Mid-Holocene (4200 kyr BP) mass mortalities in Mauritius (Mascarenes): Insular vertebrates resilient...

http://hol.sagepub.com/The Holocene

http://hol.sagepub.com/content/21/8/1179The online version of this article can be found at:

DOI: 10.1177/0959683611405236

2011 21: 1179 originally published online 18 July 2011The HoloceneBrinkkemper, Tamara Vernimmen and Anwar Janoo

Hanneke J.M. Meijer, Hubert B. Vonhof, Nick Porch, F.B. Vincent Florens, Claudia Baider, Bas van Geel, Joost Kenneth F. Rijsdijk, Jens Zinke, Perry G.B. de Louw, Julian P. Hume, Hans (J.) van der Plicht, Henry Hooghiemstra,

climatic extremes but vulnerable to human impactMid-Holocene (4200 kyr BP) mass mortalities in Mauritius (Mascarenes): Insular vertebrates resilient to

Published by:

http://www.sagepublications.com

can be found at:The HoloceneAdditional services and information for

http://hol.sagepub.com/cgi/alertsEmail Alerts:

http://hol.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://hol.sagepub.com/content/21/8/1179.refs.htmlCitations:

What is This?

- Jul 18, 2011 OnlineFirst Version of Record

- Dec 7, 2011Version of Record >>

at AMERICAN EMBASSY PANAMA on August 30, 2012hol.sagepub.comDownloaded from

http://hol.sagepub.com/

http://hol.sagepub.com/content/21/8/1179

http://www.sagepublications.com

http://hol.sagepub.com/cgi/alerts

http://hol.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://hol.sagepub.com/content/21/8/1179.refs.html

http://hol.sagepub.com/content/21/8/1179.full.pdf

http://hol.sagepub.com/content/early/2011/07/12/0959683611405236.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml


The Holocene21(8) 1179 –1194© The Author(s) 2011Reprints and permission:sagepub.co.uk/journalsPermissions.navDOI: 10.1177/0959683611405236hol.sagepub.com

Research paper

Introduction

Because of the anticipated increased frequencies of droughts and water scarcity on tropical islands (Falkland, 1994; Hoerling and Kumar, 2003; Lal et al., 2002; Wong et al., 2005), it is crucial to understand how this will affect already insular vertebrate popula-tions. The Holocene period (11.5 cal. kyr BP to present) is marked by periods of pronounced climatic extremes, i.e. severe droughts and cold periods, that lasted for several decennia (Fleitmann et al., 2008; Gasse, 2000; Marchant and Hooghiemstra, 2004; Pross et al., 2009). Stress imposed by climatic extremes may have been especially severe on island fauna, compared with continental pop-ulations, because of generally limited boundaries and intra-island refugia. Evidence for this, however, is scant and there are few documented cases of natural extinction events. The discovery of hundreds of fossil localities on oceanic islands have provided lit-tle fossil evidence for pre-anthropogenic extinctions, which sug-gests that lower rates occurred compared with the period characterized by human impact (e.g. Alcover et al., 1999; Burney, 1999; Holdaway, 1999; Martin and Steadman, 1999; Olson, 1989;

Olson and Hearty, 2003; Olson and James, 1992; Steadman, 1995, 2006). Without doubt, it was after the arrival of humans

405236 HOL21810.1177/0959683611405236Rijsdijk et al.The Holocene

1NCB-Naturalis, The Netherlands2University of Amsterdam, The Netherlands3VU University Amsterdam, The Netherlands4Subsurface and Groundwater systems Unit, Deltares, The Netherlands5Natural History Museum, UK6University of Groningen, The Netherlands7Leiden University, The Netherlands8Smithsonian Institution, USA9Deakin University, Australia10University of Mauritius, Mauritius11Mauritius Sugar Industry Research Institute, Mauritius

Received 9 August 2010; revised manuscript accepted 14 February 2011

Corresponding author:Kenneth F. Rijsdijk, NCB-Naturalis, PO Box 9517, 2300 RA Leiden, The Netherlands. Email: [email protected]

Mid-Holocene (4200 kyr BP) mass mortalities in Mauritius (Mascarenes): Insular vertebrates resilient to climatic extremes but vulnerable to human impact

Kenneth F. Rijsdijk,1,2 Jens Zinke,3 Perry G.B. de Louw,4 Julian P. Hume,5 Hans (J.) van der Plicht,6,7 Henry Hooghiemstra,2

Hanneke J.M. Meijer,8,1 Hubert B. Vonhof,3 Nick Porch,9 F.B. Vincent Florens,10 Claudia Baider,11 Bas van Geel,2 Joost Brinkkemper,2 Tamara Vernimmen1 and Anwar Janoo10

AbstractIn the light of the currently increasing drought frequency and water scarcity on oceanic islands, it is crucial for the conservation of threatened insular vertebrates to assess how they will be affected. A 4000 yr old fossil assemblage in the Mare Aux Songes (MAS), southwest Mauritius, Mascarene Islands, contains bones of 100 000+ individual vertebrates, dominated by two species of giant tortoises Cylindraspis triserrata and C. inepta, the dodo Raphus cucullatus, and 20 other vertebrate species (Rijsdijk, Hume, Bunnik, Florens, Baider, Shapiro et al. (2009) Mid-Holocene vertebrate bone Concentration-Lagerstätte on oceanic island Mauritius provides a window into the ecosystem of the dodo (Raphus cucullatus). Quaternary Science Reviews 28: 14–24). Nine radiocarbon dates of bones statistically overlap and suggest mass mortality occurred between 4235 and 4100 cal. yr BP. The mortality period coincides with a widely recognized megadrought event. Our multidisciplinary investigations combining geological, paleontological and hydrological evidence suggests the lake was located in a dry coastal setting and had desiccated during this period. Oxygen isotope data from a Uranium-series dated stalagmite from Rodrigues, an island 560 km east of Mauritius, supports this scenario by showing frequently alternating dry and wet periods lasting for decades between 4122 and 2260 cal. yr BP. An extreme drought resulted in falling water-tables at MAS and elsewhere on the island, perhaps deprived these insular vertebrates of fresh water, which led to natural mass mortalities and possibly to extirpations. In spite of these events, all insular species survived until at least the seventeenth century, confirming their resistance to climatic extremes. Despite this, the generally exponential increase of combined human impacts on islands including loss of geodiversity, habitats, and stocks of fresh water, there will be less environmental safe-haven options for insular endemic and native vertebrates during future megadrought conditions; and therefore will be more prone to extinction.

Keywordsclimate change, dodo, insular vertebrates, mass mortality, megadrought, palaeoecology, speleothems, volcanic islands



1180 The Holocene 21(8)

that almost all documented extinctions occurred (Burney and Flannery, 2005; Burney et al., 2001; Martin, 1984; Martin and Steadman, 1999; Steadman, 2006; Turvey, 2009), and that island faunas are generally resistant to natural climatic and environmen-tal change. However, it is unclear to what degree faunas were affected by climatic extremes, and how generally they were able to withstand these effects on small islands. The re-working in 2005 of a 4000 yr old natural fossil bed on the volcanic island of Mauritius (Rijsdijk et al., 2009) (Figure 1a), provided important new supporting data. The fossil bed is a marsh called the Mare aux Songes (MAS, hereafter) situated in the coastal lowlands, southwest Mauritius, at the base of a land-filled lake, and is esti-mated to comprise millions of subfossil remains, including the dodo. In addition it has abundant plant remains including well-preserved trunk and branch samples. The fossil bed pre-dates human contact by more than 3000 years.

Early Arab traders probably discovered Mauritius, followed by the Portuguese in 1516 (North-Coombes, 1980), but as far as is known, they never settled there and no written accounts exist. It was after the Dutch arrival in September 1598 (Barnwell, 1948; Moree, 1998), that the documented record began. These records have proved vital in ascertaining the original faunal composition. The Dutch abandoned the island in 1710 (Moree, 1998), and it was during this period that the dodo and most other large, terres-trial vertebrates became extinct (Cheke and Hume, 2008). It was not until the discovery of the first subfossil remains at the MAS in September 1865 (Clark, 1866; Hume et al., 2005, 2009), that the true extent of faunal diversity could be assessed. Subsequent excavations continued until the 1930s, after which time the site

fell into neglect. Because field descriptions were not taken, the precise location and context of all previously collected subfossil material is not known. The fossil bed is 500 to 1500 mm thick and spans 2 ha, and the rock valley where the site is situated com-prises four basins (varying between 0.5 and 2 ha); the fossils were discovered in basin 1 (Figure 1b). Based on our surveys we infer that subfossils discovered in the nineteenth century were derived from basin 0 located 50 m NW of basin 1 (Hume et al., 2009). About 5000 years ago a freshwater lake formed at MAS basin 1 (Figure 2b) (Rijsdijk et al., 2009), and the emergent lake fresh water, situated in an otherwise dry coastal region, plausibly attracted a diverse vertebrate fauna and led to faunal concentra-tion especially during dry spells. Radiocarbon datings (n = 12) on different faunal and floral elements indicate that the fossils accu-mulated within a relatively short interval between 4400 and 3900 cal. yr BP. The small time window and large bone concentration suggests that a mass mortality had occurred. It is hypothesized that extreme drought conditions may have led to the drying out of the freshwater lake which deprived the local fauna of fresh water; thus inducing this event. Evidence of extreme drought conditions between 4300 and 4000 cal. yr BP from Asia, East Africa and Madagascar supports this hypothesis (Booth et al., 2005; March-ant and Hooghiemstra, 2004; Staubwasser and Weiss, 2006; Thompson and Davis, 2007; Williams, 2009), but precise data for the Mascarene Islands is lacking. Dating is further compounded by the deviating age of bone fragments sampled from one excava-tion locality (TR0-2005) and in a disturbed core (CH5-1995) (Figure 1b), compared with two radiocarbon dates on 1865 MAS museum specimens of Cylindraspis spp. (Burleigh and Arnold,

