Rapid human-induced landscape transformation in Madagascar at the end of the first millennium of the Common Era

Rapid human-induced landscape transformation in Madagascar at the end of the first millennium of the Common Era

Quaternary Science Reviews 134 (2016) 92e99 Contents lists available at ScienceDirect Quaternary Science Reviews journal homepage: www.elsevier.com/...

994KB Sizes 0 Downloads 4 Views

Quaternary Science Reviews 134 (2016) 92e99

Contents lists available at ScienceDirect

Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev

Rapid human-induced landscape transformation in Madagascar at the end of the first millennium of the Common Era Stephen J. Burns a, *, Laurie R. Godfrey b, Peterson Faina c, David McGee d, Ben Hardt d, Lovasoa Ranivoharimanana c, Jeannot Randrianasy c a

Department of Geosciences, 611 North Pleasant Street, University of Massachusetts, Amherst, MA 01003, USA Department of Anthropology, 240 Hicks Way, University of Massachusetts, Amherst, MA 01003, USA D epartement de Pal eontologie et d'Anthropologie Biologique, Facult e des Sciences, B.P 906 e 101 Antananarivo, Madagascar d Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 23 October 2015 Received in revised form 8 January 2016 Accepted 12 January 2016 Available online 21 January 2016

The environmental impact of the early human inhabitants of Madagascar remains heavily debated. We present results from a study using two stalagmites collected from Anjohibe Cave in northwestern Madagascar to investigate the paleoecology and paleoclimate of northwestern Madagascar over the past 1800 years. Carbon stable isotopic data indicate a rapid, complete transformation from a flora dominated by C3 plants to a C4 grassland system. This transformation is well replicated in both stalagmites, occurred at 890 CE and was completed within one century. We infer that the change was the result of a dramatic increase in the use of fire to promote the growth of grass for cattle fodder. Further, stalagmite oxygen isotope ratios show no significant variation across the carbon isotope excursion, demonstrating that the landscape transformation was not related to changes in precipitation. Our study illustrates the profound impact early inhabitants had on the environment, and implies that forest loss was one trigger of megafaunal extinction. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Paleoecology Paleoclimatology Stable isotopes Speleothems Madagascar

1. Introduction Considerable controversy remains over the timing, causes and magnitude of forest and woodland loss in Madagascar (McConnell and Kull, 2014). For much of the past century the prevailing view was that prior to human arrival forest and woodland covered 90% of the island (Humbert, 1927), and that the loss of Madagascar's for^thie, 1921). Recent ests was due to human activity (Perrier de la Ba studies, however, suggest that Madagascar's grasslands are a nat me  re  et al., ural part of the island's ecosystem (Burney, 1987; Que 2012) and may even date to the late Miocene (Bond et al., 2008). Indeed, some authors question the entire narrative of extensive alteration of the landscape by early human activity (Klein, 2002; Kull, 2000). Whether forest loss was the result of early human activity (Gade, 1996), was post-Colonial (Jarosz, 1993) or was, in fact, not significant (Klein, 2002) also has important implications for the causes of the decline and ultimate disappearance of Madagascar's endemic megafauna, which mainly occurred prior to European

contact (Burney et al., 2004; Crowley, 2010). Answering these questions is difficult because there are few high-resolution, welldated studies of Madagascar's paleoclimate and paleoecology. Speleothems, calcium carbonate cave deposits, offer an opportunity to investigate both ecosystem and climate change in the same archive via stable carbon and oxygen isotope ratios, respectively. Carbon isotope ratios are sensitive to the ratio of surface flora that utilize the C3 (woody taxa) versus C4 (grasses) photosynthetic pathways. In the tropics, the oxygen isotope ratio of precipitation is, empirically and mechanistically, strongly correlated with the amount of precipitation, a signal that is captured in the oxygen isotope ratios of speleothem calcite. Speleothems also offer unusually robust chronological control through disequilibrium U-series dating techniques. To investigate the recent environmental and climate history of Madagascar, we collected two actively growing stalagmites, M14-AB2 and M14-AB3, from Anjohibe Cave (15.54 S, 46.89 E, 131 masl) in northwestern Madagascar in 2014.

2. Setting * Corresponding author. E-mail address: [email protected] (S.J. Burns). http://dx.doi.org/10.1016/j.quascirev.2016.01.007 0277-3791/© 2016 Elsevier Ltd. All rights reserved.

