Nitrogen Isotope (Δ15n) Patterns for Amino Acids in Lemur Bones Are

JOURNAL OF QUATERNARY SCIENCE (2018) 33(8) 958–968 ISSN 0267-8179. DOI: 10.1002/jqs.3073

Nitrogen isotope (d15N) patterns for amino acids in lemur bones are inconsistent with aridity driving megafaunal extinction in south-western Madagascar

SEAN W. HIXON,1* EMMA A. ELLIOTT SMITH,2 BROOKE E. CROWLEY,3,4 GEORGE H. PERRY,1,5 JEANNOT RANDRIANASY,6 JEAN FREDDY RANAIVOARISOA,6 DOUGLAS J. KENNETT1,7 and SETH D. NEWSOME2 1Department of Anthropology, Pennsylvania State University, State College, PA, USA 2Department of Biology, University of New Mexico, Albuquerque, NM, USA 3Department of Geology, University of Cincinnati, Cincinnati, OH, USA 4Department of Anthropology, University of Cincinnati, Cincinnati, OH, USA 5Department of Biology, Pennsylvania State University, State College, PA, USA 6Department of Biological Anthropology and Sustainable Development, University of Antananarivo, Madagascar 7Institutes for Energy and the Environment, Pennsylvania State University, State College, PA, USA Received 21 February 2018; Revised 25 September 2018; Accepted 3 October 2018

ABSTRACT: Most endemic species with body masses >10 kg on Madagascar went extinct within the past 1000 years. The extent to which human predation, anthropogenic landscape transformation and aridification may separately or together explain this extinction pattern remains controversial. We present nitrogen isotope (d15N) values of individual amino acids preserved in bones from now-extinct Pachylemur insignis and extant Propithecus verreauxi from two subfossil sites in south-western Madagascar: Tsirave and Taolambiby. The amino acid-specific approach allows us to identify environmental signals that are otherwise difficult to recognize in bulk collagen d15N values. Specifically, we use the d15N values of source amino acids (phenylalanine and lysine) as a proxy for habitat aridity between ca. 4000 years ago and present and the spacing of d15N values between trophic and source amino acids to quantify trophic levels for these two lemur species. Despite paleoenvironmental evidence for lowering water tables and the expansion of relatively arid savanna between 4000 and 1000 years ago, our isotope data suggest that these lemurs did not live in increasingly arid habitats and did not change their trophic level. Together, our results support the hypothesis that aridity alone did not play a major role in late Holocene megafaunal extinctions in south-western Madagascar. # 2018 John Wiley & Sons, Ltd.

KEYWORDS: glutamic acid; habitat baseline; habitat modification; lysine; phenylalanine; trophic level.

Introduction Humans and their associated domesticated animals (e.g. dogs, cattle, goats and sheep) spread across the island during Controversy exists regarding the drivers of past and ongoing the past 1000 years (Verin and Battistini, 1971; Dewar, 1984; extinctions of Madagascar’s endemic fauna. Some scenarios Wright et al., 1993). Landscape modification, disease and direct attribute species loss to human activities such as hunting, interspecific interactions associated with this spread may have landscape burning and the introduction of invasive species to contributed to the decline of endemic taxa (Battistini, 1971; the island (Dewar, 1984; Virah-Sawmy et al., 2016), but MacPhee, 1997; Godfrey and Jungers, 2003; Burney et al., climate change in the form of aridification may have also 2004; Crowley, 2010; Burns et al., 2016; Crowley et al., 2017). played a significant role in this process (Mahe and Sourdat, Additionally, aridification may have contributed to extirpations 1972; Burney et al., 2004; Virah-Sawmy et al., 2010). South- (Battistini, 1971; Mahe and Sourdat, 1972; Burney et al., 2004). western Madagascar is an ideal region to study late Holocene A paleohydrological record from Lac Ihotry, south-western ecological change, because this region has a high diversity of Madagascar (Figs 1 and 2; Supporting Information, Fig. S1) endemic species and some of the earliest evidence for reveals that the arrival of both humans and bovids coincided humans on the island (Perez et al., 2003; Burney et al., 2004; with a drying trend that included a lowering of the water table Douglass and Zinke, 2015). Early human colonists, who were and salinization of coastal pans and wetlands between roughly present in south-western Madagascar by at least 2300 calen- 2250 and 700 cal a BP (Vallet-Coulomb et al., 2006). dar years before present (cal a BP, MacPhee and Burney, Palynological changes in the sediment record of the saline 1991; Perez et al., 2005), encountered a diverse endemic coastal basin of Ambolisatra, south-western Madagascar (Fig. 1), megafauna that included pygmy hippos, elephant birds, and a also suggest aridification during the past several thousand years wide array of giant lemur species (Burney et al., 2004; (Burney, 1993; Virah-Sawmy et al., 2016) as do contemporane- Crowley, 2010). Most of these species went extinct after ous declines in the abundance of aquatic birds at coastal 1000 cal a BP (Crowley, 2010), and, despite oral histories subfossil sites such as Belo-sur-Mer (Goodman and Rakotozafy, and European accounts that describe relatively recent sight- 1997). These data provide support for the possibility that the ings of now-extinct megafauna (e.g. Burney, 1998; De combination of human activities and climate change triggered Flacourt, 2007), the most recent credible radiocarbon dates species declines (Burney et al., 2004; Clarke et al., 2006; Virah- > for endemic animals with body mass 10 kg are ca. 500 cal Sawmy et al., 2010). a BP (Simons, 1997; Muldoon et al., 2009). Although a variety of environmental factors influence the d15 nitrogen isotope ( N) values of soil and overlying terrestrial Correspondence: Sean W. Hixon, as above. plants, moisture availability is one of the most important E-mail: [email protected]

