{"paper_id":"233d56a2-7b8d-4a0d-a0b1-ce11b807c4de","body_text":"Full title: Baobab isotope records and rainfall forcing in southwest \nMadagascar over the last 700 years.\nShort title: Baobab isotope: a rainfall proxy records from Madagascar \nAuthors: Estelle Razanatsoa1¶*, Lindsey Gillson1, Grant Hall2, Malika Virah-Sawmy3, \nStephan Woodborne3,4¶\n1Plant Conservation Unit, Department of Biological Sciences, University of Cape \nTown, Rondebosch, South Africa\n2Stable Isotope Laboratory, Mammal Research Institute (MRI), University of Pretoria, \nSouth Africa\n3Humboldt-Universitat zu Berlin, Geography, Germany\n4iThemba LABS, Johannesburg, South Africa\nCorresponding author: \nE-mail: estellebota@gmail.com\n¶These authors contributed equally to this work.\n&These authors also contributed equally to this work.\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n1 ABSTRACT\n2 Highly resolved climate records for Madagascar are scarce but are essential for understanding \n3 of rainfall drivers over time and assessing the risks and likely trajectories of future climate \n4 change. We measured variation in the carbon isotopes of baobabs (Adansonia spp.) which \n5 reflect rainfall in southwest Madagascar. The record indicates a decreasing trend of rainfall \n6 over the last 700 years with high variability at a centennial-scale. The duration of wetter \n7 periods decreased over time with the wettest periods between 1350 – 1450 CE, after the onset \n8 of the Little Ice Age, while the driest period occurred between 1600 – 1750 CE, during the \n9 Maunder Minimum. The results suggest that decadal to centennial rainfall variability in \n10 southwest Madagascar is dominated by tropical forcing rather than subtropical forcing. Wetter \n11 periods are regulated by the movement and migration of easterly winds linked to the \n12 Intertropical Convergence Zone, while dry periods are influenced by the effect of the Pacific \n13 Decadal Oscillation linked to the El Niño Southern Oscillation and the sea surface \n14 temperature variation in the Southwestern Indian Ocean. The Southern Annular Mode is \n15 significantly correlated with the record, but its effect was only visible at the beginning of the \n16 record around 1300 CE. This evidence provides a new understanding of rainfall across \n17 southern Africa and the interaction of global forcing with regional factors. Further \n18 investigation is required to improve tree chronology from Southern Hemisphere and \n19 understand the migration of the westerlies and its potential future effect on the rainfall in \n20 Madagascar. Understanding the interplay between tropical and other rainfall forcings will be \n21 essential in assessing likely scenarios of resilience, and adaptive capacity of social-ecological \n22 systems in Madagascar. \n23 Keywords: Baobab, carbon isotopes, radiocarbon dates, rainfall drivers, ITCZ, Late Holocene, \n24 Madagascar, dendroclimatology, Pacific Decadal Oscillation, Sea Surface Temperature\n25\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n26 1. INTRODUCTION\n27 The African continent has been tangibly affected by climate change in the last decades, \n28 manifested by increasing temperatures and rainfall variability (IPCC, 2021; Baptista et al., \n29 2022; Weathering Risk, 2023; Onyeaka et al., 2024). In parts of southern Africa, decreased \n30 rainfall with more pronounced and recurrent severe drought events are expected in the near \n31 future (e.g Ingram & Dawson, 2005; Heland & Folke, 2014; Hoscilo et al., 2015; Baudoin et \n32 al., 2017; Serele et al., 2020; IPCC, 2021). Understanding rainfall drivers requires knowledge \n33 of local and regional synoptic changes that occur at decadal to centennial scales (Scroxton et \n34 al., 2017). Madagascar plays a role in the regulation of climate across southern Africa through \n35 its topographic influence on the Mozambique Channel Trough (MCT, Barimalala et al., \n36 2018). Madagascar has an east-west rainfall gradient associated with its topography which \n37 reduces the influence of easterly trade winds that bring moisture from equatorial Indian Ocean \n38 (Jury, 2016; Fig. 1). A north-south rainfall gradient derives from the north-westerly monsoon \n39 as the ITCZ crosses Madagascar to 20°S during the austral summer (Donque, 1975; Jury & \n40 Huang, 2004; Tadross et al., 2008; Scroxton et al., 2017). While the southwest region derives \n41 moisture from the monsoon and tropical cyclone events in the South Indian Ocean (east \n42 Madagascar), and the Mozambique Channel (west-southwestern Madagascar) (Ho et al., \n43 2006), it is the driest area on the island with <600 mm of rainfall per year with recurrent \n44 drought events (Ganzhorn, 1995; Tadross et al., 2008). The southern region of Madagascar \n45 has experienced limited rainfall at least in the last few decades (Serele et al., 2020; Harrington \n46 et al., 2021; Harrington et al., 2022; Otto et al., 2022). The region is also vulnerable to future \n47 global climate changes as simulations suggest more severe and longer dry seasons in the \n48 tropics, and an average global rise in temperature of 1.5°C estimated to be reached by 2030 \n49 (Christensen et al., 2007; IPCC, 2021; Barimalala et al., 2021). This forecast is associated \n50 with large uncertainty due to a lack of understanding of seasonal, annual, decadal, and \n51 centennial rainfall trends, and this prevents proper assessment of adaptation and hazard \n52 reduction in the area (Serele et al., 2020).\n53\n54 On an annual to decadal scale, droughts in southern Africa have been associated with El Niño \n55 Southern Oscillation (ENSO; Baudoin et al., 2017). Austral summer rainfall is controlled \n56 primarily by the seasonal interplay between subtropical high-pressure systems and the \n57 migration of easterly flows associated with the Intertropical Convergence Zone (ITCZ) \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n58 (Chase & Meadows, 2007; Sachs et al., 2009). The position and intensity of tropical-\n59 temperate troughs (TTTs) and their associated cloud bands that form in response to this \n60 interplay (Mason & Jury, 1997) are closely linked to sea surface temperature (SST) conditions \n61 in the Southwest Indian Ocean (SWIO) (Reason & Mulenga, 1999; Nash, 2017, Fauchereau et \n62 al., 2009). These synoptic dynamics are derived from observational data, longer term data are \n63 needed to understand trends and the interplay between the spatial dynamics of rainfall and the \n64 different drivers at decadal and centennial scales. This is especially true for Madagascar \n65 where rainfall variability is intrinsically high; there is a high dependence in rain-fed \n66 agriculture; and environmental vulnerability is high.\n67 The limited climate analyses from the island show drying trends at millennial, centennial to \n68 decadal scales (Ganzhorn, 1995; Tadross et al., 2008; Virah-Sawmy et al., 2016). ENSO and \n69 the seasonality of the ITCZ are essentially tropical forcing mechanisms, but recent evidence \n70 suggests that under glacial climate forcing, there is an influence from the sub-tropics that is \n71 related to the latitudinal position of the southern hemisphere westerlies (Hahn et al., 2021). \n72 The South Annular Mode (SAM) is a position index of the westerly vortex, but little is known \n73 about its past and current effect on rainfall in the region (Nash, 2017) although it is known to \n74 affect circulation on weekly to centennial time scales (e.g., Trenberth, 1979; Thompson & \n75 Wallace, 2000). \n76\n77\n78\n79 Fig. 1: (A) Location of the four baobabs that were analysed from southwest \n80 Madagascar (red) and austral summer synoptic features in southern Africa, including the Inter \n81 Tropical Convergence Zone (ITCZ) during the austral summer, Tropical Temperate Troughs, \n82 Southwest Madagascar Coastal Current (SMACC), Agulhas Current and Mozambique \n83 Channel Current. The locations of the SST coral records from Ifaty in southwest Madagascar \n84 and the speleothem record from Anjohibe cave in northwest Madagascar, stalagmites records \n85 from Asafora cave and additional tree records (hexagons) from southern Africa are shown. \n86 (B) Madagascar rainfall gradient with the four trees (circle) and other proxy records used in \n87 this study (rectangle). \n88 Paleoclimate proxy records  \n89 Paleoclimate records contain information at various temporal and spatial scales, which \n90 enables the interpretation of trends, variability, and the underlying processes within the \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n91 climate system. Paleoclimate proxy records derived from ice cores (e.g. Svensson et al., 2006; \n92 Thompson et al., 2003), lake and wetland sediments (e.g. Neukom & Gergis, 2011; \n93 Verschuren, 2003), and speleothems (e.g. Holmgren et al., 1999; Voarintsoa et al., 2017; \n94 Scroxton et al., 2017) provide a deep time record of global-scale climate forcing, while coral \n95 reefs (e.g. Zinke et al., 2014) and tree rings (e.g. Cook et al., 2002; Helama et al., 2005) yield \n96 shorter, more localised records of climatic responses. In temperate environments, abundant \n97 records of tree ring widths and varved sediments growth can be used to infer rainfall regimes, \n98 but records with similar resolution are rare in tropical ecosystems. Recently the possibility of \n99 reconstructing past climate variability using carbon isotope ratios (δ13C) from the growth \n100 rings of subtropical trees has emerged (Robertson et al. 2006; Woodborne et al., 2015; 2016). \n101 This increases the potential for paleohydrological reconstruction in the subtropics, and the \n102 approach holds the potential to address the climate data deficit in Madagascar which has \n103 numerous species of long-lived, dry adapted trees such as the baobab (Baum & Baum, 1995; \n104 Baum et al., 1998; Razanamaro et al., 2015).\n105 Carbon isotope in tree rings as a climate proxy \n106 The use of δ13C as a climate proxy in tree growth is based on environmental regulation of \n107 carbon isotope fractionation during photosynthesis. Fractionation is controlled by stomatal \n108 conductance, which manifests in the ratio of leaf internal CO2 concentration and the \n109 atmospheric CO2 concentration (ci/ca). Many environmental factors control stomatal \n110 conductance (Farquhar et al., 1982; Tieszen, 1991), but in ecosystems where water stress is \n111 the main control, reducing the ci/ca and increases the δ13C values of plants (more positive) in \n112 dry conditions. In low rainfall regions, water stress responses in trees relate to edaphic water \n113 availability which is directly linked to rainfall (Farquhar et al., 1982). In the southern African \n114 subtropics, the δ13C value of a specific growth ring in a tree potentially proxies the rainfall \n115 conditions during the period when the ring was formed. A negative correlation consistent with \n116 the theoretical expectation was found between rainfall and the δ13C values of baobab rings in \n117 Southern Africa (Woodborne et al., 2015). Low δ13C values infer higher rainfall while high \n118 δ13C values are indicative lower precipitation, but a direct transfer function accounting for \n119 evaporation, rechange and runoff has not been established, and so the proxy record reflects \n120 relative changes in effective rainfall through time. The isotopic records from southern African \n121 baobabs allow the investigation of the local rainfall drivers of rainfall at a higher spatial \n122 resolution, and they reveal decadal, multi-decadal and centennial variability of rainfall over \n123 the last 1000 years (Woodborne et al., 2015; 2016). In this paper, we describe new proxy \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n124 rainfall records for the southwest region of Madagascar during the last millennium based on \n125 δ13C time series from several baobab trees and evaluate the potential drivers of centennial \n126 variability of rainfall in this area.