Ajohibe Cave is part of a karst system formed within the Eocene

S.J. Burns et al. / Quaternary Science Reviews 134 (2016) 92e99


limestone plateau (Besairie and Collignon, 1972) of northwestern Madagascar (Fig. 1). Anjohibe (‘big cave’ in Malagasy) consists of 5.3 km of cave passages with over two dozen openings (Burney et al., 1997). Samples AB2 and AB3 (Fig. 2) were taken ~1 km from the main entrance to the cave and approximately 400 m distant from one another (for a map of the cave see Burney et al., 1997). The samples were collected in October, at the end of the dry season, and there were very few active drips within the cave. Neither sample was beneath an active drip at the time of collection. Both samples appeared to be still actively or very recently growing, however, based on the clean, white calcite deposits on the top of the stalagmites. The landscape overlying the cave is presently palm savannah (Fig. 1) with small patches of mesic forest in wetter areas (Burney et al., 1997). Rainfall in the region is highly seasonal. At the nearest weather station, Mahajanga, which is ~70 km from the cave, over 80% of the mean annual rainfall (1496 mm) occurs in the four month period from December through April. Mean monthly temperature varies 3.2  C throughout the year, with an annual mean of 27.2  C (data from http://www.ncdc.noaa.gov/data-access).

Fig. 2. Sample photographs. Photograph with scale of cut and polished slabs of samples AB2 and AB3 showing internal layering.

3. Methods The two speleothems were halved along the growth axis and subsampled along growth layers for radiometric dating using uranium-thorium (UeTh) techniques by multi-collector, inductively coupled plasma mass spectroscopy (MC-ICP-MS) (Cheng et al., 2013). Carbon and oxygen stable isotope ratios were measured on 266 samples from AB2 and 173 samples from AB3. Temporal resolution of the stable isotope time series is less than 10 years for all of M14-AB2 and M14-AB3. Age models (Fig. 3) were constructed using 8 age determinations from AB2 and 10 from AB3 (Table 1) and assuming an age for the top of each stalagmite of 2014 CE. Fig. 1. Madagascar ecosystems and site location. This map shows the present distribution of major ecosystems in Madagascar (Mayaux et al., 2000) and the locations of the study area and lake study sites mentioned in the text: 1. Anjohibe Cave, 2. Lake Amparihibe (Burney et al., 2004), 3. Lake Tritrivakely (Gasse and Van Campo, 1998), 4. Lake Kavitaha (Burney, 1987) and 5. Lake Mitsinjo (Matsumoto and Burney, 1994).

3.1. O and C stable isotope analyses The stable oxygen and carbon isotope ratio measurements were


S.J. Burns et al. / Quaternary Science Reviews 134 (2016) 92e99

a Finnigan Delta Plus XL ratio mass spectrometer. Results are reported as the per mil difference between sample and the Vienna Pee Dee Belemnite (VPDB) standard in delta notation where d18O ¼ Rsample/Rstandard  1)*1000, and R is the ratio of the minor to the major isotope. Reproducibility of standard materials is 0.1‰ for d18O and 0.05‰ for d13C. 3.2. UeTh dating

Fig. 3. Age-depth data. These plots show the U/Th age determinations with 2s errors versus depth for each sample.

performed at the University of Massachusetts. Subsamples were microdrilled from a cut and polished slab every 0.5e01 mm along the central growth axis of the stalagmites. The sampling interval yields sub-decadal resolution throughout the record. The samples were analyzed in an on-line carbonate preparation system linked to

UeTh dating samples were prepared and analyzed at MIT. Samples weighing 25e210 mg were combined with a 229 The233Ue236U tracer, digested, and purified via iron coprecipitation and ion exchange chromatography. Separate U and Th aliquots were analyzed using a Nu Plasma II-ES multi-collector ICP-MS equipped with a CETAC Aridus II desolvating nebulizer. Analyses were performed in static mode with 234U and 230Th measured on an ion counter and all other masses measured on Faraday cups. U and Th standard solutions bracketed each analysis to monitor mass bias and ion counter yield. CRM-112a was used to bracket U analyses, and an internally calibrated 229The230The232Th standard (MITh-1) was used for Th. Tailing was assessed for each U standard and sample by measurement of half-masses (233.5, 234.5, 236.5) and mass 237, and tails were fit to a power law. Th tailing was measured once per day by measuring masses 229.5 and 230.5 in the MITh-1 standard. Tailing was consistently 1e2 ppm at 1 amu. Uranium hydride counts were also measured each day and were consistently 2e2.5 ppm. Ionization, transmission and detection efficiency was approximately 1% for both U and Th. MIT's isotope tracer is calibrated against the University of Minnesota's isotope tracer (Cheng et al., 2013) and standard HU-1. The laboratory's age determinations produce good agreement with ages from the Berkeley Geochronology Center in an intercomparison of young (<500 y) powder splits (McGee, unpublished data). Reported isotope ratios uncertainties reflect propagated uncertainties from isotope ratio measurements, spike calibration and isotopic composition, instrument background, procedural blanks (assigned a 50% 2s uncertainty), SEM yield drift, tailing and 233UH counts on 234U. Uncertainties on U and Th half-lives are not included, but would be insignificant relative to other uncertainties. Total procedural blanks ranged from 0.02 to 0.2 fg 230Th in three of the four sample sets, with an anomalously high blank of 1.7 fg 230Th in sample set 150206, and 30e300 pg 238U for all four sample sets. The only sample for which the total uncertainty is increased by more than a decade due to the procedural blank is sample AB3-682 (512 ± 37 CE) due to the unusually high 230Th blank in sample set 150206. Ages were calculated using the 230Th and 234U half-lives of Cheng et al. (2013) and the 238U half-life from Jaffey et al. (1971). Corrections for initial 230Th assume an initial 230Th/232Th ratio of 17 ± 8.5  106 (atomic ratio). The mean value was determined from the construction of Osmond Type II isochrons from three samples at 62 mm depth in stalagmite M14-AB2 using Isoplot 4.15 (Ludwig, 1993), which gave an initial 230Th/232Th ratio of 17 ± 4  106 (Fig. 4). We chose to assign a relative uncertainty of 50% (2s) to allow for temporal variability in the initial ratio. Two pieces of evidence suggest that this mean value and uncertainty are appropriate. First, the 230Th/232Th ratio from our youngest sample (AB2-9.5) is 26.6 ± 1.0  106, suggesting that the initial 230 Th/232Th should be lower than this value. Second, an initial 230 Th/232Th ratio greater than 11  106 is required to maintain stratigraphic order for samples AB3-144 through AB3-257. Similar constraints from stratigraphic order have been used to estimate initial 230Th/232Th ratios in previous studies (Cheng et al., 2000; McGee et al., 2012). Four samples have unusually low U concentrations and produce ages that are anomalously old (AB3-68, AB3-