Figure 1. (a) The volcanic island of Mauritius. Inset map shows Africa and Madagascar, position of Mauritius in SW Indian Ocean. Black rectangle denotes position of Mare aux Songes. Black dots positions of former vents aligned NNE. Black line shows the extent of a caldera edge. (b) A digital elevation model of Mare aux Songes, showing basin 0, basin 1 and part of basin 2, basin 3 is not depicted. Borehole evidence indicates that the fossil layer extends for 2 ha within the 2 m elevation contour. Triangles indicate positions of excavation trenches (TR) and dots indicate positions of cores (BH). White arrows indicate positions of basins 0, 1 and 2



Rijsdijk et al. 1181

Figure 2. (a) Schematic cross-section of groundwater-table and emerging freshwater scenarios (after Borchiellini et al. 1999; Proag, 1995). Five freshwater springs emerge as a result of geodiversity. Surplus rainwater in the uplands infiltrates the permeable Recent Lava Series basalts (young permeable basalt in figure), flows through the subsurface and emerges as freshwater springs near the coast. Springs and lakes upslope may form as a result of faulting associated with the caldera formation, leading to steps in impermeable Older Lava Series basalts. Intrusions of volcanic dikes may create local springs and depressions in basalts (collapsed lava tunnels or blast out holes). Critically springs emerge at the coastal lowlands that are characterised by an evaporation surplus. (b) Cross-section of the rock basin 1 at MAS. Present surface level as bold black lines, vertical bold black lines show the continuation of bedrock within the subsurface, thin black line with black dots shows the depth of the top of coral sand unit B (in Figure 1b position of cross-section is indicated). BH codes indicate positions of boreholes. Dots indicate depth of top of coral sand layer B as found in boreholes, the dots mark the base of the palaeolake. Stippled line indicates positions of sea level and lake level during sea level rise. Approximately 4200 cal. yr ago lake level was lowered 75 cm due to extreme drought. Left vertical axis shows height in m mean sea level (M.S.L.). Right vertical axis shows age in cal. kyr BP of corresponding palaeo sea levels (Camoin et al., 2004)




1986). The latter had yielded significantly younger ages, between c. 1800 and 1200 cal. yr BP, which suggests that the fossil layer was diachroneous.

The aim of this paper is to test the hypothesis that the 4000 cal. yr BP mass mortality of insular vertebrates on Mauritius occurred as a result of a natural extreme drought event. We have employed a multidisciplinary approach to reconstruct the events that led to the vast accumulation of fossils at MAS and we aim to evaluate the following three questions: (1) Was the vertebrate fauna subject to instantaneous or gradual death? If the fossil bed was diachroneous, significantly older fossils were expected to occur in the deeper sediments of the basin; therefore a fresh local-ity was selected at the centre of the palaeolake and subfossils were sampled with stratigraphical context for radiocarbon dating. (2) To what degree had lake levels at MAS been affected by drought? Hydrological measurements were carried out to assess the sensitivity of the lake to present local climatic conditions. (3) Did extreme drought affect the Mascarenes? We addressed this issue by analyzing an oxygen isotope record of a 4122 year old speleothem from the Mascarene Island of Rodrigues. Both Mas-carene Islands have similar climatic conditions and their interan-nual and decadal variability in air temperature and rainfall are strongly correlated (Figure 3D, E; WMO weather station database data at http://climexp.knmi.nl).

Regional setting and climateThe volcanic island of Mauritius (20°10′S, 57°30′E) is located in the southwestern Indian Ocean (Figure 1a). The island (1865 km2) comprises of a central plateau at 400–500 m elevation represent-ing a former caldera. These uplands are gently sloping into low-lands near the coast. The island emerged from the Indian Ocean some 10 million years ago (10 Ma), and after a shield-building phase and deposition of the Old Lava Series between 7.6 Ma to 5 Ma, a period of low volcanic activity commenced with the forma-tion of a caldera (Saddul, 2002). A new phase of volcanism occurred between 3.5 and 1.7 Ma leading to the formation of the Early Lava Series. A third phase of active volcanism commenced at 0.7 Ma and lasted until 20 ka. During this period fluid basalts were erupted from a series of 25 vents aligned NNE (Figure 1a). These basalts formed the Intermediate and Recent Lava Series.

For most of the year Mauritius lies within the southeast trade winds, but during the austral summer months tropical cyclones and depressions associated with the seasonal movement of the Inter Tropical Convergence Zone affect the island. The mean annual temperature is 22°C and rainfall 2100 mm. Depending on relief and the orientation of the slopes to the prevailing wind direction, mean annual rainfall varies from 1400 mm in the east-ern coastal lowlands, to 4000 mm on the uplands, and 800 mm in the western coastal lowlands. Rainfall is seasonal, with a dry season from May to October under influence of the cool and dry easterly trade winds, and has a wetter and warmer season from November to April when the ITCZ has its southernmost position (Senapathi et al., 2010). During the driest month of October, c. 3.5% (74 mm) of the total rainfall is registered (Padya, 1989). While the central uplands annual rainfall exceeds evapotranspi-ration, it is relatively dry in the coastal regions (< 50 m eleva-tion) with evapotranspiration > rainfall, leading to an annual precipitation deficit (Padya, 1989). At the southeast coast, including the site of MAS, a mean annual evapotranspiration of

2200 mm exceeds the mean annual rainfall of 1400 mm. This deficit exists throughout the year but is especially high during the dry season. Surface drainage is scarce because of the high infiltration capacity of the Younger Lava Series. Freshwater springs emerge in the SE as a result of rainfall surplus in the uplands and surface and subsurface heterogeneity in infiltration capacity of soils and basalts (Figure 2a) (Borchiellini et al., 1999; Proag, 1995). In the SE the Older Lava Series form aqui-cludes, whereas the younger basalts of the Recent Lava Series have high infiltration capacities and form aquifers. Through subsurface flow of groundwater that is derived from surplus rainwater from central uplands and catchments, a base flow of fresh water is maintained in springs that emerge at MAS and other coastal sites (Figure 2a).

Materials and methodsSampling and chronologyTo assess the time window represented by the fossil layer, we excavated a new trench TR4-2007 by means of a mechanical dig-ger. Based on geological evidence, TR4 was located approxi-mately 50 m eastward from the first sample location TR0-2005 within the deepest part of the palaeo-lake. We sampled five fossil elements in stratigraphic context for radiocarbon dating from the digger scoop including a bone fragment of Cylindraspis sp. Radiocarbon samples of wood, seeds and charcoal were treated using standard Acid-Alkali-Acid; for bone samples, collagen was extracted. The isolated fraction was combusted into CO2 gas. For large (gram-size) samples, the 14C activity was measured by radi-ometry (GrN); for small (mg-size) samples, the 14C concentration was measured by mass AMS (GrA). The 14C dates are reported in BP, which includes correction for isotope fractionation. The back-ground level of the laboratory corresponds to 14C dates of about 50 000 14C yr BP. By means of the OxCal calibration program (Bronk Ramsey, 2009) the 14C ages were calibrated. The 14C dates are calibrated into calendar ages using the radiocarbon calibration curve for the Southern Hemisphere (McCormac et al., 2004). Control measurement on a modern gastropod demonstrated that reservoir effects are absent. We excluded bulk 14C datings of com-bined samples to avoid methodological artifacts. In addition we excluded degraded wood samples that may have grown on lake banks and fell into the lake later. Uncalibrated radiocarbon ages are reported as 14C yr BP, calibrated radiocarbon ages are reported in cal, yr BP (BP = prior to ad 1950).

Hydrological measurements

In order to determine the sensitivity of the freshwater supply at MAS to climatic conditions, we assessed the hydrological response to current dry and wet conditions from a water level log-ger located at MAS in basin 1 (Figure 1b). The present groundwa-ter levels correlate with mean ocean levels (0 m M.S.L.). Since 2007, the sensors of the logger continuously register groundwater fluctuations in a piezometer, and net rainfall and evaporation data were obtained from the Mon Desert weather station situated 500 m from MAS. The current geohydrological responses to varia-tions in precipitation and evapotranspiration of the MAS basin provide insights in the system response and help to reconstruct hydrological responses from 5000 years ago, when global sea lev-els were 1.5 m lower than present (Camoin et al., 2004).