# 2018 John Wiley & Sons, Ltd. MEGAFAUNAL EXTINCTION IN SOUTH-WESTERN MADAGASCAR 959

habitats during the Holocene. Rather, the authors hypothe- sized that agropastoralism probably put a greater strain on endemic fauna compared with aridification. However, the possibility that past aridification acted in concert with negative interactions between introduced and endemic species to drive megafaunal extinctions could not be excluded. Interpretation of bulk collagen d15N values is complicated by the fact that these values are influenced by both trophic level (TL) and the aridity-dependent isotopic baseline (d15N values in primary producers in a given habitat). Animals have higher d15N values than their diet due to physiological processes that discriminate against 14N; d15N offsets between consumer bulk collagen and diet D15 ‰ ( Nconsumer-diet) are 2–5 (Vanderklift and Ponsard, 2003). Even when the TL of an organism can be assumed with reasonable accuracy, there exists uncertainty regarding D15 the magnitude of Nconsumer-diet, because this value can vary significantly among taxa and is affected by diet quality as well as physiological condition (Vanderklift and Ponsard, 2003; Herrera et al., 2006; Miron et al., 2006; Ramı´rez and Hobson, 2006). Omnivorous lemurs have a greater potential for variable TL than obligate grazers or folivores. Crowley et al. (2017) made reasonable assumptions regarding the TL of various terrestrial vertebrates in Madagascar, and differences in diet can explain much of the variability in measured d15N values among taxa. Nevertheless, there remains the possibility that shifts in either TL or baseline d15N values d15 ( Nbaseline) over time could have contributed to the observed changes in bulk collagen d15N values, which complicates interpretation of the data. Figure 1. Map of south-western Madagascar (Laborde projection). The search for a climate-related signal can be expanded by Sites that produced bones analyzed in this study (Tsirave and measuring d15N values of individual amino acids in bone Taolambiby/Beza Mahafaly) are marked with squares. Other relevant collagen. There are significant differences in D15N sites (black circles) include important paleoecological sites such as consumer-diet Belo-sur-Mer (Goodman and Rakotozafy, 1997), Lac Ihotry (Vallet- values among individual amino acids in terrestrial organisms Coulomb et al., 2006) and Ambolisatra (Burney, 1993; Virah-Sawmy (Edgar Hare et al., 1991; Chikaraishi et al., 2011). For et al., 2016). example, glutamic acid (Glu) has consistently elevated d15N in consumer tissues relative to those in diet, and it is (Austin and Vitousek, 1998; Schulze et al., 1998; Handley consequently considered a trophic amino acid (AA, Chikar- et al., 1999; Amundson et al.,2003;Craineet al., 2009). aishi et al., 2011). Conversely, phenylalanine (Phe) in Preferential loss of 14N from the inorganic soil nitrogen consumers is isotopically similar to that in primary producers, pool due to nitrification, denitrification and ammonia and is consequently considered a source AA (Chikaraishi 15 d15 volatilization results in N-enriched soil nitrogen under et al., 2011). Thus, NPhe values in any consumer will d15 arid conditions (Austin and Vitousek, 1998). Spatial or resemble the NPhe at the base of the food web in its habitat d15 d15 temporal patterns in the N values of primary producers ( Nbaseline), and the TL of an organism can be approximated cascade up the food chains. Thus, animals that live in based partly on the offset in d15N values between trophic AA relatively arid habitats tend to have high d15N values (typically glutamic acid) and source AA (typically phenylala- relative to those at similar trophic levels that live in more nine) within a single organism (McClelland and Montoya, mesic environments (Heaton et al., 1986; Sealy et al., 2002; Bradley et al., 2014; McMahon et al., 2015b). To 1987; Cormie and Schwarcz, 1996; Murphy and Bow- estimate TL, it is necessary to identify the offset in man, 2006; Crowley et al., 2011b). It is important to note d15N values of trophic and source amino acids in primary that land use (e.g. irrigation, fertilizer application and producers at the base of the food web (b). Typically, b ¼ d15 d15 grazing intensity by herbivores) can also influence soil NPhe NGlu. This value is known to vary signifi- and associated plant or animal d15N values (Frank et al., cantly among different primary producers (Chikaraishi et al., 1995; Aranibar et al., 2008; Szpak, 2014), but intensive 2010, 2011; McMahon et al., 2015a). Thus, the accuracy of land use in south-western Madagascar probably began AA-based TL estimates follows from an accurate identification only within the past several hundred years (Erdmann, of the plants that comprise the base of the food chain for any 2003). given consumer (e.g. C3 versus C4 plants). However, the Recently, Crowley et al. (2017) used bulk bone collagen mechanisms that create these separations and the variation in d15N values from 238 radiocarbon-dated terrestrial verte- values among taxa are poorly understood (Chikaraishi et al., brates belonging to 24 extant and extinct genera in Madagas- 2011; O’Connell, 2017). Fortunately, organisms that live in car to identify temporal changes in the aridity of the habitats the same habitat will probably have source amino acids with in which these animals lived. These specimens were col- similar d15N values. lected from 17 subfossil sites that range in age from 10 000 to Here we use specific AA d15N values in bone collagen 500 cal a BP. After controlling for taxon, geographic region from two lemur species to test the hypothesis that these and coastal proximity, the authors found no evidence for endemic vertebrates lived in increasingly arid habitat during directional trends in d15N values through time. Their data the period of aridification that followed 3800 cal a BP. If thus suggest that animals did not live in increasingly arid these animals experienced increasing arid conditions through

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) 960 JOURNAL OF QUATERNARY SCIENCE

Figure 2. Radiocarbon dates for individuals used in this study (top, 3700–3960 cal a BP to present, Table S1, Crowley, 2010; Crowley et al., 2017; Godfrey et al., unpubl.), and a contemporaneous record of aridification and the salinization of Lac Ihotry (bottom, Vallet-Coulomb et al., 2006). Brackets associated with each calibrated radiocarbon date span 95.5% of each probability distribution. We estimated calendar dates for Lac Ihotry samples using an age-depth model (Fig. S1). time, then source AA d15N values should progressively as direct and indirect anthropogenic impacts were the more increase because of increases in primary producer likely triggers for endemic species declines. d15N values associated with aridity. Support for this hypothesis would leave open the possibility that aridification and Methods human activity acted synergistically to drive megafaunal Specimen and site selection extinctions. By contrast, if we find no shifts in source AA d15N values over time, then aridification cannot explain We analyzed bones from 13 Pachylemur insignis and 10 extinction, and this conclusion suggests that alternatives such Propithecus verreauxi individuals housed at the University of