Table 1 UeTh dating results. Sample ID

Sample set


19 68 87 144 186 219 257 284 359 480 682 833 9.5 62 114 150 190 455 556 674 835 988

10970 140 175 2078 2680 4644 6450 8550 7220 4831 8370 1882 7940 9160 183 7950 5560 3561 1096 10710 8770 8220

220 3 4 42 54 93 130 170 140 97 170 38 160 180 4 160 110 71 22 220 180 160

6120 1446 574 1422 3773 2787 423 444 1488 532 609 10250 1204 496 485 2867 1067 4157 274 281 331 943

120 29 12 29 76 56 9 10 30 11 25 210 25 11 11 57 22 83 6 7 7 19

2.3 5.8 3.4 2.3 3.3 3.4 3.7 2.5 0.4 0.4 0.4 0.9 9.1 10.0 10.0 12.8 3.0 7.8 7.1 6.6 3.8 0.4

1.2 2.8 2.3 1.4 1.6 1.6 1.5 1.6 1.8 1.8 1.1 1.2 1.4 0.6 2.1 1.4 1.3 1.2 2.7 3.0 0.5 7.9

0.001436 0.01544 0.01445 0.009272 0.010395 0.009989 0.009610 0.009817 0.011097 0.012327 0.01377 0.02106 0.0002539 0.001484 0.01252 0.004181 0.004359 0.008719 0.01076 0.010318 0.011386 0.014025

62 62 284 284 480 480

8730 7300 8450 10880 4007 5190

180 150 170 220 80 100

4321 2099 1299 1404 757 1788

87 42 26 28 26 36

0.3 7.9 0.5 0.2 1.3 0.9

1.7 1.8 1.8 1.9 1.2 1.9

0.002019 0.001677 0.013876 0.007362 0.011802 0.012274

U (ng/g)a

± (2s)


Th (pg/g)a

± (2s)


(per mil)b

± (2s)

(230Th/238U) (activity)

Th/232Th (atomic  106)

± (2s)

0.000013 0.00034 0.00028 0.000094 0.000042 0.000046 0.000048 0.000041 0.000052 0.000063 0.00033 0.00061 0.0000099 0.000020 0.00070 0.000018 0.000036 0.000054 0.00012 0.000053 0.000042 0.000084

41 24 70 215 117 264 2328 3000 855 1778 3001 61 26.6 435.7 75.2 184.1 360.8 118.6 683.3 6252 4789 1943

0.4 0.4 1.2 2.2 0.4 1.1 24 27 4 12 130 2 1.0 7.4 4.3 0.7 3.3 0.7 9.9 94 47 13

0.000036 0.000030 0.000071 0.000042 0.000644 0.000070

64.8 92.6 1432.4 906.2 993 565.3

1.1 1.6 9.1 5.1 61 3.2

± (2s)


Age (yr)

± (2s)

Age (yr)

± (2s)

d234U initial

± (2s)

Year C.E.

