. The r-values in bold (if p<0.05)
and in italics (if p<0.10). ........................................................................................... 50
Table 3. As in Table 1, but for dry months (JJAS) and wet months (ONDJFMAM).
The first two weeks of October correspond to the dry season and the last two, to the
wet season. .............................................................................................................. 50
NOMENCLATURE
δD Stable Hidrogen isotope ratio (2H/1H) of precipitation
δ18O Stable Oxigen isotope ratio (18O/16O) of precipitation
δ18Ocal Stable Oxigen isotope ratio (18O/16O) of calcite
DRU Degree of rainout upstream
dxs Deuterium excess
ENSO El Niño southern oscillation
GMWL Global meteoric water line
GPM Global precipitation mission
ITCZ Intertropical Convergence Zone
kyr BP Thousand years before present (1950)
LAI Leaf area index
LGM Last Glacial Maximum
LLJ Low Level Jet
m.a.g.l. Meters above ground level
m.a.s.l. Meters above sea level
PR Precipitation recycling or moisture recycling
SACZ South Atlantic Convergence Zone
SAMS South American Monsoon System
SST Sea surface temperature
TRMM Tropical rainfall measurement mission
CONTENTS
RESUMO ................................................................................................................... 6
LIST OF FIGURES .................................................................................................... 7
LIST OF TABLES .................................................................................................... 11
NOMENCLATURE................................................................................................... 12
1. INTRODUCTION ............................................................................................... 15
2. OBJECTIVES .................................................................................................... 18
2.1. SPECIFIC OBJECTIVES ................................................................................... 18
3. THEORETHICAL BASIS ................................................................................... 19
3.1. CLIMATE OF THE AMAZON AND GLOBAL IMPACTS ..................................... 19
3.2. THE AMAZON, A CONTINENTAL MOISTURE SOURCE ................................. 21
3.3. AMAZONIAN FOREST COVER SHIFTS: THE EVOLUTION OF THE FOREST
ALONG THE HOLOCENE ....................................................................................... 25
3.4. WATER STABLE ISOTOPOLOGUES IN THE WATER CYCLE ........................ 29
3.5. DEUTERIUM EXCESS: A PROXY FOR RECYCLED MOISTURE? .................. 30
3.6. SPELEOTHEMS AND FLUID INCLUSIONS ..................................................... 31
3.7. MONITORING SITE .................................................................................... 32
4. MATERIALS AND METHODS .......................................................................... 34
4.1. ISOTOPE MONITORING .................................................................................. 34
4.2. CLIMATE DATA ................................................................................................ 35
4.3. PALINOLOGYCAL RECORDS .......................................................................... 38
4.4. METHODS ........................................................................................................ 38
5. RESULTS .......................................................................................................... 41
5.1. PRECIPITATION AND BACK-TRAJECTORY ANALYSIS ................................. 41
5.2. FOREST MOISTURE FLUXES ......................................................................... 44
5.3. ISOTOPE MONITORING .................................................................................. 47
6. DISCUSSION .................................................................................................... 53
6.1. CONTROLS OF δ18O AND DXS IN PRECIPITATION ....................................... 53
7. CONCLUSION .................................................................................................. 60
8. REFERENCES .................................................................................................. 61
15
1. INTRODUCTION
The effects of rainforest evapotranspiration on the stable isotopic composition
of rainfall (δ18O, δD) is particularly important to reconstruct the hydroclimate history in
South America. In general, the isotopic composition of a precipitating air masses
moving over a continental area results from a combination of processes that mediate
the Rayleigh distillation. Raleigh distillation is the process where heavy isotope
species are progressively removed from the atmosphere through precipitation,
leaving the remaining vapor isotopically depleted (VUILLE; WERNER, 2005;
LACHNIET, 2008). While the progressive condensation of moisture leads to a
gradual lowering of heavy isotopes in the remaining vapor, re-evaporated moisture
from land compensates, at least partially, the loss of the heavy isotopes.
Several studies have shown that the Amazon basin itself constitutes a major
moisture source for the region, driven by the hydrologic regime of the rainforest
(SALATI, 1979, DRUMOND et al., 2014, MOLINA et al., 2019). Recently Staal et al.
(2018) estimated that about 32% of Amazonian rainfall originates from
evapotranspiration within the basin; two thirds of which correspond to tree
transpiration. Furthermore, the positive feedback between the forest and rainfall
promotes cascades of recycled moisture that precipitate across distant areas,
allowing for forest cover expansion. At the same time, the forest induces dry season
rainfall, necessary to sustain itself when oceanic moisture supply is limited (ZEMP et
al., 2014, STAAL et al., 2018).
Another proxy suitable to describe moisture activity is deuterium excess (dxs =
δD - 10 × δ18O), a second order parameter derived from kinetic fractionation of the
water stable isotopologues (DANSGAARD, 1964; MERLIVAR; JOUZEL, 1979). The
dxs value of an air mass is influenced by the physical conditions at the oceanic
moisture source, principally relative humidity and sea surface temperature (PFAHL;
SODEMANN, 2014). Moreover, prevailing conditions during advection across the
continent and interaction or mixing with different air masses further modify dxs
(FROEHLICH et al 2002). Particularly, over large continental areas, such as the
Amazon basin, variations in the dxs composition have the potential to serve as a
fingerprint of recycled air moisture (SALATI et al., 1979).
16
Water isotopologue data across the Amazon suggest that changes in forest
vegetation may affect the dxs values through changes in the rates of moisture
recycling associated with forest evapotranspiration (SALATI et al., 1979, ZEMP et al.,
2014 among others). Although progress has been made in understanding the major
drivers of δ18O and δD variability in the tropical region, the relationship between
moisture recycling and the rainfall isotopic composition in the Amazon basin is not
fully understood. Moreover, the climatic significance of dxs values over continental
areas in the tropics is still unclear (LANDAIS et al., 2010; RISI et al., 2013; VUILLE,
2018). In this sense, it is important to clarify the effect of land atmosphere-
interactions on the isotopic composition of rain to reliably interpret isotope-based
paleoclimate records. For instance, in the Amazon basin paleo-precipitation
reconstructions based on speleothem δ18Ocalcite records are controversial. Some
authors sustain the notion that during the Last Glacial Maximum (LGM, ~ 26.5 to 19
kyr BP) a strong east-west rainfall isotope gradient along the Amazon may have
resulted from reduced water recycling, as consequence of a widely drier Amazon,
leading to reduced plant transpiration (WANG et al., 2017). On the other hand, other
authors claim that the spatial variability in the isotope rainfall composition resulted
from an east-west precipitation dipole (CRUZ et al., 2009; SYLVESTRE et al., 1999;
CHENG et al., 2013) associated with an upper level wave response to enhanced
monsoon circulation, known as the Bolivian-High Nordeste Low system (LENTERS;
COOK, 1997; 1999, CHEN et al., 1999; SULCA et al., 2016). In this scenario, the
increased convective activity and precipitation over the western Amazon basin is
balanced by enhanced subsidence over Northeastern Brazil, leading to an
antiphased precipitation pattern between western and eastern Amazon, rather than
pervasive dry conditions.
In this scenario, the aim of this study is to clarify the role of the forest in the
Amazonian hydrologic cycle along the Holocene, capitalizing on dxs records. Based
on what is known so far about the climatic controls on the water stable isotopes and
current interpretations of the paleo-records, we proposed the following hypothesis:
H1: In continental forested areas, dxs of precipitation records recycled
precipitation.
H2: Large shifts in the forest cover modify the precipitation regimes across the
Amazon basin, leaving a distinct imprint in dxs.
17
To test the hypothesis, an empirical approach is taken, relying on δ18O and
dxs from rainfall monitored in the western Amazon basin. A description of transport
history of air masses using air parcel back-trajectory modeling is presented, together
with climatic and land data derived from satellite observations and reanalysis
products. Finally, a discussion of the paleoclimate implications of dxs variability along
the Holocene is proposed, based on published speleothem records, global vegetation
modeling and pollen from continental lake cores.
18
2. OBJECTIVES
The general objective of this thesis work is to identify possible effects of
recycled moisture on the isotopic composition of precipitation in order to improve the
climatic interpretations of speleothem records from the region.
2.1. SPECIFIC OBJECTIVES
The general objective of this work is approached in two phases. First, we
focus on modern observations of water stable isotopic composition of rainfall in the
north western Amazon basin, for what the following specific objectives were
establish:
• Constrain the effect of atmospheric circulation and regional precipitation on the
stable isotopic composition.
• Evaluate the effect of regionally recycled moisture on the stable isotopic
composition.
• Assess the possible influence of the regional forest cover on dxs.
• Evaluate the temporal variability of the relationship between δ18O and dxs.
On the second phase, we focus on published records of dxs in fluid inclusions
in speleothems from the northwestern Amazon and speleothem δ18Ocalcite across the
Amazon basin, for what the following specific objectives were established:
• Assess the dxs variability along the Holocene.
• Evaluate the variation of the regional land moisture contribution to precipitation in
the northwestern Amazon based on dxs variability.
• Evaluate the connection between δ18Ocalcite records across the Amazon basin
based on dxs variability.
19
3. THEORETHICAL BASIS
3.1. CLIMATE OF THE AMAZON AND GLOBAL IMPACTS
From the Andes cordillera to the Atlantic coast, the Amazon basin covers
approximately 6 200 000 km2, which is about one third of the continental mass of
south America (MARENGO et al., 2012). It hosts the largest tropical forest in the
planet and plays a major role in the global energy and hydrological budgets. The
potential energy derived from the large evapotranspiration through the forest drives
winds, affecting global Hadley and Walker circulations (MAKARIEVA et al., 2013;
BARICHIVICH et al., 2018). This implies that Amazonian precipitation responds to
the combination of large-scale atmospheric circulation and local water sources
(MARENGO; ESPINOZA, 2015).
Amazonian rainfall is large, about 2300 mm year-1 (DRUMOND et al., 2014).
The spatial distribution of rainfall varies largely in space, as shown in Figure 1a.
Rainfall regime is also heterogeneous across the basin, going from the ever-wet
north-western, with no distinct dry season, to the seasonal south-east, where dry
season last about five months (RESTREPO-COUPE et al., 2013). Figure 1b shows
the spatial distribution of the length of the dry season, given that in the Amazon, the
dry season is when precipitation falls below 100 mm month-1.
Figure 1. (a) Mean annual precipitation in mm for the period 1989-1995, based on the average of the products CRU, GPCC, GPCP and CPC and(b) number of months with less than 100 mm rainfall. Deforested (red dots) and non-forest (gray dots) areas are indicated.
Source: ZEMP, 2016.
Seasonal variations of precipitation in the Amazon basin are driven, in general, by the South American Monsoon Systems (SAMS). The SAMS is a seasonal mechanism that drives the moisture influx from the tropical north Atlantic to the continent, producing heavy rainfall over most of tropical South America during the
20
warm season (VERA et al., 2006). The features of the SASM in the lower atmosphere (~850 hPa) are the pervasive Low-Level Jet to the east of the Andes and the South Atlantic Convergence Zone (SACZ). In the upper troposphere (~300hPa) it appears the Bolivian High and the Nordeste Low. As a result, from late November through late February, the SASM sets the wet season for most of the Amazon basin. Contrasting conditions prevail during the dry season, from June to August. During this period, steady flow from East to West prevails in low and high atmospheric levels. The band of convection is restricted to the northern Hemisphere, reaching the northern tip of South America. Oceanic moisture input to the Amazon during the dry season is mainly from the tropical south Atlantic Ocean (DRUMOND et al., 2014), but it is less than the SAMS related influx. Climatological precipitation and atmospheric circulation are shown in Figure 2
Figure 2 for a mature phase SAMS month (January) and a month when SAMS
is absent (July).
Figure 2. Long term mean precipitation (shades) and wind circulation (vectors and stream lines). Left panels show low level (950hPa) wind circulation for (a) January and (b) July and right panels show high level (300hPa) wind circulation for (c) January and (d) July.
Source: GARREAUD et al., 2008.
Precipitation variability on interannual timescale is modulated by sea surface
temperature (SST) in the tropical Pacific and Atlantic oceans. In one hand, El Niño
Southern Oscillation (ENSO), rooted in the equatorial Pacific, affects precipitation in
21
the Amazon through atmospheric teleconnections (GARREAUD et al., 2008). Some
evidence of this are the severe droughts reported in 1912, 1926, 1983 and 1997-
1998, related to el Niño, and extreme floods in 2009, 2011 and 2012, related to la
Niña events (MARENGO; ESPINOZA, 2015). In the other hand, sea surface
temperature (SST) in the tropical Atlantic regulates directly the moisture influx to the
continent. For instance, the droughts of 1964 and 2005 were related to warm SST
anomalies in the tropical north Atlantic due to a northward displacement of the
Intertropical Convergence Zone (ITCZ) and the 2009 flooding, associated with warm
tropical South Atlantic SST (MARENGO; ESPINOZA, 2015, BARICHIVIC et al.,
2018).
Recent observations suggest that the impact of the Atlantic Ocean in the
Amazonian precipitation is even more important than the Pacific Ocean influence.
Wang et al. (2018) show the leading role of the Atlantic Ocean in the intensification of
the wet season precipitation over the equatorial Amazon for the last 30 years,
concomitant with increasing Atlantic’s SSTs. SST relatively warmer conditions
enhance moisture transport from the Atlantic Ocean to the Amazon, favoring
convergence, which results in intensified precipitation, especially in the equatorial
area. Furthermore, Barichivich et al. (2018) suggest that the severe floods in the
Amazon during the last decades, were driven by the strengthening of the Walker
circulation, which in turn is a response to the recent tropical Atlantic warming (LI et
al., 2016).
