Portal de Periódicos da CAPES

South Pacific hydrologic and cyclone variability during the last 3000 years

2016; American Geophysical Union; Volume: 31; Issue: 4 Linguagem: Inglês

10.1002/2015pa002870

1944-9186

Michael Toomey, Jeffrey P. Donnelly, Jessica E. Tierney,

Geology and Paleoclimatology Research

PaleoceanographyVolume 31, Issue 4 p. 491-504 Research ArticleFree Access South Pacific hydrologic and cyclone variability during the last 3000 years Michael R. Toomey, Corresponding Author Michael R. Toomey Jackson School of Geosciences, University of Texas at Austin, Austin, Texas, USA Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA Correspondence to: M. R. Toomey, mrt02008@gmail.comSearch for more papers by this authorJeffrey P. Donnelly, Jeffrey P. Donnelly Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USASearch for more papers by this authorJessica E. Tierney, Jessica E. Tierney Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA Department of Geosciences, University of Arizona, Tucson, Arizona, USASearch for more papers by this author Michael R. Toomey, Corresponding Author Michael R. Toomey Jackson School of Geosciences, University of Texas at Austin, Austin, Texas, USA Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA Correspondence to: M. R. Toomey, mrt02008@gmail.comSearch for more papers by this authorJeffrey P. Donnelly, Jeffrey P. Donnelly Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USASearch for more papers by this authorJessica E. Tierney, Jessica E. Tierney Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA Department of Geosciences, University of Arizona, Tucson, Arizona, USASearch for more papers by this author First published: 31 March 2016 https://doi.org/10.1002/2015PA002870Citations: 8AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract Major excursions in the position of the South Pacific Convergence Zone (SPCZ) and/or changes in its intensity are thought to drive tropical cyclone (TC) and precipitation variability across much of the central South Pacific. A lack of conventional sites typically used for multimillennial proxy reconstructions has limited efforts to extend observational rainfall/TC data sets and our ability to fully assess the risks posed to central Pacific islands by future changes in fresh water availability or the frequency of storm landfalls. Here we use the sedimentary record of Apu Bay, offshore the island of Tahaa, French Polynesia, to explore the relationship between SPCZ position/intensity and tropical cyclone overwash, resolved at decadal time scales, since 3200 years B.P. Changes in orbital precession and Pacific sea surface temperatures best explain evidence for a coordinated pattern of rainfall variability at Tahaa and across the Pacific over the late Holocene. Our companion record of tropical cyclone activity from Tahaa suggests major storm activity was higher between 2600-1500 years B.P., when decadal scale SPCZ variability may also have been stronger. A transition to lower storm frequency and a shift or expansion of the SPCZ toward French Polynesia around 1000 years B.P. may have prompted Polynesian migration into the central Pacific. Key Points SPCZ migrated and/or intensified since 3200 years B.P. in response to orbital forcing Enhanced South Pacific (SP) cyclogenesis occurred between ~2600 and 1500 years B.P. Changes in SP rainfall ~1000 years B.P. may have prompted Polynesian expansion 1 Introduction The South Pacific Convergence Zone (SPCZ) is a convective rain trough that stretches across the South Pacific with its mean axis orientated from Solomon Islands to French Polynesia [Vincent, 1994; Vincent et al., 2011]. Northward migration and zonal orientation of the SPCZ, particularly during El Niño events, is thought to drive stronger cyclogenesis in the central South Pacific on interannual time scales by reducing wind shear around French Polynesia [Revell and Goulter, 1986; Vincent et al., 2011]. Observational records