± (2s)



(per mil)e


156.3 1707 1593 1018 1143 1098 1057 1078 1216 1353 1511 2319 28 164 1387 463 478 963 1188 1139 1253 1540

91 490 1210 938 978 1028 1049 1072 1192 1340 1503 1680 10 157 1080 420 455 825 1159 1136 1249 1526

2.3 5.8 3.4 2.3 3.3 3.4 3.7 2.5 0.4 0.4 0.4 0.9 9.1 10.0 10.0 12.8 3.0 7.8 7.1 6.7 3.8 0.4

1924 1525 805 1077 1037 987 966 943 823 675 512 335 2005 1858 935 1595 1560 1190 856 879 766 489

1.5 38 31 10 5 5 6 5 6 7 37 68 1 2 78 2 4 6 14 7 5 15

33 610 190 40 83 35 6 5 12 7 37 320 9 3 160 21 11 69 15 7 5 15

Notes: Decay constants for 230Th and 234U are from Cheng et al. (2013); decay constant for 238U is from Jaffey et al. (1971). a Reported errors for 238U and 232Th concentrations are estimated to be ±1% due to uncertainties in spike concentration; analytical uncertainties are smaller. b d234U ¼ ([234U/238U]activity  1)  1000. c 230 [ Th/238U]activity ¼ 1  el230T þ (d234Umeasured/1000)[l230/(l230  l234)](1  e(l230-l234) T), where T is the age. “Uncorrected” indicates that no correction has been made for initial d Ages are corrected for detrital 230Th assuming an initial 230Th/232Th of (17 ± 8.5)  106. e d234Uinitial corrected was calculated based on 230Th age (T), i.e., d234Uinitial ¼ d234Umeasured X el234*T, and T is corrected age.


1.2 2.8 2.3 1.4 1.6 1.6 1.5 1.6 1.8 1.8 1.1 1.2 1.4 0.6 2.1 1.4 1.3 1.2 2.7 3.0 0.5 7.9

33 610 190 40 83 35 6 5 12 7 37 320 9 3 160 21 11 69 15 7 5 15

S.J. Burns et al. / Quaternary Science Reviews 134 (2016) 92e99

AB3-19 150713 AB3-68 150713 AB3-87 150713 AB3-144 150713 AB3-186 150713 AB3-219 150713 AB3-257 150713 AB3-284 150713 AB3-359 150730 AB3-480R 150730 AB3-682 150206 AB3-833 150206 AB2-9.5 150616 AB2-62 150616 AB2-114 150616 AB2-150 150713 AB2-190 150616 AB2-455 150616 AB2-556 150730 AB2-674 150616 AB2-835 150616 AB2-988 150616 Isochron analyses AB2-62L 150730 AB2-62R 150730 AB3-284L 150730 AB3-284R 150730 AB3-480 150206 AB3-480L 150730

Distance from top (mm)




S.J. Burns et al. / Quaternary Science Reviews 134 (2016) 92e99

the soil waters and not by sample-specific effects. The oxygen isotope values of both samples vary mainly between 4 and 7‰ VPDB with values for AB2 enriched by 0.3e0.5 per mil compared to AB3. The 11-point running averages (approximately century scale) display similar trends in both samples (Fig. 5). From 500 to 900 CE, both oxygen isotope time series show a decrease in average values of about 1.5‰. After 900 CE average values increase by about 0.5‰ and remain relatively constant until about 1700 CE, then steadily increase by about 2‰ over the next 300 years. The carbon and oxygen isotope time series show no correlative shifts in isotopic values. 5. Discussion

Fig. 4. Isochron data from sample AB2-62. The data indicate an initial 230Th/232Th atomic ratio of 17 ± 4  106. This value with an expanded uncertainty was used to calculate ages corrected for initial 230Th for all samples.

87, AB2-114, AB2-556.) We interpret these horizons as having suffered diagenetic U loss and do not include them in the age models. 4. Results Fig. 5 shows time series of oxygen and carbon isotope values in standard delta notation versus age for both stalagmites. The carbon isotope time series for our samples show relatively constant negative values from 500 CE to about 890 CE. For stalagmite AB2 the average d13C value over this time interval is 7.21‰, and for AB3 it is 8.77‰. The ~1.5‰ difference between the two averages is most likely due to sample-specific parameters such as different drip rates or degrees of cave ventilation (Fairchild et al., 2006; Frisia et al., 2011). At about 890 ± 20 CE an abrupt increase in d13C begins. Over the next 100 years d13C increases by ~8‰ in both stalagmites (Fig. 5). For example, based on the age model of AB2 d13C values are still below 7‰ at 885 CE and rise to 0.62 at 1020 CE. The increase is slightly more rapid in AB3, with d13C values of 8.69 at 892 CE rising to þ0.4 at 992 CE. After about 900 CE the d13C values continue to increase more slowly, with peak values of greater than 4‰ in both samples around 1800e2000 CE. A brief return to slightly negative values is seen around 1700 CE. The reproducibility of the shape, magnitude and timing of the d13C records in the two samples is a strong indication that the positive shift is driven by a regional change in the d13C values of dissolved inorganic carbon in