Mean annual low-level temperature in the Amazon basin is ~20°C, with a slight
thermal gradient from the Equator to the poles (GARREAUD et al., 2008). In general,
the annual cycle exhibits maximum temperatures in austral summer and minimum
temperatures in austral winter, with less seasonal variation near the Equator. The
exception is the southern Amazon, where the warmest temperatures appear during
austral winter, which is explained by the summer time cloud development that
shades the surface from sunlight (GARREAUD et al., 2008).
3.2. THE AMAZON, A CONTINENTAL MOISTURE SOURCE
Continental moisture is very important for the Amazonian water budget, as it
induces precipitation, especially during the dry season. The basic mechanism is
known as “moisture recycling” (Figure 3), which refers to the fraction of precipitation
22
originated form evaporation in land, within a determined region. Precipitated water
can re-evaporate and precipitate again downwind in a cycle that repeats along the
advection pathway, featuring what is referred to as “cascading moisture recycling”
(ZEMP et al., 2014; ZEMP et al., 2016; STAAL et al., 2018). A schematic
representation of moisture cascading in the Amazon basin is presented in Figure 6. It
also occurs that atmospheric moisture just flows across the continent towards areas
where convergence ensures that precipitation occurs, without exchange between
vegetation and atmosphere on the way (TRENBERTH, 1998). A combination of
these processes drives rainfall distribution in the Amazon.
Moisture recycling derives from two processes. One is the physical
evaporation from open water, soil and canopy, which in this work is referred to as
evaporation. The other process is transpiration, which is the biological evaporation
trough the leave stomata. It is worth noting that there are different terms to refer to
these processes depending on the study line. For example, in the ERA Interim
atmospheric reanalysis, the variable "evaporation" accounts for the sum of
evaporation and transpiration, while in the Large-Scale Biosphere Atmosphere
Experiment in Amazonia (LBA) project, that same variable is referred to as
evapotranspiration or latent heat flux (LE) (DA ROCHA et al., 2009). Therefore, it is
important to clarify that in this work we use the term evapotranspiration to refer to the
sum of evaporation and transpiration.
Figure 3. (a) Mean annual evapotranspiration in mm for the period from 1989 to 1995 from forty products. including field observations, reanalysis and model outputs. (b) Precipitation recycling defined as the fraction of rainfall that comes from continental evapotranspiration and vertically integrated moisture fluxes (black arrows). Deforested (red dots) and non-forest (gray dots) areas are indicated
Source: ZEMP, 2016.
Mean annual evapotranspiration for the Amazon is shown in Figure 3.
Quantifications of evapotranspiration based in numerical modeling estimate in
23
average 3.9 mmd-1(~1420 mm yr-1), while estimations based on observations range
from 3.1 to 3.7 mm d-1(1130 to 1350 mm yr-1) (SHUTTLEWORTH, 1988; DA ROCHA
et al., 2004; VON RANDOW et al.,2004; HUTYRA et al., 2007). Evapotranspiration
rates vary across the Amazon basin, with mark differences expressed during the dry
season. While the areas south and east borders, show decreased evapotranspiration
due to soil moisture limitation during the dry season, towards the north and west,
evapotranspiration rates increase, mainly controlled by atmospheric evaporative
demand (DA ROCHA et al., 2009). The reason is that the broadleaf forest with deep
root systems in the northern and western Amazon is capable to extract water from
deep soil layers, maintaining high transpiration rates, unlike the south and east
Amazon, that is dominated by Transitional forest and Cerrado (NEPSTAD et al.,
1994; OLIVEIRA et al., 2005).This relation with vegetation highlights the importance
of tree transpiration in the Amazonian water budget. Indeed, tree transpiration is key
to maintain wet conditions when oceanic moisture supply is limited. This buffer effect
is well documented in Staal et al., (2018), where it determines that the transpiration
rate that produce precipitation in the Amazon, increases as oceanic moisture influx
diminishes (Figure 4). In the Amazon, tree transpiration is a notable portion of the
evapotranspiration flux, inducing in average 20% of the annual rainfall (Figure 4),
although it reaches values up to 50% in the western end of the basin (Figure 5).
Figure 4. a) Schematic representation of tree transpiration recycling. On average 20% of rainfall in the Amazon has been transpired at least once. About half of this transpiration (51%) is recycled in the first cycle, the remainder occurs after multiple re-evaporation cycles (moisture cascading). b) Transpiration recycling ratio in the Amazon against average rainfall from oceanic origin. 2005 and 2010 are labeled because vast drought was registered in vast areas across the Amazon.
Source: STAAL et al., 2018.
24
Figure 5. a) Mean annual transpiration by trees that precipitate over land. b) Fraction of mean annual rainfall that has been transpired by trees in the Amazon basin. The Amazon basin is shown by black outline.
Source: STAAL et al., 2018.
The forest-atmosphere interaction has been evidenced in different studies. An
evaluation of rainfall sensitivity to forest cover showed that air flowing over extensive
vegetated areas produce at least twice more precipitation after some days in
comparison with air flowing over less vegetated areas (SPARKLEN et al., 2012). In
addition, changes in rainfall might also affect forest stability. For instance, the
observed alteration of canopy structure and water content in southwestern Amazonia
and slow recovery after the drought in 2005, suggests that frequent droughts (every
5-10 years) might permanently alter the forest canopy (SAATCHI et al., 2013). In fact,
during the last decades, recurrent anomalous droughts and floods have been
reported. Furthermore, hydrological data show trends towards more extreme events
in the Amazon (MARENGO; ESPINOZA, 2015). Although all model projections don’t
agree, a possible scenario is that water stress over long periods will lead to
significant forest loss which, at the same time, could enhance the intensification of
the water cycle and further forest loss through feedbacks between forest and rainfall.
A schematic representation of this process is shown in Figure 6.
.
25
Figure 6. Schematic representation of cascading recycling effects in the vegetation-rainfall system. a) System in equilibrium. b) Initial forest loss triggered by decreasing oceanic moisture influx. c) Altered rainfall regime in another location, leading to further forest loss and reduced moisture transport.
Source: ZEMP et al., 2016.
3.3. AMAZONIAN FOREST COVER SHIFTS: THE EVOLUTION OF THE FOREST
ALONG THE HOLOCENE
The Holocene Epoch is the youngest subdivision of the Geological Time
Scale, representing the current interglacial Period. Different from other chrono-
stratigraphic divisions, the boundary of the Holocene defines a climate-
stratigraphically division. Fixed in the oxygen stable isotope record of the North
Greenland Ice Core Project (NGRIP) (SINHA et al., 2013), the Holocene starts
officially at 11.700 kyr BP (Before Present refers to 1950 CE) marked by a positive
δ18O excursion related to the end Younger Dryas cold event, the last cold period
recorded during the deglacial period (SINHA et al., 2013; RASMUSSEN et al., 2014).
The Holocene has been informally subdivided into three subseries/subepochs Early-
Middle- and Late Holocene. SINHA et al. (2013) proposed that the traditional
subdivision of the Holocene would represent major changes in the North Atlantic
climate associate to major changes in the thermohaline circulation, roughly
correspondent to the Bond events (BOND et al., 1997; BOND et al., 2001). On 2018
the International Subcommission on Quaternary Stratigraphy (ISQS) ratified the
26
subdivision of the Series/Epoch of the Holocene in three Stages/Ages (WALKER et
al., 2018):
▪ Stage/Age Greenlandian: equivalent to the Lower/Early
Subseries/Subepoch Holocene the Greenlandian spans from 11.653 ±
0.004 to 8.276 Kyr BP;
▪ Stage/Age Nortgrippian: equivalent to the Middle/Mid Holocene
Subseries/Subepoch the Nortgrippian spans from 8.276 to 4200 Kyr
BP;
▪ Stage/Age Meghalayan: equivalent to the Upper/Late
Subseries/Subepoch the Nortgrippian spans from 4200 Kyr BP to the
present;
Although the proposal coined by the ISQS is still a bit controversial, for
instance the Meghalayan lacks a global expression (VOOSEN, 2018), this new
subdivision offers a formal ratification whose boundaries roughly resemblance the
informal subdivision frequently used in the scientific publications. In face of that, in
this study is used the Subseries Early, Mid- and Late Holocene following the
boundaries defied by ISQS subdivisions of 2018 (WALKER et al., 2018).
In general, the Holocene is characterized as period of reduced climate
variability when compared with the glacial climate oscillations. However, at orbital
time-scale, changes in the hydrologic regime may have imprinted considerable
changes in the forest due to the close relation between the climate and vegetation.
Therefore, the evolution of the Amazonian forest is a major issue for paleoclimate
studies in South America.
Reconstructions of vegetation composition were inferred from variations in the
type of pollen present in the phases of well dated lake cores across the Amazon. The
information that pollen brings is the biomes present at one time, whether it is Rain
forest, Seasonal forest or open Savanna. The general principal is that some types of
vegetation can only exist in certain climate conditions. For instance, pollen records
from lake cores in the Colombian Andes suggest that, during glacial conditions,
vegetation belts migrated to lower altitudes, following warmer temperatures (VAN
DER HAMMEN, 1974; VAN DER HAMMEN; ABSY, 1994; WILLIE et al., 2004).
27
However, paleoenvironmental reconstructions are complicated and
controversial. In one hand, the early study made by Colinvaux et al. (2000) based on
pollen analysis from Serra dos Carajás and Pata Lake (eastern Amazon basin)
suggests that the Amazon lowlands remained under forest, with a remarkable
stability throughout the Pleistocene. That study argues that the forest didn´t expand
or regressed, but some plant species were replaced by others with better adaptation
to the warming climate. In the other hand, evidence based mostly on lake cores and
speleothem records suggest that the Amazon forest has undergone dramatic shifts
concomitant with wet and dry conditions along the Quaternary (FONTES et al.,
2017). Indeed, what observations suggest is the general expansion of the Amazon
rainforest throughout the Holocene, interrupted by regression periods.
The Early Holocene (~11.7 to 8.3 kyr BP) is recognized as a period of forest
expansion and increased monsoon precipitation in the Amazon basin when
compared with the LGM, conditions. Speleothem records form Paraíso cave (WANG
et al., 2017), located in the eastern edge of the Amazon basin, and lake records
based on pollen and biogeochemical proxies from the same region show wetter
conditions during the Early Holocene in comparison with the LGM (BEHLING et al.,
2002; CORDEIRO et al., 2011). At the same time, δ18O records from speleothems
from the western Amazon basin (El Condor and Cueva del Diamante) show the
opposite, indicating moderately dry conditions during the Early to Mid- Holocene int
that region (CHENG et al., 2013). The slight rainfall reduction is attributed to reduced
intensity of the SAMS during the Early Holocene.
During the Mid-Holocene (~ 8.3 – 4.2 kyr BP) the South Hemisphere summer
insolation (December, January and February) underwent an increase of near 3 %
departing from ~ 882 W/m2 at about 8 kyr BP to 913 W/m2 near 4 kyr BP. Several
evidences appoint to dry conditions in the Amazon. A gap in the sedimentation hiatus
in the Comprido Lake from 7.8 to 3.0 Kyr BP (MOREIRA et al., 2012). Furthermore,
in the eastern Amazon, at Serra dos Carajás lake records showing increase in
charcoal deposition between 7.4 to 4.7 kyr BP points to episodes of large biomass
burning which reinforces the notion of dry climate (CORDEIRO et al., 2008). Also,
lake records from southwestern Amazon basin show similar results from 7.2 to 3.3
kyr BP (BUSH et al., 2007). However, there is significant variability in the chronology
of these dry envents. Furthermore, the dryness observed in the lake records from
28
eastern (MOREIRA et al., 2012) and western (BUSH et al., 2007) Amazon basin,
contrasts with the expansion in the várzea/igapó forest at ca. 7.7 kyr BP associated
with the rise of the Amazon water level and the formation of Calado Lake at central
Amazon (BEHLING et al., 2001).
The Late-Holocene (from 4.2 kyr BP to the present) is well characterized as
period of increase in humidity and forest expansion. However, the geographic
patterns of climate change and vegetation variation along the Late-Holocene is still
controversial. Speleothems records from the Amazon basin suggests a high east-
west precipitation dipole during the Late-Holocene. While speleothems from Paraíso
cave (eastern Amazon) shows stepwise δ18O enrichment along the Late Holocene
(from 4.2 to the present) speleothem records from western Amazon basin from Tigre
Perdido cave shows the opposite trend (VAN BREUKELEN et al., 2008). The same
has been observed by Cheng et al. (2013) using speleothems from Peruvian
Amazon. They explain the east-west precipitation dipole with the intensification of the
SAMS, which results in enhanced convection and upward motion in the western
Amazon which, in turn, enhances the Nordeste low, resulting in subsidence and dry
conditions in the eastern side. Moreover, in the western Amazon basin, pollen
records from Parker, Gentry and Werth Lakes show an increase in arboreal pollen
(BUSH et al., 2007) after 3.7 kyr BP. Also, pollen records from the eastern Amazon
(CSS 2 Lake, Serra dos Carajás) (ABSY et al., 1991; SIFEDDINE et al., 2001) and
Saci Lake (FONTES et al., 2017) shows a remarkable expansion of the rainforest
during the Mid-Holocene, reaching their maximum after 4.2 kyr BP, during the Late-
Holocene.
A recent study by Maksic et al., (2018) assess the climate conditions and
vegetation shifts along the Holocene through numerical simulations with an
atmospheric model coupled with surface model. Simulations show that the tropical
Atlantic SST is the most important control on the Amazonian precipitation, rather than
the orbital forcing. They show reduced intensity of the SAMS during the Holocene
and gradual intensification the towards the present. Moreover, the results show that
vegetation changes follow moisture availability, mainly controlled by the SAMS, in an
heterogenous way across the Amazon. Thus, rainforest persisted in the western
Amazon throughout the Holocene, even with reduced precipitation compared with
present. At the same time, there was an expansion of savanna and seasonal forest in
29
the eastern Amazon, and fluctuations between different biomes in the ecotone areas,
as shown in Figure 7.