\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n127 2. MATERIAL AND METHODS\n128 2.1. Study setting\n129 This research focuses on southwest Madagascar, the driest region of the island (Fig. 1). The \n130 area has a semi-arid climate with irregular rainfall (Donque, 1975; Jury, 2016) decreasing \n131 from 600 mm to less than 300 mm per year towards the south and the coast. Most of the \n132 rainfall occurs during the austral summer (November – March) with monthly rainfall between \n133 300 mm for January and <2 mm in June, July, and August with an above average dry season \n134 rainfall recorded between 1983 and 1990, while wet seasons had elevated precipitation 2000 – \n135 2015 (World Bank 2017). \n136 This region supports three baobab species, Adansonia grandidieri, A. za and, A. rubrostipa, \n137 and all three species offer the opportunity to establish paleoclimate records as they are long-\n138 lived with distinct radial growth rings (Patrut et al., 2015; Woodborne et al., 2015, 2016). \n139 2.2. Sample collection\n140 Sampling was conducted in 2015 on four living trees following a north-south transect of \n141 southwest Madagascar to cover the regional climate. The trees were coded as DFL, DFS, \n142 GTR and TSP based on their location from north within the dry forest vegetation to the spiny \n143 thickets vegetation in the south (shown in Fig. 1; Table 1). The trees were selected and cored, \n144 based on their large size (>7 m in circumference) and location (>1 km from any possible \n145 surface water source, such as marshes and lakes that could obscure the rainfall signal due to \n146 the buffering effects of local hydrology). In addition to official permits for sample collection \n147 and exportation from the Ministry of Environment and Sustainable Development of \n148 Madagascar and Madagascar National Parks, local communities were consulted and gave \n149 permission for the sampling to proceed. Cores were carefully extracted from selected trees \n150 using a Haglöf 10mm diameter increment borer at a position of 1.30 m above the ground. \n151 After core extraction from each tree, the hole was sealed with a commercial tree sealant to \n152 prevent any potential damage to the tree by insects or fungus infestation (Tsen et al., 2015). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n153 Table 1: Core information from trees sampled along a north-south transect in southwestern \n154 Madagascar.\nSpecies \nand \nsample ID\nCoordinates Elevati\non (m)\ncore \nlength \n(cm)\nTree \ncircumferenc \nat breast \nheight (m)\nNumber of \n𝛿13C \nsamples \nNumber of \n14C AMS \ndates \nA.gran 01 – \nDFL\n-20.2224167 ºS\n44.4277500 ºE\n25 97.5 8.8 433 12\nA.gran 04 – \nDFS\n-20.8207222 ºS\n44.3940000 ºE\n20\n92 11.9 445 12\nA.gran 05 – \nGTR\n-21.8602778 ºS\n43.8670556 ºE\n23\n91 12.2 450 13\nA. za – TSP\n-23.8863889 ºS\n44.2583889 ºE\n88\n139 7.7 706 15\n155\n156 2.3. Tree chronology\n157 The chronologies of the cores were determined from 52 AMS radiocarbon dates: respectively \n158 12 for DFL, 12 for DFS, 13 for GTR and 15 for TSP (Supplementary Information SI1). \n159 Samples for radiocarbon dating were selected based on conspicuous shifts in the 𝛿13C time \n160 series. Samples were processed at iThemba LABS in Johannesburg (South Africa). Samples \n161 were pre-treated using the acid-base-acid (ABA) method (Hajdas et al., 2017). This treatment \n162 consists of the use of acid 0.5M HCl at 60°C, followed by a weak base of 0.1M NaOH at \n163 60°C, to dissolves humic acids and finally a hot acid wash of 0.5M HCl at 60°C, removing \n164 any carbonates that precipitated from modern atmospheric carbon dioxide. Following each \n165 step, the samples were rinsed until a neutral pH was obtained (Hajdas et al., 2017). The dates \n166 were calibrated with the 2020 Southern Hemisphere calibration from the online calibration \n167 programs CALIB 14C version 7.10 (Hogg et al., 2020) and the Calibomb program \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n168 (http://calib.org/CALIBomb/) from Queens University, Belfast, which uses the bomb carbon \n169 dataset of Hua et al. (2013). \n170 The age model uses wiggle matching between the trees to constrain possible radiocarbon \n171 calibration intercepts, as was used in previously published tree records from southern Africa \n172 (Woodborne et al., 2015; 2016). Linear age interpolation of the chronology was used to assign \n173 an age to each 𝛿13C analysis. The approach is a pragmatic attempt to use the climate data as \n174 additional apriori constraints on the age of samples. In Bayesian age models, the relative \n175 chronology (x is known to be older than y based on stratigraphy, or in the case of trees, the \n176 radial growth) is used as an apriori constraint, and these methods provide a means of \n177 propagating analytical errors in a series of radiocarbon dates. Our approach notes that within \n178 windows of 50-100 years (the kind of error that may be expected in a radiocarbon age model) \n179 there are sequences of extreme rainfall and drought events that are common across different \n180 trees, as is the atmospheric δ13C Suess Effect, and these should be presumed to be \n181 synchronous because they are driven by large scale synoptic or earth system dynamics. In this \n182 way climatology in addition to the radiocarbon dates provide many more apriori constraints \n183 on the chronology than the traditional Bayesian models based solely on the radiocarbon dates. \n184 None of the existing Bayesian age model packages provide a convenient way to integrate \n185 these additional constraints. The argument has a risk of circularity because the chronology is \n186 required to generate the climate record, and the climate record is required to refine the \n187 chronology, but all attempts to reconcile climate sequences are constrained by the requirement \n188 that the age model must fall within the 1-sigma calibration ranges for the radiocarbon dates \n189 (with the exception of patent erroneous radiocarbon dates that show clear age inversions in \n190 the radial growth of trees, and which should be dismissed as errors, possibly in sample \n191 labeling). An implication of this approach is that a date on an individual tree becomes a \n192 constraint on the age model of all the trees for that particular time period. The approach that \n193 we use yields a chronology that differs very little from Bayesian models but provides a more \n194 coherent climate record that more accurately represents the regional rainfall variability \n195 through time. The major flaw in the approach is the inability to calculate errors on the age \n196 model, although the additional constraints from the climate record, and the demand for \n197 coherence in large scale climate events across trees likely yields a tighter chronology than a \n198 traditional Bayesian model based only on the within-tree radiocarbon analyses alone. The \n199 record obtained from the isotopic values and the age model, was therefore subject to a 21-\n200 year-biweight mean analysis which suppresses uncertainties in the age model and also the \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n201 localised effect of synoptic changes over stochastic weather processes (Woodborne et al., \n202 2015; 2016). The approach prevents any analysis of temporal patterns that are resolved at less \n203 than decadal time scales.\n204 2.4. Stable carbon isotope analysis\n205 The δ13C measurements were conducted at the Stable Isotope Laboratory at the University of \n206 Pretoria. Each core was mounted on a wood backing panel so that their axial orientation of the \n207 baobab core could be seen. The exposed half of the core was sub-sampled from the bark to the \n208 pith while being sensitive to the observable direction of the tree growth ring. The remaining \n209 half of the core are preserved as an archive. The number of samples per core is provided in \n210 Table 1. Each sample was placed in an individually labelled reaction vessel and subject to a \n211 soxhlett extraction in a 2:1 toluene/ethanol mixture to eliminate the mobile and soluble \n212 components in the wood. Followed by α-cellulose extraction in 17% sodium hydroxide \n213 (NaOH) and 10% sodium chlorite (NaClO2) at different concentrations to remove the lignins \n214 and eliminate the hemicellulose of the wood. The last steps consisted of covering the samples \n215 with 1% Hydrochloric acid (HCl) solution for 10 minutes, rinsed and dried overnight at 70 \n216 °C(Loader et al., 1997; McCarroll & Loader, 2004; Hall et al., 2009). Once dried, aliquots of \n217 the α-cellulose were weighed (0.050 - 0.060 mg) using a Mettler Toledo MK5 microbalance \n218 and folded into tin capsules before isotopic analysis. The samples were combusted at 1020°C \n219 in an elemental analyzer (Flash EA 1110 Series) coupled to a Delta V Plus stable light isotope \n220 ratio mass spectrometer, via a ConFlo III system (all equipment supplied by Thermo Fischer, \n221 Bremen, Germany). An in-house running standard of Malaysian wood (Shorea superba) with \n222 an isotopic value of -28.4‰ was used (Hall et al., 2008; 2009; Woodborne et al., 2015; 2016) \n223 to calibrate the results. A blank and in-house standard sample which was calibrated against a \n224 large number of international standards was run after every 12 samples. The results were \n225 reported in δ13C relative to Vienna-PDB (VPDB) and expressed as per mille according to the \n226 formula:\n227 δ13C = (Rsample / Rstandard -1)1000٭ \n228 Where δ13C is the isotopic composition of the sample and R indicates the ratio of 13C/12C in \n229 the sample. Sample replication including sample pre-treatment and error in the analysis were \n230 <0.2‰. This is based on a total of 1936 analyses conducted for the overall project, giving 161 \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n231 standards approximately. The precision achieved here is similar to that reported by other \n232 stables light isotope laboratories.