Figure 3. (a) The Island of Rodrigues in the Indian Ocean at 19°S, 63°E, including the surrounding coral reef and lagoon. (b) Inset shows the exact location of the La Vierge Cave. (c) The entrance of the cave is situated on the grounds of the Francois Leguat Giant Tortoise and Cave Reserve. The cave is relatively shallow, extending for approximately 100 m parallel to Anse Quitor river. Cave air humidity is nearly 100% and cave temperature is seasonally stable around 25.5°C. (d) Spatial correlations of mean annual temperatures and rainfall across the Indian Ocean. Source WMO weather station data base data at http://climexp.knmi.nl. (e) Temporal variability of mean annual temperatures at Mauritius, Réunion and Rodrigues. Source WMO weather station data base data at http://climexp.knmi.nl




Climatic reconstruction from a stalagmiteTo reconstruct past climate variability in the region, we ana-lyzed oxygen isotope ratios of a stalagmite from Rodrigues. In spite of being 560 km distant, current coastal climate variability of Rodrigues and Mauritius are strongly correlated (Figure 3D). Mean air temperature at Rodrigues is slightly higher than Mau-ritius, but interannual and decadal variability is similar across the Mascarene Islands (Figure 3E). Rainfall in Mauritius and Rodrigues is also highly correlated on annual to decadal times-cales. We therefore assume that past climate variability of Rodrigues is also representative of Mauritius. A 400 mm long stalagmite was obtained from the La Vierge Cave at Rodrigues (Figure 3), and a 10 mm thick section was cut along the growth axis. Carbonate samples were drilled every 2 mm with a hand-held drill for stable isotope analysis. Four U-series datings on speleothem carbonate samples from Rodrigues were obtained at the University of Melbourne on a MC-ICP-MS30 (methods out-lined in Hellstrom, 2003). The δ18O and δ13C values were anal-ysed on a Thermo Finnigan Delta+ mass spectrometer equipped with a GASBENCH II preparation device at the VU University Amsterdam. About 30 µm of CaCO3 sample, placed in a He-filled 10 ml exetainer vial, was digested in concentrated H3PO4 at a temperature of 45°C. Subsequently the CO2-He gas mixture was transported to the GASBENCH II by use of a He flow through a flushing needle system. In the GASBENCH, water was extracted from the gas by use of NAFION tubing, and CO2 was analysed in the mass spectrometer after separation of other gases in a GC column. Isotope values are reported as δ13C and δ18O relative to V-PDB. The reproducibility of routinely anal-ysed laboratory CaCO3 standards is better than 0.1‰ (1 SD) for both δ18O and δ13C.

ResultsDating the fossil layerThe diachroneity of the fossil layer was assessed by dating five fossil elements collected from a stratigraphical context at site TR4 (Figure 4, Table 1). The samples were taken from two scoop sam-ples that contained an overlapping stratigraphy of coral sands (B) at the base, overlain by 200 mm lake marl (C), the fossil layer (D) and anthropogenically dumped basaltic gravels and boulders (E). The stratigraphy at site TR4 was similar to site TR0 (Rijsdijk et al., 2009), but the fossil layer D was up to 1 m thick and rich in decomposed organic sediments (up to 20–50% of the volume of layer D). Wood stem fragments were abundant at the base of the fossil layer, and rootlets (A11) were sampled from the coral sand layer B just below the gyttja layer C. A seed (A10) was sampled from the top of lake marl layer C, marking the base of fossil layer D. Three fossils were sampled from the fossil layer D: a fresh wood branch fragment (A14); a shell fragment of the giant tor-toise (A13); and a seed present in the top of layer D (A12) (Table 1, Figure 4a). The uncalibrated 14C ages of these five samples are very close to the ages earlier obtained from fossils derived at sites TR0 and CH5, suggesting a lateral diachronism of c. 135 14C yr (Table 1: A1–A13). Nearly all 1 sigma ranges of the uncalibrated 14C ages of all dated bones overlap (Figure 5); therefore, on statis-tical grounds the 14C concentrations of these bones could be con-sidered as the 1 sigma variation of the same age. This implies that all animals may have died during a single mass mortality event. Averaging nine 14C datings of vertebrate bones yields an age of 3850 ± 15 14C yr BP, placing the mass mortality event between 4235 and 4100 cal. yr BP (1 sigma) (Figure 6). The 1 sigma spread of the averaged age is also generated by two peaks in the calibra-tion curve and may be smaller (Figure 6). Although statistically justified, the sedimentological data suggest that the vertebrate bones did not accumulate during a single event. The calibrated

ages of fossils from layer D (A12–A14) at site TR4 are systemati-cally c. 100 to 150 yr (visual estimation) younger than the ages from sites TR0 and CH3 (Figure 5), but these dates still overlap the other dates within their 1 sigma ranges, suggesting a statistical similarity. Vertically, seed A10 from the top of lake marl layer C is the oldest (4380–4155 cal. yr BP) of all datings, and seed A12 from the top of layer D the youngest (4145–3985 cal. yr BP) (Figure 4a). These dates suggest the fossil bed accumulated within a period of 10 to 400 years (Table 1). While the minimum time span is certainly an underestimation, the maximum time span of 400 yrs is likely an overestimation, given that the width of the time span can also be attributed to plateau effects of the cali-bration curve. The nine datings of vertebrate bones show a similar time span. However, the radiocarbon ages of the fossils are too close and their resolution too low to determine the time window of death and fossil layer formation. Based on the 1 sigma time ranges in Figure 5 we visually estimate the range in age of the bones also between 100 and 150 yr between 4260 and 4100 cal. yr BP. This narrow time window and period of death is in contrast with the younger ages of the MAS museum specimens of Cylin-draspis sp. of 1800 and 1200 cal. yr BP (Burleigh and Arnold, 1986). Likely, these fossils were retrieved from basin 0 situated > 50 m to the east from site TR0 and positioned at > 2 m higher in altitude (Hume et al., 2009). However without information on the precise sample locations of these museum specimens, the signifi-cance of their younger ages cannot be evaluated.

Hydrological data of MASLike many wetlands in Mauritius, MAS was anthropogenically in-filled with basaltic rocks in the 1940s. Records from the nine-teenth century describe the marsh as a >1.5 m deep lake (Hume et al., 2009), therefore, what we here denote as ‘groundwater level’ at MAS was in fact the lake level prior to basin-fill. Our geological borehole and groundwater level data confirms that removal of the infill would result in a permanent lake more than 2 m deep (Figure 2b). The groundwater overlies a saline ocean water wedge and increases in salinity with depth and in the direc-tion of the ocean (Rijsdijk et al., 2009). Mean groundwater levels fluctuate with the tide around mean ocean level and are on aver-age 100 mm above mean ocean level (Figure 7a), and our hydro-logical data from 2007 to 2008 clearly show that groundwater at MAS responds to local rainfall and evapotranspiration (Figure 7a). During wet periods and rainfall events with intensities larger than 100 mm/day, groundwater levels never rise more than 500 mm above surface level. On the other hand, groundwater levels do not fall below 200 mm under mean ocean level, not even after a period with a cumulative precipitation deficit of 450 mm (Figure 7b). Maximum rates of groundwater rise attributed to rainfall are 1.2 cm/h, a lower value than the tidally induced rises of c. 4 cm/h. These slow water-table rises preclude scenarios that vertebrates in MAS were caught by surprise and drowned as a result of rapidly rising water. On the other hand during dry peri-ods the rates of water level decrease are as low as 0.16 cm/h, values also lower than the highest rate of tidal falls. The hydro-logical data demonstrates that the groundwater system at MAS is sensitive to seasonal variation and consequently to droughts.

La Vierge stalagmite data from RodriguesThe age model of the La Vierge stalagmite indicates relatively continuous growth of the speleothem which began 4125 cal. yr BP and ceased 2260 cal. yr BP (Figure 8). The oxygen isotope (δ18O) record of the speleothem has, on average, a temporal reso-lution of 1 sample every 19 years and shows significant decadal- to centennial-scale cyclicity, interpreted to reflect repetitive wet to dry climate alternations between 4122 and 2260 cal. yr BP




(Figure 8). Relatively low δ18O values indicate humid conditions and high values indicate dry conditions. The carbon isotope (δ13C) record shows a general trend from higher values at 4125 cal. yr BP towards lower values at 2260 cal. yr BP. Superimposed are decadal- to centennial-scale variations in δ13C, which possibly reflect changes in drip rate, or in the dynamics of thinly developed soils overlying the cave site. δ13C values do not co-vary with δ18O values. In the period of stalagmite growth, the record shows an alternation of humid periods lasting up to ~200 years with dry periods lasting decades. The stalagmite commenced growing just after the c. 4200 cal. yr BP mass mortality event at Mauritius. Although the 4200 drought period itself is not recorded, the onset of this stalagmite’s growth may possibly follow from the subse-quent regional shift to more humid conditions. Analysis of

additional stalagmites would be required to assess the conditions for drought prior to ~4100 cal. yr BP.

Interpretation and environmental reconstructionThe formation of the fossil layer

The sedimentary sequence at MAS formed in response to sea level rise, as rising saline marine groundwater raised the freshwa-ter-table (Rijsdijk et al., 2009), and illustrates a progressive deep-ening of a freshwater lake. Coral sands B represent aeolian sands that were concentrated in the rock valley of MAS between 10 and 5 cal. kyr BP, while the lake marl layer C1 is an accumulation of

Figure 4. (a) Generalized stratigraphy of two scoop samples at TR4. Scoop length is 100 cm and width 40 cm. Stars indicate position of samples for radiocarbon dating. Captions right of scoop sketches show the radiocarbon sample Ax and calibrated age (see Table 1). Depth is in m below mean sea level (M.S.L.). Accuracy of depth is ± 25 cm. (b) Picture of scoop sample showing the woody fossil layer D1 and the bony fossil layer D2. Mechanical digger scoop is 50 cm × 100 cm. Note the bones within the organic sediments. (c) Close up of bony fossil layer D. Bones are dispersed within seeds (4–40 mm). Pelvis of dodo within organic sediment




Tabl

e 1.