Figure 7. Vegetation distribution map for different periods of the Holocene, from numerical simulation. The model used from the Center for Weather Prediction and Climate Studies (CPTEC) is the Potential Vegetation Model version 2 (CPTEC-PVM2). Two experiments were performed for the 6K period, one with present SST (6k-pvm veg type) and the other with mid Holocene SST (6k-pvm-MH-sst veg type).
Source: MAKSIC et al. (2018).
3.4. WATER STABLE ISOTOPOLOGUES IN THE WATER CYCLE
Water stable isotopologues are widely used as tracers in the hydrologic cycle.
Their concentrations are given by the ratios 2H/1H and18O/16O and expressed in δ
notation with respect to a global standard value as δ=(Rsample/Rstandard)-1 (CRAIG,
1961). The accepted standard is Vienna Standard Mean Ocean Water (VSMOW).
Because relative differences between the isotopic ratios are small, δ values are
expressed in parts per thousand (‰).
In the hydrologic cycle, isotope fractionation occurs with transitions from one
physical state to another and transport processes. Most of the fractionation in the
hydrologic cycle is mass-dependent and is of two types: thermodynamic (when the
system is equilibrium) and kinetic (one-way reaction).The fractionation rate is given
by the fractionation factor α, which in turn is defined by the equilibrium constant of the
30
reaction (K) and diminishes exponentially with increasing temperature (MELANDER,
1960).
In the tropical region, Rayleigh distillation is a major driver of the isotopic
composition of precipitation over continental areas. Rayleigh distillation is the
process by which a moist air parcel progressively releases heavy isotopes through
precipitation, as it flows over the continent, leaving the remnant vapor isotopically
lighter (Figure 8). In the Amazon basin, the heavy isotope loss is compensated by
recharge from land evapotranspiration, isotopically heavier than the atmospheric
vapor originally present.
Figure 8. Simplified Rayleigh distillation scheme in the water cycle.
Source:
http://www.geo.cornell.edu/geology/classes/Geo656/656notes11/IsotopeGeochemistryChapter8.pdf
3.5. DEUTERIUM EXCESS: A PROXY FOR RECYCLED MOISTURE?
Another approach is based on the isotopic composition of rain, specifically
deuterium excess (dxs = δD – 8 × δ18O) (DANSGAARD, 1964), which is a second
order parameter, specifically sensitive to the conditions during evaporation from the
source (MERLIVAT; JOUZEL, 1979; JOHNSEN et al., 1989; PFAHL; WERNLI,
2008). Physically, dxs reflects the slower movement of the H218O molecule during
diffusion, leading to a relative enrichment of the H2HO molecules in the phase with
weaker bonds (e.g.: gas phase during evaporation). This slower movement could
lead to measurable differences when there is not enough time for the phases to
reach isotopic equilibrium. During evaporation, kinetic conditions result from a strong
relative humidity gradient over the water surface and advection before reaching
equilibrium between the two phases (PFAHL; SODEMANN, 2014).
31
In one hand, the main application for dxs is the identification of oceanic
moisture sources. For instance, Araguas-Araguas et al. (1998) found a difference
around 5‰ between dxs values of rain in Southeast Asia associated to moisture
fluxes from the Pacific and Indian monsoons. Moreover, Masson-Delmotte al., (2005)
identified southward shifts in Greenland oceanic moisture source during cold events,
based on the relation between dxs and SST at millennial time scale.
In the other hand, observations over the continent point that dxs is related to
moisture recycling across the forest (SALATI et al., 1979). In fact, several studies
have used dxs to estimate the fraction of recycled moisture (e.g. KONG et al., 2012).
However, for practical reasons, only few models consider vegetation effect in its
simulations of the water cycle (LAI; EHLERINGER; 2010). Although there are
significant advances in the understanding of dxs, it is still necessary to clarify what if
implies for the water cycle in tropical regions.
3.6. SPELEOTHEMS AND FLUID INCLUSIONS
Speleothems present several advantages that make them remarkable tools in
paleo-climate studies. For instance, they form in stable cave environments, are
geographically widely distributed and can be precisely dated by U-series mass
spectrometry. As rainfall infiltrates the soil and epikarst, it uptakes trace elements and
dissolved calcite that will later form the speleothems through dripping and calcite
precipitation in the cave environment. Thereby, the isotopic composition of rainfall
incorporates in the calcite as the speleothem grows. Oxygen and carbon isotope
records from speleothem calcite (δ18Ocalcite, δ13Ccalcite) have been extensively used in
Paleoclimate and paleoenvironmental reconstruction of tropical areas (e.g. BURNS
et al., 2001; GRIFFITHS et al., 2010; CHENG et al., 2013; APAÉSTEGUI et al., 2014
WANG et al., 2017).
Another technique to work with speleothems is the analysis of the fossil drip
water trapped in the speleothem calcite as “fluid inclusions”, as shown in Figure 9.
The advantage of fluid inclusions is that they provide δ18O and δD composition of the
dripping water that formed the stalagmite, providing a unique opportunity to access
the past rainfall water. Since drip water isotopic composition is believed to reflect that
of rainfall recharging the cave aquifer, fluid inclusions in stalagmites provide temporal
records of rainfall isotope variations which can be related to changing rainfall patterns
32
through time (MCDERMOTT et al., 2006; VONHOF et al., 2007). Fluid inclusion
analysis of δD (in combination with corresponding δ18Ocalcite) has been used to
reconstruct paleo-temperatures (VAN BREUKELEN et al., 2008) and changes in the
source and amount of rainfall (e.g. FLEITMANN et al., 2003b; HARMON et al., 1979;
MCGARRY et al., 2004; SCHWARCZ et al., 1976).
Fluid inclusion analysis also offer the possibility to explore past dxs variability
which, as mentioned above, was suggested as a moisture recycling tracer over the
Amazon, although this interpretation has not yet been tested. Therefore, this thesis
work will attempt to interpret this isotopic tracer in fluid inclusions along the Holocene
in the northwestern Amazon.
Figure 9. Images of typical fluid inclusions in the speleothems analyzed in VAN BREUKELEN et al., 2008.
Source: VAN BREUKELEN et al., 2008
3.7. MONITORING SITE
As part of the paleoclimate reconstructions in South America, van Breukelen
et al. (2008) analyzed fluid inclusions in two precisely dated stalagmites, covering the
Holocene. These stalagmites were collected in the Tigre Perdido cave (Rioja, San
Martín, Perú), located in northeast Perú at ~1000 m.a.s.l., in the transition region
between the Amazon lowlands and the Andes. In the same region, near the Palestina
cave (5.92°S, 77.35°W, 870 m. a. s. l.), a rain collector for was placed in 2011 by
Apaéstegui et al. (2014) and reactivated for the present study. The location of the
monitoring site is presented in Figure 10.
The present-day climate is tropical humid with a mean annual temperature of
22.8°C and mean precipitation of ~1570 mm year-1, measured at the meteorological
33
station of Rioja (APAÉSTEGUI et al., 2014), located 15 KM from the Palestina
station. Precipitation regime features a long, wet season (precipitation higher than
100 mm/month) from September to May, with peaks in November and March (Figure
10).
Figure 10. Location of the Palestina station and precipitation climatology. Dark blue shadows represent the Andes mountain range and the green line, the limit of the Amazon basin taken from the HyBAM data base. Precipitation at the Palestina station was obtained from TRMM 3B42 from 1998 to 2018.
Source: AMPUERO, 2019
J F M A M J J A S O N D0
50
100
150
200
250
Local pre
cip
itation (
mm
month
-1)
34
4. MATERIALS AND METHODS
4.1. ISOTOPE MONITORING
The monitoring was made in three stages at the Palestina station. The first
stage of the monitoring goes from 06/2012 to 09/2013 by Apaéstegui et al. (2014).
The second and third stages on the monitoring were performed in the frame mark of
this study and go from 10/2013 to 05/2014 and from 10/2016 to 05/2018. To fill the
gap from July to September 2016, data from the Pomacochas station (5.84°S,
77.97°W, 2257 m.a.s.l.) was used, only for the analysis on interannual timestep.
Although there is a significant height difference, the stations are located close to
each other and the records are remarkably similar (Figure 11).
Figure 11. Records of δ18O (upper panel) and dxs (lower panel) from the stations Palestina (black) and Pomacochas (red) on biweekly timestep.
-20
-16
-12
-8
-4
0
18O
(‰
, V
SM
OW
)
Palestina
Pomacochas
1/05
/201
2
1/08
/201
2
1/11
/201
2
1/02
/201
3
1/05
/201
3
1/08
/201
3
1/11
/201
3
1/02
/201
4
1/05
/201
4
1/05
/201
6
1/08
/201
6
1/11
/201
6
1/02
/201
7
1/05
/201
7
1/08
/201
7
1/11
/201
7
1/02
/201
8
1/05
/201
8
0
4
8
12
16
20
dxs (
‰, V
SM
OW
)
Source: AMPUERO, 2019.
The rainwater sampling followed the guidelines given by the Global Network of
Isotopes in Precipitation (GNIP) from the International Atomic Energy Agency (IAEA).
A tube-dip-in-water collector with pressure equilibration was used (Figure 12). This
35
collector works as follows: Rain enters the collector trough the funnel on top and
passes through a tube that goes to the bottom of the container. As water fills the
collector, only the area inside the tube is exposed, which limits evaporation. Pressure
equilibration with the atmosphere is attained with a 2 mm hole on top of the
container. The hole is connected to a 6 m long hose, long enough to avoid
exchanges with the atmosphere. This system permits accumulation and preservation
of rain fallen within the sampling period. The samples were taken twice a month
using 8mL HDPE and amber glass bottles.
Figure 12. Scheme of the rain collector used in this study.
Source: Adapted from the GNIP manual.
Water analysis were performed at the CEPAS - Centro de Pesquisas de
Águas Subterrâneas at the University of São Paulo (IGc-USP), using a PICARRO
L2130i water analyzer and processed by LIMS for Lasers software. Values are
reported with an analytical precision of 0.09‰ for δ18O and 0.9‰ for δD relative to
Vienna Standard Mean Ocean Water (VSMOW). In total values from 86 samples
were obtained and used in this study.
4.2. CLIMATE DATA
We used satellite derived precipitation from two products to cover the
monitoring period. Precipitation data for the period from 05/2012 to 03/2014 is from
the Tropical Rainfall Measuring Mission (TRMM; http://trmm.gsfc.nasa.gov/) multi-
satellite3B42 V7 daily product, available at 0.25° x 0.25° resolution. The 3B42
36
product combines data from TRMM and other satellites to estimate precipitation
(HUFFMAN et al., 2007). In the beginning of April 2014, the instruments on TRMM
were turned off, so another precipitation product was necessary to complete the
monitoring period. For this reason, precipitation data for the period from 04/2014 to
05/2018 from the Integrated Multi-satellitE Retrievals for GPM (IMERG)(GPM;
https://pmm.nasa.gov/data-access/downloads/gpm) was used. Daily data is available
in 0.1º x 0.1º resolution with full coverage over 60ºN - 60ºS. GPM data was upscaled
spatially applying linear interpolation in order to match TRMM resolution (from 0.1° to
0.25°).
Data on Leaf Area Index (LAI) was used to represent the structure of the forest
canopy. In general, LAI is a dimensionless variable that refers to the area of
photosynthetic tissue per unit ground surface area. However, the exact definition
depends on the technique used to measure it (JONCKHEERE et al., 2004). In this
study, LAI is derived from measurements made by the Moderate Resolution Imaging
Spectroradiometer (MODIS), from the National Aeronautics and Space
Administration’s (NASA). Therefore, in LAI is defined as the one-sided green leaf
area per unit ground area in broadleaf canopies. The product used is MCD15A2H
version 6 MODIS Level 4 with 500 m pixel size
(https://lpdaac.usgs.gov/dataset_discovery/modis/modis_products_table/mcd15a2h_
v006) (MYNENI et al., 2015). The algorithm chooses the best pixel available from all
the acquisitions of both MODIS sensors located on NASA’s Terra and Aqua
satellites. The retrievals are obtained every 8 days, by averaging the best quality
values. For this study, we minimized biases by screening out pixels produced with
low confidence using the quality layer included in the product. Specifically, pixels
where bits from 5 to 7 were 011 and 100 were ruled out. Details on the bit coding are
presented in Table 1. The accepted pixels were spatially interpolated to a 0.25° grid.
There is still debate about the interpretation of LAI regarding the apparent
greening in the Amazon during the dry season. Morton et al. (2014) argues that the
apparent greening observed in optical remote sensing data is an artifact of variations
in the sun-sensor geometry and concludes that moisture availability governs
photosynthetic activity in the Amazon. However, evidence that includes modeling and
field observations appoint that, in the Amazon, LAI seasonality identified with MODIS
is real, but the amplitude of the seasonal changes might not yet be precise
37
(SALESKA et al., 2016). These evidences support that sunlight is the limiting control
on photosynthetic activity in the Amazon. A comprehensive analysis of LAI is
reported in Myneni et al. (2007) where all the biases are analyzed, showing the
consistency of the database and appointing a good relationship with
evapotranspiration.
Table 1. Values of the quality layer for LAI/FPAR (8bit) in MODIS.
Source: MODISCollection6(C6) LAI/FPAR Product User’s Guide.
Precipitation recycling (PR) was calculated using the Eulerian atmospheric
moisture tracking model WAM-2layers (Water Accounting Model – two layers)
version 2.4.8 (VAN DER ENT, 2014). The model was run with precipitation, land
evaporation and wind data from the European Centre for Medium-Range Weather
Forecasts (ECMWF) Interim Re-Analysis (ERA-Interim) (DEE et al., 2011), on a 1.5°
grid. The the first year from the results was omitted due to model spin-up. The result
in each grid cell is the ratio of precipitation that originates from land evaporation.