\n233 2.5. Isotope correction and analysis \n234 Tree physiological responses to environmental conditions are reflected in the δ13C ratios of \n235 the wood tissue (McCarroll & Loader, 2004), but the carbon isotope ratios are also affected by \n236 changes in the isotopic ratio of atmospheric CO2 over time (Farquhar et al., 1982) and \n237 variation in intrinsic water-use efficiency (iWue) in response to elevated atmospheric CO2 \n238 concentrations since the industrial revolution (Wang & Feng, 2012; Wils et al., 2016). The \n239 isotopic data derived from each tree therefore needs to be corrected to isolate the local \n240 environmental response. This correction compensates for the δ13Catm variations using the \n241 global record of Belmecheri & Lavergne (2020) and is a simple normalisation to pre-\n242 industrial δ13Catm values (McCarroll & Loader 2004). In this study this is assigned to 1748 \n243 (the inflection date between pre- and post-industrial atmospheric changes in the Belmecheri & \n244 Lavergne (2020) dataset). In addition to the measured change in δ13Catm, there is also an \n245 increase in the atmospheric concentration of CO2 (ca), which affects the intercellular CO2 \n246 concentration (ci). Since the rate of carbon assimilation, and accordingly the δ13C ratio of \n247 wood tissue from each growth ring, is linked to the ci/ca ratio (McCarroll & Loader, 2004), \n248 reduced water transpiration is associated with elevated ca during photosynthesis, which \n249 increases the iWue of the plant. The correction used in this study followed the method of \n250 Woodborne et al. (2016), which is equivalent to the δ13C pin (“preindustrial”) correction \n251 method (McCarroll et al., 2009). \n252 A 21-year biweight mean (a weighted mean that considers the δ13C values from the decade \n253 prior to, and after the date in question) was calculated to emphasize the decadal and multi-\n254 decadal pattern in the record. The biweight mean also accommodates the possible age model \n255 errors that arise from using a linear growth model to approximate a growth pattern that is \n256 likely punctuated by slight variations in growth rates.\n257 The isotopic time series provides a regional decadal record of effective rainfall for the four \n258 trees. Each tree experienced unique local climate conditions over the last 700 years; and the \n259 composite record allows evaluation of the common and wider climate forcing by emphasizing \n260 commonalities in the low frequency component of the record. Change point analysis using the \n261 package change point in R was conducted on the composite records. Following this, the \n262 composite record was compared with published data for a range of potential rainfall drivers \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n263 including the local Southwestern Indian Ocean (SWIO) SST, and also equatorial climate \n264 shifts caused by the Indian Ocean Dipole (IOD) (Table 2). \n265 Table 2: Paleoclimate forcing used for comparison with the baobab isotope records.\nData Proxy Time covered \n(AD)\nRecord length \n(years)\nReferences \nSea Surface \nTemperature \n(SST, Ifaty)\nOxygen isotopes of coral 1660-1994 334 Zinke et al., 2022\nSouthern Annual \nMode Index (SAM) \nMid-latitude to polar \ndomain proxy records \n1000-2007 1007 Abram et al., \n2014\nPacific Decadal \nOscillation (PDO)\nTree rings 993-1996 1003 MacDonald & \nCase 2005\nHadISST1.1 SST \ndataset (Index \nreferring to IOD)\nCalculated anomalous \nSST gradient between the \nwestern equatorial Indian \nOcean and the south-\neastern equatorial Indian \nOcean. Based on coral \nisotopes and Ca/Mg \nratios.\n1981-2010 29 Saji & Yamagata \n2003 \n266\n267 The identification of monotonic trends within the record was conducted using a non-\n268 parametric Mann-Kendall trend test (Nasri & Modares, 2009; Pohlert, 2018) combined with a \n269 Least Squares Regression to evaluate the rate of change in rainfall per year in the regional \n270 records. In addition, gridded datasets downloaded from GPCC monthly total precipitation \n271 from around Betioky at 2.5° x 2.5° resolution were compared to the composite record. For \n272 correlations, significance at a level of 0.05 were accepted. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n273 3. RESULTS \n274 The calibrated radiocarbon dates for the four trees suggest that they grew over the last 700 years \n275 (1302 – 2013 CE) (Table SI1). The most parsimonious age models, that reconcile the AMS dates \n276 and stable carbon isotope records, are shown in Fig. 2. All the trees show linear growth over time \n277 except for GTR, which demonstrated a hiatus from 1500 CE to 1700 CE. This is not uncommon \n278 in baobabs (Patrut et al., 2017). The age model assigned most of the AMS ages within 1-sigma \n279 error of 68%. \n280\n281 Fig. 2: Age-models for (A) DFL, (B) DFS, (C) GTR with the hiatus indicated in dashed line, and \n282 (D) TSP, based on 52 radiocarbon dates. The horizontal lines are the 1-sigma calibration \n283 intervals for the radiocarbon dates. The bold line represents the age model that best intercepts the \n284 1-sigma calibration range for the radiocarbon dates\n285 The corrected δ13C time series from the four trees range between -26.2‰ (1400 CE) and -24.5‰ \n286 (1650 CE) with a mean of -25.3‰, and a variation of about 1.7‰ (Fig. 3). The trend analysis on \n287 the composite δ13C records using the Mann-Kendall test shows that there is a marginal \n288 decreasing trend of the isotope data over time (S=-5.07, p<0.01) with a difference of about -\n289 0.7‰ between 1300 CE and 2013 CE. A least square linear regression is significant (F= 80.6, p \n290 = 0.001) but with an R2 of 0.10. The change point analysis revealed a number of major shifts in \n291 the mean values of the composite isotope data becoming either more positive or negative in the \n292 time series (significant at 95%). These occur at approximately 1400 CE, 1480 CE, 1500 CE, \n293 1630 CE, 1660 CE, 1820 CE, and 980 CE (Fig. 4). \n294\n295 Fig. 3: Corrected δ13C time series of four baobabs from southwest Madagascar with inverted y-\n296 axis indicating drier (less negative isotope value) and wetter conditions (more negative isotope \n297 value). \n298\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n299 Fig. 4: The composite record from four trees (black) with wetter periods (blue) and drier periods \n300 (yellow). Error bars represent standard errors.\n301 The isotope biweight mean composite and the GPCC monthly total precipitation both shows \n302 similar trends over time. A linear regression of precipitation data from the GPCC monthly total \n303 precipitation from 1900–present revealed a decreasing but not significant trend over time (β = -\n304 0.67, p = 0.17), with year explaining only ~1.7% of the variance (R² = 0.017). Conversely, the \n305 isotope data showed a small but significant increasing trend (β = +0.00112, p = 0.015) reflecting \n306 decreasing rainfall, though the model explains less than 1% of variance (R² = 0.0099). These \n307 results suggest subtle but differing temporal behaviours in the two proxies potentially associated \n308 with the associated ages. Both records show a slight decrease in rainfall as reflected by the \n309 precipitation value and the more positive isotope value around 1950 and then from 1970-1990 \n310 while more wetter periods are recorded around 1960 and 2000 (Fig. 5). \n311\n312 Fig. 5: Comparison of the baobab δ13C records with existing model datasets: Black indicates the \n313 baobab δ13C composite records from 1900-2015 with linear regression indicated in grey. Blue \n314 and dark blue shows GPCC total precipitation at 2.5° x 2.5° resolution from around Betioky \n315 between 1900-2013 along with the associated linear model. \n316 The correlation analysis of potential rainfall drivers in Southwest Madagascar has been \n317 calculated (Table 3). The results diverse correlation values and significance (Table 3). There is a \n318 negative correlation between SST and the isotope data an overall significant positive correlation \n319 between SAM and the isotope records with recorded weak correlation since the 16th century. \n320 Related to the Pacific Decadal Oscillation (PDO) reconstruction,  no significant correlation has \n321 been recorded during the entire period  but during the LIA negative PDO anomalies are \n322 associated with decreases in rainfall particularly between 1600 – 1750 CE (r=-0.30, p<0.001), \n323 whereas in the record before and after this period, they are associated with increases in rainfall \n324 (r=0.18, p=0.004 and r=0.18, p=0.001 respectively) (Fig. 7C). \n325\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n326 Table 3: Correlation between baobab isotope data and various environmental indices.\n327 Pearson correlation coefficients (r) and associated p-values are shown. Sample size (n) indicates \n328 the number of data points compared for each dataset.\nDataset Compared with Baobab \nIsotope Data\nTime Period \n(AD)\nSample Size \n(n)\nPearson \nCorrelation \n(r)\np-value\nSea Surface Temperature (SST, \nIfaty) 1660–1994 334 –0.22* < 0.001*\nSouthern Annual Mode Index \n(SAM) 1000–2007 1007 0.16* < 0.001*\nPacific Decadal Oscillation (PDO) 993–1996 1003 0.01 > 0.05\nHadISST1.1 SST dataset (IOD \nIndex) 1981–2010 29 –0.06 > 0.05\n329 Note: * indicates significance at α = 0.05.*\n330 4. DISCUSSION\n331 4.1. Rainfall record of southwest Madagascar for the last 700 years \n332 The similarities between the δ13C records and the GPCC precipitation data meets the theoretical \n333 expectation, and reaffirms the results obtained for baobab isotope controls on the African \n334 mainland (Woodborne et al., 2015; 2016; Fig. 5). The baobab δ13C record can thus be interpreted \n335 as a proxy for local effective rainfall in southwest Madagascar, reflecting decadal to centennial \n336 variability in the last 700 years. Accordingly, lower δ13C values indicate wetter periods while \n337 more positive isotope ratios correlate to drier conditions.\n338 The chronology of the four baobabs ranges from 1300 CE until 2013 with a hiatus in the GTR \n339 core from 1500 CE to 1700 CE. Punctuated growth models have been noted in other baobabs \n340 (Patrut et al., 2017) and may arise because of attenuated growth rates over time, or possibly due \n341 to lobate development of the stem resulting in the reallocation of resources to another part of the \n342 tree. The GTR baobab was collected near the River Mangoky in southwest Madagascar where a \n343 sediment core was taken from nearby Lake Tsizavatsy. The age-depth model of the Tsizavatsy \n344 core shows a hiatus from 1400 to 1900, suggested to be associated with a regional drought event \n345 (Razanatsoa et al., 2021). Although the other trees show positive isotope excursions (dry events) \n346 during this period, there are oscillations of wetter conditions covered by the present records (Fig. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n347 3). Notwithstanding the gap, the record from GTR prior to and after the hiatus was combined \n348 with the composite record of all trees to provide a full rainfall proxy record for the last 700 years \n349 for southwest Madagascar covering the Little Ice Age (LIA; Lechleitner et al., 2017; Putnam & \n350 Broecker, 2017) and the Anthropocene. \n351 Synchronicity in the baobab δ13C records from southwest Madagascar demonstrate regional \n352 variability with a succession of wet and dry cycles on the scale of decades to centuries (Fig. 4). \n353 The rainfall in the region was suggested to be variable with drought events recorded in \n354 Southwest Madagascar prior 1000 CE based on stalagmite records (Faina et al., 2021) with \n355 drying trends being recorded on the isotope tree records since 1300 CE. Both records, baobab \n356 δ13C and stalagmite δ¹⁸O show similar trends and patterns demonstrating high agreement in the \n357 variability of rainfall over time despite a weak negative Pearson correlation between the two \n358 variables (See Table 3). This non correlation could be explained by the uncertainties associated \n359 with the two age models. Before 1700 CE, both records show different trends with discrepancies \n360 around mid-17th century where the tree record showed drier conditions while the stalagmites is \n361 showing wetter conditions. Post 1700 CE similarities between the stalagmite 𝛿18O and 𝛿13C tree \n362 records are recorded with wet condition around 1700-1800 CE, dry condition 1800-1900 CE and \n363 a wetter condition again between 1900-2000 CE. Very recently (post-1950 CE), both records \n364 reflected a severe drought (Fig. 6). \n365 Fig. 6: Comparison between baobab δ13C records from tree rings in southwest Madagascar \n366 (black) with stalagmite δ18O records from Asafora cave from the southwest coast and just \n367 southeast of the Velondriake Marine Reserve (green, from Faina et al., 2021). \n368 The composite baobab record shows increasing δ13C values indicating marginal drying at the \n369 locations of the trees over the past 700 years, corresponding with previous findings suggesting \n370 increasing aridity since 1000 CE (e.g. Burney, 1993; Burns et al., 2016; Virah-Sawmy et al., \n371 2016, Razanatsoa et al., 2021). Pollen records of the last millennia show a synchronous \n372 desiccation from various regions in Madagascar including the southwest (Burney, 1993; Burns et \n373 al., 2016; Virah-Sawmy et al., 2016) which agrees with reduced length of wet periods in the tree \n374 records compared to regional records. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n375 4.2. Synoptic drivers of rainfall in southwest Madagascar \n376 When compared with other baobab δ13C records from South Africa (Woodborne et al., 2015; \n377 2016), the rainfall patterns for southwest Madagascar and the South African summer rainfall \n378 zone are in phase for most of the last 700 years. Wet (dry) periods in the South African Pafuri \n379 and Mapungubwe records (22 °S) corresponded to similar wet (dry) periods from southwest \n380 Madagascar. This suggests that southwest Madagascar rainfall responds to similar drivers of the \n381 summer rainfall zone in southern Africa, including Agulhas Current sea-surface temperature \n382 variations regulated by the East-West displacement of the TTTs (Woodborne et al., 2016) in \n383 addition to the regulatory effect of the island’s mountains (Donque, 1975; Jury & Huang, 2004). \n384 However, further investigation of the various drivers of rainfall is required to provide more \n385 understanding of local and regional temporal changes. \n386 4.2.1. ITCZ and SAM modulated rainfall during Early Little Ice Age between 1370 \n387 and 1500\n388 The composite baobab record from southwest Madagascar shows a wet period between \n389 approximately 1370 – 1500 CE (Fig. 4) which coincides with the second phase 1495 – 1833 CE \n390 of a wet-neutral-wet cycle recorded in northwest Madagascar (Scroxton et al., 2017). The \n391 movement of the ITCZ has a significant impact on rainfall variability in Madagascar (Haug et \n392 al., 2001; Liu et al., 2003; Verschuren et al., 2000; Schneider et al., 2014). A southward shift was \n393 recorded at the beginning of the Little Ice Age around 1300 CE (Chiang & Bitz, 2005; Broccoli \n394 et al., 2006; Lechleitner et al., 2017; Putnam & Broecker, 2017) and this may have resulted in \n395 increased rainfall for southwestern Madagascar. Wetter conditions are also evident in several \n396 East African sediment records during this period, including Lake Chilwa (Crossley et al., 1984), \n397 Lake Malawi (Johnson et al., 2001), Lake Massoko (Barker et al., 2000), Lake Tanganyika (Alin \n398 & Cohen 2003), Lake Victoria (Stager et al., 2005), Lake Naivasha (Verschuren et al., 2000; \n399 Tyson et al., 2001; Tierney et al., 2013) suggesting a common driver of rainfall, most likely the \n400 southwards movement of the ITCZ. \n401 The wetter conditions from 1370 – 1500 CE in southwestern Madagascar occur when the \n402 Southern Annular Mode (SAM) was at its most extreme negative phase in the fifteenth century \n403 (Abram et al., 2014; Fig. 7A). Negative SAM indices imply an expansion of the westerly \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n404 circumpolar vortex, which has significant impacts on temperature and precipitation over all four \n405 Southern Hemisphere continents (Gillett et al., 2006), and particularly over Africa south of 25°S, \n406 where an increase in precipitation is associated with the northward migration of the westerlies \n407 during the austral winter. The response is most pronounced along the east coast, where it is \n408 associated with anomalous easterly winds advecting more moisture off the SWIO (Gillett et al., \n409 2006). No significant response to SAM has been reported for Madagascar on the basis of \n410 instrumental data (Gillett et al., 2006) although heavy rainfall events have been associated to \n411 atmospheric circulation displays a Southern Annular Mode-like pattern throughout the \n412 hemisphere (Randriamahefasoa & Reason, 2017). The relationship between SAM and rainfall in \n413 the baobab record presented here is not consistent with an overall significant positive correlation \n414 (r=0.16, p<0.001). However, since the 16th century the relationship is weak and non-significant \n415 suggesting that the southward contraction of the westerly winds during the positive SAM reduces \n416 their influence on the region. When the westerlies migrate southward during a positive SAM \n417 phase, they cease to be a driver of rainfall, and decadal to centennial rainfall variability responds \n418 to other forcing. The lack of consistent correlation throughout the records also supports recent \n419 findings related to the lack of a forced response in SAM variability prior to the 20th century \n420 (King et al 2023). \n421 Our composite record suggests that rainfall responds to subtropical forcing during extreme \n422 negative SAM phases as the westerly winds migrate northwards bringing wetter condition to the \n423 subtropics including southern Madagascar (Fig. 7A), but otherwise it responds to tropical forcing \n424 determined by the position of the ITCZ during the austral summers. Evidence of this mechanism \n425 operating during glacial periods is derived from marine records and model analysis of the late \n426 Glacial maximum over southern Africa (Anderson et al., 2009; Sigman et al., 2010; Miller et al., \n427 2019b; Engelbrecht et al., 2019; Hahn et al., 2021). The mechanism may explain the rainfall \n428 maximum during the 14th and 15th centuries, but further research is needed to elucidate the \n429 influence of the SAM on southern African climate during the Little Ice Age. \n430 Fig. 7: Comparison of the baobab δ13C records (black) with (A) the Southern Annual Mode \n431 (SAM) index (orange) (Abram et al. 2014), (B) the Pacific Decadal Oscillation (PDO) index \n432 (blue) (MacDonald & Case 2005), (C) Sea Surface Temperature (SST) from Ifaty, southwest \n433 Madagascar (grey) (Zinke et al. 2022), and (D) HadISST1.1 SST dataset that refers to Indian \n434 Ocean Dipole variability  (red, Saji & Yamagata 2003) \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n435 4.2.2. Regional and localised rainfall drivers during the LIA 1600-1750\n436 Wet conditions during the 14th and 15th centuries are followed by extremely dry conditions 1600 \n437 – 1750 CE in southwest Madagascar. This period is characterised by a positive phase of the \n438 SAM that commences at 1500 CE (Abram et al., 2014) implying reduced effect of the westerlies. \n439 It is also a period of reduced sunspot activity known as the Maunder Minimum (1645 – 1715) \n440 (Mann et al., 2009; Scroxton et al., 2017) with decreased global temperatures at the maximum of \n441 the Little Ice Age (LIA; Tyson et al., 2001). It has been speculated that there was a southward \n442 migration of the ITCZ during the maximum of the LIA (Jury & Huang, 2004; Russel et al., 2007; \n443 Tadross et al., 2008; Voarintsoa et al., 2017). Our results show drier condition during this period \n444 starting around 1600 CE and has also been experienced in northwest of Madagascar around 1700 \n445 CE (Scroxton et al., 2017). Dry periods were also experienced across the African continents \n446 including East Africa (Russell & Johnson, 2007; Tierney et al., 2013), and the southern African \n447 summer rainfall area (Huffman, 2004; PAGES 2k Consortium, 2013; Macron et al., 2014; \n448 Chevalier & Chase, 2015; Huffman & Woodborne, 2016; Woodborne et al., 2016). Pollen \n449 evidence from the Lake Longiza in southwest Madagascar suggested an increase in grass and \n450 decrease in trees and shrubs such as Arecaceae, Pandanus, and Acacia potentially associated \n451 with the drying in the region (Matsumoto & Burney, 1994; Razanatsoa et al., 2022). Lake \n452 Tsizavatsy, from the same region, showed a hiatus in its sediment deposition between 1400 – \n453 1900 CE (Razanatsoa et al., 2021) suggesting the drying of the lake during this period, which is \n454 consistent with the tree records. The evidence of dry conditions from records across southern \n455 Africa does not support southward migration of the ITCZ during this period and if the migration \n456 occurred, there might have been other drivers that inhibited its effect leading to a decrease in \n457 rainfall during this period.\n458 The Pacific Decadal Oscillation (PDO) is linked to ENSO in relation to drought patterns across \n459 Africa through teleconnections that influence the longitudinal position of climate systems. The \n460 (PDO) reconstruction shows that during the LIA negative PDO anomalies are associated with \n461 decreases in rainfall particularly between 1600 – 1750 CE (r=-0.30, p<0.001), whereas in the \n462 record before and after this period, they are associated with increases in rainfall (r=0.18, p=0.004 \n463 and r=0.18, p=0.001 respectively) (Fig. 7C). Despite the change in the sign of the correlation, \n464 this evidence suggests that climate forcing is dominated by tropical forcing. (Thompson et al., \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n465 2003; MacDonald & Case 2005; Hoell et al., 2017). Changes in the TTTs position which tend to \n466 propagate eastward, from southern Africa to the Mozambique Channel and southern Madagascar \n467 is known to have a strong influence on intra-seasonal and even interannual rainfall variability in \n468 the region (Macron et al., 2014). This has particularly been suggested to be dictated by the \n469 migration of the ITCZ (Chase & Meadows, 2007; Sachs et al., 2009) and to increase during La \n470 Nina conditions (Manhique et al., 2011; Ratna et al., 2012; Macron et al., 2014). Moreover, its \n471 persistence was suggested to be maintained by variation of SST anomalies over the Agulhas \n472 Current (Manhique et al., 2011; Vigaud et al., 2007). Comparison of the baobab δ13C records \n473 from 1660 – 1994 CE with the 300-year Agulhas Current Sea surface temperature (SST) record \n474 from Ifaty in southwest Madagascar (Zinke et al., 2014) shows a significant negative correlation \n475 (r=-0.22, p<0.001). Positive (negative) SST corresponds with higher (lower) rainfall in the \n476 baobab record (Fig. 7B). The coolest oceanic temperatures in the coral record, with anomalies of \n477 -0.3 – -0.5 °C between 1675 – 1760 CE, correspond to the driest period in