also link the behavior of the SPCZ to sea surface temperature (SST) anomalies associated with the Interdecadal Pacific Oscillation [Folland et al., 2002] and solar insolation [Meehl et al., 2009]. However, these inferred relationships are based on short data sets ( 30 m) occupy coastal embayments and capture terrestrial sediments [Zinke et al., 2001] likely shed by nearby slopes during runoff events. Such sites are widespread and have high potential as valuable paleoclimate archives. Previous coring and geophysical efforts indicate that deposition in back-barrier reef lagoons has been fast (up to 6.9 mm/yr) [Montaggioni, 2005] and spans much of the Holocene as well as previous interglacials [Le Roy et al., 2008]. Here we use sedimentary records of late Holocene tropical cyclone and SPCZ variability from Apu Bay, offshore the island of Tahaa, French Polynesia, to answer two main questions: (1) how is the SPCZ's position or strength impacted by different climatic regimes (e.g., SST, insolation) and (2) is there a statistically significant relationship between SPCZ behavior and tropical cyclone activity in the central South Pacific on geologic time scales? 2 Study Site Tahaa is a highly eroded volcanic ocean island, formed 2–3 Myr ago [Guillou et al., 2005] as part of the Society Islands hotspot chain (Figures 1a and 1b). Apu Bay is ~40 m deep drowned valley incised into the volcanic upland and flanked on three sides by steep (~25 % grade) ridges (Figure 1c). Submerged karst, visible in aerial imagery of the barrier (16°41.241′S 151°29.648′W) fronting Apu Bay, suggests that the geometry of the site has not changed substantially since the glacial transgression. Core TAH VC10 (~3.8 m in length, including the core catcher) was taken in the center of the bay (39 m water depth), ~500 m from the shore (16°40.116′S, 151°29.508′W). Seismic surveying at the site (Figure 1d) and 14C dating indicates that a large, relatively uniform sedimentary pile (shaded blue) has accumulated in Apu Bay since deglacial sea level inundation. Figure 1Open in figure viewerPowerPoint (a) Satellite (CMAP [Xie and Arkin, 1997]—NOAA/OAR/ESRL) observed precipitation over the past 30 years (1979–2009 C.E.). Circles indicate locations of climate records used in Figures 5 and 6: Makassar Strait (MK, green), Tahaa (red), Mangaia (orange), Rimatara (yellow), Laguna Pallcacocha (LP, light blue), and El Junco (EJ, dark blue). (b) Location of Tahaa relative to other islands in the Society Islands Chain. Position of rain gauge station on Bora Bora is shown by purple circle. Apu Bay and Bora Bora both sit in a slight rain shadow topographic highs on Tahaa (maximum elevation = 590 m) relative to the prevailing trade winds. (c) Topographic (Aster 30 m resolution) and bathymetric (British Admiralty chart #1103) map of southwestern Tahaa centered on Apu Bay, our coring site, TAH VC10. (d) Seismic trace running north-south through Apu Bay which shows deposition of a large wedge of Holocene sediment as well as several older reflectors. The location of TAH VC10 is indicated by a white dashed line. Tahaa is located on the southeastern edge of the modern SPCZ trough and experiences major seasonal as well as interannual changes in precipitation (Figures 2a and 2b). El Niño (La Niña) events are associated with a northeastern (southwestern) displacement of the SPCZ [Widlansky et al., 2011] and increased (decreased) precipitation at Tahaa. A rain gauge record from Bora Bora (38 km NW of our site) shows a ~30% increase in precipitation during months when the Southern Oscillation Index (www.cgd.ucar.edu/cas/catalog/climind/soi/html)—the sea level pressure difference between Tahiti, French Polynesia, and Darwin, NE Australia—is negative (El Niño) compared with months that the Southern Oscillation Index (SOI) is positive (La Niña). The difference is mostly seen during the austral summer (DJF, December-January-February) when, on average, negative SOI corresponds with nearly 100 