38
Data on Vertically Integrated Divergence of the Moisture Flux (VIDMF) from
ERA Interim was used for complementary analysis. This variable is computed with
the specific humidity (q), which is a measure of the atmospheric moisture, and the
orthogonal components of the wind field (u and v), integrated from 1000 hPa to 300
hPa levels in the atmosphere. VIDMF is often used to identify atmospheric moisture
sources and sinks (e.g. DRUMOND et al., 2014), so that the positive values point
moisture sources or surface evaporation regions, and the negative values, moisture
sinks or precipitation regions.
4.3. PALINOLOGYCAL RECORDS
A rain forest index was calculated based on available pollen frequency records
from five lakes in the eastern and western edges of the Amazon: Iriri, Altamira - PA
(SANTOS, 2019); Ilha Arapujá, Altamira - PA (SANTOS, 2019); Laguna Chochos,
Peru (BUSH et al. 2005); Saci, central-south Amazon (FONTES et al., 2017) and
Serra Sul Carajás Lake, CSS2 (ABSY et al., 1991; SIFEDDINE et a., 2001). The
records are shown in Figure 28 and the geographic location of the lakes in shown in
Figure 25. The records were interpolated in time in order to calculate the average
(Figure 27).
4.4. METHODS
Local precipitation on daily time step was estimated based on TRMM and
GPM products. To obtain the local precipitation values, the average of 25 tiles was
calculated, considering the tile over the Palestina station in the center and the
adjacent tiles around it.
Air parcel back-trajectories were modeled with the HYbrid Single-Parcel
Lagrangian Integrated Trajectory 4 (HySPLIT 4) model (STEIN et al., 2015; ROLPH
et al., 2017). The model was run with wind fields from Era Interim on native spatial
resolution of 0.75°. Trajectories were tracked back in time for seven days. As ten
days is the mean residence time of water vapor in the atmosphere (NUMAGUTI,
1999), it was considered that seven-day back trajectories cover an air mass pathway
back to the point of the last saturation, hence, the relevant isotopic fractionation
processes along the atmospheric moisture transport (HURLEY et al., 2012). Back-
trajectories were initiated at the Palestina station at 500, 1000, 1500 and 2000
39
meters above ground level (m.a.g.l.), representing the boundary layer and the lower
troposphere, where the main moisture transport to our study site takes place. Back-
trajectories were computed at 0, 6, 12 and 18 UTM from May 2012 to May 2018.
All the climate data (precipitation, LAI and PR) were summed up along the
path of the back-trajectories, excluding the grid cells over the ocean. The schematic
representation of the selection of tiles along the back-trajectories is presented in
Figure 13. Thereby, the daily indices of degree of rainout upstream (DRU),
accumulated LAI upstream and Precipitation recycling upstream (PR) were
computed. The resulting units for DRU are mm traj-1 while LAI and PR are
dimensionless. Similar approach has been used in other studies (FIORELLA et al.,
2015; BAKER et al., 2016).
Figure 13. Schematic representation of one back-trajectory (black arrow). The back-trajectory is divided in seven segments, each one covering one day. The beginning of the back-trajectory is marked with a star at the Palestina station on the most recent day (0), and then progresses back in time until completing 7 days (-7). Climate data on precipitation, LAI and PR was selected from the tiles spanned by the back-trajectory on the corresponding day. In this example, on the sixth day (between -5 and -6), the trajectory goes over the ocean (blue shadow), so information on those tiles is excluded. The dashed line represents the coast line.
Source: AMPUERO, 2019.
Daily indexes were averaged across the four initialization times and vertical
levels mentioned above. To better assess the effect of these indexes in the isotopic
signature, all values were weighted by local precipitation on bi-weekly timestep,
which is the frequency of the water isotopologue sampling. This approach permits us
to better estimate the extent by which the indices affect the isotope values of rainfall
at the monitoring site. For instance, by weighting DRU by the local precipitation, a
high value of DRU associated with an event of reduced local precipitation will have a
minor importance given that it corresponds to a small fraction of the total water that
reached the study site. Hence, this approach allows us to estimate how much of the
40
isotopic signals can be related to DRU, LAI and PR of our monitoring site. From here
on, DRU, LAI and PR will be used to refer to the bi-week weighted values of the
accumulated data along back-trajectories. Moreover, as we know that our isotopic
composition data and calculated parameters (DRU, LAI and PR), exhibit a seasonal
cycle, seasonality was removed by means of calculating anomalies with respect to
the monitoring period from June 2012 to May 2018. For example, within that period,
there are six values of each calculated parameter for the 1st of July and four values
of the isotopic composition due to the monitoring gaps. The average of those six or
four values, corresponding to the same day and month every year, was calculated for
each parameter and then subtracted from each value to get the anomalies.
To assess atmospheric transport variability, cluster analysis of the back-
trajectories was performed with the package included in HySPLIT4. This cluster
analysis technique is more accurate than the monthly clustering because low-level
large-scale circulation in the Amazon vary on intra-seasonal time scale (PACCINI et
al., 2017). This means that, although a circulation pattern may be more frequent
during one season, it can appear during other periods of the year as well. Thereby,
with the cluster analysis it is possible to group and visualize the back-trajectories
more clearly than with the monthly approach. Moreover, as part of the clustering
process, each back-trajectory is tagged with a cluster number, which permits to
calculate the frequency of occurrence of each cluster. It is also possible to relate
each trajectory with the amount of local rainfall that fell on the starting day of the
back-trajectory. Based on this association, the amount of local rainfall that co-occur
with each cluster was estimated. This study presents only clusters of back-
trajectories initiated at 1500 m.a.g.l. at 12:00 UTM, as considering other heights and
time lead to similar results.
41
5. RESULTS
5.1. PRECIPITATION AND BACK-TRAJECTORY ANALYSIS
Local precipitation accounts for ~1570 mmyear-1 and corresponds to a tropical
regime with a marked peak in March and dry season from July to August
(precipitation <100 mm month-1). Because dry season precipitation is significant at
our site, it is important to consider its influence on the mean annual δ18O and dxs
composition of rainfall.
Trade winds dominate large-scale atmospheric transport, varying from
northeast during austral summer to southeast during austral winter. Eleven clusters
of back-trajectories summarize the atmospheric transport (Figure 14). Air masses
crossing the Amazon basin (clusters 2, 3, 4, 6 and 7) are associated with 40% of
annual rainfall at the collecting site, while 34% result from northeast trajectories
(clusters 1, 5 and 11) and 14.5% from short-length trajectories (cluster 8) that appear
to be related to regional-scale transport. Although trajectories originating over the
Pacific Ocean (cluster 10) account for 8.5% of the annual rainfall, it is unlikely that
those air masses transport moisture across the Andes. Instead, associated
precipitation likely results from local moisture. Precipitation associated with
extratropical cold fronts, known to significantly affect precipitation in the south-central
Peruvian Andes (HURLEY et al., 2015) are associated with the remaining 3% of local
rainfall (cluster 9).
Based on daily data from back-trajectories and precipitation it is possible to
compute the degree of rainout upstream (DRU) associated with each rain event. In
general, the DRU annual cycle is similar with the one of local precipitation, although
there is an offset from March to July (Figure 15). The differences become larger on
shorter time scales; in fact, at biweekly time steps, DRU and local precipitation are
correlated with r=0.51, p<0.01. Those differences arise because DRU integrates
areas with different precipitation regimes across the continent, while local
precipitation reflects only the local regime.
42
Figure 14. Clusters of seven-day back trajectories at 1500 m.a.g.l. from June 2012 to May 2018.The bars under each map show the frequency of back-trajectory clusters per month.
(Continue)
J F MA M J J A S ON D
0
10
20
30
40
50 C1
J F MA M J J A S ON D
0
10
20
30
40
50 C2
J F MA M J J A S ON D
0
10
20
30
40
50 C3
J F MA M J J A S ON D
0
10
20
30
40
50 C4
J F MA M J J A S ON D
0
10
20
30
40
50 C5
J F MA M J J A S ON D
0
10
20
30
40
50 C6
Source: AMPUERO, 2019.
43
Figure 14. Clusters of seven-day back trajectories at 1500 m.a.g.l. from June 2012 to May 2018.The bars under each map show the frequency of back-trajectory clusters per month.
(End)
J F MA M J J A S ON D
0
10
20
30
40
50 C7
J F MA M J J A S ON D
0
10
20
30
40
50 C8
J F MA M J J A S ON D
0
10
20
30
40
50 C9
J F MA M J J A S ON D
0
10
20
30
40
50 C10
J F MA M J J A S ON D
0
10
20
30
40
50 C11
Source: AMPUERO, 2019.
44
Figure 15. Annual cycle of the degree of rainout upstream (DRU) and local precipitation from 2012 to 2018. DRU is the accumulated precipitation along the back-trajectories that initiate on precipitation days at the Palestina station. Local precipitation is calculated from TRMM3B42 and GPM by averaging the nearest tiles to the Palestina station.
J F M A M J J A S O N D0
200
400
600
800
1000
1200
1400D
RU
(m
m tra
j-1
)
0
50
100
150
200
250
Lo
ca
l p
recip
ita
tio
n (
mm
mo
nth
-1)
Source: AMPUERO, 2019.
5.2. FOREST MOISTURE FLUXES
Accumulated Leaf area Index values along the back-trajectories (LAI) present
the lowest values in February and increase continuously until reaching a peak around
August and September (Figure 16). Although cloud cover during the rainy season
might lead to lower than expected LAI, leading to a negative relationship with
precipitation, LAI is not consistently correlated with DRU. This could be related to the
fact that the main moisture transport to our monitoring site does not consistently
cross the region of most intense convection upstream. For instance, during austral
summer, strongest convection locates over the central and southern part of the
Amazon basin, while advection to our site occurs farther north. The opposite is true
during austral winter. These observations, derived from the back-trajectory approach,
suggest that LAI in the back-trajectories pathways to the monitoring site in this study
is not significantly affected by cloud cover.
45
Figure 16. Annual cycle of the degree of rainout upstream (DRU), Leaf area index on the back-trajectories (LAI) and Precipitation Recycling on the back-trajectories (PR) for the period from June 2012 to May 2018.
J F M A M J J A S O N D0
200
400
600
800
1000
1200
1400
DR
U (
mm
tra
j-1
)
0
200
400
600
800
1000
1200
LA
I
3
4
5
6
7
PR
Source: AMPUERO, 2019.
Moisture recycling along trajectories (PR) was also analyzed. It is worth noting
that PR display similar annual cycle with LAI from March to December, and similar
with DRU from January to April (Figure 16). Based in this observation, it appears that
evapotranspiration is the most important control on PR during the dry season and the
beginning of the wet season and precipitation is most important during the end of the
wet season. Hence, it makes sense that the lowest PR values appear during austral
summer and the higher values at the end of the dry season. It is worth noting that the
most intense precipitation recycling rates are in the western and southern Amazon
and are higher from June to November (Figure 17).
Analogously, during austral summer, the main moisture influx comes from
tropical north Atlantic (DRUMOND et al., 2014), consistent with the highest frequency
of north east back-trajectories observed during austral that period. Moreover, during
austral winter, evapotranspiration in the central, south and eastern Amazon is
maximum (DA ROCHA et al. 2009), concomitant the prevalence of south-easterly
back-trajectories (Figure 18).
46
Figure 17. Precipitation recycling (PR) computed with WAM2lyr and data from ERA Interim. a) Mean annual PR. b-e) Seasonal anomalies.
Source: AMPUERO 2019
Figure 18. The maps show mean moisture recycling (PR) per season in percentage (%). Only values in the dominant back-trajectory areas are shown. Panels under the maps show the meridional average PR. Clusters 1, 5 and 11 are frequent during austral summer; clusters 2 and 3, during austral autumn and winter; and clusters 6 and 7, during austral winter.
Source: AMPUERO, 2019.
a)
d) e)
b) c)
47
5.3. ISOTOPE MONITORING
The isotopic composition of the local precipitation fits well with the Global
Meteoric Water Line (GMWL) (Figure 19), presenting a slope of 8.40 and an intercept
(dxs) of16.74. The δ18O values range from -18‰ to 0‰, with the most depleted
values in March-April-May. Dxs values range from 8.4‰ to 20.4‰ with the highest
values in July-August-September and higher seasonal amplitude than its analytical
uncertainty (0.94‰). During the study period, dxs and δ18O show a positive
correlation (r=0.50, p<0.01, Figure 20). dxs increases with δ18O at about
∂dxs/∂δ18O=0.40, on two-week time step. This slope is high compared with the global
average, although that value is based on long term annual averages from GNIP
stations (0.1-0.2) (FROEHLICH et al., 2002). It also appears that the relation
between dxs and δ18O changes with the seasons, with the best covariation during the
austral winter months (Figure 21).
Figure 19. Biweekly δ18O andδD of precipitation at the Palestina station. The GMWL is represented by a solid red line and deviations of ±5‰ are represented with dashed lines.
-20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2
-140
-120
-100
-80
-60
-40
-20
0
20
D
(‰
, V
SM
OW
)
18
O (‰, VSMOW)
Palestina station
GMWL
Source: AMPUERO, 2019.
48
Figure 20. Correlation between δ18O and dxs of precipitation at the Palestina station on biweekly timestep.
-20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4
6
8
10
12
14
16
18
20
22
24
dxs (
‰, V
SM
OW
)
18
O (‰, VSMOW)
r = 0.50
p < 0.01
Source: AMPUERO, 2019.
Figure 21. Correlation between of δ18O and dxs of precipitation at the Palestina station on biweekly timestep for the seasons (a)JJA, (b)SON, (c)DJF and (d)MAM.