Rad

ioca

rbon

sam

ples

, inc

ludi

ng t

hose

pub

lishe

d in

Rijs

dijk

et

al. (

2009

)

Labe

lsG

roni

ngen

la

b co

deSp

ecim

ens

DR

P sa

mpl

e co

deFo

ssil

Spec

ies

Dup

loLo

catio

nLa

yer

Age

1 si

gma

Cal

ibra

ted

age

rang

eR

efer

ence

yr B

Cyr

BP

RC

A-M

AS

A0

GrA

-321

07M

ollu

sca

Car

bona

te-

Snai

lPa

chys

tyla

bico

lor

TR

0La

yer

D40

7035

2570

2480

4520

4430

Rijs

dijk

et

al. 2

009

RC

A-M

AS

Al

GrA

-313

62si

ngle

bon

eC

olla

gen

-Fe

mur

Raph

uscu

culla

tus

1T

R 0

Laye

r D

3865

4023

4021

4542

8540

90R

ijsdi

jk e

t al

. 200

9R

CA

-MA

S A

2G

rA-3

1363

sing

le b

one

Col

lage

n-

Fem

urRa

phus

cucu

llatu

slb

TR

0La

yer

D39

0540

2455

2205

4405

4155

Rijs

dijk

et

al. 2

009

RC

A-M

AS

A3

GrA

-313

64si

ngle

bon

eC

olla

gen

-T

ibio

tars

usRa

phus

cucu

llatu

s2

TR

0La

yer

D38

5040

2295

2140

4245

4090

Rijs

dijk

et

al. 2

009

RC

A-M

AS

A4

GrA

-313

67si

ngle

bon

eC

olla

gen

-T

ibio

tars

usRa

phus

cuct

illatu

s2b

TR

0La

yer

D38

6040

2335

2140

4285

4090

Rijs

dijk

et

al. 2

009

RC

A-M

AS

A5

GrA

-313

68se

edW

ood

-Se

eds

Side

roxy

lon

gran

diflo

rum

TR

0La

yer

D38

7540

2345

2150

4295

4095

Rijs

dijk

et

al. 2

009

RC

A-M

AS

A6

GrA

-313

72si

ngle

bon

eC

olla

gen

-C

arap

ace

Cylin

dras

pis

sp.

CH

5La

yer

D?

(510

cm

)38

9040

2430

2200

4380

4150

Rijs

dijk

et

al. 2

009

RC

A-M

AS

A7

GrA

-313

73si

ngle

bon

eC

olla

gen

-C

arap

ace

Cylin

dras

pis

sp.

CH

5La

yer

D?

(510

cm

)38

7540

2345

2150

4295

4095

Rijs

dijk

et

al. 2

009

RC

A-M

AS

A8

GrA

-313

74si

ngle

bon

eC

olla

gen

-C

arap

ace

Cylin

dras

pis

sp.

CH

5La

yer

D?

(398

cm

)38

4040

2290

2140

4240

4090

Rijs

dijk

et

al. 2

009

RC

A-M

AS

A9

GrA

-313

69si

ngle

bon

eC

olla

gen

-C

arap

ace

Cylin

dras

pis

sp.

CH

5La

yer

D?

(398

cm

)38

8540

2400

2200

4350

4145

Rijs

dijk

et

al. 2

009

RC

A-M

AS

A10

GrA

-396

22se

edW

ood

MA

S 07

-380

Seed

Side

roxy

lon

gran

diflo

rum

TR

4La

yer

D b

ase

3900

3024

3522

0543

8041

55ne

wR

CA

-MA

S A

11G

rA-3

9694

root

letb

Woo

dM

AS

07-3

81R

ootle

tsU

nide

ntifi

edro

otle

tsT

R4

Laye

r B

3815

3022

7520

6042

2540

05ne

wR

CA

-MA

S A

12G

rA-3

9689

seed

Woo

dM

AS

07-3

79Se

edSi

dero

xylo

ngr

andi

floru

mT

R4

Laye

r D

top

3775

3022

0020

3541

4539

85ne

wR

CA

-MA

S A

13G

rA-3

9756

sing

le b

one

Col

lage

nM

AS

07-3

79Bo

neCy

lindr

aspi

ssp

.T

R4

Laye

r D

top

3770

3021

9520

3541

4539

85ne

wR

CA

-MA

S A

14G

rA-3

9621

fres

h w

ood

Woo

dM

AS

07-3

79W

ood

Uni

dent

ified

bran

chT

R4

Laye

r D

top

3785

3022

0020

4041

5039

90ne

wR

CB-

MA

S Bl

GrN

-300

69tw

o bo

nes

com

bine

dcC

olla

gen

-2

bone

sCy

lindr

aspi

ssp

.T

R 0

Laye

r D

3700

4021

1519

5040

6539

00R

ijsdi

jk e

t al

. 200

9

RC

B-M

AS

B2G

rN-3

0070

woo

d st

umps

c ’ d

Woo

d-

Woo

dU

nide

ntifi

edbr

anch

esT

R 0

Laye

r D

4030

2525

6524

6545

1544

15R

ijsdi

jk e

t al

. 200

9

In t

ext

sam

ples

are

den

oted

by

‘A0’

etc

. as

show

n in

the

labe

ls c

olum

n. R

adio

carb

on s

ampl

es in

gre

y ro

ws

are

excl

uded

from

ass

essi

ng a

ge o

f fos

sil b

ed a

nd m

orta

lity

even

t as

the

y ar

e co

mbi

ned

bulk

sam

ples

and

/or

may

pre

-dat

e bo

ne b

ed fo

rmat

ion.

a Ter

rest

rial

land

sna

il sp

ecie

s en

dem

ic t

o M

auri

tius

and

Réu

nion

.b R

ootle

ts p

rese

nt in

laye

r B

(fern

or

palm

) m

ay b

e fr

om p

lant

tha

t ha

d gr

own

in la

yer

D.

c Bul

k da

ting

of c

ombi

ned

sam

ples

.d B

ranc

hes

and

stem

s m

ay p

re-d

ate

the

bone

bed

laye

r as

tre

e gr

owth

may

hav

e co

mm

ence

d pr

ior

to fo

ssil

bed

form

atio

n.




Figure 5. Calibrated age ranges of dated fossil. A codes (A1–A14) correspond with A labels of radiocarbon dates present within first column of Table 1. Grey frequency curves represent probability of age distributions, black horizontal bar lines underneath curves represent 1 sigma spread. Output from OxCal calibration programme (Bronk Ramsey, 2009)

Figure 6. Averaged age of nine vertebrate bone datings (A1, A2, A3, A4, A6, A7, A8, A9, A13). Vertical: Radiocarbon determination (BP), radiocarbon age (14C yr BP); horizontal calibrated date calBP, calibrated age (cal. yr BP). Output from OxCal calibration programme (Bronk Ramsey, 2009). Black horizontal bar lines mark 1 sigma (67.5%) range. Grey undulating curve show the radiocarbon measurements on the tree rings with vertical error bars (plus and minus one standard deviation) and are part of the radiocarbon calibration curve for the Southern Hemisphere (McCormac et al., 2004)




Figure 7. (a) Groundwater level at MAS (hourly data) basin 1 referenced to present mean ocean level (M.S.L.) and daily precipitation surplus (precipitation minus Penman evaporation). (b) Cumulative precipitation deficit (precipitation minus Penman evaporation). Data from meteorological station Mont Desert, 500 m from MAS, measured between 5 August 2007 and 3 October 2008

fine decomposed organic debris and calcium carbonates. The sediments are the oldest lake deposits at MAS formed during an early phase of stagnation of shallow (<0.50 m) low salinity fresh water. Compared with 500 mm at TR0 and TR1 (see Rijsdijk et al., 2009) the fossil layer D at TR4 is thicker (> 1 m) and matrix supported, with decomposed organic sediment forming the matrix. The organic sediment of D1-3 is interpreted as decom-posed plant debris (gyttja) that accumulated under tranquil condi-tions in the lake. The woody basal layer represents the drowning

and death of trees due to the deepening of the lake over time. Trees may already have grown at MAS prior to standing water conditions; thus fossil wood likely predates the bone bed (e.g. sample B2), and the organic-rich sediment that forms the matrix of the fossil layer (Figure 4b, c) must have accumulated gradually. It is certain that the organic sediment did not accumulate simulta-neously with the vertebrate carcasses, as the organic lake sedi-ments required standing water for accumulation. This suggests that the bones could only accumulate during dry conditions when




the lake floor was exposed; allowing terrestrial vertebrates to enter the lake floor, but it is likely that the bones dispersed within the organic matrix did not becoming deposited during a single event. Furthermore, it cannot be excluded that the vertebrate fos-sils originally formed a single bone layer and later became dis-persed within the peat layer as a result of their displacement by basaltic blocks from layer E (the disturbed layer in Figure 4a). The radiocarbon datings (Figures 4a, 5) confirm that the fossil layer is diachronic, which is supported by the sedimentological data, and also indicate that the time window for its formation is very short; their temporal resolution too low to resolve the exact timing of fossil accumulation. Here we have estimated that the lateral and vertical diachroneity of the fossil layer may be less than 150 years. The radiocarbon datings confirm that the forma-tion of the fossil layer coincided with megadrought conditions prevailing between 4300 and 4000 cal. yr BP.