southwest \n478 Madagascar. Similar patterns of rainfall were noted in the summer rainfall area of the adjacent \n479 African mainland (Woodborne et al., 2015). Variation in SST in the western Indian Ocean are \n480 determinant in the IOD with a suggested negative relationship with eastern African rainfall \n481 responses (Hoell et al., 2017; Taylor et al., 2021). \n482 4.2.3. Mixed effect of changes in ITCZ position, human land-use and climate change \n483 from 1750 – 2013 CE\n484 Around 1750 CE until early 1800 CE, at the end of the LIA, there is a relatively wet period \n485 recorded in the baobab δ13C data. A relatively dry period after 1860 CE is similar to conditions \n486 experienced over the summer rainfall zone in southern Africa. These periods coincided with \n487 more extreme ENSO warm phases (Nash, 2017) with the warmest period in the Agulhas SST \n488 record between 1880 CE and 1900 CE and a northward migration of the ITCZ (Zinke et al., \n489 2014; Railsback et al., 2018). \n490 The comparison of the composite baobab δ13C record with the Dipole Mode Index (DMI) that \n491 were used as an index of the IOD (1870 – 2013) shows very low to no correlation (Table 3, Fig. \n492 7D) but with a noticeable positive but not significant correlation since 1980 CE (r= 0.27, p= \n493 0.08). The effect that the IOD has on equatorial climate forcing is similar to the ENSO or PDO \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n494 effects as it is driven by an equatorial SST differential across the Indian Ocean while ENSO is \n495 driven by a gradient in the Pacific Ocean. The effect in the subtropical region of southwestern \n496 Madagascar appears to be dominated by the Pacific Ocean influences on global climate \n497 (ENSO/PDO) which is influential in the latitudinal position of the TTTs system.\n498 Around 1950 CE, conditions in southwest Madagascar are as wet as at any time in the record. \n499 This corresponds to a positive SST anomaly, positive phase of IOD and a negative PDO phase. \n500 Despite suggested changes in the impact of ENSO cycles on the SST in the region of the SWIO \n501 since 1970 (Zinke et al., 2014), our results show typical PDO/rainfall phasing with more \n502 negative PDO corresponding to high rainfall while the inverse is not always true. Records \n503 suggest that the IOD intensified following the onset of global warming during the 20th century \n504 along with forced response of SAM (Abram et al., 2008; Namakura et al., 2009; Watanabe et al., \n505 2019; King et al 2023), with increased evidence of human induced climate change (IPCC, 2021), \n506 evidence of increased river runoff and shifts in human land-use through slash and burn \n507 agriculture were recorded in coral records from eastern Madagascar (Grove et al., 2013). Pollen \n508 records from the region suggest a decrease in the tree component including Arecaceae coinciding \n509 with an increase in Poaceae and pioneer taxa such as Asteraceae mostly likely associated with \n510 tree cutting associated with agriculture expansion (Razanatsoa et al., 2022). These suggest that \n511 changes in rainfall in the region led to changes in land-use. There is a return to drier conditions \n512 around 1980 called “belt of iron” also recorded in historical records peaking at the beginning of \n513 the 1990s (Von Heland & Folke, 2014). This was followed by a trend towards wet conditions in \n514 the past 20 years similar the instrumental record (Tadross et al., 2008). \n515 4.3. Implications for future climate change risk and adaptation \n516 The last 300 years of the baobab composite record show the interacting global, regional and local \n517 drivers that have influenced rainfall variability in southwest Madagascar. The dominant effect of \n518 the position of the ITCZ during the austral summer is evident, and this is troubling in the context \n519 of forecast climate change. The position and zone of the ITCZ is predicted to narrow with a \n520 northward shift over eastern Africa and the Indian Ocean and a southward shift in the eastern \n521 Pacific and Atlantic oceans by 2100, which would severely reduce rainfall in the region \n522 (Mamalakis et al., 2021). In terms of climate risk and adaptation, southwest Madagascar will \n523 likely experience a drier climate with more frequent and prolonged droughts as already predicted \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n524 in the IPCC 6th assessment report (IPCC, 2021). Climatic factors are an important driver of \n525 economic, environmental and societal decisions (Boutin & Smit, 2016) and would be crucial for \n526 the future of these dry areas. Indeed, the population are dependent on rainfall as a source of \n527 water and for agriculture due to the lack of infrastructure and the limited permanent water ponds \n528 (Hänke et al., 2017; Carriere et al., 2018). The effect of drought and lack of rainfall has already \n529 led the Mikea forager communities to diversify their livelihoods with seasonal agriculture to \n530 ensure food security (Razanatsoa et al., 2021). Some adaptations that have been established \n531 elsewhere and could be conducted in the region include the introduction of new drought resistant \n532 crops, (e.g. Thomas et al., 2007; Yaro et al., 2014). Multiple species livestock herding with cattle \n533 and goat pastoralism (Kaufmann & Tsirahamba, 2006; Hänke & Barkmann, 2017) and livelihood \n534 diversification including work that is not farming (e.g. crafts and services for local markets) were \n535 suggested to be a major adaptive strategy under drying conditions in a short and long term and \n536 buffer livelihoods in the face of environmental change (e.g. Kuiper  et al., 2007; Tambo & \n537 Abdoulaye, 2013). Further understanding of the effect of future climate on these populations and \n538 their surrounding environment is critical in planning strategies of adaptation in terms of \n539 livelihoods but also water provision also in the coming years. \n540 5. CONCLUSION\n541 Baobab δ13C data from southwest Madagascar are a proxy for changing rainfall over the last 700 \n542 years. The inferred wettest period was between 1370 – 1500 CE while the driest period occurred \n543 between 1600 – 1750 CE. High centennial variability was recorded with a decreasing rainfall \n544 trend and reducing duration of wet periods over time. The comparison of the records with \n545 existing records of rainfall drivers at local, regional, and global scales shows that the baobab \n546 rainfall proxy record is not dominated by the influence of any single forcing over the entire \n547 record. The westerlies may play a role during extreme negative phases of the SAM, while \n548 latitudinal shifts of the ITCZ are the dominant low frequency driver of rainfall. At a more local \n549 level, the role of SST seems to dominate variability possibly through longitudinal influences on \n550 the position of the TTTs system which is also influenced by the PDO/ENSO system. Localised \n551 climate forcing in relation to the Southwest Madagascar Coastal Current (SMACC) within the \n552 greater Agulhas Current system has been suggested (Ramanantsoa et al., 2018). What emerges \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n553 from comparisons with other rainfall proxy records on the island of Madagascar, and from the \n554 adjacent African mainland, is that the temporal trends are not consistent, probably reflecting the \n555 contrasting dominance of different drivers in different regions. A rainfall dipole exists between \n556 southern Africa and Madagascar (Jury et al., 2015; 2016; Woodborne et al., 2016; Barimalala et \n557 al., 2018) with increases in precipitation over southern Africa extending from Mozambique to \n558 Angola coincident with a decrease in rainfall most of Madagascar (Barimalala et al., 2018) but \n559 not in the southwest region. The mountains that extend from the north to the south of \n560 Madagascar (>1500 m elevation) reduce the direct transport of moisture from the Indian Ocean \n561 toward southern Africa (Barimalala et al., 2018) and southwest Madagascar. The Agulhas \n562 current SST forcing of Madagascar rainfall is opposite to that in southern African where negative \n563 SST anomalies were associated with wetter conditions over the southern African interior \n564 (Woodborne et al., 2015). These contradictions suggest that the variation in rainfall is not a \n565 simple intensification or weakening of the existing climate patterns, but rather a response in the \n566 synoptic systems to tropical (ENSO/PDO), extratropical (SAM), and localised (SST) forcing. \n567 Why some forcing appears to dominate at certain periods and not at others is unclear, and the \n568 evidence presented here suggests that synergistic effects might be explored in global climate \n569 models.\n570 The potential effect of climate change and land-use change were also recorded at the near present \n571 period, as well as possible effects of SAM if there is a northward migration of westerlies similar \n572 to what happened around 1300 CE. The data generated here provide the opportunity to unravel \n573 the relative importance and interaction between global, regional, and local drivers across the \n574 southern and eastern African region. These findings are crucial in the simulations of rainfall \n575 projections to help evaluate the impacts and trends of migration of the westerlies and \n576 anthropogenic induced climate change on the African continents that are not fully understood. \n577 For southwest Madagascar with an expected drier climate and increasing occurrence of severe \n578 drought conditions predicted for the near future, understanding the risks, and establishing \n579 adaptation strategies particularly in terms of livelihood could avoid disastrous famine. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n1. ACKNOWLEDGEMENT \nWe would like to thank the Ministry of Environment and Sustainable Development of \nMadagascar and Madagascar National Park for providing the permission for the field campaign, \nsampling and exportation.  We acknowledge ESSA-Forêts Mention Foresterie et Environnement \nde l’Ecole Supérieure des Sciences Agronomiques, Université d’Antananarivo – \nMADAGASCAR for collaborating with the obtention of the research permit. We also \nacknowledge all the field assistants that have participated in retrieving the cores and Tsilavo \nRazafimanantsoa for his input on the map. \n1. DATA REFERENCES \nZinke, J., B. Loveday, C. Reason, W.-C. Dullo, and D. Kroon. (2014). Madagascar corals track \nsea surface temperature variability in the Agulhas Current core region over the past 334 \nyears. Scientific Reports, 4, 4393. doi: 10.1038/srep04393. NOAA's National Centers for \nEnvironmental Information (NCEI)\nAbram, N. J., Mulvaney, R., Vimeux, F., Phipps, S. J., Turner, J., & England, M. H. (2014). \nEvolution of the Southern Annular Mode during the past millennium. Nature Climate \nChange, 4(7), 564–569. doi.org/10.1038/nclimate2235. NOAA's National Centers for \nEnvironmental Information (NCEI)\nJinbao Li, Shang-Ping Xie, Edward R. Cook, Gang Huang, Rosanne D'Arrigo, Fei Liu, Jian Ma, \nand Xiao-Tong Zheng. 2011. Interdecadal modulation of El Niño amplitude during the past \nmillennium. Nature Climate Change. 1(2) 114-118.  