-20 -16 -12 -8 -4 0
8
12
16
20
dxs
18
O
dxs=0.41*18
O+17.78
R=0.48
-20 -16 -12 -8 -4 0
8
12
16
20
a)
c)
b)dxs=0.24*
18O+16.36
R= 0.13
dxs
18
O
a)
-20 -16 -12 -8 -4 0
8
12
16
20
d)dxs=0.22*
18O+14.90
R=0.03
dxs
18
O
-20 -16 -12 -8 -4 0
8
12
16
20 dxs=0.24*18
O+14.60
R=0.30
dxs
18
O
Source: AMPUERO, 2019.
Although δ18O and dxs are correlated, significant differences arise in
response to distinct land-atmosphere processes. Table 2 summarizes the linear
49
correlation coefficients (r-values) between the isotope values of δ18O and dxs with
the local precipitation at the monitoring site and calculated parameters in the back-
trajectories (DRU, LAI and PR). Linear correlations were calculated directly with the
original data and with data with the seasonality removed by means of calculating
anomalies. Timeseries for local precipitation, DRU and LAI are shown in Figure 22. It
is important to make this distinction because the parameters used follow seasonal
cycles that may induce apparent correlations which do not represent causality
relations.
The main control on δ18O is the precipitation upstream the monitoring site.
When the complete original records are considered, the best correlation is with DRU
(r=-56, p<0.05), followed by the local precipitation (r=-0.31, p<0.05). δ18O also
presents a positive correlation with LAI (r=0.29, p<0.05), possibly related to the fact
that higher LAI values in the Amazon represent more leaf area, with the potential to
transpire more water from the soil that is enriched in heavy isotopes compared with
the atmospheric vapor. However, when seasonality is removed, only DRU is
significantly correlated with δ18O (r=-0.21, p<0.10). This result confirms that DRU,
which represents the air mass precipitation history, is the most accurate metric of the
physical processes affecting the isotopic fractionation compared with the other
parameters evaluated.
The analysis of the original records of dxs shows the best positive correlation
with LAI (r=0.43, p<0.05) and weaker but still positive correlation with PR (r=0.31,
p<0.05). There are also negative correlations with precipitation-based parameters,
but these are weaker than with LAI. However, when seasonality is removed, only PR
exhibit a significant correlation with dxs (r=0.28, p<0.05).
Correlations during the wet (ONDJFMAM) and the extended dry season
(JJAS) were also calculated separately and are presented in Table 3. During the dry
season, strong correlation between δ18O and DRU based on the original data (r=-
0.81, p<0.05) an on the data with seasonality removed (r=-0.67, p<0.05) were found.
Weaker correlations than with DRU, but still significant, were found with local
precipitation based on the original data (r=-0.40, p<0.10) and with removed
seasonality (r=-0.39, p<0.10). Significant correlation with PR only appears in the
original data (r=0.41, p<0.05). Moreover, during the same period, only LAI and PR
presented significant correlations with dxs in both the original data (both r=0.57,
50
p<0.05) and when seasonality is removed (r=0.49, p<0.05 and r=0.71, p<0.05
respectively). During the wet season significant correlations only appear when the
original data is considered; both δ18O and dxs are correlated with DRU (r=-0.44 and
r=-0.29, p<0.05 respectively).
Since the focus of this study is dxs, the interannual variability was assessed,
focusing on the most relevant parameters (LAI and PR) related with dxs, and
separating the wet and dry seasons as shown in Figure 23. Although there are only
four values per year representing the seasonal averages, it is worth noting the
consistent positive relations during the dry season.
Table 2.Linear correlation coefficient (r) and p-value between isotope records and potential climatic controls based on biweekly data. The r-values in bold (if p<0.05) and in italics (if p<0.10).
Original data Seasonality removed
δ18O Dxs δ18O dxs
Local precipitation -0.31 -0.20 -0.10 -0.01
DRU -0.56 -0.38 -0.21 -0.18
LAI 0.29 0.43 0.06 0.15
PR 0.16 0.31 0.04 0.28
Source: AMPUERO, 2019.
Table 3.As in Table 1, but for dry months (JJAS) and wet months (ONDJFMAM). The first two weeks of October correspond to the dry season and the last two, to the wet season.
Original data Seasonality removed
δ18O dxs δ18O dxs
Dry
Local precipitation -0.40 -0.18 -0.39 -0.33
DRU -0.81 -0.36 -0.67 -0.34
LAI 0.22 0.57 -0.14 0.49
PR 0.41 0.57 0.07 0.71
Wet
Local precipitation -0.12 0.01 -0.05 0.05
DRU -0.44 -0.29 -0.10 -0.14
LAI -0.13 0.05 0.10 0.08
PR -0.06 0.10 0.03 0.19
Source: AMPUERO, 2019.
51
Figure 22. Panels on the left show the original records at the Paletina station and the ones in the right, the records without seasonality (anomalies). (a) and (b) show δ18O of precipitation and local precipitation accumulated along the water sampling period. (c) and (d) show δ18O of precipitation and calculated degree of rainout upstream (DRU), weighted by local precipitation for the water sampling period. (e) and (f) show deuterium excess (dxs) of precipitation and Leaf area index accumulated along the back-trajectories (LAI). LAI was weighted by local precipitation for the water sampling period.
1/05
/201
2
1/08
/201
2
1/11
/201
2
1/02
/201
3
1/05
/201
3
1/08
/201
3
1/11
/201
3
1/02
/201
4
1/05
/201
4
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
0
-5
-10
-15
-20
18O
(‰
, V
SM
OW
)
a)
0
50
100
150
200
Lo
ca
l pre
cip
itatio
n (m
m)
1/05
/201
2
1/08
/201
2
1/11
/201
2
1/02
/201
3
1/05
/201
3
1/08
/201
3
1/11
/201
3
1/02
/201
4
1/05
/201
4
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
10
5
0
-5
-10
Local p
recip
itatio
n (m
m)
18O
(‰
, V
SM
OW
)
b)
-100
-50
0
50
100
1/05
/201
2
1/08
/201
2
1/11
/201
2
1/02
/201
3
1/05
/201
3
1/08
/201
3
1/11
/201
3
1/02
/201
4
1/05
/201
4
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
0
-5
-10
-15
-20
18O
(‰
, V
SM
OW
)
0
500
1000
1500
2000
2500
3000
3500
DR
U
c)
1/05
/201
2
1/08
/201
2
1/11
/201
2
1/02
/201
3
1/05
/201
3
1/08
/201
3
1/11
/201
3
1/02
/201
4
1/05
/201
4
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
10
5
0
-5
-10
18O
(‰
, V
SM
OW
)
d)
-2000
-1500
-1000
-500
0
500
1000
1500
2000
DR
U
1/05
/201
2
1/08
/201
2
1/11
/201
2
1/02
/201
3
1/05
/201
3
1/08
/201
3
1/11
/201
3
1/02
/201
4
1/05
/201
4
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
8
12
16
20
dxs (
‰,
VS
MO
W)
0
200
400
600
800
1000
1200
1400
LA
I
e)
1/05
/201
2
1/08
/201
2
1/11
/201
2
1/02
/201
3
1/05
/201
3
1/08
/201
3
1/11
/201
3
1/02
/201
4
1/05
/201
4
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
-6
-4
-2
0
2
4
6
dxs (
‰,
VS
MO
W)
f)
-400
-200
0
200
400
LA
I
Source: AMPUERO, 2019.
52
Figure 23. Time series of dxs, PR in the upper panels and LAI in the lower panels. Lines represent the series in biweekly time step dots, the seasonal averages for (a,c) wet season and (b,d) extended dry season. The gray shadow indicates the period filled with dxs data from Pomacochas station.
1/06
/201
2
1/09
/201
2
1/12
/201
2
1/03
/201
3
1/06
/201
3
1/09
/201
3
1/12
/201
3
1/03
/201
4
1/06
/201
4
1/09
/201
4
1/12
/201
4
1/03
/201
5
1/06
/201
5
1/09
/201
5
1/12
/201
5
1/03
/201
6
1/06
/201
6
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
1/09
/201
8
1/12
/201
8
6
8
10
12
14
16
18
20
22
Annual mean: r = 0.77
15 days sampling: r = 0.13
dxs
Mean annual dxs
Precipitation Recylcing (PR)
Mean annual PR
dxs (
‰)
Wet season (ONDJFMAM)
0
1
2
3
4
5
6
7
8
9
Pre
cip
itatio
n R
ecylc
ing (P
R)
a)
1/12
/201
1
1/03
/201
2
1/06
/201
2
1/09
/201
2
1/12
/201
2
1/03
/201
3
1/06
/201
3
1/09
/201
3
1/12
/201
3
1/03
/201
4
1/06
/201
4
1/09
/201
4
1/12
/201
4
1/03
/201
5
1/06
/201
5
1/09
/201
5
1/12
/201
5
1/03
/201
6
1/06
/201
6
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
10
12
14
16
18
20
22
Annual mean: r = 0.78
15 days sampling: r = 0.48
dxs
Mean annual dxs
Precipitation Recylcing (PR)
Mean annual PR
dxs (
‰)
Dry season (JJAS)
3
4
5
6
7
8
9
Pre
cip
itatio
n R
ecylc
ing
(PR
)
b)
1/01
/201
2
1/04
/201
2
1/07
/201
2
1/10
/201
2
1/01
/201
3
1/04
/201
3
1/07
/201
3
1/10
/201
3
1/01
/201
4
1/04
/201
4
1/07
/201
4
1/10
/201
4
1/01
/201
5
1/04
/201
5
1/07
/201
5
1/10
/201
5
1/01
/201
6
1/04
/201
6
1/07
/201
6
1/10
/201
6
1/01
/201
7
1/04
/201
7
1/07
/201
7
1/10
/201
7
1/01
/201
8
1/04
/201
8
1/07
/201
8
1/10
/201
8
1/01
/201
9
6
8
10
12
14
16
18
20
22
Annual mean: r = -0.56
15 days sampling: r = 0.02
dxs
Mean annual dxs
LAI
Mean annual LAI
dxs (
‰)
Wet season (ONDJFMAM)
200
400
600
800
1000
LA
I
c)
1/12
/201
1
1/03
/201
2
1/06
/201
2
1/09
/201
2
1/12
/201
2
1/03
/201
3
1/06
/201
3
1/09
/201
3
1/12
/201
3
1/03
/201
4
1/06
/201
4
1/09
/201
4
1/12
/201
4
1/03
/201
5
1/06
/201
5
1/09
/201
5
1/12
/201
5
1/03
/201
6
1/06
/201
6
1/09
/201
6
1/12
/201
6
1/03
/201
7
1/06
/201
7
1/09
/201
7
1/12
/201
7
1/03
/201
8
1/06
/201
8
1/09
/201
8
1/12
/201
8
10
12
14
16
18
20
22
Annual mean: r = 0.61
15 days sampling: r = 0.50
dxs
Mean annual dxs
LAI
Mean annual LAId
xs (
‰)
Dry season (JJAS)
600
800
1000
1200
1400
LA
I
d)
Source: AMPUERO, 2019
53
6. DISCUSSION
The Amazon basin leaf area observations during our monitoring period (2012-
2018) show a consistent seasonality, presenting an increment of about 20% during
the dry season, relative to the wet season. These results are consistent with previous
findings (MYNENI et al., 2007). The forest leaf area exerts a strong control on
moisture exchange between the forest and the atmosphere, with the potential to
modulate the Rayleigh distillation of water isotopologues over the Amazon basin.
Indeed, the positive spatial correlation between LAI and the Vertically Integrated
Divergence of Moisture Flux (VIDMF) (Figure 24) suggests that the moisture input
from the forest to the atmosphere, increases with denser forest canopy. Moreover,
both LAI and PR on the back-trajectories present a consistent increase of about 140
% and 44 % respectively during the dry season, which might result from a
combination of larger routes over the continent and denser forest canopy.
Figure 24. Spatial correlation between LAI and VIDMF on monthly time step for the period from June 2012 to May 2018. Seasonality was removed from both datasets. Shadows show only significant grid cells (p<0.05).
Source: AMPUERO, 2019.
6.1. CONTROLS OF δ18O AND DXS IN PRECIPITATION
The correlations of δ18O and dxs with the original calculated parameters
showed higher coefficients than the ones based on the data without seasonality. The
apparent high correlations might result in part from the similar seasonal cycles, rather
than from physical processes. However, the correlations found when seasonality is
54
removed are significant and consistent with the mechanisms documented in previous
studies.
The negative correlation between δ18O and DRU reflects the removal of heavy
isotopic species through precipitation during air mass transport as in a Rayleigh
model. The correlation with DRU is stronger than with local precipitation, suggesting
that at our monitoring site, a strong local influence on the isotopic composition of
rainwater can be ruled out. This result is consistent with previous studies from the
eastern flank of the Andes, where regional precipitation and large-scale atmospheric
transport are considered the main drivers on δ18O (VIMEUX et al., 2005; VILLACÍS et
al., 2008; INSEL et al., 2012; WINDHORST et al. 2013; FIORELLA et al., 2015;
SAMUELS-CROW et al. 2014; HURLEY et al. 2015). We also assume that below
clouds, secondary evaporation has a negligible effect in our study site, given that
frequent heavy rainfall, typical for tropical rain forests, would suppress that effect
(PENG et al. 2007). Meanwhile, dxs is best correlated with LAI along the air parcel
trajectories, however, the relationship is only significant when the seasonality is
included, while correlation with PR is significant in both series.
When wet and dry months are analyzed separately (Table 3) the time series
without seasonality do not show significant correlations with δ18O or dxs during the
wet months. From back trajectory analysis it is known that the atmospheric transport
from the tropical north Atlantic dominates during the wet season months and
precipitation is mostly of oceanic in origin. As shown in Figure 14, during austral
summer, the tropical North Atlantic pathways (Clusters 1, 5 and 11) feature the
lowest moisture recycling and contributes nearly 34% of the total annual precipitation
falling over the western Amazon basin. During this period, there is reduced re-supply
of moisture through evaporation, thereby, precipitation is not significantly affected by
land processes.