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) MEGAFAUNAL EXTINCTION IN SOUTH-WESTERN MADAGASCAR 961

Antananarivo or the University of Massachusetts Amherst. Montoya, 2002; Styring et al., 2010), and we thus categorize Pachylemur insignis (giant ruffed lemur) is an extinct species threonine as a metabolic AA after Germain et al. (2013). We of the family Lemuridae (Seligsohn and Szalay, 1974; Kistler do not report results for hydroxyproline or proline, because et al., 2015) that is reconstructed to have had a pronograde we were unable to consistently separate them. We derivat- terrestrial/arboreal locomotor strategy (Jungers, 1980). The ized samples in batches of 8–17 along with an internal dental morphology of P. insignis suggests dedicated frugivory reference material containing all amino acids for which we or possibly trophic omnivory (Godfrey et al., 2012). Bone obtained d15N values (Sigma Aldrich, Co., St Louis, MO, collagen carbon isotope (d13C) values suggest that this species USA). For all samples, 1 mL of derivatized material (using generally, if not exclusively, relied on C3 vegetation (Crowley dichloromethane as a carrier solvent) was injected into a et al., 2011a; Crowley and Godfrey, 2013). Propithecus BPx5 gas chromatograph column for AA separation (0.32 ID, verreauxi (Verreaux’s sifaka) is an extant member of the 1.0 mm film thickness, SGE Analytical Science, Ringwood, family Indriidae (Pastorini et al., 2001). Behavioral and Victoria, Australia) in a TRACE 1310 gas chromatograph, d13C data confirm that this species is an arboreal frugivore– combusted via a Thermo Scientific GC Isolink II and then folivore that also relies primarily on C3 vegetation (Yamashita, analyzed with a Thermo Scientific Delta V Plus isotope ratio 2002; Crowley et al., 2011a). mass spectrometer. We conducted all AA chemical proce- The bone specimens used in this study came from two dures and analyses at the University of New Mexico Center subfossil sites and one modern site in south-western Mada- for Stable Isotopes (Albuquerque, NM, USA). Nitrogen gascar: Tsirave, located in the Mangoky River drainage; isotope data are expressed as d values, d15 ¼ ½ð Þ= Taolambiby, located in the Onilahy River drainage about N 1000 Rsample Rstandard Rstandard , where Rsample 15 14 220 km south-west of Tsirave; and Beza Mahafaly, a modern and Rstandard refer to the ratio of the heavy ( N) to light ( N) reserve located <10 km to the east of Taolambiby that stable isotope in the sample and standard, respectively. The d15 borders the Sakamena River (Fig. 1). South-western Madagas- internationally accepted standard for N is atmospheric N2 car is characterized by xeric spiny and succulent vegetation gas. Samples were measured in triplicate; we present mean and experiences an extended dry season due to an orographic d15N values SD. We did not record data for amino acids rain shadow in the region and the absence of the Intertropical from samples that had chromatograms with coeluted amino Convergence Zone during the austral-summer (Wells, 2003; acids or small AA peak amplitudes (<100 mV). We bracketed Burgess et al., 2004). All the P. insignis specimens and three triplicate measurements on samples with internal reference of the subfossil P. verreauxi included in this study came from materials. Our internal reference materials are homogenous Tsirave. Four subfossil P. verreauxi came from Taolambiby amino acids that we purchased individually from Sigma and three modern P. verreauxi individuals were recovered Aldrich, analyzed via elemental analysis-IRMS, and refer- from Beza Mahafaly. Beza Mahafaly protects both riparian enced against internationally accepted standards. Standard forest and spiny thicket. All the Beza Mahafaly specimens deviations in d15N values among standards ranged from included in this study were found within or adjacent to the 0.4‰ for leucine and lysine to 1.1‰ for alanine. The mean riparian corridor that borders the river. Bulk collagen d15N precision among the standards for all amino acids was d13C and d15N values for all individuals have been published 0.6‰. previously (Crowley et al., 2010, 2012, 2017; Crowley and Godfrey, 2013; Godfrey et al., unpubl.), and all individuals Data analysis have been previously radiocarbon dated (Fig. 2; Crowley, 2010; Crowley et al., 2017; Godfrey et al., unpubl.). We Statistical analyses were conducted in SigmaPlot 14.0 with calibrated all dates in Oxcal 4.3 using the Southern Hemi- significance set to a ¼ 0.05. We examined data distributions sphere calibration curve SHCal13 (Hogg et al., 2013). using Shapiro–Wilk tests for normality and Brown–Forsythe tests for equal variance. We used Mann–Whitney U-tests to Amino acid nitrogen isotope (d15N) analysis compare AA-specific d15N values between groups (e.g. comparison of d15N values between lemur genera), Krus- d15 Phe AA-specific N analysis was conducted via gas chromato- kal–Wallis H-tests to compare sites, Pearson product-moment graph combustion isotope ratio mass spectrometry (GC-C- correlation coefficients (r) to assess relationships between IRMS). We worked with previously extracted and purified bulk collagen and AA-specific d15N values, and Spearman collagen. Collagen isolation techniques followed existing correlation coefficients (rs) to explore relationships between standards (Brown et al., 1988; Stafford et al., 1988) but 14C dates and d15N values. differed slightly among samples (Table S1). As part of initial We calculated TL (where primary producers have TL ¼ 1) chemical pretreatment, three samples had been previously using the empirically derived relationship from Chikaraishi hydrolyzed in 6 N hydrochloric acid at 100 ˚C for approxi- et al. (2009): mately 24 h and passed through XAD resin. We hydrolyzed 3–10 mg of extracted collagen from all other specimens in d15 d15 þ b ¼ NGlu NPhe þ ð Þ 1 mL of 6 N hydrochloric acid at 110 ˚C for approximately TL ðÞD D 1 1 20 h. Next, for all samples, we completed carboxyl esterifica- Glu Phe tion with a 4:1 mixture of 2-isopropanol/acetyl chloride at d15 d15 110 ˚C for 1 h followed by amine acetylation with trifluoro- where consumer NGlu and NPhe values appear in the acetic acid anhydride at 110 ˚C for 10 min (Engel and Hare, numerator, and b represents the aforementioned constant d15 d15 1985; Fantle et al., 1999). This derivatization protocol value of NPhe NGlu in the primary producers at the allowed us to measure d15N values of 11 amino acids: one base of the food web. We used the b value (mean SD) of ‰ metabolic AA (threonine), four source amino acids (glycine, 8.4 1.4 observed for C3 plants from Japan and Thailand lysine, phenylalanine and serine), and six trophic amino acids (Chikaraishi et al., 2010, 2011), because stable carbon (alanine, aspartic acid, glutamic acid, isoleucine, leucine and isotope values from the bulk bone of P. insignis and P. valine). Note that threonine d15N values are unlike those in verreauxi (Crowley and Godfrey, 2013) suggest that these other amino acids given that they are consistently lower in lemurs primarily consumed C3-based resources throughout consumers than in diet (Hare et al., 1991; McClelland and the period of interest.

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) 962 JOURNAL OF QUATERNARY SCIENCE