doi: 10.1038/nclimate1086. NOAA's \nNational Centers for Environmental Information (NCEI)\nScroxton, N., Burns, S. J., Mcgee, D., Hardt, B., Godfrey, L. R., Ranivoharimanana, L., & Faina, \nP. (2017). Hemispherically in-phase precipitation variability over the last 1700 years in a \nMadagascar speleothem record. 164. doi.org/10.1016/j.quascirev.2017.03.017. National \nCenters for Environmental Information (NCEI)\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nMacDonald & Case. 2005. Pacific Decadal Oscillation Reconstruction for the Past Millennium. \nGRL 32, L08703. doi:10.1029/2005GL022478. National Centers for Environmental \nInformation (NCEI)\nSaji, N.H., & Yamagata, T., (2003). Possible impacts of Indian Ocean Dipole mode events on \nglobal climate. CLIMATE RES, 25 (2): 151-169. \nhttps://psl.noaa.gov/gcos_wgsp/Timeseries/DMI/ \n2. DATA AVAILABILITY\nDatasets related to this article can be found at 10.25375/uct.16590035 on Zivahub, an open-\nsource online data repository hosted by the University of Cape Town. The DOI will become \nactive upon publication of the manuscript. \n3. REFERENCES\nAbram, N. J., Gagan, M. K., Cole, J. E., Hantoro, W. S. & Mudelsee, M. (2008). Recent \nintensification of tropical climate variability in the Indian Ocean. Nature Geoscience 1, \n849–853. \nhttps://doi.org/10.1038/ngeo357 \nAbram, N. J., Mulvaney, R., Vimeux, F., Phipps, S. J., Turner, J., & England, M. H. (2014). \nEvolution of the Southern Annular Mode during the past millennium. Nature Climate \nChange, 4(7), 564–569. https://doi.org/10.1038/nclimate2235\nAlin, S. R., & Cohen, A. S. (2003). Lake-level history of Lake Tanganyika, East Africa, for the \npast 2500 years based on ostracode-inferred water-depth reconstruction. Palaeogeography, \nPalaeoclimatology, Palaeoecology, 199(1–2), 31–49. https://doi.org/10.1016/S0031-\n0182(03)00484-X\nAnderson, R. F., Ali, S., Bradtmiller, L. I., Nielsen, S. H. H., Fleisher, M. Q., Anderson, B. E., \nand Burckle, L. H. (2009). Wind-Driven Upwelling in the Southern Ocean and the \nDeglacial Rise in Atmospheric CO2, Science. 323, 1443–1448, \nhttps://doi.org/10.1126/science.1167441.\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nBaptista, D. M. S., Farid, M., Fayad, D., Kemoe, L., Lanci, L. S., Mitra, P., Muehlschlegel, T. S., \nOkou, C., Spray, J. A., Tuitoek, K., & Unsal, F. D. (2022). Climate change and chronic \nfood insecurity in sub-Saharan Africa. International Monetary Fund Library, 2022(16), 1. \nhttps://doi.org/10.5089/9798400218507.087\nBarimalala, R., Desbiolles, F., Blamey, R. C., & Reason, C. (2018). Madagascar Influence on the \nSouth Indian Ocean Convergence Zone, the Mozambique Channel Trough and Southern \nAfrican Rainfall. Geophysical Research Letters, 45(20), 11,380-11,389. \nhttps://doi.org/10.1029/2018GL079964 \nBarimalala, R., Raholijao, N., Pokam, W., & Reason, C. J. C. (2021). Potential impacts of 1.5 \n°C, 2 °C global warming levels on temperature and rainfall over Madagascar. \nEnvironmental Research Letters, 16(4). https://doi.org/10.1088/1748-9326/abeb34  \nBarker, P., Telford, R., Merdaci, O., Williamson, D., Taieb, M., Vincens, A., & Gibert, E. \n(2000). The sensitivity of a Tanzanian crater lake to catastrophic tephra input and four \nmillennia of climate change. Holocene, 10(3), 303–310. \nhttps://doi.org/10.1191/095968300672848582\nBaudoin, M.A., Vogel, C., Nortje, K., Naik, M. (2017).  Living with drought in South Africa: \nlessons learnt from the recent El Niño drought period. International Journal of Disaster \nRisk Reduction., 23 , 128-137. https://doi.org/10.1016/j.ijdrr.2017.05.005\nBaum, D. A., Bombacaceae, A., & Baum, D. A. (1995). A Systematic Revision of Adansonia \n(Bombacaceae) 82(3), 440–471. https://doi.org/10.2307/2399893\nBaum, D. A., Small, R. L., & Wendel, J. F. (1998). Biogeography and flora evolution of Baobabs \n(Adansonia, Bombacaceae) as infered from multiple data sets. Systematic Biology, 47(2), \n181–207. \nhttps://doi.org/10.1080/106351598260879\nBeck, H. E., Wood, E. F., Pan, M., Fisher, C. K., Miralles, D. M., van Dijk, A. I. J. M., McVicar, \nT. R., and Adler, R. F. MSWEP V2 global 3-hourly 0.1° precipitation: methodology and \nquantitative assessment Bulletin of the American Meteorological Society 100(3), 473–500, \n2019\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nBelmecheri, S., & Lavergne, A. (2020). Compiled records of atmospheric CO2 concentrations \nand stable carbon isotopes to reconstruct climate and derive plant ecophysiological indices \nfrom tree rings. Dendrochronologia, 63(August), 125748. \nhttps://doi.org/10.1016/j.dendro.2020.125748\nBroccoli, A. J., Dahl, K. A., & Stouffer, R. J. (2006). Response of the ITCZ to Northern \nHemisphere cooling. Geophysical Research Letters, 33(1), 1–4. \nhttps://doi.org/10.1029/2005GL024546\nBurney, D. (1993). Late Holocene Environmental changes in arid southwestern Madagascar. \nQuaternary Research, 40, 98–106. https://doi.org/10.1006/qres.1993.1060\nBurns, S. J., Godfrey, L. R., Faina, P., McGee, D., Hardt, B., Ranivoharimanana, L., & \nRandrianasy, J. (2016). Rapid human-induced landscape transformation in Madagascar at \nthe end of the first millennium of the Common Era. Quaternary Science Reviews, 134, 92–\n99. https://doi.org/10.1016/j.quascirev.2016.01.007\nCarrière, S.D., Chalikakis, K., Ollivier, C. et al. (2018).  Sustainable groundwater resources \nexploration and management in a complex geological setting as part of a humanitarian \nproject (Mahafaly Plateau, Madagascar). Environmental Earth Science 77, 734 \nhttps://doi.org/10.1007/s12665-018-7909-1\nChase, B. M., & Meadows, M. E. (2007). Late Quaternary dynamics of southern Africa’s winter \nrainfall zone. Earth-Science Reviews, 84(3–4), 103–138. \nhttps://doi.org/10.1016/j.earscirev.2007.06.002\nChevalier, M., & Chase, B. M. (2015). Southeast African records reveal a coherent shift from \nhigh- to low- latitude forcing mechanisms along the east African margin across last glacial \ne interglacial transition. Quaternary Science Reviews, 125, 117–130. \nhttps://doi.org/10.1016/j.quascirev.2015.07.009\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nChiang, J. C. H., & Bitz, C. M. (2005). Influence of high latitude ice cover on the marine \nIntertropical Convergence Zone. Climate Dynamics, 25(5), 477–496. \nhttps://doi.org/10.1007/s00382-005-0040-5\nChristensen J.H., Hewitson B., Busuioc A., Chen A., Gao X., Held I., Jones R., Kolli R.K., \nKwon W-T., Laprise R., Magana Rueda V., Mearns L., Menendez C.G., Raisanen J., Rinke \nA., Sarr A., Whetton P. (2007). Regional climate projections. In: Solomon S., Qin D., \nManning M., Chen Z., Marquis M., Averyt A.B., Tignor M., Miller H.L. (eds) Climate \nchange 2007: the physical science basis. Contribution of Working Group I to the Fourth \nAssessment Report of the Inter-governmental Panel on Climate Change. Cambridge \nUniversity Press, Cambridge\nConnolly-Boutin, L. & Smit, B. (2016). Climate change, food security, and livelihoods in sub-\nSaharan Africa. Regional Environmental Change, 16: 385-399. \nhttps://doi.org/10.1007/s10113-015-0761-x\nCook, E. R., Palmer, J. G., Cook, B. I., Hogg, A., & D’Arrigo, R. D. (2002). A multi-millennial \npalaeoclimatic resource from Lagarostrobos colensoi tree-rings at Oroko Swamp, New \nZealand. Global and Planetary Change, 33(3–4), 209–220. https://doi.org/10.1016/S0921-\n8181(02)00078-4\nCrossley, R., Davison-Hirschmann, S., Owen, R.B., Shaw ,P.A. (1984). Lake level fluctuations \nduring the last 2000 years in Malawi. J. Vogel (Ed.), Late Cainozoic Palaeoclimates of the \nSouthern Hemisphere, A.A. Balkema, Rotterdam pp. 305-316. http://pascal-\nfrancis.inist.fr/vibad/index.php?action=getRecordDetail&idt=7268013 \nDonque, G. (1972). The climatology of Madagascar. In: Biogeography and ecology of \nMadagascar. R. Battistini and G. Richard-Vindard (eds). Junk, The Hague. 87–144. \nhttps://doi.org/10.1007/978-94-015-7159-3_3 \nEngelbrecht, F. A., Marean, C. W., Cowling, R. M., Engelbrecht, C. J., Neumann, F. H., Scott, \nL., Nkoana, R., O’Neal, D., Fisher, E., Shook, E., Franklin, J., Thatcher, M., McGregor, J. \nL., Van der Merwe, J., Dedekind, Z., & Difford, M. (2019). Downscaling Last Glacial \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nMaximum climate over southern Africa. Quaternary Science Reviews, 226, 105879. \nhttps://doi.org/10.1016/j.quascirev.2019.105879\nFaina, P., Burns, S. J., Godfrey, L.R., Crowley, B.E., Scroxton, N., McGee, D., Sutherland, \nM.R., & Ranivoharimanana, L. Comparing the paleoclimates of northwestern and \nsouthwestern Madagascar during the late Holocene: Implications for the role of climate in \nmegafaunal extinction. Malagasy nature, 15 (). Retrieved from \nhttps://par.nsf.gov/biblio/10316730.\nFarquhar, G., O’Leary, M. & Berry, J. (1982). On the Relationship between Carbon Isotope \nDiscrimination and the Intercellular Carbon Dioxide Concentration in Leaves. Australian \nJournal of Plant Physiology. 9(2), 121-137. https://doi.org/10.1071/PP9820121  \nGanzhorn, J. U. (1995). Cyclones over Madagascar: fate or fortune? Ambio. 24,124–125. \nhttp://www.jstor.org/stable/4314308 \nGillett, N. P., Kell, T. D., & Jones, P. D. (2006). Regional climate impacts of the Southern \nAnnular Mode. Geophysical Research Letters, 33(23), 1–4. \nhttps://doi.org/10.1029/2006GL027721\nGrove, C. A., Zinke, J., Peeters, F., Park, W., Scheufen, T., Kasper, S., Randriamanantsoa, B., \nMcCulloch, M. T., & Brummer, G. J. A. (2013). Madagascar corals reveal a multidecadal \nsignature of rainfall and river runoff since 1708. Climate of the Past, 9(2), 641–656. \nhttps://doi.org/10.5194/cp-9-641-2013\nHahn, A., Schefuß, E., Groeneveld, J., Miller, C., & Zabel, M. (2021). Glacial to interglacial \nclimate variability in the southeastern African subtropics (25–20° S). Climate of the Past, \n17, 345–360. https://doi.org/10.5194/cp-2019-158\nHajdas, I., Hendriks, L., Fontana, A., & Monegato, G. (2017). Evaluation of Preparation \nMethods in Radiocarbon Dating of Old Wood. Radiocarbon, 59(3), 727–737. \nhttps://doi.org/10.1017/RDC.2016.98\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nHall, G., Woodborne, S., & Pienaar, M. (2009). Rainfall control of the δ 13 C ratios of \nMimusops caffra from KwaZulu-Natal, South Africa. The Holocene, 19(2), 251–260. \nhttps://doi.org/10.1177/0959683608100569\nHall, G., Woodborne, S., & Scholes, M. (2008). Stable carbon isotope ratios from archaeological \ncharcoal as palaeoenvironmental indicators. 247, 384–400. \nhttps://doi.org/10.1016/j.chemgeo.2007.11.001\nHänke, H. & Barkmann, J. (2017). Insurance Function of Livestock: Farmer’s Coping Capacity \nwith Regional Droughts in South-Western Madagascar. World Development. 