mm/month more precipitation than during the same months when SOI is positive. During the rainiest months in the record, rainfall can exceed 500 mm/month above background. Figure 2Open in figure viewerPowerPoint (a) Monthly mean precipitation (KNMI; http://climexp.knmi.nl/) observed for the island of Bora Bora (38 km NW of our site on Tahaa) (1951–2000 C.E.). Cross signs mark statistical outliers. (b) Rain gauge record from Bora Bora. Time series smoothed with 5 year Gaussian filter shown in purple. (c) XRF Ti and Ca counts for discrete surface samples collected from onshore the volcanic island (maroon cross) and reef flat (yellow dot). Grey dots show XRF Ti and Ca intensity for TAH VC10. (d) Ti/Ca time series (blue dots, 1 cm, ~5 year smoothed: grey line) in core top of TAH VC10 based on the highest probability sedimentation 210Pb rate compared to Bora Bora rainfall (5 year smoothed: purple line). 3 Materials and Methods In January 2009 Common Era (C.E.) we collected a core (TAH VC10) using a Rossfelder P-3 vibracoring system from Apu Bay on the island of Tahaa (17°S 151°W; Figure 1), French Polynesia, 230 km northeast of Tahiti. Seismic profiles (e.g., Figure 1d) were taken using a Benthos CHIRP-II (2–7 kHz). Following recovery, the core was shipped to the Woods Hole Oceanographic Institution (WHOI), split, and refrigerated prior to analysis. Major stratigraphic changes were identified at 10 cm intervals based on measurements of grain size (Beckman-Coulter LS13320), fixed volume dry bulk density and loss on ignition [Heiri et al., 2001] analyses (Table 1). Table 1. Major Stratigraphic Changes in Downcore Composition and Grain Size Depth (cm) Organic (%) CaCO3 (%) d50 Grain Size (µm) Std. Deviation (µm) Density (g/cm3) 5–6 6.2 81.7 24.6 3.0 1.4 15–16 5.9 83.1 26.5 3.0 1.4 25–26 5.6 83.8 27.4 3.3 1.5 35–36 5.6 84.0 26.2 3.1 1.4 45–46 5.7 83.2 24.9 3.1 1.5 55–56 6.2 80.5 25.5 3.1 1.4 65–66 6.2 79.9 24.2 3.1 1.4 75–76 6.3 80.3 24.9 3.2 1.5 85–86 6.0 82.2 24.9 3.2 1.5 95–96 5.9 82.3 24.2 3.2 1.5 105–106 6.2 82.7 19.0 3.0 1.5 115–116 6.2 80.9 19.5 3.1 1.4 125–126 6.1 81.3 19.0 3.0 1.5 135–136 6.0 82.6 24.8 3.3 1.5 145–146 6.0 83.4 23.9 3.3 1.6 155–156 6.0 83.5 18.9 3.5 1.7 165–166 6.1 81.5 22.0 3.4 1.6 175–176 5.9 83.4 23.1 3.4 1.7 185–186 6.0 83.0 20.8 3.2 1.5 195–196 6.1 83.1 23.5 3.4 1.6 205–206 6.0 82.3 22.3 3.4 1.7 215–216 5.9 83.2 22.5 3.5 1.8 225–226 5.7 83.9 23.3 3.4 1.6 235–236 5.7 84.3 23.7 3.4 1.6 245–246 5.9 84.5 18.9 3.5 1.6 255–256 5.0 84.6 23.2 3.4 1.6 265–266 6.9 79.9 12.8 3.6 1.6 275–276 5.2 85.3 19.9 3.5 1.7 285–286 5.1 85.1 19.6 3.4 1.6 295–296 5.2 85.9 22.1 3.7 1.5 305–306 5.2 85.6 22.5 3.5 1.6 315–316 5.0 85.8 21.2 3.7 1.7 325–326 5.0 86.3 22.1 3.5 1.6 335–336 5.2 85.3 22.1 3.6 1.7 345–346 5.1 83.2 21.6 3.5 1.7 355–356 5.2 83.5 23.1 3.8 1.6 365–366 5.0 85.9 21.8 3.7 1.6 Higher-resolution (1 cm) downcore grain size changes were measured using standard 63 µm, 250 µm, and 2 mm sieves. Sediment was extracted from the core at 1 cm intervals before being dried at 100°C for 5 h. After the dry mass was obtained, each sample was wet sieved at 63 µm. The >63 µm fraction was then dried and subsequently dry sieved at 250 µm and 2 mm. Color and mineral properties of the core were measured using optical reflectance spectrometry (Konica Minolta spectrophotometer CM-2600d). Cores were scanned using a Geotek core scanner at 5 mm intervals for wavelengths between 360 and 740 nm (10 nm resolution). We then extracted the primary mineral constituents from this data: following previously published methods [Ji et al., 2005], we (1) computed the first derivative values, (2) determined the principal components, and (3) compared them to the spectral profiles of known minerals in the USGS spectral library (http://speclab.cr.usgs.gov/spectral-lib.html). In order to estimate the input of terrestrial material downcore we used high-resolution (0.5–1.0 mm) X-ray fluorescence (XRF). Relative elemental abundance measurements were made