MAS sensitivity to drought

The hydrological data demonstrate that the groundwater system at MAS is sensitive to seasonal variation and consequently to droughts. During periods of net rainfall deficits, the groundwater-table slowly falls up to 200 mm below present sea level, and dur-ing severe droughts this may even be lower. A lake level below ocean levels will induce a reverse landward flow from saline ocean water towards MAS, leading to a stronger salt intrusion, and subsequently to salinization of fresh water. Between 4500 and 4000 yr ago a sea level rise occurred with 0.5 mm/yr from −1.5 m to −1.25 m M.S.L. (Camoin et al., 2004; Zinke et al., 2003), and it is likely that at these times the mean groundwater-table at MAS was equivalent to the present ocean level. The top of the coral sand layer B (Figure 4a) can be taken as the lowermost floor of the lake. Borehole evidence indicates that this top ranges between −1.5 and −2.5 m M.S.L. This signifies that between 4300 and 4100 cal. yr BP with sea level at −1.3 m M.S.L., the mean water-table was more than 700 mm above the sand. For terrestrial

vertebrates, especially giant tortoises and flightless birds, to access the top of the deepest part of the exposed lake floor, as is indicated by the presence of their bones at TR4, a lowering of the water-table by more than 700 mm should have occurred, and the maximum lowering of water levels that led to complete desicca-tion of the lake must have been reached during the dry season. Given an annual maximum 200 mm fall measured during normal dry season conditions, lake level fall lower than 500 mm could only be achieved by an extreme dry season. Such water level drops can only be explained when evaporation rates largely exceed groundwater discharge rates, coinciding with reduced upland derived groundwater base flow as a result of less rainfall in the highlands. The 14C datings indicate that no vertebrate bones accumulated after 4000 cal. yr BP. This suggests that the lake bot-tom was no longer exposed after this date, and must have been permanently too deep > 750 mm for terrestrial vertebrate access. Clearly these permanently high lake levels after 4000 cal. yr BP cannot be attributed to the 4200 cal. yr BP continuous sea level rise of 0.5 mm/yr (Camoin et al., 2004). It suggests instead that the extreme drought period had ceased, humidity levels had increased, and a permanent deep lake had returned. Furthermore, a water-table lowering induced by a marine regression after 4200 cal. yr BP can also be excluded (Camoin et al., 2004). In contrast to reef sites across the Pacific, and at continental margins where a high sea level stand occurred between 6000 and 4000 cal. yr BP followed by a regression (Fleming et al., 1998; Grossman et al., 1998), no such trend is observed for the Mascarene Islands (Camoin et al., 2004; Milne and Mitrovica, 2008; Zinke et al., 2003). Subsequently, sea levels rose with constant rates until reaching present positions between 2000 and 3000 cal. yr BP (Camoin et al., 2004; Davies and Montaggioni, 1985; Zinke et al., 2003), and modeling studies confirm that these rates are common to all southwestern Indian Ocean islands surrounding Madagascar (Milne and Mitrovica, 2008; Milne, personal communication, 2009). From hydrological data we can therefore infer that at least

Figure 8. Stable isotope data of the La Vierge cave stalagmite record. Age model is based on 4 ICP-MS derived U/Th ages (analyses performed at the University of Melbourne, Australia; see Hellstrom, 2003, for further analytical detail). Age model is reported as cal. yr BP relative to 1950 for direct comparison with AMS 14C datings. *, individual U/Th ages used in the age model




seasonally, the MAS became completely desiccated and exposed during the dry period, and at the end of this dry period at 4000 cal. yr BP, the lake bottom was never exposed again.

DiscussionScenarios explaining massive deathBased on our multidisciplinary data, we exclude volcanic events or tsunamis (see Rijsdijk et al., 2009), as we infer instead that multiple rather than single mass mortality events occurred, which were induced by extreme droughts between 4300 and 4100 cal. yr BP. During these droughts, springs and rivers dried out within the area, and the animals may have been attracted by the presence of standing fresh water. In the seasonal droughts the MAS lake dried out completely, providing opportunity for the fauna to enter the exposed lake floor to obtain water from shrinking pools and pud-dles. As lake water level fell below palaeo-ocean levels, landward intrusion of the saline ocean water occurred, and this coupled with greater evaporation induced salinization of the lake. The slow lowering of the freshwater-table combined with a slow increased salinization may have progressively negatively affected the fauna; therefore freshwater depletion and salinization may have caused mass mortalities (see also Darwin, 1860; Hart et al., 1991). Additionally within the warm and shallow MAS lake water, guanotrophication induced by the excrements of a more concentrated fauna may have led to toxic blooms of cyanobacte-ria and potentially contributed to mass mortality (Codd et al., 2005). In addition and as illustrated by the abundance of their remains, terrestrial vertebrates such as flightless birds, tortoises and smaller reptiles that entered the exposed lake floor to drink or wallow in remaining water pools, became trapped in the lake sediments and subsequently died. The latter scenario is better sup-ported by taphonomic analysis of dodo bones (Meijer et al., unpublished data, 2006–2008). The death scenarios warrant fur-ther research to test and assess whether or not they operated simultaneously. In spite of this uncertainty, it is clear that all these scenarios depended on the sensitivity of the lake system to drought and could not operate when the lake was refilled. When assuming that an extreme drought period coincided with a mass mortality event, it probably lasted several decades between 4235 and 4100 cal. yr BP (Figure 5), but during the wet seasons the lake may have refilled, creating faunal turnover, while remaining car-casses were submerged and buried under the organic lake sedi-ments. At the end of the extreme drought period, normal humidity conditions prevailed and a permanent lake again returned at MAS, and an ongoing sea level rise after 4000 cal. yr BP, and by c. 2000 cal. yr BP, had increased elevation of the MAS mean water-table from −1.25 m M.S.L. to present M.S.L. The alkaline and anoxic conditions of the permanent lake promoted the conservation of the submerged bones.

Death rates

In total 461 individuals were found at the excavation localities TR0, TR1, and TR3 on 19 m2. As the full fossil bed extends for 18 632 m2 (see Rijsdijk et al., 2009) it is suggested that c. 485 000 vertebrate individuals are part of the bone bed when a linear extrapolation is applied. This is certainly an underestimation, as not one complete sample of the excavation surface has been pro-cessed, and the samples counted showed a strong bias towards bones larger than 4 mm. This signifies a gross underestimation of

smaller vertebrates such as microchiropteran bats, saurian lizards and passerines, and a bias towards larger (non-avian) macrover-tebrates. Regardless, it still implies that around 4200 years ago, at least half a million vertebrate individuals accumulated in a 2 ha site within c. 150 years. From the Minimum Number of Indi-viduals (MNI) estimates (see Rijsdijk et al., 2009), at least 63% or c. 300 000 comprise giant tortoises Cylindraspis spp., while 7% or c. 34 000 are dodo individuals. For the dodo, however, natural densities and death rates are unknown. The Aldabra giant tortoise Aldabrachelys gigantea and Cylindraspis spp. possibly share a common ancestor (Austin and Arnold, 2002), and although the former can be larger in the carapace, cranial and post-cranial bones, it represents the closest analogue species available to estimate death rates and population density. On Grand Terre, Aldabra, a total population of 147 000 tortoises occur on 116 km2 (Swingland and Lessells, 1979), but from our calculations twice as many are found at MAS in 2 ha. Using a minimum and maximum temporal windows of mortality between 10 and 400 years, it is clear that inferred death rates per hectare per day vastly exceed any known background death rate and sig-nify for tortoises at least one or more mass mortalities had occurred at MAS in this period. While the dominance of the ter-restrial vertebrates can be explained by their lower mobility, the presence of Pteropus fruit bats (7.8%) and passerines (10%) can perhaps be explained by the fact that the area had permanently abundant fruiting trees, in an otherwise dry, seasonal region. Whereas the passerines may have been restricted to small territo-ries, even in periods of water scarcity, water-dependent birds such as flamingos (Phoenicopterus ruber) are relatively rare in the fossil record because of their migratory habits.

Resilience to climate change

The mass mortality event at MAS indicates that insular vertebrate populations were vulnerable to climatic extremes, which must have greatly reduced populations of large vertebrate species within catchments. Despite insufficient evidence from other oce-anic islands, we postulate that this phenomenon was unlikely an exclusive event. Climate driven shifts in hydrological cycles including the 4200 cal. yr BP event and other megadroughts around 9200, 8200, and 5200 cal. yr BP (de Menocal, 2001; Fleitmann et al., 2008; Pross et al., 2009; Staubwasser and Weiss, 2006 and ref. therein), may have had similar effects elsewhere. Regardless, the dodo and other vertebrate species survived until human colonization of the island in the seventeenth century, while some more mobile vertebrates, e.g. Mauritius kestrel (Falco punctatus) and Echo parakeet (Psittacula echo), are still extant (Cheke and Hume, 2008; Groombridge et al., 2000). This sup-ports the idea that the indigenous insular vertebrate species were adapted to withstand climatic extremes. On the other hand the fossil evidence of MAS demonstrates that climatic extremes did induce mass mortalities of species otherwise suitably adapted to their environment, and given the large numbers of casualties at MAS, drought may have significantly increased risks of natural species extinction by adversely reducing populations. We suggest that the survival of species during climatic extremes was not only due to their high resilience to climate change, but also to the diversity of the abiotic environments. We postulate that while local populations may have been affected, the geodiversity of Mauritius in providing alternative freshwater sources in the uplands (Figure 2a), benefiting other subpopulations during




extreme droughts. It is envisaged that ongoing geohydrological research and multiscale modeling will assess the availability of fresh water on island scale under extreme climate change scenar-ios including drought.