96: 264–275. \nhttps://doi.org/10.1016/j.worlddev.2017.03.011\nHänke, H., Barkmann, J., Coral, C., Enforskaustky, E., & Marggraf, R. (2017). Social-ecological \ntraps hinder rural development in Southwestern Madagascar. Ecology and Society, 22(1). \nhttps://doi.org/10.5751/ES-09130-220142\nHaug, G., Hughen, K., Sigman, D., Peterson, L., & Ro, U. (2001). Southward Migration of the \nITCZ holocene. Science, 293(August), 1304–1309. https://doi.org/10.1126/science.1059725 \nHelama, S., Timonen, M., Lindholm, M., Meriläinen, J., & Eronen, M. (2005). Extracting long-\nperiod climate fluctuations from tree-ring chronologies over timescales of centuries to \nmillennia. International Journal of Climatology, 25(13), 1767–1779. \nhttps://doi.org/10.1002/joc.1215\nHo, C. H., Kim, J. H., Jeong, J. H., Kim, H. S., & Chen, D. L. (2006). Variation of tropical \ncyclone activity in the South Indian Ocean: El Nino-Southern Oscillation and Madden-\nJulian Oscillation effects. Journal of Geophysical Research-Atmospheres, 111(D22), \nD22101. https://doi.org/Artn D22101\\nDoi 10.1029/2006jd007289\nHoell, A., Funk, C., Zinke, J., & Harrison, L. (2017). Modulation of the Southern Africa \nprecipitation response to the El Niño Southern Oscillation by the subtropical Indian Ocean \nDipole. Climate Dynamics, 48(7–8), 2529–2540. https://doi.org/10.1007/s00382-016-\n3220-6\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nHogg A.G., Hua Q., Blackwell P.G., Niu M., Buck C.E., Guilderson T.P., Heaton T.J., Palmer \nJ.G. et al. (2013) SHCal13 Southern Hemisphere calibration, 0–50,000 cal yr BP. \nRadiocarbon. 55: 1889–1903. https://doi.org/10.2458/azu_js_rc.55.16783 \nHolmgren, K., Karlén, W., Lauritzen, S. E., Lee-Thorp, J. A., Partridge, T. C., Piketh, S., \nRepinski, P., Stevenson, C., Svanered, O., & Tyson, P. D. (1999). A 3000-year high-\nresolution stalagmite-based record of palaeoclimate for northeastern South Africa. \nHolocene, 9(3), 295–309. https://doi.org/10.1191/095968399672625464\nHoscilo, A., Balzter, H., Bartholomé, E., Boschetti, M., Brivio, P. A., Brink, A., Clerici, M., & \nPekel, J. F. (2015). A conceptual model for assessing rainfall and vegetation trends in sub-\nSaharan Africa from satellite data. International Journal of Climatology, 35(12), 3582–\n3592. https://doi.org/10.1002/joc.4231\nHua, Q., Barbetti, M., & Rakowski, A. (2013). Atmospheric Radiocarbon for the Period 1950–\n2010. Radiocarbon, 55(4), 2059-2072. https://doi.org/10.2458/azu_js_rc.v55i2.16177 \nHuffman, T. N., & Woodborne, S. (2016). Archaeology, baobabs and drought: Cultural proxies \nand environmental data from the Mapungubwe landscape, southern Africa. Holocene, \n26(3), 464–470. https://doi.org/10.1177/0959683615609753\nHuffman, T.N. (2004) The archaeology of the Nguni past. Southern African Humanities. \n16(1929): 79–111. https://hdl.handle.net/10520/EJC84742 \nIngram, J. C., & Dawson, T. P. (2005). Inter-annual analysis of deforestation hotspots in \nMadagascar from high temporal resolution satellite observations. International Journal of \nRemote Sensing, 26(7), 1447–1461. https://doi.org/10.1080/01431160412331291189\nIPCC (2021): Climate Change (2021) The Physical Science Basis. Contribution of Working \nGroup I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change \n[Masson-Delmotte, V., Zhai, P., Pirani, A., Connors, S.  L., Péan, C., Berger, S. , Caud, N.,  \nChen, Y., Goldfarb, L., Gomis, M. I. , Huang, M., Leitzell, K., Lonnoy, ., E., Matthews, J. \nB.  R., Maycock, T. K., Waterfield, T., Yelekçi, O., Yu, R. & Zhou, B. (eds.)]. Cambridge \nUniversity Press. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nJohnson, T.C., Barry, S.L., Chan, Y., Wilkinson, P. 2001. Decadal record of climate variability \nspanning the past 700 yr in the Southern Tropics of East Africa. Geology  29 (1): 83–86. \nhttps://doi.org/10.1130/0091  \nJury, M. R. (2016). Summer climate of Madagascar and monsoon pulsing of its vortex. \nMeteorology and Atmospheric Physics, 128(1), 117–129. https://doi.org/10.1007/s00703-\n015-0401-5\nJury, M. R., & Huang, B. (2004). The Rossby wave as a key mechanism of Indian Ocean climate \nvariability. Deep Sea Research Part I: Oceanographic Research Papers, 51(12), 2123–2136. \nhttps://doi.org/10.1016/j.dsr.2004.06.005\nKaufmann, J. C. & Tsirahamba, S. (2006). Forests and Thorns: Conditions of Change Affecting \nMahafale Pastoralists in Southwestern Madagascar. Conservation and Society. 4(2): 231–\n261.\nKing, J., Anchukaitis, K.J., Allen, K. et al. Trends and variability in the Southern Annular Mode \nover the Common Era. Nat Commun 14, 2324 (2023). https://doi.org/10.1038/s41467-023-\n37643-1\nKuiper, M., Meijerink, G., & Eaton, D. (2007). Rural livelihoods: Interplay between farm \nactivities, non-farm activities and the resource base. Science for Agriculture and Rural \nDevelopment in Low-Income Countries, 77–95. https://doi.org/10.1007/978-1-4020-6617-\n7_5\nLechleitner, F. A., Breitenbach, S. F. M., Rehfeld, K., Ridley, H. E., Asmerom, Y., Prufer, K. \nM., Marwan, N., Goswami, B., Kennett, D. J., Aquino, V. V., Polyak, V., Haug, G. H., \nEglinton, T. I., & Baldini, J. U. L. (2017). Tropical rainfall over the last two millennia: \nEvidence for a low-latitude hydrologic seesaw. Scientific Reports, 7(June 2016), 1–9. \nhttps://doi.org/10.1038/srep45809\nLi, J., Xie, S.P., Cook, E. et al. (2011). Interdecadal modulation of El Niño amplitude during the \npast millennium. Nature Climate Change. 1, 114–118. \nhttps://doi.org/10.1038/nclimate1086\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nLiu, Z., Otto-Bliesner, Z., Kutzbach, J., Li, L. C. (2003). Shields Coupled climate simulation of \nthe evolution of global monsoons in the Holocene. Journal of Climate, 16,  2472-2490.  \nhttps://doi.org/10.1175/1520-0442(2003)016<2472:CCSOTE>2.0.CO;2 \nLoader, N. J., Robertson, I., Barker, A. C., Switsur, V. R., & Waterhouse, J. S. (1997). An \nimproved technique for the batch processing of small wholewood samples to a-cellulose. \nChemical Geology, 136, 313–317. https://doi.org/10.1016/S0009-2541(96)00133-7\nMacDonald, G. M., & Case, R. A. (2005). Variations in the Pacific Decadal Oscillation over the \npast millennium. Geophysical Research Letters, 32(8), 1–4. \nhttps://doi.org/10.1029/2005GL022478\nMacron, C., Pohl, B., Richard, Y., & Bessafi, M. (2014). How do tropical temperate troughs \nform and develop over Southern Africa? Journal of Climate, 27(4), 1633–1647. \nhttps://doi.org/10.1175/JCLI-D-13-00175.1\nMamalakis, A., Randerson, J.T., Yu, JY. et al. (2021). Zonally contrasting shifts of the tropical \nrain belt in response to climate change. Nature Climate Change 11, 143–151. \nhttps://doi.org/10.1038/s41558-020-00963-x\nManhique, A. J., Reason, C. J. C., Rydberg, L. & Fauchereau, N. (2011). ENSO and Indian sea \nsurface temperatures with tropical temperate troughs over Mozambique and the southwest \nIndian Ocean. International Journal of Climatology. 31, 1–13. \nhttps://doi.org/10.1002/joc.2050 \nMann, M. E., Zhang, Z., Rutherford, S., Bradley, R. S., Hughes, M. K., Shindell, D., Ammann, \nC., Faluvegi, G., & Ni, F. (2009). Global Signatures and Dynamical Origins of the Little \nIce Age and Medieval Climate Anomaly. Science, 326(5957), 1256–1260. \nhttps://doi.org/10.1126/science.1177303\nMason, S.  J.  & Jury, M.  R. (1997).  ‘Climatic change and variability over Southern Africa:  A \nReflection on Underlying Processes’, Progress in Physical Geoggraphy. 21, 23–50. \nhttps://doi.org/10.1177/030913339702100103\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nMatsumoto, K. & Burney, D. A. (1994). Late Holocene environments at Lake Mitsinjo, \nnorthwestern Madagascar. Holocene, 4(1), 16–24. \nhttps://doi.org/10.1177/095968369400400103\nMcCarroll, D., & Loader, N. J. (2004). Stable isotopes in tree rings. Quaternary Science \nReviews, 23(7–8), 771–801. https://doi.org/10.1016/j.quascirev.2003.06.017\nMiller, C., Finch, J., Hill, T., Peterse, F., Humphries, M., Zabel, M., and Schefuß, E. (2019) Late \nQuaternary climate variability at Mfabeni peatland, eastern South Africa, ClimAte of the \nPast, 15,1153–1170, \nhttps://doi.org/10.5194/cp-15-1153-2019\nNakamura, N., Kayanne, H., Iijima, H., McClanahan, T.R., Behera, S.K., Yamagata, T., (2009). \nMode shift in the Indian Ocean climate under global warming stress. Geophysical Research \nLetters. 36 http://dx.doi.org/10.1029/2009GL040590\nNash. (2017). Changes in Precipitation Over Southern Africa During Recent Centuries. Climate \nScience. https://doi.org/10.1093/acrefore/9780190228620.013.539\nNasri, M., & Modarres, R. (2009). Dry spell trend analysis of Isfahan Province, Iran. \nInternational Journal of Climatology, 29, 1430–1438. https://doi.org/10.1002/joc\nNeukom, R., & Gergis, J. (2012). Southern Hemisphere high-resolution palaeoclimate records of \nthe last 2000 years. The Holocene, 22(5), 501–524. \nhttps://doi.org/10.1177/0959683611427335\nNicholson, S. E., Klotter, D., & Dezfuli, A. K. (2012). Spatial reconstruction of semi-quantitative \nprecipitation fields over Africa during the nineteenth century from documentary evidence \nand gauge data. Quaternary Research, 78(1), 13–23. \nhttps://doi.org/10.1016/j.yqres.2012.03.012\nPAGES 2k Consortium. 2013. Continental-scale temperature variability during the past two \nmillennia. Nature Geoscience. 6: 339-346. http://dx.doi.org/10.1038/1849t  \nPatrut, A., Reden, K. F. Von, Danthu, P., Pock-tsy, J. L., Rakosy, L., Patrut, R. T., Lowy, D. A., \n& Margineanu, D. (2015). Nuclear Instruments and Methods in Physics Research B AMS \nradiocarbon dating of very large Grandidier ’ s baobabs ( Adansonia grandidieri ). Nuclear \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nInst. and Methods in Physics Research, B, 361, 591–598. \nhttps://doi.org/10.1016/j.nimb.2015.04.044\nOnyeaka, H., Nwauzoma, U. M., Akinsemolu, A. A., Tamasiga, P., Duan, K., Al-Sharify, Z. T., \n& Siyanbola, K. F. (2024). The ripple effects of climate change on agricultural \nsustainability and food security in Africa. Food and Energy Security, 13, e567. \nhttps://doi.org/10.1002/fes3.567 \nPatrut, A., Woodborne, S., Von Reden, K. F., Hall, G., Patrut, R. T., Rakosy, L., Danthu, P., \nPock-Tsy, J. M. L., Lowy, D. A., & Margineanu, D. (2017). The growth stop phenomenon \nof baobabs (Adansonia spp.) Identified by radiocarbon dating. Radiocarbon, 59(2), 435–\n448. \nhttps://doi.org/10.1017/RDC.2016.92\nPohlert, T. (2018). R Package “trends.” 18. https://cran.r-\nproject.org/web/packages/trend/vignettes/trend.pdf\nPutnam, A. E., & Broecker, W. S. (2017). Human-induced changes in the distribution of rainfall. \nScience Advances, 3(5), 1–14. https://doi.org/10.1126/sciadv.1600871\nRailsback, L. B., Brook, G. A., Liang, F., Voarintsoa, N. R. G., Cheng, H., & Edwards, R. L. \n(2018). A multi-proxy climate record from a northwestern Botswana stalagmite suggesting \nwetness late in the Little Ice Age (1810–1820 CE) and drying thereafter in response to \nchanging migration of the tropical rain belt or ITCZ. Palaeogeography, Palaeoclimatology, \nPalaeoecology, 506(April), 139–153. https://doi.org/10.1016/j.palaeo.2018.06.029\nRamanantsoa, J. D., Penven, P.,Krug, M., Gula, J., & Rouault, M. (2018). Uncovering a new \ncurrent: The Southwest Madagascar Coastal Current. Geophysical Research Letters, 45, \n1930–1938. \nhttps://doi.org/10.1002/2017GL075900\nRatna, S. B., Behera, S., Ratnam, J. V. , Takahasgi, K., & Yamagata, T. (2012) An index for \ntropical temperate troughs over southern Africa. Climate Dynamics, 41, 421–441. \nhttp://dx.doi.org/10.1007/s00382-012-1540-8 . \nRazanamaro O., Rasoamanana E., Rakouth B., Randriamalala J.R., Rabakonadrianina E., \nClément-Vidal A., Leong Pock Tsy J.M., Menut C., Danthu P. (2015). Chemical \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\ncharacterization of floral scents in the six endemic baobab species (Adansonia sp.) of \nMadagascar. Biochemical Systematics and Ecology, 60 : 238-248 \nhttp://dx.doi.org/10.1016/j.bse.2015.04.005\nRazanatsoa, E., Gillson,L., Virah-Sawmy, M., and Woodborne, S. (2021) Pollen records of the \n14th and 20th centuries AD from Lake Tsizavatsy in southwest Madagascar. \nPalaeoecology of Africa, 35, pp. 309-315. http://dx.doi.org/10.1201/9781003162766-20  \nRazanatsoa, E., Gillson, L., Virah-Sawmy, M., and Woodborne, S. (2022) Synergy between \nclimate and human land-use maintained open vegetation in southwest Madagascar over the \nlast millennium. The Holocene. 