The carbon isotope ratios of speleothem calcite reflect a mixture of three carbon sources: plant root-respired CO2 in the soil zone, atmospheric CO2, and dissolution of the host carbonate bedrock (McDermott, 2004). The carbon isotopic ratios of the latter two sources vary little, with CO2 of preindustrial atmosphere having a d13C value of 6.5‰ (Leuenberger et al., 1992) and Eocene marine limestones such as that which hosts Anjohibe cave generally ranging from þ1 to þ2‰ (Shackleton, 1987). The d13C value of plant respired CO2 varies primarily as a function of the photosynthetic pathway used by the plants. In landscapes dominated by plants photosynthesizing via the Calvin cycle (C3 plants) e the great majority of temperate plants including trees and shrubs e plantrespired CO2 averages approximately 26‰ VPDB (Farquhar et al., 1989). Plant root respiration also acidifies the soil water, which then equilibrates with the host bedrock and partially exchanges with the atmosphere. In such settings, typical d13C values for total dissolved inorganic carbon (DIC) in the groundwater and for speleothems growing beneath such landscapes lie between 8 and 12‰ VPDB (Fairchild et al., 2006; McDermott, 2004). In regions dominated by plants using the Hatch-Slack photosynthetic pathway (C4 plants such as most dry region grasses), plant respired CO2 averages about 14‰ VPDB (Farquhar et al., 1989). In these landscapes groundwater DIC and speleothem d13C values are enriched by 8e10‰ (Fairchild et al., 2006; McDermott, 2004) compared to regions overlain by C3 plants. For our samples, the d13C values prior to 890 CE are typical for speleothems deposited in caves that are overlain by C3-dominated ecosystems. Our data do not rule out the presence of C4 grasses, but grasses likely made up less than 25% of the flora. Indeed, grass pollen are present in low percentages in lacustrine sediments in several areas of Madagascar well prior to 900 CE (Burney, 1987; Gasse and Van Campo, 1998). We interpret the 8e10‰ enrichment in the carbon isotope values that occurs from 890 to 990 CE to be the result of a rapid, nearly complete transformation of the ecosystem in the surrounding region from one primarily composed of C3 vegetation to one dominated by C4 grasses with very little remnant C3 vegetation. The d13C values of our samples after 990 CE are among the most enriched reported from speleothems. In particular, the d13C values of the past two centuries are enriched beyond what is to be expected from even complete C4 ecosystems. We are not sure of the origin of such enriched values, but one possibility is that gradual drying of the region, as evidenced by a trend of increasing speleothem d18O over the past ~200 years, led to a gradual reduction in drip rate. A decreased drip rate would allow more extensive degassing of isotopically depleted CO2 leading to more enriched speleothem d13C values. In any case, the C4-dominated ecosystem has remained constant over the past ~950 years and comprises the modern landscape. Supporting evidence for our interpretation comes from carbon isotopic analyses of vertebrate bone collagen from fossils in Anjohibe Cave. Extinct taxa that lived in this region before 500 CE have average collagen d13C values of 23‰,

S.J. Burns et al. / Quaternary Science Reviews 134 (2016) 92e99


Fig. 5. Stable isotope time series. Time series of oxygen and carbon stable isotope ratios for stalagmites AB2 (orange) and AB3 (blue) from Anjohibe Cave, Madagascar. The age scale is in calendar years in the Common Era (CE).

indicating a C3 plant diet, while more recently introduced taxa (rats and shrews) have d13C values of 15‰ and higher, indicating an almost exclusively C4 diet (Crowley and Samonds, 2013). Together with environmental indicators from lakes in the region, our results further suggest that the landscape transformation was the result of the expansion of the use of fire as a means of increasing the new growth of grass as forage for cattle. Sediments from lakes across Madagascar, including Kavitaha, Tritrivakely, Mitsinjo and Amparihibe, show marked increases in charcoal and Gramineae pollen percentages sometime between 700 and 1100 CE (Burney, 1987; Burney et al., 2004; Gasse and Van Campo, 1998; Matsumoto and Burney, 1994). These studies indicate important links between landscape burning and grassland expansion, but they do not provide precise estimates of the timing and rate of landscape change, nor have their proxy data allowed quantitative estimates of the increase in grass abundance. Our results place tight constraints on the timing of these changes, indicate that they occurred over a short time interval between 890 and 990 CE, and demonstrate that a complete floral succession from a dominantly forested landscape to the present palm savannah grassland took place in northwestern Madagascar at that time. Burney et al. (2003) note increases in Sporormiella, a fungus found on the dung of large herbivores, in Lakes Kavitaha and Amparihibe at approximately the same time as the increase in charcoal and grass pollen abundance. We posit that the introduction of Bos indicus during the middle to late first millennium (Allibert et al., 1989; Beaugard, 2011; Burney et al., 2003; Fuller and Boivin, 2009) resulted in a shift in human economy from dominant foraging (with its greater dependence on bushmeat hunting and wild plant collecting) to a dedicated agro-pastoralist subsistence strategy, as seen at early large settlement sites such as Irodo (Dewar and Wright, 1993) and Mahilaka (Radimilahy, 1998). The changes in land use led to major habitat modification including major loss or fragmentation of existing forest. Our results suggest that C3 dominated forest remained widespread prior to 900 CE, but was rapidly reduced in extent thereafter. The loss of forest habitat could only have increased the environmental pressure on remaining endemic species that rely on relatively connected forest cover for survival.