The denominator of Equation (1) is frequently termed the Amino acid d15N values trophic discrimination factor (TDF; McMahon et al., 2015a, Mean AA d15N values for P. verreauxi range from 4.0‰ 2015b), where D and D represent the absolute differences Glu Phe (lysine) to 13.3‰ (phenylalanine). The range is slightly in d15N values between consumer and diet glutamic acid and smaller for P. Insignis: from 5.2‰ (lysine) to 12.4‰ (glycine). phenylalanine respectively. Phenylalanine and glutamic acid With the exceptions of glycine, phenylalanine and serine, have traditionally been considered the most appropriate there are no significant differences in AA-specific d15N values source AA and trophic AA, respectively, due to their significant between species (Fig. 3). Glycine d15N values (mean SD) and consistent differences in 15N enrichment per TL (Hoen are significantly higher for P. insignis (n ¼ 13, 12.4 0.8‰) et al., 2014; Nielsen et al., 2015). The TDF is assumed to be than for grouped P. verreauxi (n ¼ 10, 8.8 1.6‰; U ¼ 6.5, constant in terrestrial food webs (Chikaraishi et al., 2011), but p < 0.001). Similarly, d15N values are significantly higher it is known to vary as a function of diet quality in certain Ser for P. insignis (n ¼ 13, 10.0 1.0‰) than grouped P. ver- marine consumers (Bradley et al., 2015; McMahon et al., reauxi (n ¼ 10, 7.4 2.4‰, U ¼ 22.0, p ¼ 0.008). In fact, 2015b; Nielsen et al., 2015). Based on data from terrestrial most AA-specific d15N values are elevated in P. insignis and marine consumers with low TL, we use a TDF value of compared to grouped P. verreauxi. Despite this tendency, 7.6 1.2‰ (Chikaraishi et al., 2010, 2011). Note that uncer- d15N values are significantly higher in grouped P. ver- tainty is associated with each variable in Equation (1), and this Phe reauxi (n ¼ 10, 13.3 1.9‰) than in P. insignis (n ¼ 13, uncertainty has contributed to controversy regarding the use of 11.5 1.2‰; U ¼ 27.0, p ¼ 0.020). This inequality holds AA-based TL estimates (Naito et al., 2018; O’Connell and when comparing P. insignis and P. verreauxi within Tsirave Collins, 2018). We followed the approach of Jarman et al. (U ¼ 2.0, p ¼ 0.022). However, Tsirave P. insignis d15N (2017) to propagate uncertainty (Table S1). Phe values are indistinguishable from values for modern Beza Mahafaly P. verreauxi (U ¼ 19.0, p ¼ 1.0). Although P. d15 Results verreauxi NPhe values tend to be higher at Tsirave and d15 Taolambiby than at Beza Mahafaly, there are no statistically Radiocarbon and bulk collagen N values significant differences in d15N values among sites (H(2) ¼ 4.8, The lemurs considered in this study all date to the late p ¼ 0.09). d15 d15 Holocene (3700–3960 cal a BP to present, Table S1; Fig. 2). Bulk collagen N correlates significantly with NGly ¼ ¼ < d15 ¼ ¼ At Tsirave, dates for P. insignis range between 3700–3960 (n 23, r 0.933, p 0.0001), NSer (n 23, r 0.742, < d15 ¼ ¼ ¼ and 737–905 cal a BP, and dates for P. verreauxi range p 0.0001) and NGlu (n 23, r 0.567, p 0.00482). 15 between 1933–2142 and 1420–1570 cal a BP. Dates for P. There is no relationship between bulk d N values and d15 d15 ¼ ¼ verreauxi from Taolambiby range between 936–1061 and N for other amino acids, such as NPhe (n 23, r 805–961 cal a BP. 0.155, p ¼ 0.480). Regarding TL estimates, the mean (SD) 15 d15 ̶d15 ¼ ‰ Descriptive statistics for all d N values are given in Table 1 value for NGlu NPhe for P. insignis (n 13, 0.2 1.2 ) and are summarized in Fig. 3. Bulk bone collagen is significantly higher than for grouped P. verreauxi (n ¼ 10, d15N values for P. insignis are consistently higher than those 2.2 1.2‰, U ¼ 3.0, p < 0.001). TL estimates for P. insignis for the entire P. verreauxi sample (Mann–Whitney U ¼ 5.5, range between 2.0 0.3 and 2.5 0.3, and those for P. p < 0.001). This tendency is not significant when comparing verreauxi range between 1.5 0.3 and 2.0 0.3 (Table 2). P. insignis and P. verreauxi within Tsirave (U ¼ 5.5, The mean TL (SD) estimate for P. insignis (n ¼ 13, 2.1 0.2) p ¼ 0.068). Within the P. verreauxi sample, bulk collagen is significantly higher than the mean TL estimate for P. d15N values for Tsirave are apparently higher than those for verreauxi (n ¼ 10, 1.8 0.2, U ¼ 6.0, p < 0.001). Within the Beza Mahafaly, but there are no significant differences among P. verreauxi sample, there are no significant differences in TL sites (Kruskal–Wallis H(2) ¼ 5.3, p ¼ 0.06). among sites (H(2) ¼ 1.7, p ¼ 0.424).

Table 1. Descriptive statistics for bulk collagen and amino acid-speciﬁc d15N values by taxon and site. Tsirave Tsirave Taolambiby Beza Mahafaly P. insignis P. verreauxi P. verreauxi P. verreauxi

d15N(‰) d15N(‰) d15N(‰) d15N(‰)

Material n min. max. mean SD n min. max. mean SD n min. max. mean SD n min. max. mean SD

Bulk Collagen 13 9.5 10.9 10.4 0.4 3 8.7 10.3 9.3 0.9 4 6.8 8.9 8.2 0.9 3 6.6 7.9 7.1 0.7 Alanine 13 8.1 10.4 9.5 0.7 3 8.8 12.4 10.2 1.9 4 8.8 11.0 9.7 1.0 3 6.7 8.4 7.4 0.9 Aspartic acid 11 9.6 12.8 11.1 0.9 3 11.4 12.6 11.8 0.7 4 8.7 12.3 10.7 1.8 3 7.8 10.1 8.6 1.3 Glutamic acid 13 9.6 13.5 11.6 1.1 3 11.8 13.9 12.6 1.1 4 10.7 11.1 10.9 0.2 3 8.9 11.0 9.7 1.2 Isoleucine 13 7.8 10.3 9.0 0.8 3 10.8 11.7 11.3 0.5 2 10.6 12.6 11.6 1.4 3 7.9 10.4 9.1 1.3 Leucine 11 7.7 10.2 9.0 0.7 3 9.2 11.1 10.2 1.0 4 9.6 10.6 10.1 0.5 3 7.6 9.7 8.3 1.2 Valine 13 9.2 13.3 11.1 1.2 3 11.1 13.6 11.9 1.4 4 10.8 13.8 12.0 1.3 3 8.6 10.9 9.4 1.3 Glycine 13 11.0 13.7 12.4 0.8 3 9.1 12.2 10.4 1.6 4 8.1 10.0 8.8 0.8 3 6.3 8.0 7.3 0.9 Lysine 12 3.7 6.4 5.2 0.9 3 4.5 7.7 5.7 1.8 1 2.2 2.2 2.2 3 1.1 4.3 2.9 1.5 Phenylalanine 13 9.9 14.2 11.5 1.2 3 13.1 16.6 14.5 1.9 4 11.8 15.1 13.8 1.5 3 10.3 12.4 11.4 1.1 Serine 13 7.9 12.1 9.9 1.0 3 8.1 10.7 9.2 0.9 4 6.5 9.8 8.3 1.7 3 4.0 5.0 4.5 0.5

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) MEGAFAUNAL EXTINCTION IN SOUTH-WESTERN MADAGASCAR 963

Figure 3. Comparison of summary (mean SD) d15N data for P. insignis and P. verreauxi by site. Note that d15N values for bulk collagen and d15 most amino acids are relatively high in P. insignis. Despite this tendency, NPhe values are signiﬁcantly higher for Tsirave and Taolambiby P. verreauxi than for P. insignis. Lysine data exist for only one individual of Taolambiby P. verreauxi.