The opposite is true for the extended dry season. Advection across open
forest permits land-atmospheric interactions, including moisture recycling and large-
scale transport. In fact, correlations of dxs with LAI and PR are only significant during
the dry months. The positive relations of dry season dxs with LAI and PR are also
consistent on interannual timescale (Figure 23). These results, together with the high
moisture recycling observed, supports the idea of forest influence on dxs. Since LAI
is a good index for evapotranspiration in the Amazon basin (MYNENI et al., 2007), it
55
is possible to infer that high values of dxs imply an increased contribution of recycled
moisture from the forest to precipitation at the monitoring site. At the same time, the
high negative correlation between δ18O and DRU suggests that, despite the moisture
recharge from land to the atmosphere, the effect of Rayleigh distillation on the
isotopic composition of water is remarkable. It is worth noting that, even though
rainfall is enriched in heavy isotopes compared with the wet season, the fractionation
process related to precipitation upstream is still dominant. Indeed, the larger
exposure to vegetation during this season could favor both land-atmosphere
interactions and heavy isotope loss due to rainout upstream.
The analysis of the relation between δ18O and dxs points toward significant
seasonal differences (Figure 21). The relationship is best during JJA, which might be
explained by a more constant moisture source placed in the Amazon forest and more
stable atmospheric conditions compared with the wet season. In contrast, during
DJF, a combination of moisture sources from the North Atlantic and from the Amazon
forest might complicate the relation between δ18O and dxs, resulting in a weak
correlation, with large scatter of the values.
6.2. PALEOCLIMATE IMPLICATIONS
The anti-phased behavior observed in the Holocene δ18O records from
speleothems from the eastern and western edges of the Amazon basin (Figure 25)
point to a see-saw effect in the monsoon precipitation at orbital time scale. In one
hand, this pattern is consistent with the notion that variations in the intensity of the
Bolivian-High - Nordeste low system serve as the main driver of precipitation
variability in the Amazon basin at orbital time-scale (CHENG et al., 2013). On the
other hand, Wang et al. (2017) argue that plant transpiration may bias the
paleoclimate interpretation based on the degree of rainout upstream. In this sense,
the Holocene and glacial rainfall stable isotopic composition over the western
Amazon basin was largely modulated by the forest evapotranspiration, which buffer
the effect of Rayleigh fractionation.
56
Figure 25. Geographic location of the geochemical and palynological records referenced in this study.
Source: AMPUERO, 2019.
The isotopic records, based on fluid inclusions of speleothems from the
western Amazon, provide important clues regarding the moisture flux pathways and
recycling processes. The calculated dxs of speleothem fluid inclusions from Tigre
Perdido cave, shows an antiphase behavior through the Holocene, when compared
with the δ18Ocalcite (VAN BREUKELEN et al., 2008). The antiphase response between
dxs and δ18Ocalcite suggests that other processes besides rainout upstream drove
changes in the isotopic composition variability during the Holocene. It is noteworthy
that the Holocene dxs variations are similar with variations in the difference between
δ18O from the western and eastern edges of the Amazon basin, δ18OTigrePerdido-Paraíso
(Figure 26).
A negative δ18OTigrePerdido-Paraíso implies more negative rainfall δ18O over the
western Amazon compared with its eastern border. Assuming a common moisture
flux pathway for both sites, the observed δ18Ocalcite gradient cannot be explained by
the Rayleigh fractionation processes alone. Indeed, if moisture reaching the western
Amazon basin were the same as the one that passes over the central Amazon, the
57
δ18Ocalcite values from the western border should always be a slightly more negative
than those from the eastern edge. Instead, the east and west edges show similar
isotopic compositions from the early to the mid- Holocene, followed by the increasing
of the east-west gradient towards the present.
Figure 26. Comparison between stable isotope record from speleothems from the western and eastern Amazon edges: a) dxs reconstruction from Tigre Perdido speleothem record (VAN BREUKELEN et al., 2008); b) Tigre Perdido - Paraíso δ18O; c) 20-yr interpolated δ18O record from Tigre Perdido cave speleothem (VAN BREUKELEN et al., 2008). The δ18O values from Tigre Perdido are corrected by 1.4 ‰ to account for temperature variations between the caves following the procedure of Wang et al. (2017); d) as in (c) 20-yr interpolated δ18O record from Paraíso cave speleothems (WANG et al., 2017).
-9
-8
-7
-6
-5
1
0
-1
-2
-3
-4
0 1 2 3 4 5 6 7 8 9 10 11
-9
-8
-7
-6
-5c)
d)
b)
Less
PR
Age (kyr BP)
Tig
re P
erd
ido
1
8O
(te
mpe
ratu
re c
orr
ecte
d)
More
PR
gradient
increase
Para
íso
18O
(‰
, VPD
B)
Tig
re P
erd
ido-P
ara
íso
18O
(‰
, VPD
B)
6
8
10
12
14
a)
Flu
id I
nclu
sio
n D
Excess (
‰, V
SM
OW
)
Source: AMPUERO, 2019.
From the early to the mid- Holocene, dxs values are low compared with the
present, ranging from 7 to 10 ‰. The low values suggest reduced moisture recycling
along the atmospheric transport pathway, consistent with the notion of reduced forest
cover during this period. Since the forest is the main moisture source for present
winter precipitation, it is likely that from the early to the mid- Holocene, precipitation
was restricted to the wet season. This implies that the dominant atmospheric
transport was along northeast pathways, as in the present austral summer back-
trajectory clusters observed in this study. In this scenario, it is likely that parallel
58
atmospheric transport routes for Tigre Perdido and Paraiso were dominant, with
precipitation mostly of oceanic origin, resulting in similar isotopic compositions over
the western and eastern edges of the Amazon basin. Moreover, since these
atmospheric fluxes travel short distances over the continent until the precipitation
sites, they are subject to only limited Rayleigh fractionation.
As we move towards the late Holocene, δ18OTigrePerdido-Paraíso becomes more
negative as a result of more negative rainfall δ18O over the western Amazon,
concomitant with increased δ18O values in the eastern Amazon. This pattern
suggests enhanced Rayleigh distillation at that time. The dxs values from Tigre
Perdido follow the same trend (Figure 26), pointing to an increase in moisture
recycling associated with forest expansion, which allows for a greater contribution of
winter precipitation feed by forest evapotranspiration. The results are also consistent
with the establishment of a Rayleigh distillation mechanism as observed when
analyzing winter isotopic composition in present-day rainfall.
As shown in Figure 27, the increasing trend in dxs along the Holocene is
concomitant with the increased rainforest pollen frequency recorded in lakes from
western, south-central and eastern Amazon, supporting the notion of forest
expansion during the mid- and late Holocene (BUSH et al 2015; SANTOS, 2019;
FONTES et al., 2017; ABSY et al., 1991; SIFEDDINE et al., 2001). The complete
available palynological records referenced in this study is shown in Figure 28.
Furthermore, the dxs increase observed from about 6 KBP to present agrees with
numerical simulations of Amazon vegetation (MAKSIC et al., 2018), which show a
clear expansion of the evergreen vegetation in the Amazon basin towards the
present.
59
Figure 27. Comparison between calculated dxs from Tigre Perdido stalagmite (VAN BREUKELEN et al., 2008) with the integrated record of rainforest pollen frequency from five lakes: Iriri, Altamira - PA (SANTOS , 2019); Ilha Arapujá, Altamira - PA (SANTOS, 2019);Laguna Chochos, Peru (BUSH et al. 2005);Saci, central-south Amazon (FONTES et al., 2017) and Serra Sul Carajás Lake, CSS2 (ABSY et al., 1991; SIFEDDINE et a., 2001).
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
-2
-1
0
1
2
Arboreal Pollen
Fluid Inclusion dxs
Age (kyr BP)
Arb
ore
al P
olle
n
6
7
8
9
10
11
12
13
14
Flu
id In
clu
sio
n D
Excess (‰
, VS
MO
W)
Source: AMPUERO, 2019.
Figure 28. Normalized interpolated frequency of arboreal pollen from lakes used in this study and average value.
Source: AMPUERO, 2019.
-3
-2
-1
0
1
2
3
4
5
0 2 4 6 8 10 12 14 16 18
Freq
uen
cy
Age (kyr BP)
Lago Iriri
Lago Arapujá
Lago Chochos
Serra Sul Carajas
Lago Saci
MEDIA
60
7. CONCLUSION
This study investigated the role of vegetation in driving changes in the rainfall
isotopic composition. The tropical North Atlantic and the Amazon forest are the main
moisture sources for precipitation at the monitoring site. Their moisture contributions
vary seasonally with air parcel transport patterns, identified by means of cluster
analysis. The main driver on the isotopic composition of precipitation is remote
precipitation upstream, along atmospheric transport pathways. Precipitation derived
from oceanic moisture with minimum exchange with surface evaporation has heavier
isotopic composition than precipitation originating from air parcels that travel over
land and are affected by DRU through the Rayleigh distillation mechanism.
Furthermore, dxs at the monitoring site is a good indicator of the remote contribution
to precipitation from forest evapotranspiration. Although the dry season isotopic
composition is heavier than the wet season and there is major moisture contribution
from the forest through evapotranspiration, there is a well stablished Rayleigh
distillation mechanism resulting from long travel distance over the continent. The
relationship of dxs with δ18O is sensitive variability of moisture sources, showing the
best coupling between them during austral winter, when the main moisture source is
the Amazon forest. Paleo-records show that during the early to the mid- Holocene,
the main moisture source to the monitoring site was located over the tropical North
Atlantic with limited moisture contribution from land due to the reduced forest cover at
the time. As a result, the water supply was limited during the dry season. After the
mid-Holocene, the contribution from winter precipitation gains importance as the
expansion of the forest permits more moisture recycling.
61
8. REFERENCES
ABSY, M.L.; CLEEF, A.M.; FOURNIER, M.; MARTIN, L.; SERVANT, M.; SIFEDDINE, A.; FERREIRA DASILVA, M.; SOUBIÈS, F.; SUGUIO, K.; TURCQ, B.; VAN DER HAMMEN, T., Mise en evidence de quatre phases d’ouverture de la foret dense dans le sudest de l'Amazonie au cours de 60,000 dernieres annees. Premiere comparaison avec dautres regions tropicales, C. R. Acad. Sci., v. 312, p. 673-678, 1991.
APAÉSTEGUI, J.; CRUZ, F.W.; SIFEDDINE, A.; VUILLE, M.; ESPINOZA, J.C.; GUYOT, J.L.; KHODRI, M.; STRIKIS, N.; SANTOS, R.V.; CHENG, H.; EDWARDS, L.; CARVALHO, E.; SANTINI, W., Hydroclimate variability of the northwestern Amazon Basin near the Andean foothills of Peru related to the South American Monsoon System during the last 1600 years, Clim. Past., v. 10, p. 1967–1981, 2014, doi: 10.5194/cp-10-1967-2014.
ARAGUAS-ARAGUAS, L.; FROEHLICH, K.; ROZANSKI, K., Stable isotope composition of precipitation over Southeast Asia. J. Geophys. Res., v. 103, n. d22, p. 721 – 724, 1998, doi: https://doi.org/10.1029/98JD02582.
BAKER, J.C.A.; GLOOR, M.; SPRACKLEN, D.V.; ARNOLD, S.R.; TINDALL, J.C.; CLERICI, S.J.; LENG, M.J.; BRIENEN R.J.W., What drives interannual variation in tree ring oxygen isotopes in the Amazon? Geophys. Res. Lett., v. 43, p. 11, 831–11, 840, 2016, doi: 10.1002/2016GL071507.
BARICHIVICH, J.; GLOOR, E.; PEYLIN, P.; BRIENEN, R.J.W.; SCHÖNGART, J.; ESPINOZA, J.C.; PATTNAYAK, K.C., Recent intensification of Amazon flooding extremes driven by strengthened Walker circulation, Science Advances, p. 1-7, 2018, doi: 10.1126/sciadv.aat8785.
BEHLING, Hermann et al., Holocene environmental changes in the central amazon basin inferred from Lago Calado (Brazil), Palaeogeography, Palaeoclimatology, Palaeoecology, v. 173, p. 87–101, 2001, doi: 10.1016/S0031-0182(01)00321-2.
BEHLING, H., Carbon storage increases by major forest ecosystems in tropical South America since the Last Glacial Maximum and the early Holocene, Global and Planetary Change, v. 33, n. 1-2, p. 107–116, 2002, doi: https://doi.org/10.1016/S0921-8181(02)00065-6.
BOND, G. et al., A Pervasive Millennial-Scale Cycle in North Atlantic Holocene and Glacial Climates, Science, v. 278, n. 5341, p. 1257–1266, 1997, doi: 10.1126/science.278.5341.1257.
BOND, G. et al. Persistent Solar Influence on North Atlantic Climate During the Holocene. Science. v. 294, n. 5549, p. 2130-2136, 2001, doi: 10.1126/science.1065680.
62
BURNS, S.J. et al. Speleothem evidence from Oman for continental pluvial events during interglacial periods. Geological Society of America. v. 29, n. 7, p. 623-626, 2001, doi: 10.1130/0091-7613(2001)029<0623:SEFOFC>2.0.CO;2.
BUSH, M.B.; SILMAN, M.R.; LISTOPAD, C.M.C.S., A regional study of Holocene climate change and human occupation in Peruvian Amazonia, Journal of Biogeography, v. 34, p. 1342–1356, 2007.
CHEN, T.; WENG, S.; SCHUBERT, S., Maintenance of austral summertime upper-tropospheric circulation over tropical South America: The Bolivian high–Nordeste low system, J. Atmos. Sci., v. 56, p. 2081–2100, 1999, doi: https://doi.org/10.1175/1520-0469(1999)056<2081:MOASUT>2.0.CO;2.