Prior assessments of the paleohabitat in the region of Anjohibe have been based on limited data. Burney et al. (1997) suggested that palm savannah might have existed in the region over the past 40,000 years. This assessment, however, derives from only six analyses of pollen contained in speleothems from Anjohibe, of which two were undated, three were deemed securely dated, and one (the most recent) less securely dated at under four thousand years old (Burney et al., 1997). All six samples contain endemic palm and grasses, both of which still occur in the region, together with other trees and shrubs, though in low quantities (Burney et al., 1997). We can infer with confidence, however, that other trees and shrubs were well represented in the area before the megafauna disappeared because of the nature of the vertebrate subfossil assemblage in the cave, which included Prolemur simus, Palaeopropithecus kelyus, and Babakotia radofilai (Godfrey et al., 1999) with known arboreal locomotor habits and food preferences (Jungers et al., 2002). Although humans are generally thought to be the primary cause of the loss of Madagascar's endemic species recent studies of the decline of forest cover have suggested that climate change may have also played an important role. In particular, periods of drought may have been the primary or at least complementary cause of the forest loss, grassland expansion and megafauna disappearance across Madagascar (Virah-Sawmy et al., 2010). The oxygen isotope results from our work provide important data that can address this question. Oxygen isotope ratios of speleothem calcite are primarily controlled by two factors: cave temperature through the temperature dependence of the fractionation between water and calcium carbonate minerals, and the isotopic composition of rainfall (Lachniet, 2009). Tropical temperature variability in the study region over the past 2000 years was likely on the order of one degree C (Weldeab et al., 2014), which would result in 0.25‰ variation in speleothem d18O. Thus, the primary driver of changes in speleothem oxygen isotope variability in our samples is the isotopic composition of precipitation. In tropical regions dominated by summer convective rainfall, interannual changes in d18O of rainfall and of speleothem calcite are inversely correlated with regional precipitation (Bony et al., 2008; Rozanski et al., 1993). Our oxygen isotope time series, while


S.J. Burns et al. / Quaternary Science Reviews 134 (2016) 92e99

displaying up to 2 per mille variability on decadal and centennial timescales, shows no monotonic trend over the past 1500 years. Importantly, the d18O values of our samples do not increase at the time of the very large shift in d13C and, therefore, do not indicate a climatic shift to drier conditions at the time of the inferred ecosystem shift. If anything, the decreasing trend in d18O suggests a trend of increasing rainfall before and continuing briefly after the observed large carbon isotope shift. We thank the Department of Paleontology and Biological Anthropology at the University of Antananarivo, Madagascar, for its help in facilitating this research. 6. Conclusions Stable isotopic time series from two well-dated speleothems from Anjohibe Cave in northwestern Madagascar provide insights into the paleoecology and paleoclimate of the region over the past 1800 years. From 300 CE to 890 CE the vegetation of the region was dominated by plants using the C3 photosynthetic pathway, and is interpreted to have been an open forest landscape with a minor component of C4 grasses. Beginning at 890 CE and over the course of the next 100 years the region changed to one heavily dominated by C4 grasses, similar to the modern landscape. Rather than climate, our results suggest that human activity, specifically expanded use of fire to produce fodder for cattle, resulted in rapid loss of forest habitat. These changes very likely increased environmental pressures on Madagascar's megafauna and accelerated their disappearance. Acknowledgements This research was partially supported by funding from the University of Massachusetts Natural History Collections (LRG). DM acknowledges support from NSF award EAR-1439559 and the MIT Ferry Fund. We greatly appreciate the support and cooperation of the Madagascar Ministry of Art and Culture, Ministry of Mines and Petroleum and Ministry of Higher Education and Scientific Research in sample collection. References ologique de Dembeni (Mayotte), Allibert, C., Argant, A., Argant, J., 1989. Le site arche an Indien 11, 61e172. Archipel des Comores. Etudes Oce Beaugard, P., 2011. The first migrants to Madagascar and their introduction of plants: linguistic and ethnological evidence. Azania Archaeol. Res. Afr. 46, 169e189. ologie de Madagascar, 1. Les terrains sedimenBesairie, H., Collignon, M., 1972. Ge ologiques Madag 35, 1e463. taires. Ann. Ge Bond, W.J., Silander Jr., J.A., Ranaivonasy, J., Ratsirarson, J., 2008. The antiquity of Madagascar's grasslands and the rise of C4 grassy biomes. J. Biogeogr. 35, 1743e1758. http://dx.doi.org/10.1111/j.1365-2699.2008.01923.x. Bony, S., Risi, C., Vimeux, F., 2008. Influence of convective processes on the isotopic composition (d18O and dD) of precipitation and water vapor in the tropics: 1. Radiative-convective equilibrium and Tropical OceaneGlobal AtmosphereeCoupled Ocean-Atmosphere Response Experiment (TOGA-COARE) simulations. J. Geophys. Res. Atmos. 113, D19305. http://dx.doi.org/10.1029/ 2008JD009942. Burney, D.A., 1987. Late Holocene vegetational change in central Madagascar. Quat. Res. 28, 130e143. http://dx.doi.org/10.1016/0033-5894(87)90038-X. Burney, D.A., Burney, L.P., Godfrey, L.R., Jungers, W.L., Goodman, S.M., Wright, H.T., Jull, A.J.T., 2004. A chronology for late prehistoric Madagascar. J. Hum. Evol. 47, 25e63. http://dx.doi.org/10.1016/j.jhevol.2004.05.005. Burney, D.A., Robinson, G.S., Burney, L.P., 2003. Sporomiella and the Late Holocene extinctions in Madagascar. Proc. Natl. Acad. Sci. 100, 10800e10805. http:// dx.doi.org/10.1073/pnas.1534700100. Burney, D., James, H., Grady, F., Rafamantanantsoa, J.-G., Ramilisonina, Wright, H., Cowart, J., 1997. Environmental change, extinction and human activity: evidence from caves in NW Madagascar. J. Biogeogr. 24, 755e767. http:// dx.doi.org/10.1046/j.1365-2699.1997.00146.x. Cheng, H., Adkins, J., Edwards, R.L., Boyle, E.A., 2000. U-Th dating of deep-sea corals. Geochim. Cosmochim. Acta 64, 2401e2416. http://dx.doi.org/10.1016/S00167037(99)00422-6. Cheng, H., Lawrence Edwards, R., Shen, C.-C., Polyak, V.J., Asmerom, Y., Woodhead, J.,