Trends in amino acid d15N values through time phenylalanine and, to a lesser extent, on lysine, because d15N and d15N most consistently reflect d15N There is a very slight decrease in d15N values through time Phe Lys baseline Phe (Chikaraishi et al., 2011; O’Connell, 2017). Glycine and for grouped P. insignis and P. verreauxi at Tsirave, but this serine are commonly categorized as source amino acids trend is not significant (Fig. 4, n ¼ 16, Spearman’s rank (McClelland and Montoya, 2002; Styring et al., 2010; Jarman correlation coefficient r ¼ 0.318, p ¼ 0.224). A similar lack of s et al., 2017), but there is significant variability in directional change in d15N values through time is observed Phe D15N for these two amino acids (O’Connell, if only P. insignis from Tsirave is considered (n ¼ 13, r consumer-diet s 2017). ¼ 0.258, p ¼ 0.382). Additionally, in the grouped Tsirave d15 sample, no directional change exists for NLys values ¼ ¼ ¼ d15 ¼ (n 15, rs 0.439, p 0.098), NSer values (n 16, rs Discussion ¼ ¼ d15 ¼ ¼ 15 0.222, p 0.403) and NGly values (n 16, rs 0.324, We use source AA d N values in P. insignis and P. verreauxi p ¼ 0.215). The small sample sizes for P. verreauxi at Tsirave, d15 as a proxy for Nbaseline and examine these values through d15 Taolambiby and Beza Mahafaly (particularly for NLys time to test the hypothesis that these lemurs lived in values) do not allow us to identify meaningful changes in increasingly arid habitats after 3800 cal a BP. We find little d15 Nbaseline within sites. However, when the entire P. support for this prediction. There is no increase in source AA verreauxi sample is considered as a single group, there are (Phe and Lys) d15N values for either lemur species at Tsirave d15 ¼ ¼ significant decreases in NPhe values (n 10, rs 0.632, between 3700–3960 and 737–905 cal a BP, which is ¼ d15 ¼ ¼ ¼ p 0.043), NGly values (n 10, rs 0.730, p 0.013), inconsistent with P. insignis and P. verreauxi living in d15 ¼ ¼ ¼ NSer values (n 10, rs 0.767, p 0.007) and bulk colla- increasingly arid habitat. Moreover, the trends we observe for d15 ¼ ¼ ¼ gen N values (n 10, rs 0.620, p 0.005) through time. P. verreauxi suggest this species has lived in increasingly Among the source amino acids, we focus our attention on mesic habitats over time. This shift towards increasingly

Table 2. Phenylalanine and glutamic acid d15N values and associated TL estimates for species grouped by site.

d15 d15 Site Species n NGlu NPhe TL (‰, mean SD) (‰, mean SD) (mean SD)

Tsirave P. insignis 13 11.6 1.1 11.5 1.2 2.1 0.2 Tsirave P. verreauxi 3 12.6 1.1 14.5 1.9 1.9 0.1 Taolambiby P. verreauxi 4 10.9 0.2 13.8 1.5 1.7 0.2 Beza Mahafaly P. verreauxi 3 9.7 1.2 11.4 1.1 1.9 0.1

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) 964 JOURNAL OF QUATERNARY SCIENCE

d15 Figure 4. Bivariate plot showing NPhe values (mean SD) for P. insignis and P. verreauxi by site and as a function of mean calendar years BP. mesic habitat could reflect a combination of regional increase As Fig. 3 illustrates, only a subset of the variability in AA in moisture availability or a shift towards the greater use of d15N values is reflected in bulk collagen d15N values, and this mesic microhabitats such as gallery forests. It is possible that highlights the strength of the AA-specific approach in recognizing d15 d15 aridity dramatically increased after 737–905 cal a BP and that subtle changes in Nbaseline. The fact that some AA Nvalues extirpation of P. insignis occurred so rapidly that it left little have trends that match that of bulk collagen (specifically glycine, trace in the material record. However, we think this scenario glutamic acid and serine) is not surprising, because these three d15 is unlikely. Tsirave P. insignis NPhe values are indistin- amino acids in total contain 35% of bone collagen nitrogen; guishable from values for modern P. verreauxi at Beza glycine comprises 26% of the nitrogen in collagen, glutamic Mahafaly, which suggests that Tsirave P. insignis lived in a acid contributes 6% and serine contributes 3% (Eastoe, relatively mesic habitat comparable to the riparian forest at 1955). While phenylalanine carries important information d15 Beza Mahafaly (20 km from the rainfall station at Betioky, regarding Nbaseline, this AA comprises only about 1% of the d15 where mean annual precipitation is 666 mm; Dewar and nitrogen in collagen (Eastoe, 1955), and NPhe signatures are Richard, 2007). thus significantly attenuated in bulk collagen d15Nvalues. d15 d15 Higher NPhe values for ancient P. verreauxi from Tsirave AA-specific N values also reveal a previously unrecog- d15 and Taolambiby in comparison to modern P. verreauxi from nized difference in Nbaseline values for P. insignis and Beza Mahafaly suggest that this species once lived in habitat P. verreauxi at Tsirave. Interspecific differences in source AA that was drier than the riparian habitat at Beza Mahafaly. d15N suggests that P. verreauxi frequented drier habitats than Because P. verreauxi is extirpated from Tsirave and because P. insignis, and, to the best of our knowledge, this is the first d15 P. verreauxi NPhe values in the region significantly documentation of spatial partitioning among co-occurring decrease through time, the data suggest that P. verreauxi in lemur species in the past. Additionally, spatial partitioning south-west Madagascar has come to live in increasingly emphasizes the fact that both behavior and aridity influence d15 mesic habitats. This trend matches a previously documented NPhe values in an organism and highlights the advantage of decrease in bulk collagen d15N values for P. verreauxi, the AA-specific approach for disentangling TL and changes in d15 Microcebus griseorufus, Lemur catta and Lepilemur leucopus Nbaseline.Crowleyet al. (2011a) noted that elevated in south-western Madagascar over the past 900 years (Crow- d15N values in the bulk collagen of several extinct giant lemur ley et al., 2012). Crowley et al. (2012) interpreted this genera (including Pachylemur) may be explained by drought d15 decrease as evidence of ecological retreat, in which activities tolerance. However, relatively low NPhe values for Tsirave by humans and human-introduced species have limited P. insignis support the interpretation that elevated bulk colla- lemurs to protected areas that are typically dominated by gen d15N values in this species are more probably a product of more humid gallery forests. Additional AA-specific analysis of diet (e.g. TL) rather than habitat aridity. AA d15N values in samples from other sites near Tsirave and Taolambiby (e.g. other giant lemur species may reveal similar results. Because Ambolisatra) may clarify temporal trends and help identify elevated bulk collagen d15N values in P. insignis cannot be d15 regional variability in Nbaseline values, but the existing data explained by habitat, our isotope data suggest that P. insignis support direct and indirect human pressures rather than may have been more omnivorous than P. verreauxi, and this aridification as the primary stressors on lemurs. finding is bolstered by AA d15N-based estimates of TL.

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) MEGAFAUNAL EXTINCTION IN SOUTH-WESTERN MADAGASCAR 965

Figure 5. Comparison of published d15N values (mean SD) for bulk tissue and speciﬁc amino ¼ acids for C3 plants (n 26, 19 different species) and C4 plants (n ¼ 7, ﬁve different species) from Japan and Thailand (Chikaraishi et al., 2010, 2011). Valine data exist for only one C4 plant.