using ITRAX X-ray fluorescence (XRF) core scanners at WHOI and the University of Massachusetts at Amherst. Core-scanning XRF is ideally suited for this work as it is fast and sensitive to relative changes in metals. The natural logarithm ratio between Ti and Ca was used in part to limit the effects of machine error (e.g., differences in tube strength) and changes in lithology (e.g., grain size, water content, and porosity) on XRF measurements. Age control in the core was established using 210Pb and radiocarbon dating (Table 2). Measurement of 137Cs did not define a discernable peak in activity related to atmospheric nuclear weapons testing or its subsequent end following the Nuclear Test-Ban Treaty (1963 C.E.). The 210Pb chronology was calculated assuming a constant flux constant sedimentation model [Appleby and Oldfield, 1983] as follows: (1) bulk material was extracted from the cores, (2) samples were crushed and homogenized, and (3) 210Pb activity was measured using gamma spectroscopy. We then iteratively (1000 times) resampled the probability distribution for the supported 210Pb levels (~24–30 cm, Figure 3a, light grey dots) and the total error (counting and gamma ray attenuation) on points with excess 210Pb (Figure 3a, black dots) before finding the best linear fit to our log transformed excess 210Pb curve. This produced a distribution of possible maximum apparent accumulation rates (Figure 3b), which we then used to generate time series for our runoff record from Tahaa that could be compared to instrumental records of SOI and rainfall. Both instrumental and XRF data were smoothed to 5 year resolution for this procedure. Table 2. Radiocarbon Dates for Core TAH VC10a Depth (cm) 14C Age (years) 14C Error (years) ΔR (years) Material Method δ13C (o/oo) Max Prob. Age (years B.P.) 34–35 750 25 225 Bulk Org C AMS −16.9 140 80–82 1270 30 225 Bulk Org C AMS −17.6 630 119–120 1730 25 225 Bulk Org C AMS −17.7 1050 171–172 1930 30 225 Bulk Org C AMS −18 1265 196–197 2100 30 225 Bulk Org C AMS −17.8 1405 203–205 3050 30 225 Bulk Org C AMS −19.2 2615 255–257 2370 30 225 Bulk Org C AMS −19.2 1720 292–293 2600 25 300 Charcoal AMS −28.4 2315 340–342 3130 30 225 Bulk Org C AMS −19.4 2705 360–361 3500 30 225 Bulk Org C AMS −17.1 3105 a The data in bold at 203–205 cm represent a major age reversal and was not used in our age model calculation. 14C Age, 1σ 14C error and R are given in radiocarbon years. 14C error is calculated as the larger of the internal statistical error and external reproducibility of the measurement (www.whoi.edu/nosams/radiocarbon-data-calculations). A maximum probable age in years B.P. (1950 C.E.) was calculated for each radiocarbon date using the MARINE13 or SHCal13 calibration curves (5 year resolution) [Reimer et al., 2013] and the ΔR value listed here. Figure 3Open in figure viewerPowerPoint 210Pb and radiocarbon chronology for TAH VC10. (a) Top gives 210Pb profile for upper 30 cm of TAH VC10. Circles and error bars show the measured 210Pb activity and 1 sigma error range. Solid grey line shows the mean sedimentation rate. Medium grey circles (shaded grey area) are likely mixed and were not used in calculating our age model. Grey dashed line indicates level of supported 210Pb in the sediments. (b) Inset shows a histogram of probable sedimentation rates generated by resampling the uncertainty on 210Pb measurements 1000 times. (c) Mean chronology and errors for the radiocarbon chronology of TAH VC10. Bars indicate 1 sigma (black) and 2 sigma (grey) ranges for corrected radiocarbon dates. Conventional Accelerator Mass Spectrometry (AMS) radiocarbon dating of bulk organic material and charcoal was used to develop a long-term chronology for TAH VC10. Measurements were made at the National Ocean Sciences Accelerator Mass Spectrometry facility in Woods Hole, Massachusetts, and Beta Analytic in Miami, Florida. An age model was calculated using 14C dates in three main steps: (1) probability distributions of ΔR-corrected (discussed further below) dates were calibrated following the procedures outlined for Calib 7.1 (http://calib.qub.ac.uk/calib/); (2) the resulting probability distributions were then used to iteratively (10,000 times) generate time index points, then (3) linearly fit between points ending at the 210Pb chronology and assuming superposition, similar to methods developed by Anchukaitis and Tierney [2013]. A cubic fit (Figure 3c) could also be used to capture the transitions between a high sedimentation regime from ~1000 to 2000 years B.P. (~2 mm/yr) and slower apparent sedimentation rates prior to 2000 years B.P. (~1 mm/yr), approximately when barrier reefs on nearby Moorea [Montaggioni, 2005] and by extension, Tahaa, caught up to SL, as well as after 1000 years B.P., following a 1–2 m SL fall from the mid-Holocene Highstand [Grossman et al., 1998] which likely resulted in restriction of the lagoon. Bulk organic radiocarbon samples (Table 2) from Tahaa show a δ13C signature (−17 to −19‰) consistent with a marine source [Stuiver and Polach, 1977] and were calibrated using the Marine13 curve [Reimer et al., 2013]. We apply a ΔR of 225 years to these samples consistent with (1) the observed offset between our 210Pb chronology and adjacent radiocarbon dates and (2) radiocarbon dating of deposit feeders from other South Pacific islands [Petchey et al., 2008] which likely draw their carbon from a similar 14C pool. A built in age of ~300 years was estimated for a large, single grain, of wood charcoal (292–293 cm) based on the average apparent built-in ages in dated charcoal from archeological sites on the nearby island of Huahine (50 km east) [Anderson and Sinoto, 2002]. Preliminary gas bench dates [von Reden et al., 2012] made on detrital carbonate were not used due to a large apparent offset (ΔR > 1000 radiocarbon years) likely due to the reworking of old material from nearby relic Pleistocene fringing reefs. 4 Proxy Development Color changes in the sediment core, seen using reflectance spectrophotometry, indicate that terrestrial material is periodically mobilized from the adjacent volcanic slopes during heavy rainfall events and deposited in the lagoon. Principal component analysis of reflectance peaks suggests that Ti-oxides (Figure 4e) constitute a significant component of the sediments and correspond to visible red-brown streaks (centimeter scale) in the core. Previous work in the Society Islands chain documents Ti-oxides (e.g., ilmenite and rutile) as significant components of soils (Tahiti: 3–4%) [Parkes et al., 1992] and the volcanic basement rocks (2–3%) at Tahaa [Schiano et al., 1992]. Heavy mineral components typical of ocean island basalts (olivine and pyroxenes) are lacking in TAH VC10 sediments, likely reflecting transport limitations. The matrix material is made up of grey carbonate mud (e.g., bioeroded fragments and aragonite needles) and glauconite (green) produced in situ or on the reef. Coarse material in the core is dominantly detrital carbonate (e.g., shell fragments and worm tubes), likely swept off nearby reefs during storms. Together, this evidence suggests that soil material is episodically delivered to the lagoon, deposited into sediments that are otherwise dominated by carbonate mud and may be used to reconstruct a record of precipitation back through time. Figure 4Open in figure viewerPowerPoint Color and principal component analysis of optical spectrophotometry data for TAH VC10. (a) Color image. Triangles show depth of 14C index points (Table 2). (b) Comparison of a* to XRF-derived ln(Ti/Ca) data (a* is a standard metric of the red to green color ratio of the sediments [Blum, 1997]). Reflectance spectroscopy is insensitive to changes in carbonate (white) material. (c) Scree plot showing the percent of variance accounted for by each principal component. First two principal components are marked by dark grey bars. (d) Comparison of Δ Reflectance for first principal component (grey) and glauconite (dashed) using data (interpolated to 10 nm spacing) from the USGS spectral reflectance library [Clark et al., 2007]. (e) Comparison of Δ Reflectance for second principal component (grey) and ilmenite (dashed). Core tops (~10 cm) of sections 1 and 2 were not used for spectral analysis due to possible disturbance during shipment. Intact areas of the core were subsampled using U-channels and scanned again separately for XRF analysis (medium grey shading in Figure 4b). Comparison of Ti and Ca intensities from discrete soil samples collected on Tahaa and carbonate material taken from the reef flat (Figure 2c) demonstrates a clear elemental division between these end-members. Carbonate reef material is depleted in titanium whereas terrestrial samples contain little calcium. Likewise, downcore increases in ln(Ti/Ca) XRF intensities—an approximation of actual log ratios of elemental concentrations which sensitivity studies indicate have an error of a few percent [e.g., Weltje and Tjallingii, 2008]—closely track red sediment color (Figure 4b), consistent with deposition of Ti-oxides and other volcanic weathering products. Therefore, we interpret titanium-rich material recovered from Apu Bay (Figures 2d and 5b) as terrestrially derived. Figure 5Open in figure viewerPowerPoint Long-term changes in tropical Pacific precipitation during the late Holocene. (a) δDwax precipitation proxy data from the Indo-Pacific Warm Pool [Tierney et al., 2010]. Raw data are given by grey line, while red line shows the three-point moving average. (b) Ti/Ca ratios over the past 3200 years recorded in sediments from Apu Bay, Tahaa. Black line shows 100 year Gaussian filter of Ti/Ca data (medium grey). (c) Changes in silt-sized and silt + sand grain size fraction observed from El Junco Lake, Galapagos [Conroy et al., 2008]. (d) Fifty year composite IPWP SST (Mg/Ca) given by black line. Raw data shown in grey [Oppo et al., 2009]. Grey dashed line shows the calculated daily, austral summer (NDJF) incoming solar radiation at 15°S (http://www.people.fas.harvard.edu/~phuybers/Mfiles/Toolbox/). Blue and red shading highlight periods of increased rainfall or drought over zthe past ~1000 years. Thin dashed line shows earliest probable migration of Polynesians into French Polynesia [Wilmshurst et al., 2011]. Covariance (Figure 4b) of upland runoff (Ti-oxide) relative to both glauconite (a*, which is insensitive to changes in the amount of white, carbonate, sediment) and carbonate material (XRF ln(Ti/Ca), which is insensitive to glauconite (K,Na)(Fe3+,Al,Mg)2(Si,Al)4O10(OH)2 effectively reduces TAH VC10 sediments to a two-component (terrestrial/marine) system. Changes in XRF ln(Ti/Ca), referred to as Ti/Ca from now on, could potentially result from variable deposition of either the lagoon-derived (carbonate + glauconite + marine organics) or terrestrial (oxides + terrestrial organics) component. Comparison of XRF Ti intensity in TAH VC10 to soil samples (~2.5%, Figure 2c), suggests that upland-derived material makes up only a small fraction of the core, in line with measurement of HCl burn insoluble residues (≤5% weight). Increased downcore density (Table 1), correlated to sediment grain size standard deviation (sorting/porosity) and changes in sedimentation rate are therefore likely driven by variable delivery of marine material which comprises upward of 90% of TAHVC10 sediments. However, if concentration or dilution of Ti-oxides were dominating the XRF Ti/Ca signal, we would expect it to be inversely related to sedimentation rate, which it is not. For example, background XRF Ti/Ca increases and wt % CaCO3 decreases from the bottom of the core through 145 cm despite sedimentation increasing and minimal change in bulk density. Likewise, the large increase (~4 times) in Ti oxides between ~125 (>1200 XRF Ti counts/s) and 145 cm (~300 counts/s), occurs when sedimentation is high and there is relatively little change in density (~5–7%). This transition (~1300–1000 