€tl, C., Wang, X., Calvin Alexander Jr., E., 2013. Hellstrom, J., Wang, Y., Kong, X., Spo Improvements in 230Th dating, 230Th and 234U half-life values, and UeTh isotopic measurements by multi-collector inductively coupled plasma mass spectrometry. Earth Planet. Sci. Lett. 371e372, 82e91. http://dx.doi.org/10.1016/ j.epsl.2013.04.006. Crowley, B.E., 2010. A refined chronology of prehistoric Madagascar and the demise of the megafauna. Quat. Sci. Rev., Special Theme Case Stud. Neodymium Isotopes Paleoceanogr. 29, 2591e2603. http://dx.doi.org/10.1016/ j.quascirev.2010.06.030. Crowley, B.E., Samonds, K.E., 2013. Stable carbon isotope values confirm a recent increase in grasslands in northwestern Madagascar. Holocene 23, 1066e1073. http://dx.doi.org/10.1177/0959683613484675. Dewar, R.E., Wright, H.T., 1993. The culture history of Madagascar. J. World Prehistory 7, 417e466. http://dx.doi.org/10.1007/BF00997802. Fairchild, I., Smith, C., Baker, A., Fuller, L., Spotl, C., Mattey, D., McDermott, F., EIMP, 2006. Modification and preservation of environmental signals in speleothems. Earth Sci. Rev. 75, 105e153. http://dx.doi.org/10.1016/j.earscirev.2005.08.003. Farquhar, G.D., Ehleringer, J.R., Hubick, K.T., 1989. Carbon isotope discrimination and photosynthesis. Annu. Rev. Plant Physiol. Plant Mol. Biol. 40, 503e537. http:// dx.doi.org/10.1146/annurev.pp.40.060189.002443. €tl, C., Borsato, A., 2011. Carbon Frisia, S., Fairchild, I.J., Fohlmeister, J., Miorandi, R., Spo mass-balance modelling and carbon isotope exchange processes in dynamic caves. Geochim. Cosmochim. Acta 75, 380e400. http://dx.doi.org/10.1016/ j.gca.2010.10.021. Fuller, D.Q., Boivin, N., 2009. Crops, cattle and commensals across the Indian Ocean. an Indien 43e43, 13e46. Etudes Oce Gade, D.W., 1996. Deforestation and its effects in highland Madagascar. Mt. Res. Dev. 16, 101e116. http://dx.doi.org/10.2307/3674005. Gasse, F., Van Campo, E., 1998. A 40,000-yr pollen and diatom record from Lake Tritrivakely, Madagascar, in the Southern Tropics. Quat. Res. 49, 299e311. http://dx.doi.org/10.1006/qres.1998.1967. Godfrey, L.R., Jungers, W.L., Simons, E.L., Chatrath, P.S., Rakotosamimanana, B., 1999. Past and present distributions of lemurs in Madagascar. In: Rakotosamimanana, B., Rasamimanana, H., Ganzhorn, J.U., Goodman, S.M. (Eds.), New Directions in Lemur Studies. Kluwer, New York, pp. 19e53. Humbert, H., 1927. Destruction d’une flore insulaire par les feux: principaux aspects ge tation a  Madagascar. Me m. L’Acade mie Malgache 5, 1e80. de la ve Jaffey, A.H., Flynn, K.F., Glendenin, L.E., Bentley, W.C., Essling, A.M., 1971. Precision measurement of half-lives and specific activities of 235U and 238U. Phys. Rev. C 4, 1889e1906. http://dx.doi.org/10.1103/PhysRevC.4.1889. Jarosz, L., 1993. Defining and explaining tropical deforestation: shifting cultivation and population growth in Colonial Madagascar (1896-1940). Econ. Geogr. 69, 366e379. http://dx.doi.org/10.2307/143595. Jungers, W.L., Godfrey, L.R., Simons, E.L., Wunderlich, R.E., Richmond, B.G., Chatrath, P.S., 2002. Ecomorphology and behavior of giant extinct lemurs from Madagascar. In: Plavcan, J.M., Kay, R.F., Jungers, W.L., van Schaik, C.P. (Eds.), Reconstructing Behavior in the Primate Fossil Record. Kluwer, pp. 371e411. Klein, J., 2002. Deforestation in the Madagascar highlands e established ‘truth’ and scientific uncertainty. GeoJournal 56, 191e199. http://dx.doi.org/10.1023/A: 1025187422687. Kull, C.A., 2000. Deforestation, erosion, and fire: degradation myths in the environmental history of Madagascar. Environ. Hist. 6, 423e450. http://dx.doi.org/ 10.3197/096734000129342361. Lachniet, M.S., 2009. Climatic and environmental controls on speleothem oxygenisotope values. Quat. Sci. Rev. 28, 412e432. Leuenberger, M., Siegenthaler, U., Langway, C., 1992. Carbon isotope composition of atmospheric CO2 during the last ice age from an Antarctic ice core. Nature 357, 488e490. http://dx.doi.org/10.1038/357488a0. Ludwig, K.R., 1993. UISO; a Program for Calculation of 230Th-234U-238U Isochron Ages (USGS Numbered Series No. 93e531). Open-File Report. U.S. Geological Survey. Matsumoto, J., Burney, D.A., 1994. Late Holocene environments at Lake Mitsinjo, northwestern Madagascar. Holocene 4, 16e24. http://dx.doi.org/10.1177/ 095968369400400103. , E., 2000. A near real-time forest cover map of Mayaux, P., Gond, V., Bartholome Madagascar derived from spot vegetation data. Int. J. Remote Sens. 21, 3139e3144. McConnell, W.J., Kull, C.A., 2014. Deforestation in Madagascar: debates over the island's forest cover and challenges of measuring forest change. In: Conservation and Environmental Management in Madagascar. Routledge, London, pp. 67e104. McDermott, F., 2004. Palaeo-climate reconstruction from stable isotope variations in speleothems: a review. Quat. Sci. Rev. 23, 901e918. McGee, D., Quade, J., Edwards, R.L., Broecker, W.S., Cheng, H., Reiners, P.W., Evenson, N., 2012. Lacustrine cave carbonates: novel archives of paleohydrologic change in the Bonneville Basin (Utah, USA). Earth Planet. Sci. Lett. 351e352, 182e194. http://dx.doi.org/10.1016/j.epsl.2012.07.019. ^thie, H., 1921. La ve ge tation malgache. Ann. Mus. Colon. Marseille 9, Perrier de la Ba 1e266.  me re , E., Amelot, X., Pierson, J., Crouau-Roy, B., Chikhi, L., 2012. Genetic data Que suggest a natural prehuman origin of open habitats in northern Madagascar and question the deforestation narrative in this region. Proc. Natl. Acad. Sci. 109, 13028e13033. http://dx.doi.org/10.1073/pnas.1200153109. Radimilahy, C., 1998. Mahilaka: an archaeological investigation of an early town in northwestern Madagascar. Uppsala. In: Studies in African Archaeology, 15,