Water balances of small volcanic islands (< 2000 km2 ) like Mauritius are sensitive to climatic change, sea level rise, and human-induced water extraction, insomuch that insular verte-brates can become detrimentally stressed because of the com-bined effects (Falkland, 1992; Lal et al., 2002; Wong et al., 2005). Further warming of the tropical oceans along with naturally occurring interannual El Niño-Southern Oscillation (ENSO) events could potentially enhance future megadrought conditions (Funk et al., 2008; Hoerling and Kumar, 2003) affecting insular vertebrates. Combined with increasing anthropogenic stress, and habitat destruction in particular, faunal species will have less environmental safe-haven options in the future and will be more prone to extinction (Caujapé-Castells et al., 2010). In Mauritius 98% of the original forest has been destroyed and the isolated patches that remain are under severe pressure from invasive spe-cies (Baider and Florens, 2006; Cheke and Hume, 2008; Florens, 2008). Conservation of insular vertebrates must therefore include hydrological sensitivity assessments including extreme drought scenario evaluations.

Mauritius drought in a regional context

Based on the multidisciplinary evidence presented here we con-clude that mass mortalities at MAS were induced by extreme drought and compare our evidence with other data in the Indian Ocean area. Although the La Vierge stalagmite does not provide exclusive evidence for a single extreme drought, it supports the occurrence of multiple drought events. Furthermore, from 4000 cal. yr BP, there is evidence for a return to higher humidity and possibly the extreme drought which occurred prior to the growth of the stalagmite (Figures 8 and 9). There is also regional evi-dence, e.g. in Asia, East Africa and Madagascar between 4300 and 4000 cal. yr BP, that similar extreme drought conditions pre-vailed. These droughts led to the lowering of water-tables and desiccation of lakes, increased sea surface temperature (SST) and salinity of ocean water due to higher evapotranspiration, and increases in atmospheric and deltaic marine dust fluxes, all of which affected ecosystems and civilizations (Booth et al., 2005; Marchant and Hooghiemstra 2004; Staubwasser and Weiss, 2006; Thompson et al. 2002; Williams, 2009). Owing to the subconti-nental scale, it is usually referred to as the ‘4200 yr BP mega-drought event’. However, the low temporal resolution of the palaeoclimatological records and possible difference in climato-logical sensitivity of various proxies, the drought peaks region-ally vary from between 4300 to 4000 cal. yr BP. The nearest site to Mauritius where drought conditions are recorded is Lake Tritri-vakely, Madagascar, in which diatom communities show a switch to relative dry conditions around 4000 cal. yr BP (Burney, 1993; Gasse, 2000; Gasse and Van Campo, 2001), whereas most pro-nounced drying in East Africa at 4200 cal. yr BP has been observed around and north of the equator (Gasse, 2000). Further-more, various East African lakes show significant lowering of lake levels and ultimately desiccation in this period (Gasse and Van Campo, 2001; Marchant and Hooghiemstra, 2004; Street-Perrot and Perrot, 1993). Megadrought conditions are reflected as major dust peaks in ice cores from Kilimanjaro evidencing an increase in dust flux during this period derived from desiccated

regions in Africa (Thompson et al., 2002; Wolff et al., 2006) (Figure 9), and the Central African Rift lakes show a similar response to increased dust fluxes. Lake Malawi for example shows a decline of diatom productivities after 4000 cal. yr BP indicating a decrease of dust (ash) input due to increased humidity after a dry period. Prior to 4000 cal. yr BP, increased dust input was generated by dry conditions with the dust transported by strong northerly winds from cool and dry volcanic areas (Figure 9). Oxygen isotope records of the foraminifer Globigerinoides ruber in Arabian Sea cores at the mouth of the Indus River 4200 cal. yr BP, suggest increased salinity because of reduced discharge of the Indus River (Staubwasser et al., 2003). Increased dust transport from a northern Arabian source is inferred from core M5-422 in the Gulf of Oman from dolomite concentrations (Staubwasser and Weiss, 2006), and megadrought conditions may have been enhanced by a La Niña-like mean state in the equatorial Pacific that affected global SST patterns increasing SST in the SW Indian Ocean waters (Koutavas et al., 2006; Rodbell et al., 1999; Senapathi et al., 2010). This seems to be confirmed by a decreased zonal SST gradient across the tropical Pacific as recon-structed from marine cores between 4 and 5 kyr cal. BP (Koutavas et al., 2006). This gradient is associated with changes in the pre-cessional cycle of insolation inducing higher ENSO variability. Palaeorecords show a prolonged shift in the Indian Ocean tropical SST pattern towards similar conditions after 4300 cal. yr BP, reflecting the effect of the Indian Ocean dipole (Abram et al., 2009). A speleothem record from Flores, Indonesia, shows dry conditions prevailing before 4300 cal. yr BP and wet conditions around 4000 cal. yr BP (Griffiths et al., 2009). Also Eastern Indian Ocean corals off Sumatra indicate cool SST between 4700 and 4100 cal. yr BP and warming at 4000 cal. yr BP (Abram et al., 2009), while temperatures in East Africa and the western Indian Ocean indicate opposing trends (Figure 9; Bard et al., 1997; Castañeda et al., 2007; Thompson et al., 2002). Currently a SST configuration of warm conditions in the eastern part of the Indian Ocean and cooler conditions in the west, a so-called negative Indian Ocean Dipole, is associated with relatively dry conditions in East Africa and in the Mascarenes, and wet conditions in Indo-nesia (Saji et al., 1999). Finally an abrupt and prolonged weakening of the Asian monsoon is observed after 4300 cal. yr BP (Figure 9; Fleitmann et al., 2007; Morill et al., 2003; Staubwasser et al., 2003; Wang et al., 2005) that was associated with a warming and extension of the western Pacific warm pool (Koutavas et al., 2006; Stott et al., 2004). These palaeoclimatological data indicate that an extensive megadrought affected the southwest Indian Ocean area and likely included the Mascarenes.

Further research

Although we have demonstrated that drought may have led to mass mortality of insular vertebrates in Mauritius, further research is required to ascertain a more precise timing and frequency of mass mortalities, the most plausible causes of death, and to assess how widespread and severe drought conditions were in the Mas-carenes. Modelling of the MAS hydrologic system may provide quantitative estimations of climatic parameters prevailing during the extreme droughts. Ongoing research on records of climate change from Mauritian sites will provide a better understanding of how intense the megadrought conditions were, how long they prevailed for, and will shed light on how climatic change affected insular fauna (Van der Plas et al., 2011). Further research will also




Figure 9. Compilation of paleoclimatic proxies from the tropical Indian Ocean covering the last 7000 years compared with the La Vierge stalagmite that commenced growing just after 4122 cal. yr BP (versus 1950). (a) Eastern Indian Ocean (EIO) coral Sr/Ca off Sumatra (Abram et al., 2009) as a proxy for mean EIO SST, (b) δ18O of the La Vierge (Rodrigues) stalagmite as a proxy for rainfall in the Mascarene plateau region. (c) The Kilimanjaro ice core δ18O record as a proxy for East Africa air temperature (Thompson et al., 2002). (d) An alkenone derived SST reconstruction from marine sediment core MD85674 from the western equatorial Indian Ocean (3°N, 50°E; Bard et al., 1997). (e) The TEX86 lake sediment record from Lake Malawi (Castañeda et al., 2007) as a proxy for East Africa lake temperature. (f) The lake level changes (black bar, low; grey bar, high) in Ethiopian lakes (Gasse, 2000). The mass mortality event at MAS is indicated by the vertical light-grey bar




ascertain to what extent the high heterogeneity of the landscapes (geodiversity) in Mauritius aided species survival by providing alternative refugia to vertebrates.

ConclusionsOur research at the MAS has shown that a natural mass mortal-ity of insular vertebrates induced by extreme drought had taken place, and the timing of this event coincides with a megadrought that occurred between 4300 and 4000 cal. yr BP, affecting a wide area in and around the Central Indian Ocean (Asia, East Africa, and Madagascar). Hydrological data also suggest that mortality was related to the drying out of a freshwater lake located in the dry leeward coastal region, and that this lake was an important freshwater resource for insular vertebrates. Stalagmite data from Rodrigues shows that alternating humid and dry periods, some of which lasted for several decades, affected the water balances of the oceanic islands in this region. During extreme droughts, it can be envisaged that insular vertebrates species at MAS were repeatedly subjected to mass mortalities. Despite this, during the past 10 000 years that several megadrought events occurred, there is little evidence to suggest that insular vertebrates, includ-ing those of Mauritius, had become extinct until after human interference, which supports the assumption that they are resil-ient to climatic extremes. In the light of the fossil evidence and ongoing habitat loss, we question the likelihood of the survival of remaining insular vertebrates that are contained in isolated forest fragments and reserves, when subject to potential extreme droughts. Now mass mortality at MAS is better understood, it becomes clear that conservation efforts must include hydrologi-cal sensitivity assessments from catchment scale to aquifer scale; and at the minimum provide safe haven options based on these parameters.