Understanding the diet and associated TL of extinct taxa is photosynthetic pathway (Crowley and Godfrey, 2013). This critical for distinguishing among extinction scenarios that minor consumption of CAM plants introduces an unknown involve changes in resource availability and interactions degree of uncertainty to TL estimates for P. verreauxi, because among consumers. TL estimates for the sampled P. insignis no AA d15N values for CAM plants currently exist. and P. verreauxi are all 2 (Tables 2 and S1), which indicates primarily herbivorous foraging ecology. TL estimates are Conclusions consistent with studies of dental morphology and microwear d15 analyses (Godfrey et al., 2012) and with behavioral studies in AA-specific N values in lemur bone suggest that the the case of P. verreauxi (Simmen et al., 2003; Crowley and demise of Madagascar’s megafauna was not triggered by Godfrey, 2013). Based on tooth morphology, Godfrey et al. aridification. This conclusion supports that of Crowley et al. (2012) concluded that P. insignis was a dedicated frugivore (2017). The present study does not rule out the possibility that that consumed tougher foods than members of the closely currently unknown synergistic effects of aridification and related genus of extant ruffed lemurs (Varecia). Consequently, human behavior contributed to the Late Holocene extinction P. insignis may have been an important seed disperser, and of P. insignis and other megafauna in Madagascar. However, d15 the extinction of these frugivores probably had negative the absence of any increase in NPhe values for P. insignis, d15 consequences for certain plant species (Godfrey et al., 2008; and a decrease in NPhe values for P. verreauxi over time Crowley et al., 2011a). Estimated TL for P. insignis is slightly make aridification an unlikely primary driver of past lemur higher than for P. verreauxi, suggesting some degree of extinctions. This conclusion is quite plausible given that omnivory in the former species was possible. However, this endemic megafauna such as P. insignis survived desiccation result should be treated with caution. Several TL estimates for during the Last Glacial Maximum as well as extreme changes P. verreauxi are slightly below the theoretical herbivore value in moisture availability during previous glacial and intergla- of 2, probably due to the preliminary b value (8.4‰) that we cial periods (Burney, 1996; Burney et al., 2004). Our comparison of d15N values for individual source amino acids used for C3-based food webs as well as variation in AA- specific trophic discrimination factors (Chikaraishi et al., (Phe and Lys) and bulk bone collagen demonstrates the utility 2011). Data from a limited collection (n ¼ 26) of Japanese of AA-specific isotope analysis in expanding inferences d15 and Thai plants indicate that b values are in the 8–9‰ range regarding both diet and temporal changes in Nbaseline values, and allows us for the first time to identify both spatial for plants that use the C3 photosynthetic pathway (Fig. 5; Chikaraishi et al., 2010, 2011). However, taxon-specific partitioning and dietary differences among coexisting lemur patterns in b and AA-specific trophic discrimination factors taxa. Using AA-based TL estimates is expedient because d15 need to be investigated across a broader range of plants N values for both source and trophic amino acids can be before they can be confidently applied to estimate TL of measured from a single tissue sample (McClelland and terrestrial consumers. Additionally, both direct observations Montoya, 2002). However, several challenges must be and isotope data suggest that P. verreauxi can consume some addressed to increase the accuracy of TL estimates. Future d15 plants that use the Crassulacean acid metabolism (CAM) research should consider changes in Nbaseline in additional