CHENG, Hai et al., Climate change patterns in Amazonia and biodiversity, Nature communications, v. 4, p. 1-6, 2013, doi: 10.1038/ncomms2415.
COLINVAUX, P.A.; DE OLIVEIRA, P.E.; BUSH, M.B., Amazonian and neotropical plant communities on glacial time-scales: The failure of the aridity and refuge hypotheses, Quaternary Science Reviews, v. 19, n. 1–5, p. 141–169, 2000.
CORDEIRO, R.C. et al., Holocene fires in East Amazonia (Carajás), new evidences, chronology and relation with paleoclimate, Global and Planetary Change, v. 61, n. 1–2, p. 49–62, 2008, doi: https://doi.org/10.1016/j.gloplacha.2007.08.005.
CORDEIRO, R. C. et al., Biogeochemical indicators of environmental changes from 50 Ka to 10 Ka in a humid region of the Brazilian Amazon, Palaeogeography, Palaeoclimatology, Palaeoecology, v. 299, n. 3–4, p. 426–43C, 2011, doi: https://doi.org/10.1016/j.palaeo.2010.11.021.
CRAIG, H. Standard for Reporting Concentrations of Deuterium and Oxygen-18 in Natural Waters, Science, v. 133, p. 1833-1834,1961.
CRUZ, F.W.; VUILLE, M.; BURNS, S.J.; WANG, X.; CHENG, H.; WERNER, M.; EDWARDS, R.L.; KARMANN, I.; AULER, A.S.; NGUYEN, H., Orbitally driven east–west antiphasing of South American precipitation, Nat. Geosci, v. 2, n. 3, p. 210–214, 2009, doi: 10.1038/NGEO444.
DANSGAARD, W., Stable isotopes in precipitation, Tellus, v. 16, p. 436–468, 1964.
DA ROCHA, H. R.; GOULDEN, M.; MILLER, S.; MENTON, M.; PINTO, L.; FREITAS, H.; FIGUEIRAA. S., Seasonality of water and heat fluxes over a tropical forest in eastern Amazonia, Ecol. Appl., v. 1 n. 4, p. 22–32, 2004, doi: https://doi.org/10.1890/02-6001.
DA ROCHA, H. R., MANZI A. O.; CABRAL, O. M.; MILLER, S. D.; GOULDEN, M. L.; SALESKA, S. R.; COUPE, N. R. et al., Patterns of water and heat flux across a biome gradient from tropical forest to savanna in Brazil, Journal of Geophysical Research, v. 114, n.G00B12, p. 1-8, 2009, doi: 10.1029/2007JG000640.
63
DEE, D. et al., The ERA-Interim reanalysis: configuration and performance of the data assimilation system, Q. J. R. Meteorol. Soc., v. 137, p. 553–597, 2011, doi: 10.1002/qj.828.
DRUMOND, A.; MARENGO, J.; AMBRIZZI, T.; NIETO, R.; MOREIRA L.; GIMENO, L., The role of the Amazon Basin moisture in the atmospheric branch of the hydrological cycle: a Lagrangian analysis, Hydrol. Earth Syst. Sci., v. 18, p. 2577–2598, 2014, doi: 10.5194/hess-18-2577-2014.
FIORELLA, R.P.; POULSEN, C.J.; PILLCOZOLÁ, R.S.; BARNES, J.B.; TABOR, C.R.; EHLERS, T.A., Spatiotemporal variability of modern precipitation δ18O in the central Andes and implications for paleoclimate and paleoaltimetry estimates, J. Geophys. Res. Atmos., v. 120, p. 4630–4656, 2015, doi: 10.1002/2014JD022893.
FLEITMANN, D.; BURNS, S.J.; NEFF, U.; MANGINI, A.; MATTER, A., Changing moisture sources over the last 330,000 years in Northern Oman from fluid-inclusion evidence in speleothems, Quat. Res., v. 60, p. 223-232, 2003b.
FONTES, D. et al., Paleoenvironmental dynamics in South Amazonia, Brazil, during the last 35,000 years inferred from pollen and geochemical records of Lago do Saci, Quaternary Science Review, v. 176, p. 161-180, 2017, doi: http://dx.doi.org/10.1016/j.quascirev.2017.08.021.
FROEHLICH, K.; GIBSON, J.J.; AGGARWAL, P., Deuterium excess in precipitation and its climatological significance, in: Study of Environmental Change Using Isotope Techniques, International Atomic Energy Agency, Vienna, p. 54–66, 2002.
GARREAUD, R.D.; VUILLE, M.; COMPAGNUUCCI, R.; MARENGO, J., Present day South American climate, Palaeogeography, Palaeoclimatology, Palaeoecology, v. 281, p. 180-195, 2008, doi: 10.1016/j.palaeo.2007.10.032.
GRIFFITHS, M.L. et al., Younger Dryas–Holocene temperature and rainfall history of southern Indonesia from δ18O in speleothem calcite and fluid inclusions, Earth and Planetary Science, v. 295, p. 30-36, 2010, doi: 10.1016/j.epsl.2010.03.018.
HARMON, R.S.; SCHWARCZ, H.P.; ONEIL, J.R., D-H ratios in speleothem fluid inclusions – guide to variations in the isotopic composition of meteoric precipitation, Earth Planet. Sc. Lett., v. 42, p. 254–266, 1979, doi:10.1016/0012-821x(79)90033-5.
HUFFMAN, G.J. et al., The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-Global, Multiyear, Combined-Sensor Precipitation Estimates at Fine Scales, Journal of Hydrometeorology, v. 8, p. 38-55, 2007, doi: 10.1175/JHM560.1.
HURLEY, J.V.; GALEWSKY, J.; WORDEN, J.; NOONE, D.A., Test of the advection-condensation model for subtropical water vapor using stable isotopologue observations from Mauna Loa Observatory, Hawaii, J. Geophys. Res, v. 117, n. D19118, 2012, doi; https://doi.org/10.1029/2012JD018029.
64
HURLEY, J.V.; VUILLE, M.; HARDY, D.R.; BURNS, S.; THOMPSONL.G., Cold air incursions, δ18O variability and monsoon dynamics associated with snow days at Quelccaya Ice Cap, Peru, J. Geophys. Res., v. 120, p. 7467-7487, 2015, doi: https://doi.org/10.1002/2015JD023323.
HUTYRA, L.; MUNGER, J.; SALESKA, S.; GOTTLIEB, E.; DAUBE, B.; DUNN, A.; AMARAL, D.; CAMARGO, P.; WOFSY, S., Seasonal controls on the exchange of carbon and water in an Amazonian rain forest, J. Geophys. Res., v. 112, n. G03008, p. 1-16, 2007, doi: 10.1029/2006JG000365.
INSEL, N.; POULSEN, C.J.; STURM, C.; EHLERS, T.A., Climate controls on Andean precipitation δ18O interannual variability, J. Geophys. Res. Atmos, v. 118, p. 9721–9742, 2013, doi: 10.1002/jgrd.50619.
JONCKHEERE, I.; FLECK S.; NACKAERTS, K.; MUYSA, B.; COPPIN P.; WEISS, M.; BARET, F., Review of methods for in situ leaf area index determination Part I. Theories, sensors and hemispherical photography, Agricultural and Forest Meteorology, v. 121, p. 19-35, 2004, doi: 10.1016/j.agrformet.2003.08.027.
JOHNSEN, S.J.; DANSGAARD, W.; WHITE, J.W.C., The origin of Arctic precipitation under present and glacial conditions, Tellus B, v. 41, p. 452-468, 1989. doi: 10.1111/j.1600-0889.1989.tb00321.x.
KONG, Y.; PANG, Z.; FROEHLICH, K., Quantifying recycled moisture fraction in precipitation of an arid region using deuterium excess, Tellus B, v. 65, p. 19251, 2012, doi: http://dx.doi.org/10.3402/tellusb.v65i0.19251.
LACHNIET, M.S., Climatic and environmental controls on speleothem oxygen-isotope values, Quaternary Science Reviews, v. 28, n. 5-6, p.412–432, 2008, doi: https://doi.org/10.1016/j.quascirev.2008.10.021.
LAI, C.; EHLERINGER, J.R., Deuterium excess reveals diurnal sources of water vapor in forest air, Ecosystem Ecology, v. 165, p. 213-23, 2010, doi: 10.1007/s00442-010-1721-2.
LANDAIS, A.; RISI, C.; BONY, S.; VIMEUX, F.; DESCROIX, L.; FALOURD, S.; BOUYGUES, A., Combined measurements of 17O-excess and d-excess in African monsoon precipitation: Implications for evaluating convective parameterizations. Earth and Planetary Science Letters, v. 298, p. 104-112, 2010, doi: 10.1016/j.epsl.2010.07.033.
LI, X.; XIE, S.-P., GILLE, S.T.; YOO, C., Atlantic-induced pan-tropical climate change over the past three decades. Nat. Clim. Change, v. 6, p. 275–279, 2016, doi: https://doi.org/10.1038/nclimate2840.
LENTERS, J.D.; COOK, K.H., On the origin of the Bolivian high and related circulation features of the south American climate, J. Atmos. Sci., v. 54, p. 656-678, 1997, doi: https://doi.org/10.1175/1520-0469(1997)054<0656:OTOOTB>2.0.CO;2.
65
LENTERS, J.D.; COOK, K.H., Summer time precipitation variability over South America: role of the large-scale circulation, Mon. Weather Rev., v. 127, p. 409–431, 1999, doi: https://doi.org/10.1175/1520-0493(1999)127<0409:SPVOSA>2.0.CO;2.
MAKARIEVA, A.M.; GORSHKOV, V.G.; SHEIL, D.; NOBRE, A.D.; LI, B.-L., Where do winds come from? A new theory on how water vapor condensation influences atmospheric pressure and dynamics, Atmos. Chem. Phys., v. 13, p. 1039–1056, 2013, doi: https://doi.org/10.5194/acp-13-1039-2013.
MAKSIC, J. et al., Simulation of the Holocene climate over South America and impacts on the vegetation, The Holocene, p. 1-13, 2018, doi: 10.1177/0959683618810406.
MARENGO, J. A.; ESPINOZA, J.C., Extreme seasonal droughts and floods in Amazonia: causes, trends and impacts, Int. J. Climatol., v. 36, p. 1033–1050, 2015, doi: 10.1002/joc.4420.
MARENGO, J. A.; TOMASELLA, J.; SOARES, W. G; ALVES, L. M.; NOBRE, C.A. Extreme climatic events in the Amazon basin. Theor Appl Climatol., v. 107, p. 73–85, 2012, doi: 10.1007/s00704-011-0465-1.
MASSON-DELMOTTE, V.; JOZUEL, J.; LANDAIS, A.; STIEVENARD, M.; JOHNSEN, S. J.; WHITE, J. W. C.; WERNER, M.; SVEINBJORNSDOTTIR, A.; FUHRER, K., GRIP Deuterium Excess Reveals Rapid and Orbital-Scale Changes in Greenland Moisture Origin, Science, v. 309, p. 118-121, 2005. doi: 10.1126/science.1108575.
MCDERMOTT, F.; SCHWARCZ, H.P.; HARMON, R.S.; THOMPSON, P.; FORD, D.C. Isotopes in speleothems. In: Leng, M.J. (Ed.), Isotopes in Palaeoenvironmental Research. Developments in Paleoenvironmental Research. Dordrecht, the Netherlands: Springer, 2006. p. 185–225. doi: 10.1007/1-4020-2504-1_05.
MCGARRY, S.; BAR-MATTHEWS, M.; MATTHEWS, A.; VAKS, A.; SCHILMAN, B.; AYALON, A., Constraints on hydrological and paleotemperature variations in the Eastern Mediterranean region in the last 140ka given by the δD values of speleothem fluid inclusions, Quaternary Science Reviews, v. 23, p. 919-934, 2004, doi: 10.1344/105.000001646.
MELANDER, L., Isotope Effects on Reaction Rates: Modern concepts in chemistry, Michigan, Ronald Press, 1960, p. 181.
MERLIVAT, L.; JOUZEL, J., Global climatic interpretation of the deuterium–oxygen-18 relationship for precipitation, Journal of Geophysical Research-Oceans and Atmospheres, v. 84, n. C8, p. 5029–5033, 1979, doi: https://doi.org/10.1029/JC084iC08p05029.
MOLINA, R.D.; SALAZAR, J.F; MARTÍNEZ, J.A.; VILLEGAS, J.C.; ARIAS, P.A., Forest‐induced exponential growth of precipitation along climatological wind
66
streamlines over the Amazon, J. Geophys. Res., 2019, doi: https://doi.org/10.1029/2018JD029534.
MOREIRA, L.S. et al., Holocene paleoenvironmental reconstruction in the Eastern Amazonian Basin: Comprido Lake, Journal of South American Earth Sciences, v. 44, p. 55–62, 2012, doi: http://dx.doi.org/10.1016/j.jsames.2012.12.012.
MORTON, D.C. et al., Amazon forests maintain consistent canopy structure and greenness during the dry season, Nature, v. 506, p. 221-224, 2014, doi: 10.1038/nature13006.
MYNENI, R. B.; YANG, W.; NEMANI, R. R.; HUETE, A. R.; DICKINSON, R. E.; KNYAZIKHIN, Y. et al., Large seasonal swings in leaf area of Amazon rainforests, Proceedings of the national academy of sciences of the united states of America, v. 104, n. 12, p. 4820–4823, 2007, doi: https://doi.org/10.1073/pnas.0611338104.
MYNENI, R.; KNYAZIKHIN, Y.; PARK, T., MCD15A2H MODIS/Terra+Aqua Leaf Area Index/FPAR 8-day L4 Global 500m SIN Grid V006, NASA EOSDIS Land Processes DAAC, 2015, doi: http://doi.org/10.5067/MODIS/MCD15A2H.006.