Acknowledgements

We thank the staff of the The Francois Leguat Giant Tortoise and Cave Reserve, and Shoals Rodrigues for logistical support during field sampling of the stalagmite in La Vierge Cave. We thank J. Hellstrom from the University of Melbourne (Australia) for providing the U/Th datings on the stalagmite from La Vierge cave. We also thank Suzan Warmerdam, Charlotte Spliethoff and Alexander Prick from the VU University Amsterdam for sampling and analytical support for the stable isotope measure-ments of the stalagmites. We thank Leo Kriegsman and Menno Schilthuizen, and four anonymous reviewers for their insightful feedback on earlier versions of this manuscript. We thank John de Vos for discussions in the field and Astrid Kromhout and Nico Staring for stimulating feedback. We thank Omnicane, World Wildlife Fund, Gen Foundation, Percy Sladen Centenary Fund, the Treub Stichting for interdisciplinary research in the tropics, Deltares and The Geological Survey of the Netherlands-TNO for financing the field work in 2007.

ReferencesAbram NJ, McGregor HV, Gagan MK, Hantoro WS and Suwrgadi BW (2009)

Oscillations in the southern extent of the Indo-Pacific warm pool during the mid-Holocene. Quaternary Science Reviews 28: 2794–2803.

Alcover JA, Seguí B and Bover P (1999) Extinctions and local disappearances of vertebrates in the western Mediterranean islands. In: MacPhee RDE (ed.) Extinctions in Near Time. New York: Kluwer Academic, 165–188.

Austin JJ and Arnold EN (2002) Ancient mitochondrial DNA and morphology elucidate an extinct island radiation of Indian Ocean giant tortoises (Cyl-indraspis). Proceedings of the Royal Society of London 268: 2515–2523.

Baider C and Florens FBV (2006) Current decline of the ‘Dodo-tree’: A case of broken-down interactions with extinct species or the result of new interac-tions with alien invaders? In: Laurance W and Peres C (eds) Emerging Threats to Tropical Forests. Chicago: Chicago University Press, 199–214.

Bard E, Rostek F and Sonzogni C (1997) Interhemispheric synchrony of the last deglaciation inferred from alkenone palaeothermometry. Nature 385: 707–710.

Barnwell PJ (1948) Visits and Dispatches 1598–1948. Port Louis, Mauritius: Standard Printing Establishment.

Booth RK, Jackson ST, Forman SL, Kutzbach JE, Bettis EA, Kreigs J et al. (2005) A severe centennial-scale drought in midcontinental North America 4200 years ago and apparent global linkages. The Holocene 15: 321–328.

Borchiellini S, Delucchi L and Giorgi L (1999) Ile Maurice, Carte Géologique au 1:50 000, Schéma hydrogéologique. Maurice: EU Cooperation Française.

Bronk Ramsey C (2009) Bayesian analysis of radiocarbon dates. Radiocarbon 51: 337–360.

Burleigh R and Arnold EN (1986) Age and dietary differences of recently extinct Indian Ocean tortoises (Geochelone s. lat.) revealed by carbon isotope analysis. Proceedings of the Royal Society of London. Series B, Biological Sciences 227: 137–144.

Burney DA (1993) Late Holocene environmental changes in arid southwestern Madagascar. Quaternary Research 40: 98–106.

Burney DA (1999) Rates, patterns and processes of landscape transformation and extinction in Madagascar. In: MacPhee RDE (ed.) Extinctions in Near Time. New York: Kluwer Academic, 145–164.

Burney DA and Flannery TF (2005) Fifty millennia of catastrophic extinctions after human contact. Trends in Ecology & Evolution 20: 395–401.

Burney DA, James HF, Burney LP, Olson SL, Kikuchi W, Wagner WL et al. (2001) Fossil evidence for a diverse biota from Kaua’i and its transforma-tion since human arrival. Ecological Monographs 71: 615–641.

Camoin GF, Montaggioni LF and Braithwaite CJR (2004) Late glacial to post gla-cial sea levels in the western Indian Ocean. Marine Geology 206: 119–146.

Castañeda IS, Werne JP and Johnson TC (2007) Wet and arid phases in the southeast African tropics since the Last Glacial Maximum. Geology 35: 823–826.

Caujapé-Castells J, Tye A, Crawford DJ, Santos-Guerra A, Sakai A, Beaver K et al. (2010) Conservation of oceanic island floras: Present and future global challenges. Perspectives in Plant Ecology, Evolution and Systemat-ics 12: 107–129.

Cheke A and Hume JP (2008) Lost Land of the Dodo. London: T & AD Poyser.Clark G (1866) Account of the late discovery of dodo remains in the island of

Mauritius. Ibis 2: 141–146.Codd GA, Lindsay J, Young FM, Morrison LF and Metcalf JS (2005) Harm-

ful cyanobacteria. From mass mortalities to management measures. In: Huisman J, Matthijs HCP and Visser PM (eds) Harmful Cyanobacteria. Dordrecht: Springer, 1–23.

Darwin CR (1860) The Voyage of the Beagle. 2nd Edition. New York: Every-mans Library (see pp. 145–147 for eyewitness account on mass mortalities also associated with saline water drinking during drought in Buenos Aires and St Fé region).

Davies PJ and Montaggioni LF (1985) Reef growth and sea level change: The environmental signature. Proceedings of the 6th International Coral Reef Symposium 3: 477–511.

De Menocal PB (2001) Cultural responses to climate change during the late Holocene. Science 292: 667–673.

Falkland AC (1992) Small Tropical Islands. Water Resources of Paradises Lost. IHP Humid Tropics Programme Series No. 2, UNESCO.

Fleitmann D, Burns SJ, Mangini A, Mudelsee M, Kramers J, Villa I et al. (2007) Holocene ITCZ and Indian monsoon dynamics recorded in sta-lagmites from Oman and Yemen (Socotra). Quaternary Science Reviews 26: 170–188.

Fleitmann D, Mudelsee M, Burns SJ, Bradley RS, Kramers J and Matter A (2008) Evidence for a widespread climatic anomaly at around 9.2 ka before present. Paleoceanography 23: doi:10.1029/2007PA001519.

Fleming K, Johnston P, Zwartz D, Yokoyama Y, Lambeck K and Chappell J (1998) Refining the eustatic sea-level curve since the Last Glacial Maxi-mum using far- and intermediate-field sites. Earth and Planetary Science Letters 163: 327–342.

Florens FBV (2008) Ecologie des forêts tropicales de l’Ile Maurice et impact des espèces introduites envahissantes. Unpublished PhD thesis, Université de la Réunion.

Funk C, Dettinger MD, Michaelsen JC, Verdin JP, Brown ME, Barlow M et al. (2008) Warming of the Indian Ocean threatens eastern and southern African food security but could be mitigated by agricultural development. Proceedings of the National Academy of Sciences 105: 11 081–11 086.

Gasse F (2000) Hydrological changes in the African tropics since the Last Glacial Maximum. Quaternary Science Reviews 19: 189–211.




Gasse F and Van Campo E (2001) Late Quaternary environmental changes from a pollen and diatom record in the southern tropics (Lake Tritrivakely, Madagascar). Palaeogeography, Palaeoclimatology, Palaeoecology 167: 287–308.

Griffiths ML, Drysdale RN, Gagan MK, Zhao J-X, Ayliffe LK, Hellstrom JC et al. (2009) Increasing Australia-Indonesian monsoon rainfall linked to early Holocene sea-level rise. Nature Geoscience 2: 636–639.

Groombridge JJ, Jones CG, Bruford MW and Nichols RA (2000) ‘Ghost’ alleles of the Mauritius kestrel. Nature 403: 616.

Grossman EE, Fletcher CH and Richmond BM (1998) The Holocene sea-level highstand in the equatorial Pacific: analysis of the insular paleosea-level database. Coral Reefs 17: 309–327.

Hart BT, Bailey P and Edwards R (1991) A review of the salt sensitivity of the Australian freshwater biota. Hydrobiologia 210: 105–144.

Hellstrom JC (2003) Rapid and accurate U/Th dating using parallel ion-count-ing multicollector ICP-MS. Journal of Analytical Atomic Spectrometry 18: 146–1351.

Hoerling M and Kumar A (2003) The perfect ocean for drought. Science 299: 691–694.

Holdaway RN (1999) Introduced predators and avifaunal extinction in New Zealand. In: MacPhee RDE (ed.) Extinctions in Near Time. New York: Kluwer Academic, 189–235.

Hume JP (2005) Contrasting taphofacies in ocean island settings: The fossil record of Mascarene vertebrates. In: Alcover JA and Bover P (eds) Pro-ceedings of the International Symposium ‘Insular Vertebrate Evolution: The Paleontological Approach’. Monografies de la Societat d’Historia Natural de les Balears 12: 129–144.