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) 966 JOURNAL OF QUATERNARY SCIENCE regions of Madagascar (e.g. coastal western Madagascar), Burns SJ, Godfrey LR, Faina P et al. 2016. Rapid human-induced where aridification may have been more substantial (Mahe landscape transformation in Madagascar at the end of the first and Sourdat, 1972). Additionally, the details of anthropogenic millennium of the Common Era. Quaternary Science Reviews 134: disturbance that may have contributed to the loss of the 92–99. Chikaraishi Y, Ogawa NO, Doi H et al. 2011. 15N/14N ratios of megafauna and ecological retreat of P. verreauxi and other amino acids as a tool for studying terrestrial food webs: a case extant lemur species are largely unknown. Landscape modifi- study of terrestrial insects (bees, wasps, and hornets). Ecological cation for agriculture and charcoal production (Burney, Research 26: 835–844. 1993), the introduction of bovids and dogs (Verin and Chikaraishi Y, Ogawa NO, Kashiyama Y et al. 2009. Determination Battistini, 1971), and hunting (Perez et al., 2005) in the region of aquatic food-web structure based on compound-specific nitro- over the past 2000 years are factors worth considering in gen isotopic composition of amino acids. Limnology and Oceanog- future research. raphy: Methods 7: 740–750. Chikaraishi Y, Ogawa NO, Ohkouchi N. 2010. Further evaluation of the Acknowledgements. We thank the Universite d’Antananarivo pa- trophic level estimation based on nitrogen isotopic composition of leontology laboratory for assistance in collecting bone samples and amino acids. Earth, life, and isotopes. Kyoto University Press: Kyoto. Laurie Godfrey, Brendan Culleton and Richard Bankoff for contribut- Clarke SJ, Miller GH, Fogel ML et al. 2006. The amino acid and ing samples to this study. We also thank John Whiteman, Laura stable isotope biogeochemistry of elephant bird (Aepyornis) egg- Burkemper and Viorel Atudorei for their support during our analyses shells from southern Madagascar. Quaternary Science Reviews 25: and two anonymous reviewers for their helpful comments. General 2343–2356. support for the Paleoecology and Isotope Geochemistry Laboratory at Cormie AB, Schwarcz HP. 1996. Effects of climate on deer bone d15 d13 d15 Penn State was provided by the NSF Archeometry program BCS- N and C: lack of precipitation effects on N for animals 1460369 to D.J.K. (AMS 14C and bulk isotope work). This material is consuming low amounts of C4 plants. Geochimica et Cosmochi- also based upon work supported by the National Science Foundation mica Acta 60: 4161–4166. Graduate Research Fellowship Program under Grant No. DGE- Craine JM, Elmore AJ, Aidar MP et al. 2009. Global patterns of foliar 1255832. nitrogen isotopes and their relationships with climate, mycorrhizal fungi, foliar nutrient concentrations, and nitrogen availability. New Phytologist 183: 980–992. Supporting Information Crowley BE. 2010. A refined chronology of prehistoric Madagascar Table S1. Specimen attributes, radiocarbon and stable isotope and the demise of the megafauna. Quaternary Science Reviews 29: 2591–2603. data, and AA-based trophic level estimates. Crowley BE, Carter ML, Karpanty SM et al. 2010. Stable carbon and Figure S1. Ihotry age–depth model. nitrogen isotope enrichment in primate tissues. Oecologia 164: 611–626. Abbreviations. AA, amino acid; CAM, Crassulacean acid metabolism; Crowley BE, Godfrey LR. 2013. Why all those spines? Anachronistic GC-C-IRMS, gas chromatograph combustion isotope ratio mass defences in the Didiereoideae against now extinct lemurs. South spectrometry; TDF, trophic discrimination factor; TL, trophic level. African Journal of Science 109: 1–7. Crowley BE, Godfrey LR, Bankoff RJ et al. 2017. Island-wide aridity References did not trigger recent megafaunal extinctions in Madagascar. Ecography 40: 901–912. Amundson R, Austin AT, Schuur EAG et al. 2003. Global patterns of Crowley BE, Godfrey LR, Guilderson TP et al. 2012. Extinction and the isotopic composition of soil and plant nitrogen. Global ecological retreat in a community of primates. Proceedings of the Biogeochemical Cycles 17: 1031. Royal Society B: Biological Sciences 279: 3597–3605. Aranibar J, Anderson I, Epstein H et al. 2008. Nitrogen isotope Crowley BE, Godfrey LR, Irwin MT. 2011a. A glance to the past: composition of soils, C 3 and C 4 plants along land use gradients subfossils, stable isotopes, seed dispersal, and lemur species loss in in southern Africa. Journal of Arid Environments 72, 326–337. southern Madagascar. American Journal of Primatology 73: 25–37. Austin AT, Vitousek PM. 1998. Nutrient dynamics on a precipitation Crowley BE, Thoren S, Rasoazanabary E et al. 2011b. Explaining gradient in Hawai’i. Oecologia 113: 519–529. geographical variation in the isotope composition of mouse lemurs Battistini R. 1971. Conditions de gisements des sites littoraux de (Microcebus). Journal of Biogeography 38: 2106–2121. subfossiles et causes de la disparition de la faune des grands De Flacourt E. 2007. Histoire de la grande isle Madagascar. animaux dans le sud-ouest et l’extreme^ sud de Madagascar. Taloha KARTHALA Editions. 4: 7–18. Dewar RE. 1984. Extinctions in Madagascar: the loss of the subfossil Bradley CJ, Madigan DJ, Block BA et al. 2014. Amino acid isotope fauna, In Quaternary Extinctions: a Prehistoric Revolution, Martin incorporation and enrichment factors in Pacific bluefin tuna, PS, Klein RG (eds). The University of Arizona Press. Thunnus orientalis. PLoS ONE 9: e85818. Dewar RE, Richard AF. 2007. Evolution in the hypervariable Bradley CJ, Wallsgrove NJ, Choy CA et al. 2015. Trophic position environment of Madagascar. Proceedings of the National Academy estimates of marine teleosts using amino acid compound specific of Sciences of the United States of America 104: 13723–13727. isotopic analysis. Limnology and Oceanography: Methods 13: Douglass K, Zinke J. 2015. Forging ahead by land and by sea: 476–493. archaeology and paleoclimate reconstruction in Madagascar. Afri- Brown TA, Nelson DE, Vogel JS et al. 1988. Improved collagen can Archaeological Review 32: 267–299. extraction by modified Longin method. Radiocarbon 30: 171–177. Eastoe JE. 1955. The amino acid composition of mammalian collagen Burgess N, Hales JA, Underwood E et al. 2004. Terrestrial Ecoregions and gelatin. Biochemical Journal 61: 589–600. of Africa and Madagascar: a Conservation Assessment. Island Press: Hare PE, Fogel ML, Stafford TW et al. 1991. The isotopic composition Washington. of carbon and nitrogen in individual amino acids isolated from Burney DA. 1993. Late Holocene environmental changes in arid modern and fossil proteins. Journal of Archaeological Science 18: southwestern Madagascar. Quaternary Research 40: 98–106. 277–292. Burney DA. 1996. Climate change and fire ecology as factors in the Engel M, Hare P. 1985. Gas-liquid chromatographic separation of quaternary biogeography of Madagascar. Biogeographie de Mada- amino acids and their derivatives. In Chemistry and Biochemistry gascar: 49–58. of the Amino Acids, Barrett GC (ed). Springer: Berlin; 462–479. Burney DA, Burney LP, Godfrey LR et al. 2004. A chronology for late Erdmann T. 2003. The dilemma of reducing shifting cultivation. In prehistoric Madagascar. Journal of Human Evolution 47: 25–63. The Natural History of Madagascar, Goodman S, Benstead J (eds). Burney DA, Ramilisonina. 1998. The Kilopilopitsofy, Kidoky, and University of Chicago Press; 134–139. Bokyboky: accounts of strange animals from Belo-sur-mer, Mada- Fantle MS, Dittel AI, Schwalm SM et al. 1999. A food web analysis of gascar, and the megafaunal ‘extinction window’. American Anthro- the juvenile blue crab, Callinectes sapidus, using stable isotopes in pologist 100: 957–966.