NEPSTAD, D.C.; DE CARVALHO, C.R.; DAVIDSON, E.A.; JIPP, P.H.; LEFEBVRE, P.A.; NEGREIROS, G.H.; DA SILVA, E.D.; STONE, T.A.; TRUMBORE, S.E.; VIEIRA, S., The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures, Nature, v. 372, p. 666–669, 1994, doi: https://doi.org/10.1038/372666a0DO
NUMAGUTI, A., Origin and recycling processes of precipitating water over the Eurasian continent: Experiments using an atmospheric general circulation model, J. Geophys. Res., v. 104, p. 1957- 1972, 1999, doi: https://doi.org/10.1029/1998JD200026.
OLIVEIRA, R.S.; DAWSON, T.E.; BURGESS, S.S.O.; NEPSTAD, D.C., Hydraulic redistribution in three Amazonian trees, Oecologia, v. 145, p. 354–363, 2005, doi: 10.1007/s00442-005-0108-2.
PACCINI, L.; ESPINOZA., J. C.; RONCHAIL, J.; SEGURA, H., Intra‐seasonal rainfall variability in the Amazon basin related to large‐scale circulation patterns: a focus on western Amazon–Andes transition region, nt. J. Climatol., v. 38, p. 2386–2399, 2017, doi: https://doi.org/10.1002/joc.5341.
PENG, H.; MAYER, B.; HARRIS, S.; H. KROUSE, R., The influence of below-cloud secondary effects on the stable isotope composition of hydrogen and oxygen in precipitation at Calgary, Alberta, Canada, Tellus B: Chemical and Physical Meteorology, v. 59, n. 4, p. 698-704, 2007, doi: https://doi.org/10.1111/j.1600-0889.2007.00291.x.
PFAHL, S.; SODEMANN, H., What controls deuterium excess in global precipitation? Clim. Past., v. 10, p. 771–781, 2014, doi: 10.5194/cp-10-771-2014.
67
PFAHL, S.; WERNLI, H., Air parcel trajectory analysis of stable isotopes In water vapor in the eastern Mediterranean, Journal of Geophysical Research, v. 113, p. 1-16, 2008, doi: 10.1029/2008JD009839.
RASMUSSEN, S.O. et al., Stratigraphic framework for abrupt climatic changes during the Last Glacial period based on three synchronized Greenland ice-core records: refining and extending the INTIMATE event stratigraphy, Quaternary Science Reviews, v. 106, p. 14-28, 2014, doi: https://doi.org/10.1016/j.quascirev.2014.09.007.
RESTREPO-COUPE et al., What drives the seasonality of photosynthesis across the Amazon basin? A cross-site analysis of eddy flux tower measurements from the Brasil flux network, Agricultural and Forest Meteorology, v. 182-183, p. 128–144, 2013, doi: https://doi.org/10.1016/j.agrformet.2013.04.031.
RISI, C.; LANDAIS, A.; WINKLER, R.; VIMEUX, F., Can we determine what controls the spatio-temporal distribution of d-excess and 17O-excess in precipitation using the LMDZ general circulation model? Climate of the Past, v. 9, p. 2173-2193, 2013, doi: https://doi.org/10.5194/cp-9-2173-2013.
ROLPH, G.; STEIN, A.; STUNDER, B., Real-time Environmental Applications and Display sYstem: READY, Environmental Modelling and Software, v. 95, p. 210-228, 2017, doi: http://dx.doi.org/10.1016/j.envsoft.2017.06.025.
SAATCHI, S.; ASEFI-NAJAFABADY, S.; MALHI, Y.; ARAGÃO, L.E.O.C.; ANDERSON, L.O.; MYNENI, R.B.; NEMANI, R., Persistent effects of a severe drought on Amazonian forest canopy, Proceedings of the National Academy of Sciences, v. 110, p. 565–570, 2013, doi: https://doi.org/10.1073/pnas.1204651110.
SALATI, E.; DALL’OLIO, A.; MATSUI, E.; GAT, J.R., Recycling of water in the Amazon Basin: An Isotopic Study, Water resources research, Res., v. 15, p. 1250–1258, 1979, doi: https://doi.org/10.1029/WR015i005p01250.
SALESKA, S.R.; WU, J; GUAN, K.; ARAUJO A.C.; HUETE, A.; NOBRE, A.D.; RESTREPO-COUPE, N., Dry-season greening of Amazon forest, Nature, v. 531, 2016, doi: https://doi.org/10.1038/nature16457.
SAMUELS-CROW, K.E.; GALEWSKY, J; HARDY, D.R.; SHARP, Z.; WORDEN, J.; BRAUN, C., Upwind convective influences on the isotopic composition of atmospheric water vapor over the tropical Andes, Journal of Geophysical Research, v. 119, n. 12, p. 7051-7063, 2014, doi: https://doi.org/10.1002/2014JD021487.
SANTOS, R.A., Relações paleoclimáticas e Paleoambientais durante o Holoceno no lesta da Amazônia na Região da volta Grande do Rio Xingu, São Paulo, 2019, 119 p., Tese (doutorado em Geociências) - Universidade de São Paulo, São Paulo, 2019.
SCHWARCZ, H.P.; HARMON, R.S.; THOMPSON, P.; FORD, D.C., Stable isotope studies of fluid inclusions in speleothems and their paleoclimatic significance,
68
Geochimica et Cosmochimica Acta, v. 40, n. 6, p. 657-665, 1976, doi: https://doi.org/10.1016/0016-7037(76)90111-3.
SHUTTLEWORTH, W.J., Evaporation from Amazonian rain forest, Proc. R. Soc. London, v. 233, p. 321–346, 1988, doi: https://doi.org/10.1098/rspb.1988.0024.
SIFEDDINE, A.; MARTIN, L.; TURCQ, B.; RIBEIRO, C.V.; SOUBIES, F.; CORDEIRO, R.C.; SUGUIO, K., Variations of the Amazonian rainforest environment: a sedimentological record covering 30,000 years, Palaeogeogr. Palaeoclimatol. Palaeoecol., v. 168, p. 221-235, 2001, doi: https://doi.org/10.1016/S0031-0182(00)00256-X.
SINHA, D.K.; SINGH, A.K.; MALLICK, K.; SINGH, V.P., Holocene: Definition and current stratigraphic status in the Geological time-scale. In: Holocene, Bahadur Singh Kotlia, 2013.
SPRACKLEN, D.V.; ARNOLD, S.R.; TAYLOR, C.M., Observations of increased tropical rainfall preceded by air passage over forests, Nature, v. 489, p. 282-285, 2012, doi: https://doi.org/10.1038/nature11390.
STAAL, A.; TUINENBURG, O.A.; BOSMANS, J.H.C.; HOLMGREN, M.; VAN NES, E. H.; SCHEFFER, M.; ZEMP, D.C.; DEKKER, S.C., Forest-rainfall cascades buffer against drought across the Amazon, Nature Climate Change, v. 8, p. 539–543, 2018, doi: https://doi.org/10.1038/s41558-018-0177-yDO.
STEIN, A.F. et al., NOAA’s HySPLIT atmospheric transport and dispersion modeling system, American Meteorological Society, p. 2059-2077, 2015, doi: 10.1175/BAMS-D-14-00110.1.
SULCA, J.; VUILLE, M.; SILVA, Y; TAKAHASHI, K., Teleconnections between the Peruvian central Andes and Northeast Brazil during extreme rainfall events in austral
summer, J. Hydrometeorol., v. 17, p. 499-515, 2016, doi: https://doi.org/10.1175/JHM-D-15-0034.1.
SYLVESTRE, F.; SERVANT, M.; SERVANT-VILDARY, S.; CAUSSE, C.; FOURNIER, M.; YBERT, J-P., Lake-Level Chronology on the Southern Bolivian Altiplano (18°–23°S) during Late-Glacial Time and the Early Holocene, Quaternary Research, v. 51, n. 1, p. 54-66, 1999, doi: https://doi.org/10.1006/qres.1998.2017.
TRENBERTH, K.E., Atmospheric Moisture Recycling: Role of Advection and Local Evaporation, Journal of Climate, v. 12, p. 1368-1381, 1998, doi: https://doi.org/10.1175/1520-0442(1999)012<1368:AMRROA>2.0.CO;2.
VAN BREUKELEN, M.R. et al., Fossil drip water in stalagmites reveals Holocene temperature and rainfall variation in Amazonia, Earth Planet. Sci. Lett., v. 275, n. 1-2, p. 54-60, 2008, doi: https://doi.org/10.1016/j.epsl.2008.07.060.
VAN DER ENT, R. J., A new view on the hydrological cycle over continents. Delft, 2014, 106 p., Thesis (Doctor in Civil Engineering and Geosciences-Water Management) - Delft University of Technology, Delft, 2014.
69
VAN DER HAMMEN, T., The Pleistocene changes of vegetation and climate in tropical South America, J. Biogeogr, v.1, p. 3-26, 1974, doi: 10.2307/3038066.
VAN DER HAMMEN, T.; ABSY, M. L., Amazonia during the last glacial. Palaeogeogr. Palaeoclimatol. Palaeoecol., v. 109, p. 247-261, 1994, doi: https://doi.org/10.1016/0031-0182(94)90178-3.
VERA, C et al., Toward a Unified View of the American Monsoon Systems, Journal of climate, v. 19, p. 4977-5000, 2006, doi: https://doi.org/10.1175/JCLI3896.1
VILLACÍS, M.; VIMEUX, F.; TAUPIN, J. D. Analysis of the climate controls on the isotopic composition of precipitation (δ18O) at Nuevo Rocafuerte, 74.5°W, 0.9°S, 250m, Ecuador.C. R. Geosci, v. 340, n. 1, p. 1–9, 2008, doi: https://doi.org/10.1016/j.crte.2007.11.003.
VIMEUX, F.; GALLAIRE, R.; BONY, S.; HOFFMANN, G.; CHIANG, J.C.H, What are the climate controls on δD in precipitation in the Zongo Valley (Bolivia)? Implications
for the Illimani ice core interpretation, Earth Planet. Sci. Lett., 240, p. 205–220, 2005, doi: 10.1016/j.epsl.2005.09.031.
VONHOF, H.B. et al., Fluid inclusion hydrogen and oxygen isotope analyses using the “Amsterdam Device”: a progress report, Geophysical Research Abstracts, v. 9, 2007.
VON RANDOW, C. et al., Comparative measurements and seasonal variations in energy and carbon exchange over forest and pasture in southwest Amazonia, Theor. Appl. Climatol., v. 78, p. 5-26, 2004, doi: 10.1007/s00704-004-0041-z
VOOSE, P., New geological age comes under fire, Science, v. 361, p. 537-538, 2018, doi: 10.1126/science.361.6402.537.
VUILLE, M.; WERNER, M., Stable isotopes in precipitation recording South American summer monsoon and ENSO variability - observations and model results, Climate Dyn., v. 25, n. 4, p. 401-413, 2005, doi: https://doi.org/10.1007/s00382-005-0049-9.
VUILLE, M., Current state and future challenges in stable isotope applications of the tropical hydrologic cycle (Invited Commentary), Hydrological Processes, v. 32, p. 1313-1317, 2018, doi: https://doi.org/10.1002/hyp.11490.
WANG, X.; EDWARDS, R. L.; AULER, A.S.; CHENG, H.; KONG, X.; WANG, Y.; CRUZ, F. W.; DORALE, J. A.; CHIANG H., Hydroclimate changes across the Amazon lowlands over the past 45,000 years, Nature, v. 541, p. 201-207, 2017, doi: https://doi.org/10.1038/nature20787.
WANG, X.; LI, W; ZHU, J.; TANAJURA, C.A.S., The strengthening of Amazonian precipitation during the wet season driven by tropical sea surface temperature forcing, Environ. Res. Lett., v. 13, p. 1-12, 2018, doi: https://doi.org/10.1088/1748-9326/aadbb9.
70
WALKER M. et al., Formal ratification of the subdivision of the Holocene Series/Epoch (Quaternary System/Period): two new Global Boundary Stratotype Sections and Points (GSSPs) and three new stages/subseries, Episodes, v.41, n. 4, p. 213–223, 2018, doi: https://doi.org/10.18814/epiiugs/2018/018016.
WILLE, M.; HOOGHIEMSTRA, H.; VAN GEEL, B.; BEHLING, H.; DE JONG, A.; VAN DER BORG, K., Submillennium-scale migrations of the rainforest-savanna boundary inColombia: 14C wiggle-matching and pollen analysis of core Las Margaritas, Palaeogeogr. Palaeoclimatol. Palaeoecol., v. 193, n. 2, p. 201-223, 2001, doi: https://doi.org/10.1016/S0031-0182(03)00226-8.
WINDHORST, D.; WALTZ, T.; TIMBE, E.; FREDE, H.-G.; BREUER, L., Impact of elevation and weather patterns on the isotopic composition of precipitation in a tropical montane rainforest, Hydrol. Earth Syst. Sci., v. 17, p. 409–419, 2013, doi: https://doi.org/10.5194/hess-17-409-2013.
ZEMP, D.C.; SCHLEUSSNER, C.-F.; BARBOSA, H.M.J.; VAN DER ENT, R.J.; DONGES, J. F.; HEINKE, J.; SAMPAIO, G.; RAMMIG, A., On the importance of cascading moisture recycling in South America, Atmos.Chem.Phys., v. 14, n. 23, p. 13337–13359, 2014, doi: https://doi.org/10.5194/acp-14-13337-2014.
ZEMP, D.C., Feedbacks between vegetation and rainfall in the Amazon basin. Berlin, 2016, 160 p. Dissertation (Doctor in Geography) - Humboldt-Universität zu Berlin, Berlin., 2016.
ZEMP, D.C. et al., Self-amplified Amazon forest loss due to vegetation-atmosphere feedbacks, Nature Communications, v. 8, p. 1-10, 2016, doi: 10.1038/ncomms14681.