31(12). https://doi.org/10.1177/09596836211041731\nReason, C. J. C., & Mulenga, H. (1999). Relationships between South African rainfall and SST \nanomalies in the southwest Indian Ocean. International Journal of Climatology, 19(15), \n1651–1673. \nhttps://doi.org/10.1002/(SICI)1097-0088(199912)19:15<1651::AID-\nJOC439>3.0.CO;2-U\nRobertson, I., Loader, N., Froyd, C., Zambatis, N., Whyte, I., & S, W. (2006). The potential of \nthe baobab (Adansonia digitata L.) as a proxy climate archive. Applied Geochemistry, 21, \n1674–1680. https://doi.org/10.1016/j.apgeochem.2006.07.005\nRussell, J. M. & Johnson, T. C. (2007). Little ice age drought in equatorial Africa: inter- tropical \nconvergence zone migrations and El Niño southern oscillation variability. Geology. 35: 21-\n24. http://dx.doi.org/10.1130/G23125A.1.\nRussell, J. M., Verschuren, D., & Eggermont, H. (2007). Spatial complexity of “Little Ice Age” \nclimate in East Africa: Sedimentary records from two crater lake basins in western \nUganda. Holocene, 17(2), 183–193. https://doi.org/10.1177/0959683607075832\nSachs, J. P., Sachse, D., Smittenberg, R. H., Zhang, Z., Battisti, D. S., & Golubic, S. (2009). \nSouthward movement of the Pacific intertropical convergence zone AD 1400-1850. Nature \nGeoscience, 2(7), 519–525. https://doi.org/10.1038/ngeo554\nSaji, N. H., & Yamagata, T. (2003). Possible impacts of Indian Ocean Dipole mode events on \nglobal climate. Climate Research, 25(2), 151–169. https://doi.org/10.3354/cr025151\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nSchneider, T., Bischoff, T. & Haug, G. (2014).  Migrations and dynamics of the intertropical \nconvergence zone. Nature, 513, 45–53.  https://doi.org/10.1038/nature13636  \nScroxton, N., Burns, S. J., Mcgee, D., Hardt, B., Godfrey, L. R., Ranivoharimanana, L., & Faina, \nP. (2017). Hemispherically in-phase precipitation variability over the last 1700 years in a \nMadagascar speleothem record. 164. https://doi.org/10.1016/j.quascirev.2017.03.017\nSerele, C., Pérez-Hoyos, A., & Kayitakire, F. (2020). Mapping of groundwater potential zones in \nthe drought-prone areas of south Madagascar using geospatial techniques. Geoscience \nFrontiers, 11(4), 1403–1413. https://doi.org/10.1016/j.gsf.2019.11.012\nSigman, D. M., Hain, M. P., and Haug, G. H. (2010). The polar ocean and glacial cycles in \natmospheric CO2 concentration, Nature, 466, 47–55, https://doi.org/10.1038/nature09149,  \n201  \nStager, J. C., Ryves, D., Cumming, B. F., David Meeker, L., & Beer, J. (2005). Solar variability \nand the levels of Lake Victoria, East Africa, during the last millenium. Journal of \nPaleolimnology, 33(2), 243–251. https://doi.org/10.1007/s10933-004-4227-2\nSvensson, A., Andersen, K. K., Bigler, M., Clausen, H. B., Dahl-Jensen, D., Davies, S. M., \nJohnsen, S. J., Muscheler, R., Rasmussen, S. O., Röthlisberger, R., Peder Steffensen, J., & \nVinther, B. M. (2006). The Greenland Ice Core Chronology 2005, 15-42 ka. Part 2: \ncomparison to other records. Quaternary Science Reviews, 25(23–24), 3258–3267. \nhttps://doi.org/10.1016/j.quascirev.2006.08.003\nTadross, M., Randriamarolaza, L., Rabefitia, Z., & Ki Yip, Z. (2008). Climate change in \nMadagascar; recent past and future. … DC (World Bank), February, 18. \nhttp://www.mediagrapher.org/gripweb/sites/default/files/disaster_risk_profiles/Madagascar \nClimate Report.pdf\nTambo, J.A. & Abdoulaye, T. (2013). Smallholder farmers’ perceptions of and adaptations to \nclimate change in the Nigerian savanna. Regional Environmental Change 13, 375–388 \nhttps://doi.org/10.1007/s10113-012-0351-0\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nTaylor, A. K., Berke, M. A., Castañeda, I. S., Koutsodendris, A., Campos, H., Hall, I. R., \nHemming, S. R., LeVay, L. J., Sierra, A. C., & O’Connor, K. (2021). Plio‐Pleistocene \nContinental Hydroclimate and Indian Ocean Sea Surface Temperatures at the Southeast \nAfrican Margin. Paleoceanography and Paleoclimatology, 36(3), 1–18. \nhttps://doi.org/10.1029/2020pa004186\nThomas, D.S.G., Twyman, C., Osbahr, H. et al. (2007).  Adaptation to climate change and \nvariability: farmer responses to intra-seasonal precipitation trends in South Africa. Climatic \nChange 83, 301–322 https://doi.org/10.1007/s10584-006-9205-4\nThompson, D. W. J., & Wallace, J. M. (2000). Annular modes in the extratropical circulation. \nPart I: Month-to-month variability. Journal of Climate, 13(5), 1000–1016. \nhttps://doi.org/10.1175/1520-0442(2000)013<1000:AMITEC>2.0.CO;2\nThompson, L. G., Mosley-Thompson, E., Davis, M. E., Lin, P. N., Henderson, K., & Mashiotta, \nT. A. (2003). Tropical glacier and ice core evidence of climate change on annual to \nmillennial time scales. Climatic Change, 59(1–2), 137–155. \nhttps://doi.org/10.1023/A:1024472313775\nTierney, J. E., Smerdon, J. E., Anchukaitis, K. J. & Seager, R. (2013). Multidecadal variability in \nEast African hydroclimate controlled by the Indian Ocean. Nature. 493: 389-392. \nhttp://dx.doi.org/10.1038/nature11785  \nTieszen, L. L. (1991). Natural variations in the carbon isotope values of plants: implications for \narchaeology, ecology and palaeoecology. Journal of Archaeological Science. 18: 227–248. \nhttps://doi.org/10.1016/0305-4403(91)90063-U\nTrenberth, K.E. (1979). Interannual variability of the 500 mb zonal-mean flow in the Southern \nHemisphere. Monthly Weather Reviews.  107, 1515–1524. https://doi.org/10.1175/1520-\n0493(1979)107<1515:IVOTMZ>2.0.CO;2\nTsen, E.W.J., Sitzia, T. and Webber, B.L. (2016). To core, or not to core: the impact of coring on \ntree health and a best-practice framework for collecting dendrochronological information \nfrom living trees. Biological Review, 91: 899-924. https://doi.org/10.1111/brv.12200\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nTyson, P. D., Odada, E. O., & Partridge, T. C. (2001). Late quaternary environmental change in \nsouthern Africa. South African Journal of Science, 97(3–4), 139–150. \nhttps://hdl.handle.net/10520/EJC97298\nVerschuren D. (2003). Lake-based climate reconstruction in Africa: progress and challenges. In: \nMartens K. (eds) Aquatic Biodiversity. Developments in Hydrobiology, vol 171. Springer, \nDordrecht. https://doi.org/10.1007/978-94-007-1084-9_22\nVerschuren, D., Laird, K. R., & Cumming, B. F. (2000). Rainfallanddroughtinequatorialeast \nAfrica during the past 1, 100 years. Solar Cells, 403(January). \nhttps://doi.org/10.1038/35000179 \nVigaud, N., Richard, Y., Rouault, M. & Fauchereau, N.  (2007). Water vapour transport from the \ntropical Atlantic and summer rainfall in tropical southern Africa. Climate Dynamics, 28, \n113–123. https://doi.org/10.1007/s00382-006-0186-9 \nVirah-Sawmy, M., Gillson, L., Gardner, C. J., Anderson, A., Clark, G., & Haberle, S. (2016). A \nlandscape vulnerability framework for identifying integrated conservation and adaptation \npathways to climate change: the case of Madagascar’s spiny forest. Landscape Ecology, \n31(3), 637–654. https://doi.org/10.1007/s10980-015-0269-2\nVoarintsoa, N. R. G., Wang, L., Railsback, L. B., Brook, G. A., Liang, F., Cheng, H., & \nEdwards, R. L. (2017). Multiple proxy analyses of a U/Th-dated stalagmite to reconstruct \npaleoenvironmental changes in northwestern Madagascar between 370 CE and 1300 CE. \nPalaeogeography, Palaeoclimatology, Palaeoecology, 469, 138–155. \nhttps://doi.org/10.1016/j.palaeo.2017.01.003\nVon Heland, J., & Folke, C. (2014). A social contract with the ancestors-Culture and ecosystem \nservices in southern Madagascar. Global Environmental Change, 24(1), 251–264. \nhttps://doi.org/10.1016/j.gloenvcha.2013.11.003\nWang, G., & Feng, X. (2012). Response of plants’ water use efficiency to increasing atmospheric \nCO2 concentration. Environmental Science & Technology, 46(16), 8610–8620. \nhttps://doi.org/10.1021/es301323m\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nWang, L., Brook, G. A., Burney, D. A., Voarintsoa, N. R. G., Liang, F., Cheng, H., & Edwards, \nR. L. (2019). The African Humid Period, rapid climate change events, the timing of human \ncolonization, and megafaunal extinctions in Madagascar during the Holocene: Evidence \nfrom a 2m Anjohibe Cave stalagmite. Quaternary Science Reviews, 210, 136–153. \nhttps://doi.org/10.1016/j.quascirev.2019.02.004\nWatanabe, T. K., Watanabe, T., Yamazaki, A., Pfeiffer, M., & Claereboudt, M. R. (2019). Oman \ncoral δ 18 O seawater record suggests that Western Indian Ocean upwelling uncouples \nfrom the Indian Ocean Dipole during the global-warming hiatus. Scientific Reports, 9(1), \n1–9. \nhttps://doi.org/10.1038/s41598-018-38429-y\nWeathering Risk (W.R). (2023). Climate Risk Profile for Southern Africa.\nWils THG, Robertson I, Woodborne S, Hall G, Koprowski M, Eshetu Z. 2016. Anthropogenic \nforcing increases the water-use efficiency of African trees. Journal of Quaternary Science \n31, 386-390. https://doi.org/10.1002/jqs.2865   \nWoodborne S, Gandiwa P, Hall G, Patrut A, Finch J (2016) A Regional Stable Carbon Isotope \nDendro-Climatology from the South African Summer Rainfall Area. PLoS ONE 11(7): \ne0159361. https://doi.org/10.1371/journal.pone.0159361\nWoodborne, S., Hall, G., Robertson, I., Patrut, A., Rouault, M., Loader, N. J., & Hofmeyr, M. \n(2015). A 1000-year carbon isotope rainfall proxy record from South African baobab trees \n(Adansonia digitata L.). PLoS ONE, 10(5), 1–18. \nhttps://doi.org/10.1371/journal.pone.0124202\nWorld Bank Group (2017). World Bank Annual Report (English). Washington, D.C. :. \nhttp://documents.worldbank.org/curated/en/143021506909711004/World-Bank-Annual-\nReport-2017\nYaro, J. A., Teye, J., & Bawakyillenuo, S. (2014). Local institutions and adaptive capacity to \nclimate change/variability in the northern savannah of Ghana. Climate and Development, \n7(3), 235–245. \nhttps://doi.org/10.1080/17565529.2014.951018\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\nZinke, J., Loveday, B. R., Reason, C. J. C., Kroon, D., Ocean, S. I., & Age, L. I. (2014). \ntemperature variability in the Agulhas Current core region over the past 334. Scientific \nreports. pp. 1–8. https://doi.org/10.1038/srep04393\n4. FUNDING \nThis project has been funded as part of the Faculty PhD fellowship (University of Cape Town, \nR.E.) 2015-2018 and the Applied Centre for Climate and Earth Systems Science (ACCESS NRF \nUID 98018, R.E.) project, the UCT University Research Committee accredited (URC) and \nCOVID supplemental support from the University of Cape Town [URC, 2019-2020] and the \nNRF/SASSCAL (Southern African Science Service Centre for Climate Change and Adaptive \nLand Management, grant number 118589), the NRF/African Origins Platform (grant number \n117666), and NRF Competitive Programme for Rated Researchers (Grant Number 118538).\nSupplementary Information \nSI1: Radiocarbon dates of each tree replicate from the four trees collected in southwest \nMadagascar\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint \n\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.15.670475doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}