Hume JP, Cheke AS and McOran-Campbell A (2009) How Owen ‘stole’ the Dodo: Academic rivalry and disputed rights to a newly-discovered sub-fossil deposit in nineteenth century Mauritius. Historical Biology 21: 33–49.

Koutavas AG, Olive C and Lynch-Stieglitz J (2006) Mid-Holocene El Niño-Southern Oscillation (ENSO) attenuation revealed by individual foramin-ifera in eastern tropical Pacific sediments. Geology 34: 993–996.

Lal M, Harasawa H and Takahashi K (2002) Future climate change and its impacts over small island states. Climate Research 19: 179–192.

Marchant R and Hooghiemstra H (2004) Rapid environmental change in African and South American tropics around 4000 years before present: A review. Earth-Science Reviews 66: 217–260.

Martin PS (1984) Prehistoric overkill: The global model. In: Martin PS and Klein RG (eds) Quaternary Extinctions: A Prehistoric Revolution. Tus-con: Arizona University Press, 354–403.

Martin P and Steadman DW (1999) Prehistoric extinctions on islands and continents. In: MacPhee RDE (ed.) Extinctions in Near Time. New York: Kluwer Academic, 17–55.

McCormac FG, Hogg AG, Blackwell PG, Buck CE, Higham TFG and Reimer PJ (2004). SHCal04 Southern Hemisphere calibration, 0–11.0 cal kyr BP. Radiocarbon 46: 1087–1092.

Milne GA and Mitrovica JX (2008) Searching for eustasy in deglacial sea-level histories. Quaternary Science Reviews 27: 2292–2302.

Moree P (1998) A Concise History of Dutch Mauritius, 1598–1710. London and New York: Kegan Paul International.

Morill C, Overpeck JT and Cole JE (2003) A synthesis of abrupt changes in the Asian summer monsoon since the last deglaciation. The Holocene 13: 465–476.

North-Coombes A (1980) La découverte des Mascareignes par les Arabes et les Portuguais. Retrospective et mise à point. Port Louis, Mauritius: Ser-vice Bureau.

Olson S (1989) Extinctions of islands, Man as a catastrophe. In: Western D and Pearl M (eds) Conservation for the Twenty First Century. London: Oxford University Press, 50–53.

Olson SL and Hearty PJ (2003) Probable extirpation of a breeding colony of Short-tailed Albatross (Phoebastria albatrus) on Bermuda by Pleistocene sea-level rise. Proceedings of the National Academy of Sciences 100: 12 825–12 829.

Olson SL and James HF (1992) Fossil birds from the Hawaiian Islands: Evi-dence for wholesale extinction by man before Western contact. Science 217: 633–635.

Padya BM (1989) Weather and Climate of Mauritius. Geography of Mauritius Series, Mahatma Ghandi Instititute. Moka: Mahatma Gandhi Institute Press.

Proag V (1995) The Geology and Water Resources of Mauritius. Geography of Mauritius Series, Mahatma Ghandi Instititute. Moka: Mahatma Gandhi Institute Press.

Pross J, Kotthoff U, Müller UC, Peyron O, Dormoy I, Schmiedl G et al. (2009) Massive perturbation in terrestrial ecosystems of the Eastern Mediterra-nean region associated with the climatic event 8.2 kyr B.P. Geology 37: 887–890.

Rijsdijk KF, Hume JP, Bunnik F, Florens FBV, Baider C, Shapiro B et al. (2009) Mid-Holocene vertebrate bone Concentration-Lagerstätte on oce-anic island Mauritius provides a window into the ecosystem of the dodo (Raphus cucullatus). Quaternary Science Reviews 28: 14–24.

Rodbell DT, Seltzer GO, Anderson DM, Abbott MB, Enfield DB and Newman J (1999) An 15,000-year record of El Niño-driven alluviation in southwestern Ecuador. Science 283: 516–520.

Saddul P (2002) Mauritius. A Geomorpholical Analysis. Geography of Mauri-tius Series, Mahatma Gandhi Institute. Moka: Mahatma Gandhi Institute Press.

Saji HH, Goswami BN, Vinayachandran PN and Yamagata T (1999) A dipole mode in the tropical Indian Ocean. Nature 401: 360–363.

Senapathi D, Underwood F, Black E, Nicoll MAC and Norris K (2010) Evi-dence for long-term regional changes in precipitation on the East Coast Mountains in Mauritius. International Journal of Climatology doi: 10.1002/joc.1953.

Staubwasser M and Weiss H (2006) Holocene climate and cultural evolution in late prehistoric–early historic West Asia. Quaternary Research 66: 372–387.

Staubwasser M, Sirocko F, Grootes PM and Segl M (2003) Climate change at the 4.2 ka BP termination of the Indus valley civilization and Holo-cene south Asian monsoon variability. Geophysical Research Letters 30: doi:10.1029/2002GL016822.

Steadman DW (1995) Prehistoric extinctions of Pacific Island birds: biodiver-sity meets zooarcheology. Science 267: 1123–1131.

Steadman DW (2006) Extinction and Biogeography of Tropical Pacific Birds. London: The University of Chicago Press.

Stott L, Cannariato K, Thunell R, Haug GH, Koutavas A and Lund S (2004) Decline of surface temperature and salinity in the western tropical Pacific Ocean in the Holocene epoch. Nature 431: 56–59.

Street-Perrot FA and Perrot RA (1993) Holocene vegetation changes, lake levels and climate of Africa. In: Wright T Jr Kutzbach JE, Webb III T, Ruddiman WF, Street-Perrott FA and Bartlein PJ (eds) Global Climates Since the Last Glacial Maximum. Minnesota: University of Minnesota Press, 318–356.

Swingland IR and Lessells CM (1979) The natural regulation of giant tortoise populations on Aldabra atoll. Movement polymorphism, reproductive suc-cess and mortality. Journal of Animal Ecology 48: 639–654.

Thompson LG and Davis ME (2007) Ice core records, Africa. In: Elias SA (ed.) Encyclopedia of Quaternary Science. Oxford: Elsevier, 1220–1225.

Thompson LG, Mosley-Thompson E, Davis ME, Henderson KA, Brecher HH, Zagorodnov VS et al. (2002) Kilimanjaro Ice core records: Evidence of Holocene climate change in Tropical Africa. Science 298: 589–593.

Turvey S (2009) In the shadow of megafauna: Prehistoric mammal and bird extinctions across the Holocene. In: Turvey ST (ed.) Holocene Extinc-tions. Oxford: Oxford University Press, 17–40.

Van der Plas G, De Boer E, Hooghiemstra H, Florens FBV, Baider C and Van der Plicht HJ (2011) Mauritius since the last glacial: Environmental and climatic reconstruction of the last 38,000 years from Kanaka Crater. Jour-nal of Quaternary Science in press.

Wang P, Clemens S, Beaufort L, Braconnot P, Ganssen G, Jian Z et al. (2005) Evolution and variability of the Asian monsoon system: State of the art and outstanding issues. Quaternary Science Reviews 24: 595–629.

Williams MAJ (2009) Human impact on the Nile basin: Past, present, future. In: Dumond HJ (ed.) The Nile. Monographiae Biologicae, 89. Springer 771–779.

Wolff EW, Fischer H, Fundel F, Ruth U, Twarloh B, Littot GC et al. (2006) Southern Ocean sea-ice extent, productivity and iron flux over the past eight glacial cycles. Nature 440: 491–496.

Wong PP, Marone E, Lana P, Agard J, Fortes M, Moro D et al. (2005) Island systems. In: Mooney HA and Cropper A (eds) Ecosystems and Human Well-being: Millennium Ecosystem Assessment. Washington: Island Press, 663–680.

Zinke J, Reijmer JJG, Thomassin BA and Dullo W-C (2003). Postglacial flood-ing history of Mayotte Lagoon (Comoro Archipelago, southwest Indian Ocean). Marine Geology 194: 181–196.



The Holocene21(8) 1279© The Author(s) 2011Reprints and permission:sagepub.co.uk/journalsPermissions.navDOI: 10.1177/0959683611428412hol.sagepub.com

428412 HOLXXX10.1177/0959683611428412CorrigendumInternational Journal of Music Education

Corrigendum

Kenneth F. Rijsdijk, Jens Zinke, Perry G.B. de Louw, Julian P. Hume, Hans (J.) van der Plicht, Henry Hooghiemstra, Hanneke J.M. Meijer, Hubert B. Vonhof, Nick Porch, F.B. Vincent Florens, Claudia Baider, Bas van Geel, Joost Brinkkemper, Tamara Vernimmen and Anwar Janoo (2011) Mid-Holocene (4200 kyr BP) mass mortalities in Mauritius (Mascarenes): Insular vertebrates resilient to climatic extremes but vulnerable to human impact. The Holocene. DOI: 10.1177/0959683611405236.

The title of this article should refer to 4200 yr BP rather than 4200 kyr BP. The title should read as follows:

Mid-Holocene (4200 yr BP) mass mortalities in Mauritius (Mascarenes): Insular vertebrates resilient to climatic extremes but vulnerable to human impact



Mid-Holocene (4200 kyr BP) mass mortalities in Mauritius (Mascarenes): Insular vertebrates resilient...

Documents

Transcript of Mid-Holocene (4200 kyr BP) mass mortalities in Mauritius (Mascarenes): Insular vertebrates resilient...