S.J. Burns et al. / Quaternary Science Reviews 134 (2016) 92e99 p. 293, 165 figs., 38 plates. ISSN 0284-5040, ISBN 91-506-1313-8. s, L., Gonfiantini, R., 1993. Isotopic patterns in modern Rozanski, K., Aragu as-Aragua global precipitation. In: Climate Change in Continental Isotopic Records. American Geophysical Union. Shackleton, N.J., 1987. The carbon isotope record of the Cenozoic: history of organic carbon burial and of oxygen in the ocean and atmosphere. Geol. Soc. Lond. Spec. Publ. 26, 423e434. http://dx.doi.org/10.1144/GSL.SP.1987.026.01.27.


Virah-Sawmy, M., Willis, K.J., Gillson, L., 2010. Evidence for drought and forest declines during the recent megafaunal extinctions in Madagascar. J. Biogeogr. 37, 506e519. http://dx.doi.org/10.1111/j.1365-2699.2009.02203.x. €nsli, H., Schneider, R.R., 2014. Links between southWeldeab, S., Lea, D.W., Oberha western tropical Indian Ocean SST and precipitation over southeastern Africa over the last 17 kyr. Palaeogeogr. Palaeoclimatol. Palaeoecol. 410, 200e212. http://dx.doi.org/10.1016/j.palaeo.2014.06.001.