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) MEGAFAUNAL EXTINCTION IN SOUTH-WESTERN MADAGASCAR 967

whole animals and individual amino acids. Oecologia 120: discrimination in blood of a New World nectarivorous bat. Journal 416–426. of Experimental Biology 209: 541–548. Frank AB, Tanaka DL, Hofmann L et al. 1995. Soil carbon and Muldoon KM, de Blieux DD, Simons EL et al. 2009. The subfossil nitrogen of Northern Great Plains grasslands as influenced by long- occurrence and paleoecological significance of small mammals at term grazing. Journal of Range Management 48: 470–474. Ankilitelo Cave, southwestern Madagascar. Journal of Mammalogy Germain LR, Koch PL, Harvey J et al. 2013. Nitrogen isotope 90: 1111–1131. fractionation in amino acids from harbor seals: implications for Murphy BP, Bowman DMJS. 2006. Kangaroo metabolism does not compound-specific trophic position calculations. Marine Ecology cause the relationship between bone collagen d15N and water Progress Series 482: 265–277. availability. Functional Ecology 20: 1062–1069. Godfrey LR, Jungers WL. 2003. The extinct sloth lemurs of Madagas- Naito YI, Chikaraishi Y, Drucker DG, et al. 2018. Reply to “Comment car. Evolutionary Anthropology: Issues, News, and Reviews 12: on” Ecological niche of Neanderthals from Spy Cave revealed by 252–263. nitrogen isotopes of individual amino acids in collagen. “[J. Hum. Godfrey LR, Jungers WL, Schwartz GT et al. 2008. Ghosts and Evol. 93 (2016) 82-90]” [J. Hum. Evol. 117 (2018) 53-55]. Journal Orphans, Elwyn Simons: A Search for Origins. Springer: Berlin; of Human Evolution 117: 56. 361–395. Nielsen JM, Popp BN, Winder M. 2015. Meta-analysis of amino acid Godfrey LR, Winchester JM, King SJ et al. 2012. Dental topography stable nitrogen isotope ratios for estimating trophic position in indicates ecological contraction of lemur communities. American marine organisms. Oecologia 178: 631–642. Journal of Physical Anthropology 148: 215–227. O’Connell TC. 2017. ‘Trophic’ and ‘source’ amino acids in trophic Goodman S, Rakotozafy L. 1997. Subfossil birds from coastal sites in estimation: a likely metabolic explanation. Oecologia 184: western and southwestern Madagascar: A paleoenvironmental 317–326. reconstruction. In Natural Change and Human Impact in Madagas- O’Connell TC, Collins MJ. 2018. Comment on “Ecological niche of car, Goodman S, Patterson B (eds). Smithsonian Institution Press; Neanderthals from Spy Cave revealed by nitrogen isotopes of 257–279. individual amino acids in collagen” [J. Hum. Evol. 93 (2016) 82- Handley LL, Austin AT, Stewart GR et al. 1999. The 15N natural 90]. Journal of Human Evolution 117: 53. abundance (d15N) of ecosystem samples reflects measures of Pastorini J, Forstner MR, Martin RD. 2001. Phylogenetic history of water availability. Australian Journal of Plant Physiology 26: sifakas (Propithecus: Lemuriformes) derived from mtDNA sequen- 185–199. ces. American Journal of Primatology 53: 1–17. Heaton THE, Vogel JC, von La Chevallerie G et al. 1986. Climatic Perez V, Burney D, Godfrey L et al. 2003. Butchered sloth lemurs. influence on the isotopic composition of bone nitrogen. Nature Evolutionary Anthropology 12: 260–260. 322: 822–823. Perez VR, Godfrey LR, Nowak-Kemp M et al. 2005. Evidence of early Herrera LGM, Hobson KA, Martıńez JC et al. 2006. Tracing the origin butchery of giant lemurs in Madagascar. Journal of Human of dietary protein in tropical dry forest birds. Biotropica 38: Evolution 49: 722–742. 735–742. Schulze E-D, Williams RJ, Farquhar GD et al. 1998. Carbon and Hoen DK, Kim SL, Hussey NE et al. 2014. Amino acid 15N trophic nitrogen isotope discrimination and nitrogen nutrition of trees enrichment factors of four large carnivorous fishes. Journal of along a rainfall gradient in northern Australia. Australian Journal of Experimental Marine Biology and Ecology 453: 76–83. Plant Physiology 25: 413–425. Hogg AG, Hua Q, Blackwell PG et al. 2013. SHCal13 Southern Sealy JC, Van Der Merwe NJ, Thorp JAL et al. 1987. Nitrogen isotopic Hemisphere calibration, 0–50,000 years cal BP. Radiocarbon 55: ecology in southern Africa: implications for environmental and 1889–1903. dietary tracing. Geochimica et Cosmochimica Acta 51: 2707–2717. Jarman CL, Larsen T, Hunt T et al. 2017. Diet of the prehistoric Seligsohn D, Szalay F. 1974. Dental occlusion and the masticatory population of Rapa Nui (Easter Island, Chile) shows environmental apparatus in Lemur and Varecia: their bearing on the systematics of adaptation and resilience. American Journal of Physical Anthropol- living and fossil primates. In Prosimian Anatomy, Biochemistry, ogy 164: 343–361. and Evolution, Martin R, Doyle G, Walker A (eds). Duckworth: Jungers WL. 1980. Adaptive diversity in subfossil Malagasy prosi- London; 543–561. mians. Zeitschrift fur€ Morphologie und Anthropologie 71: Simmen B, Hladik A, Ramasiarisoa P. 2003. Food intake and dietary 177–186. overlap in native Lemur catta and Propithecus verreauxi and Kistler L, Ratan A, Godfrey LR et al. 2015. Comparative and introduced Eulemur fulvus at Berenty, Southern Madagascar. population mitogenomic analyses of Madagascar’s extinct, giant International Journal of Primatology 24: 949–968. ‘subfossil’ lemurs. Journal of Human Evolution 79: 45–54. Simons E. 1997. Lemurs: old and new. In Natural Change and Human MacPhee RD. 1997. The 40,000-year plague: humans, hyperdisease, Impact in Madagascar, Smithsonian Institution Press; 142–156. and first-contact extinctions. In Natural Change and Human Impact Stafford TW, Brendel K, Duhamel RC. 1988. Radiocarbon, 13C and in Madagascar, Goodman S, Patterson B (eds). Smithsonian Institu- 15N analysis of fossil bone: removal of humates with XAD-2 resin. tion Press; 169–217. Geochimica et Cosmochimica Acta 52: 2257–2267. MacPhee RDE, Burney DA. 1991. Dating of modified femora of Styring AK, Sealy JC, Evershed RP. 2010. Resolving the bulk extinct dwarf Hippopotamus from southern Madagascar: implica- d15N values of ancient human and animal bone collagen via tions for constraining human colonization and vertebrate extinction compound-specific nitrogen isotope analysis of constituent amino events. Journal of Archaeological Science 18: 695–706. acids. Geochimica et Cosmochimica Acta 74: 241–251. Mahe J, Sourdat M. 1972. Sur l’extinction des vertebr es subfossiles et Szpak P. 2014. Complexities of nitrogen isotope biogeochemistry in l’aridification du climat dans le sud-ouest de Madagascar. Bulletin plant-soil systems: implications for the study of ancient agricultural de la Societ eG eologique de France 14: 295–309. and animal management practices. Frontiers in Plant Science 5: McClelland JW, Montoya JP. 2002. Trophic relationships and the DOI: 10.3389/fpls.2014.00288. nitrogen isotopic composition of amino acids in plankton. Ecology Vallet-Coulomb C, Gasse F, Robison L et al. 2006. Hydrological 83: 2173–2180. modeling of tropical closed Lake Ihotry (SW Madagascar): sensitiv- McMahon KW, Polito MJ, Abel S et al. 2015a. Carbon and nitrogen ity analysis and implications for paleohydrological reconstructions isotope fractionation of amino acids in an avian marine predator, over the past 4000 years. Journal of Hydrology 331: 257–271. the Gentoo penguin (Pygoscelis papua). Ecology and Evolution 5: Vanderklift MA, Ponsard S. 2003. Sources of variation in consumer- 1278–1290. diet d15N enrichment: a meta-analysis. Oecologia 136: 169–182. McMahon KW, Thorrold SR, Elsdon TS et al. 2015b. Trophic Verin P, Battistini R. 1971. Les anciens habitats de Rezoky et discrimination of nitrogen stable isotopes in amino acids varies d’Asambalahy. Taloha 4: 29–49. with diet quality in a marine fish. Limnology and Oceanography Virah-Sawmy M, Gillson L, Gardner CJ et al. 2016. A landscape 60: 1076–1087. vulnerability framework for identifying integrated conservation and Miron LLM, Herrera M LG, Ramı´rez P N et al. 2006. Effect of diet adaptation pathways to climate change: the case of Madagascar’s quality on carbon and nitrogen turnover and isotopic spiny forest. Landscape Ecology 31: 637–654.

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018) 968 JOURNAL OF QUATERNARY SCIENCE

Virah-Sawmy M, Willis KJ, Gillson L. 2010. Evidence for drought and Wright HT, Rakotoarisoa JA, Heurtbize G et al. 1993. The forest declines during the recent megafaunal extinctions in Mada- evolution of settlement systems in the Efaho River Valley, Anosy gascar. Journal of Biogeography 37: 506–519. (Madagascar). Bulletin of the Indo-Paciﬁc Prehistory Association Wells N. 2003. Some hypotheses on the Mesozoic and Cenozoic 13: 61–93. paleoenvironmental history of Madagascar. In The Natural History Yamashita N. 2002. Diets of two lemur species in different micro- of Madagascar, Goodman S, Benstead J (eds). University of habitats in Beza Mahafaly Special Reserve, Madagascar. Interna- Chicago Press; 16–34. tional Journal of Primatology 23: 1025–1051.

# 2018 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 33(8) 958–968 (2018)