years B.P.) to increased terrestrial deposition also predates the likely arrival of Polynesian settlers to the Society Islands (~925–830 years B.P.) [Wilmshurst et al., 2011]. Well-dated archeological remains on nearby Huahine (50 km east) cluster around ~750 years B.P. [Anderson and Sinoto, 2002] suggesting a lag of several centuries between colonization and the eventual buildup of substantial human settlements in the Leeward Society Islands. Several other large magnitude changes in Ti/Ca also occur over the past 1000 years (e.g., ~400 years B.P.) and are unaccompanied by major changes in density, sedimentation, or macroscopic charcoal deposition potentially related to land clearance. Therefore, we suggest that variable Ti/Ca in TAH VC10 is largely driven by the deposition of Ti-oxides related to changes in rainfall, not marine sedimentation or human disturbance. Delivery of upland material to the lagoon likely represents a threshold process with generation of overland flow requiring soil saturation and sustained heavy rainfall. Volcanic rocks/soils typically have high infiltration capacity, and surface runoff may occur during only a small subset of strong storms [Peterson, 1972]. Peaks in Ti/Ca likely represent episodic delivery of terrestrial material by heavy rainfall, predominantly during the austral summer (DJF), which are then smoothed (~5–10 years) by mixing processes at the seafloor. Mean background rainfall conditions may modulate centennial scale changes in terrestrial sediment supply to the lagoon by pacing groundwater recharge and therefore chemical weathering rates [Schopka and Derry, 2012]. Extreme precipitation during passage of tropical systems can also result in runoff/extreme flooding [Galewsky et al., 2006; Silverberg et al., 2007]; however, instrumental observations from Bora Bora (Figures 2a and 2b) show that storms also commonly produce little change in rainfall near our site. For example, precipitation for November 1997 C.E. totaled only 262 mm despite a direct strike by Osea (CAT 2), the closest cyclone to pass our site during the instrumental era (>1950 C.E.) [Knapp et al., 2010]. Three of the seven tropical systems to pass within 100 km of the Bora Bora rain gauge station (>1950 C.E.) occurred during December (Lisa, TS) February, and March (Prema, TD; Rewa, CAT 3) of 1982/1983 C.E., but rainfall levels during these months (356, 213, and 122 mm, respectively) diverge little from their monthly means (Figure 2a). However, disturbance (tree mortality/fires) of coastal forests due to intense storm strikes [Walker, 1991; Whigham et al., 1991] could promote runoff under conditions of comparable rainfall. Large charcoal fragments (~5 mm diameter), likely indicating significant fires, were found between ~292–293 and 334–335 cm in TAH VC10. 5 Results and Discussion 5.1 Pacific Rainfall Variability The derived record of terrestrial runoff at our site demonstrates a strong correspondence to observed changes in historic local rainfall. In general, we observe an increase in Ti/Ca after 1950 C.E. which likely corresponds with three periods (~1950–1954, 1960–1967, and 1979–1988 C.E.) of increased rainfall at Bora Bora. Likewise, we see episodes of higher rainfall around 1900 C.E., consistent with observational records showing several large negative SOI excursions around this time. However, we note that Tahaa is located just to the northeast of the nodal region for historic El Niño–Southern Oscillation (ENSO) SST/wind variability which potentially tempers its sensitivity to changes in rainfall related to SOI. Of the thousand potential sedimentation rates (Figure 3b) generated by resampling the uncertainties in our 210Pb chronology, more than 88% yield a negative correlation (Pearsons "r") between Ti/Ca and SOI, although only a small subset (~16%) are significant at the p < 0.1 level after taking into account autocorrelation of the time series [Ebisuzaki, 1997]. While the short length of instrumental re